Ultrasound-Assisted Hydroxyapatite-Decorated Breath-Figure Polymer-Derived Ceramic Coatings for Ti6Al4V Substrates.

The introduction of nanoparticles (NPs) into the breath-figure-templated self-assembly (BFTSA) process is an increasingly common method to selectively decorate a surface porous structure. In the field of prosthetic devices, besides controlling the morphology and roughness of the structure, NPs can enhance the osteointegration mechanism because of their specific ion release. Among the most widely used NPs, there are silica and hydroxyapatite (HAp). In this work, we propose a novel one-stage method to fabricate NP-decorated surface porous structures that are suitable for prosthetic coating applications. This technique combines the classical direct BFTSA process with the cavitation effect induced by an ultrasonic atomizer that generates a mist of water droplets with embedded NPs. Coatings were successfully obtained by combining a UV cross-linkable polymer precursor, alkoxy silicone, with synthesized HAp NPs, on Ti6Al4V alloy discs. The cross-linked polymeric surface porous structures at selected concentrations were then pyrolyzed in an ammonia atmosphere to obtain a silicon oxynitride (SiON) ceramic coating. Herein, we report the chemical and morphological analyses of both the polymeric and ceramic coatings as well as the effect of NPs at the interface.

Introduction

Self-assembly techniques can be defined as bottom-up strategies to produce patterned porous structures, suitable for applicaclass="Chemical">tions at the nanoscale and microscale.[1] Among these techniques, the breath-figure (BF) class="Chemical">process has come to the attenpan class="Chemical">tion of the scientific community only over the past 2 decades,[2] even though the mechanism had already been studied and theorized by Lord Rayleigh in 1911.[3] The breath-figure-templated self-assembly (BFTSA) process is a low-cost, one-stage technique to achieve regular porous patterned structures.[1,2] It is based on the sinking and evaporation of water droplet arrays from the surface of a polymeric solution. Droplets are generated by moisture condensation, which is induced by the evaporation of a volatile solvent over a cold substrate under a humid environment.[4−6]

The produced patterns, which are highly organized class="Species">honeycomb structures, can be exploited for a wide range of class="Chemical">potenpan class="Chemical">tial applications, for instance, as membranes,[7] sensors,[8] and optoelectronic devices[9] and also in the biomedical field.[10] Within the tissue engineering and prosthesis fields, it is well recognized that porosity and roughness control are two key factors to promote cell growth, adhesion, and proliferation.[11−15] Surface roughness in the range of 1–10 μm[16] is indeed beneficial for the synthesis, recruitment, and formation of an adsorbed layer of proteins (i.e., integrins), which enhances the cell metabolism and growth.[12] On the other hand, surface pores, or surface patterns, in the range of 40–250 μm[17] are effective in affecting the contact guidance of cells.[18] As a matter of fact, in the cascade of events that leads to osseointegration, osteoblast migration and cell colony formation are more prominent in the presence of 2D–3D patterns, such as pits or pores, than flat surfaces.

The BFTSA structures can be obtained with a wide range of class="Chemical">polymeric solupan class="Chemical">tions. Particularly, silicon-based polymers have gained increasing interest as coatings for bulk prosthetic devices because of the possibilities to add elements into the Si network (i.e., nitrogen or carbon) which can enhance the osteoinductive and osteoconductive properties as well as the mechanochemical properties.[19] Indeed, silicon-based polymers can be used as precursors to produce, via the polymer-derived ceramic (PDC) process,[19] inorganic coatings for prosthetic devices, as also demonstrated by Carlomagno et al.(20,21)

The authors performed pyrolyclass="Chemical">sis of the class="Chemical">porous pan class="Chemical">polymer structures under different atmospheres: pure air, nitrogen, and ammonia. Accordingly, different PDC materials with controlled surface pore structures were obtained depending on the pyrolysis atmosphere:[20] silicon dioxide, SiO2 (air atmosphere); silicon oxycarbide, SiOC (N2 atmosphere); and silicon oxynitride, SiON (NH3 atmosphere). The SiON BFTSA was further evaluated for bioapplications.[21] Silicon-ion release from the SiON materials was shown to be beneficial to promote osteoblast activation,[22] a fundamental stage of osseointegration that occurs at the prosthetic–bone interface.[15] Furthermore, SiON coatings show a thermal expansion coefficient that matches the coefficient of medical-graded titanium alloys (i.e., Ti6Al4V), thus reducing excessive stresses and failure at the interface with the solid titanium surface.[23]

It is worth menclass="Chemical">tioning that surface coapan class="Chemical">tings can be formed via vapor-phase processes such as plasma spraying, hot isostatic pressing, and thermal spraying[26,27] or via imbibition-based processes such as dip coating[28] and Langmuir–Blodgett deposition.[29] In comparison, the BFTSA process offers additional benefits such as selectively decorating the pore structures by inorganic nanoparticles (NPs).[24,25] In recent years, this procedure has gained more interest because of either the benefits of forming hybrid polymeric–inorganic patterned films[25] or the Pickering emulsion phenomenon. Pickering emulsion is another self-assembly phenomenon that takes place at the liquid–liquid interface of a water/oil or oil/water emulsion.[30] The combination of the Pickering emulsion effect with the BFTSA technique leads to an NP adsorption at the air–water–polymer three-phase contact that mechanically hinders the droplet coalescence. This effect, in turn, increases both the pore regularity and circularity.[24] Examples of the reported inorganic particles are silica,[31] gold,[32] TiO2,[33] and hydroxyapatite (HAp) NPs.[34] For biomedical applications, HAp is highly desirable because it releases calcium and phosphate ions in situ.[35,36] These ions, combined with Si4+, can enhance the osseointegration, thus guaranteeing a better mechanical interlocking between the prosthetic device and the surrounding bone tissue.[37]

The main challenge of Nclass="Chemical">P decorapan class="Chemical">tion via the direct-breath-figure method is the choice of solvent. BFTSA typically employs low-boiling-point solvents that are often difficult to disperse inorganic NPs. To overcome this challenge, surface modifications have been proposed either on NPs[31] or on the precursor polymeric structure.[38] Several studies on the generation of amphiphilic or Janus NPs[39−42] were carried out to enhance the emulsion stability.[30] This approach results in a proper NP decoration of the edges and the inner bottom areas of the pores.[24,31] Nevertheless, surface modifications are still solvent-dependent, labor-intensive, and chemistry-specific. To the authors’ knowledge, the literature has yet to offer an alternative, one-stage process that allows selective NP decoration simultaneously during the formation of the pore structures.

In this work, we developed a novel, economical one-stage technique to create a surface porous self-assembled coaclass="Chemical">ting, pan class="Chemical">simultaneously guaranteeing NP decoration. The basis of this idea is to implement the conventional BFTSA set up by means of an ultrasonic atomizer, which, by cavitation, can form NP-embedded water droplets. As in the works of Carlomagno et al.,[20,21] a ceramic prosthetic coating was evaluated as a possible application for this technique. A medical-grade UV-cross-linkable alkoxy silicone was used as a PDC precursor to generate the UV-cross-linked BFTSA structure, decorated with synthetized HAp NPs. This organic coating was further transformed into SiON by an organic-to-inorganic pyrolysis process at 900 °C in an NH3 environment. Both the NP-decorated polymer structures and the final ceramic structures were physically and chemically characterized.

Experimental Section

Materials

The class="Chemical">silicon class="Chemical">preceramic pan class="Chemical">polymer used in this study is Loctite 5248 (AG & Co., Germany, Henkel), a UV-curable alkoxy silicone with thixotropic behavior and a viscosity range of 50–80 cP at 25 °C. The Si/O/C ratio calculated by XPS spectra was 19:27:54 (see Figure S1 in the Supporting Information), while the number-average molecular weight was = 55 kDa, as measured from static light scattering analysis (SLS, Zetasizer Nano ZEN3600, Malvern Instruments). Loctite 5248, which is a straw-colored one-component liquid solution, was directly dissolved in ethyl acetate (vapor pressure = 73 mmHg at 20 °C; Sigma-Aldrich) to prepare BFTSA precursor solutions at 2 and 5% w/v (mg/mL). The solution was stirred overnight at room temperature and wrapped into an aluminum foil to prevent curing from ambient light. Double-side mechanically polished Ti6Al4V discs of a diameter of 10 mm and a thickness of 2 mm (provided by Eurocoating SpA, Italy) were used as substrates. Calcium nitrate, Ca(NO3)2 (Sigma-Aldrich, USA); diammonium phosphate, (NH4)2HPO4 (Sigma-Aldrich, USA); gelatin from cold water fish skin (Sigma-Aldrich, USA); deionized (DI) water; and ammonium hydroxide (28%, Millipore, USA) were used to synthesize HAp NPs.

Breath-Figure Setup and Process

The breath-figure setup was adaclass="Chemical">pted from the work of Maniglio et al.(43) It conpan class="Chemical">sisted of a N2 flowmeter (0.4–1.34 L/min), connected to a gas bubbler (water chamber) filled with an NP/DI water (or only DI water) solution. Inside the gas bubbler, an ultrasonic atomizer (a power of 19 W, 1700 ± 50 kHz, and a maximum flow rate of 3000 mm/h) was placed to generate the mist by the cavitation effect. An orbital shaker (KS 125 basic, IKA LABORTECHNIK, Germany) was positioned below the gas blubber to keep the solution properly mixed throughout all the process time. Through a pipe connection, the moisty flow was transported into a 6061 aluminum alloy (MC-Master Carr, USA) reaction chamber, where samples were placed at direct contact with the flow itself. The chamber was sealed with a UV-light transparent PLEXIGLASS Solar, with a hole to allow the casting of the precursor solution by means of a micropipette. Moreover, the setup was equipped with a hygrometer sensor to detect the temperature and relative humidity (RH) inside the reaction chamber. The cross-linking was induced by UV irradiation onto the specimen after a pre-exposure period using a SpotLed365/15 (Photo Electronics Srl, Italy) UV lamp (the emission peak at 365 nm). The ultrasonic atomizer was switched off just before turning on the UV light.

The main process parameters were oclass="Chemical">ptimized starpan class="Chemical">ting from the works of Carlomagno et al.(20,21) and by a trial-and-error approach, supported by optical and electronic microscopy analyses. The process parameters were optimized to obtain a surface porous structure with an equivalent pore diameter ranging between 40 and 250 μm based on a morphological design suitable for cell adhesion[17] and previous work of Carlomagno et al.(21) on biological responses to BFTSA-derived bone prosthetic coatings. Herein, the adopted parameters, either for the direct BFTSA method or for the ultrasonication-assisted BFTSA method, are reported. In the first case, a 5% Loctite 5248–ethyl acetate solution was used with a pre-exposure time (the time after casting the solution and before turning on UV light) of 8 min and a UV exposure time of 5 min. The flow rate and the RH in the chamber were, respectively, set to 0.8 L/min and 99%.

In the second case, a 2% class="Chemical">Loctite 5248–ethyl acetate solupan class="Chemical">tion and a 5% Loctite 5248–ethyl acetate solution were used. The settings adopted, for both concentrations, were 1 min 30 s of pre-exposure time, 2 min of UV exposure, a flow rate of 1.0 L/min, and a 99% RH. In both cases, 40 μL of Loctite 5248–ethyl acetate solution droplets was used for film casting.

HAp NP Synthesis

class="Chemical">HAp synthepan class="Chemical">sis was adapted from the work of Chen et al.(44) Aqueous solutions of calcium nitrate (7.84 g in 100 mL of DI water) and diammonium phosphate (3.16 g in 100 mL of DI water) were respectively prepared, and the pH was adjusted to 11 by a dropwise addition of ammonium hydroxide. An aqueous solution of 12 g of gelatin in 200 mL of DI water was prepared and then added to the calcium nitrate solution at 80 °C under vigorous stirring. The diammonium phosphate solution was dropped into the reaction solution at a rate of 2 mL/min using a peristaltic pump (Rainin Dynamax RP-1, Marshall Scientific, USA) under vigorous stirring at 80 °C. Afterward, the solution was left for 2 h at 80 °C before cooling it down to room temperature. It was subsequently centrifuged for 30 min at 4000 rpm (centrifuge, Sigma 2-5, USA) and washed with DI water five times. Finally, the pellet was placed into a 45 °C oven to dry overnight.

Heat Treatment

The heat treatment was performed on the class="Chemical">polymeric BFTSA specimens (pan class="Chemical">Hap-decorated and nondecorated samples) to induce the organic-to-inorganic conversion. As described in Section , also, HAp NPs alone had been heat-treated with the same procedure to evaluate if any changes could occur to the NPs. The heat treatment was performed in a silica tubular furnace (Thermo Scientific Heraeus) at 900 °C in an NH3 atmosphere. A purging time of 1 h at room temperature was promoted before starting the heating cycle. The cycle adopted consisted of a heating ramp until 120 °C at 5 °C/min, after which it was held for 2 h at 120 °C. Subsequently, a further heating ramp, still at 5 °C/min, was promoted to reach 900 °C. After holding for 1 h at 900 °C, a cooling ramp at 5 °C/min was set to go back to room temperature. The NH3 flow was set at 0.4 L/min (LPM).

Characterization

class="Chemical">HAp Npan class="Chemical">Ps were characterized by scanning electron microscopy (SEM), FESEM, (Zeiss supra 40, Germany), and transmission electron microscopy (TEM; Talos F200 S G2, Thermo Fisher Scientific, USA) to evaluate their size and morphological features. X-ray diffraction (XRD) analysis with an IPD 3000 (scan from 5° to 120°, Co kα = 1.7902 Å radiation, scan step = 0.03°, and an applied voltage and a current of 40 V and 30 mA, respectively) was performed to determine the HAp structure before and after the pyrolysis stage in an ammonia environment. XRD raw data were then elaborated by MAUD, XRD software based on Rietveld refinement. The patterns of the reference phases were supplied by the crystallography open database (COD). The surface and morphology of the coating, either in the polymeric stage or in the ceramic stage, were inspected by FESEM and by energy-dispersive X-ray spectrometry (EDXS) microanalysis, with a Jeol IT300 equipped with a Bruker EDXS. The latter had also been performed to obtain a semiquantitative analysis of the chemical composition of the coating and to detect the NP presence by mapping the calcium and phosphorus elements. All the specimens were previously coated with a platinum/palladium (Pt/Pd, 80:20) conductive thin film (Q150T ES, Quorum Technologies, UK) before the FESEM and EDXS analyses. For XPS, an Axis DLD Ultra (Kratos—UK) analyzer equipped with a monochromatized Al kα X-ray source of 1486.6 eV was used to evaluate the chemical composition and the bonding state of the ceramic coatings. Spectra were analyzed with homemade software based on the R platform.[45] Sessile contact angle (CA) measurements were carried out to evaluate the surface physical properties of the BFTSA specimens after ceramic conversion. CA analyses were performed with a homemade equipment and a 3 μL DI water droplet.

Statistical Analysis

Staclass="Chemical">tispan class="Chemical">tical analyses were conducted to study the pore size and its distribution as well as the effect of the Pickering emulsion. The pore analyses were carried out using ImageJ by thresholding techniques on the SEM images. Pore area values were obtained from different images of the same specimen and then analyzed with Origin Pro Lab. Histograms, showing the pore area distribution, were generated with a bin size ranging from 600 to 800 μm2. The distributions of the pore size were analyzed using a log–normal function for the more regular areas and an exponential function for regions where the coalescence dominated. For the Pickering emulsion evaluation, a perfect circularity was assumed to express the equivalent diameter values, which have a more direct understanding of the pore dimensions. Concerning the pore size statistical analyses, the median, quartiles (Q1 and Q3), and interquartile range (IQR) were reported. Circularity tests were performed using the above-mentioned software and a box chart diagram has been reported, showing the trend as function of an increase of the HAp content in the specimen.

Results and Discussion

Breath-Figure Structures

Breath-figure structures, obtained with the method reported in Secclass="Chemical">tion upan class="Chemical">sing a solution of 5% (w/v) Loctite 5248–ethyl acetate, have been investigated and compared to specimens obtained via the direct-breath-figure method. The latter is based on the experiments described in the work of Carlomagno et al.(20) As reported in Figure A,B, a morphological comparison of the obtained patterns was carried out by a statistical analysis of the SEM images (Figure C–E) to determine the pore areas and their distribution. The main difference between the two adopted BFTSA methods lies in the central areas of the specimens where the mist flow generated by the ultrasonic atomizer directly impacts the casted Loctite 5248–ethyl acetate solution. As illustrated in Figure , two different regions of pore structures were observed: a peripheral region (red box, number 1) with smaller and more regular pore sizes and a central region (blue box, number 2) with bigger and less spherical pores. The differences in pore morphology and dimension are related to two distinct physical mechanisms that occur on the specimen. In the central region, water droplet condensation is also assisted by a water droplet deposition onto the polymeric solution because of the mist flow produced by the ultrasonic atomizer. Collectively, the two phenomena lead to a more pronounced coalescence, which promotes the formation of larger and less regular pores.[2,4] On the other hand, in the peripheral region, the droplet formation mechanism is dominant by condensation of water droplets, induced by the humid environment in the reaction chamber. This hypothesis is supported by the histograms reported in Figure C–E. The peripheral region shows a log-normal distribution (see Figure D), similar to that obtained by the direct BFTSA method (Figure C), while the central region shows an exponential distribution (Figure D).

Figure 1

Comparison between the direct BFTSA and ultrasonic atomizer BFTSA methods. (A,B) SEM images of samples obtained from the direct BFTSA method and from the ultrasonic atomizer BFTSA method (scale bars: 100 μm). (C–E) Histograms and pore area distribution of structures as labeled.

Figure 2

3D (A) and top-view image (B) of the chamber setup with the sample placed inside. In (C), the top-view sketch of the chamber is reported, highlighting the two generated BFTSA regions with different morphological structures. The SEM image in the blue square is referred to the area directly hit by the mist flow, and the one in the red square is instead the peripheral area of the substrate. Scale bar: 100 μm.

Comparison between the direct BFTSA and ultrasonic atomizer BFTSA methods. (A,B) SEM images of samples obtained from the direct BFTSA method and from the ultrasonic atomizer BFTSA method (scale bars: 100 μm). (C–E) Histograms and pore area distribuclass="Chemical">tion of structures as labeled.

3D (A) and top-view image (B) of the chamber setup with the sample placed inclass="Chemical">side. In (C), the top-view sketch of the chamber is reported, highlighpan class="Chemical">ting the two generated BFTSA regions with different morphological structures. The SEM image in the blue square is referred to the area directly hit by the mist flow, and the one in the red square is instead the peripheral area of the substrate. Scale bar: 100 μm.

Concerning the phyclass="Chemical">sical mechanisms that lead to the BFTSA structures, it is also necessary to take into conpan class="Chemical">sideration the wetting/dewetting phenomena. Dewetting can occur at a solid–liquid interface (i.e., titanium–Loctite 5248–ethyl acetate solution), when spontaneous spreading of the liquid is not developed and liquid withdrawal and retraction are observed on the surface.[46] In this context, it is worth noting that dewetting of the Loctite 5248–ethyl acetate solution during the pore-forming process was not observed as both the solute and solvent are of low surface tension. As a matter of fact, complete spreading of the solution on the high-energy substrate occurs during all experiments. Furthermore, local dewetting within the large coalesced pores was also not relevant as XPS measurements (see Section ) of the converted inorganic coating only show ∼1% Ti. Therefore, regardless of the size and shape of the pores formed, the coating layer covers the entire substrate instead of forming a gridlike network structure due to local dewetting. Furthermore, photo cross-linking during the BFTSA process hinders the mobility of growing polymer chains and generates elastic forces that suppress the dewetting tendency.[47] Combining the abovementioned factors, it is unlikely that dewetting plays any role in the pore structure formation in the BFTSA process.

As given by the histogram in Figure D, the median value of the fitted curve is of 4233 μm2, while the quarclass="Chemical">tiles (Q1 and Q3) and the IQR are, respecpan class="Chemical">tively, 2768, 6474, and 3706 μm2. The histogram in Figure E instead reports a larger population in a small range of values, as deducible from Q1 (9672 μm2) and Q3 (46,607 μm2), showing 25% (blue area) and 75% (blue + green area) of the measurements. Considering the statistical analyses performed on the entire specimen, the exponential decay is the prevailing distribution with a Q1 and a Q3 of 4665 and 22,480 μm2, respectively. These structures were produced with the parameters reported in Section , accurately chosen to reach a pore dimension suitable for a prosthetic coating application, as reported in ref (17). The following discussion regards the analyses of the more regular regions because these are the prevalent areas of the samples. Furthermore, to increase regularity, specimens were placed off the center of the reaction chamber because in the center of the reaction chamber, the direct droplet deposition and the corresponding coalescence are more pronounced.

HAp NPs

TEM analyclass="Chemical">sis reveals that pan class="Chemical">HAp NPs have a rodlike morphology with dimensions of 52 ± 22 and 16 ± 6 nm (Figure A). As given in the diffractogram of Figure B, obtained from the selected area electron diffraction (SAED) analysis, the peak matching between the NP structure and a reference HAp spectrum reported in the literature was confirmed.[48] The synthesis route also involved a gelatin coating in order to increase the dispersibility and stability of HAp in DI water, which nevertheless did not affect the crystalline structure of the HAp NPs. For a more detailed particle characterization, FT-IR, TGA, and DTA are provided in the Supporting Information (Figure S2).

Figure 3

(A) TEM image of the rodlike shape of the HAp NPs. The scale bar is 200 nm. (B) X-ray diffractogram obtained by SAED analysis in comparison with the reference HAp peaks.

(A) TEM image of the rodlike shape of the pan class="Chemical">HAp NPs. The scale bar is 200 nm. (B) X-ray diffractogram obtained by SAED analysis in comparison with the reference HAp peaks.

class="Chemical">HAp Npan class="Chemical">Ps were also pyrolyzed at 900 °C in an NH3 environment for 1 h to evaluate if any reaction could have occurred once implemented in the BFTSA structures. As a matter of fact, NH3 heat treatment of HAp NPs was performed to provide the environment for organic-to-inorganic transformation as the BFTSA samples. Accordingly, XRD analysis was performed on the HAp nanopowder before and after the heat treatment. In Figure , the comparison between the two XRD spectra is reported, showing no changes in the HAp structure, neither due to the NH3 environment nor due to the process temperatures, and both spectra match data reported in COD-ID: 4317043 (HAp).

Figure 4

XRD spectra of the as-synthesized HAp nanopowder (black spectrum) and the NH3 heat-treated HAp nanopowder (red spectrum).

XRD spectra of the as-syntheclass="Chemical">sized pan class="Chemical">HAp nanopowder (black spectrum) and the NH3 heat-treated HAp nanopowder (red spectrum).

A class="Chemical">HAp Npan class="Chemical">Ps–DI water suspension at 20% (w/v) was then produced and introduced into the water chamber via the atomizer in order to evaluate the NP transport into the reaction chamber.

The tests revealed that Nclass="Chemical">Ps were successfully depopan class="Chemical">sited, without affecting the deposition time of the process. As a matter of fact, varying the deposition time, coffee-ring-like particle deposition was always observed. Each ring represents the trace of a coalesced droplet prior to the drying, as shown in Figure .

Figure 5

FESEM images of the HAp NP deposition onto a glass substrate in a coffee-ring-like shape. The scale bar is set to 100 μm for the main SEM image and 4 μm for the magnified view. The coffee-ring-like mechanism can be divided into four different stages, namely, droplet transport, droplet deposition, water evaporation, and coffee-ring formation, as depicted in the sketches.

FESEM images of the class="Chemical">HAp Npan class="Chemical">P deposition onto a glass substrate in a coffee-ring-like shape. The scale bar is set to 100 μm for the main SEM image and 4 μm for the magnified view. The coffee-ring-like mechanism can be divided into four different stages, namely, droplet transport, droplet deposition, water evaporation, and coffee-ring formation, as depicted in the sketches.

These coffee-ring structures show an assembly of class="Chemical">HAp Npan class="Chemical">Ps into multiple layers at the edges of the evaporated water droplets. In addition, NP depositions at the inner part of the droplet are observed. The overall NP distribution within each isolated region is inhomogeneous, which is attributed to both the coalescence of droplets during droplet formation and the subsequent water evaporation. First, significant droplet coalescence is expected from both direct droplet deposition from the mist flow and condensation, which results in the overall droplet shape and corresponds to the overall trace of the NP deposition. Second, within each droplet trace, the distribution of NPs is dictated by the energy minimization mechanism and transport behavior during water evaporation. Particularly, the enhanced deposition at the droplet edge is by a combination of the transport of NPs to, and stabilization of NPs at the three-phase contact line where the water evaporation rate is the highest. Similar to a typical BFTSA process, the Marangoni flow associated with the temperature-induced surface tension gradients will carry and deposit some NPs toward the droplet center.[49] Such a nonuniform deposition pattern, often considered a limitation for coffee-ring applications,[49] could be adopted to selectively decorate edges and the inner part of pores. Nevertheless, such initial analysis confirms that the home-built setup is suitable for the proposed ultrasonication-assisted BFTSA method.

Decorated BFTSA Structures and the Pickering Emulsion Effect

The process variables invesclass="Chemical">tigated to opan class="Chemical">ptimize the NP decoration of the BFTSA structures are reported in Table . Two parameters have been evaluated as principal factors affecting the process, namely, Loctite 5248 concentration and HAp amount in the DI water solution. Having a sufficiently high precursor concentration is important to avoid the gravity-driven sedimentation of the particles. Increasing concentration from 2 to 5% leads to an overall increase of the precursor solution viscosity, thus reducing NP mobility prior to the cross-linking stage and therefore keeping the particles on the surface and limiting the embedment into the matrix. In addition, the 5% solution is expected to reach gelation during the evaporation/cross-linking quicker than the 2% solution, which results in a reduced time for NP sedimentation and often formation of a denser skin layer.[50] On the other hand, the increment of concentration of HAp NPs in the gas bubbler results in NP-enriched droplets, which will likely increase the level of pore decoration. Different HAp concentrations were therefore inspected, varying from an initial 1.5 mg/mL to 5 mg/mL and eventually to 20 mg/mL. The optimum BFTSA structure decoration was obtained with a 5% (w/v) Loctite 5248–ethyl acetate solution and a 20% (w/v) HAp NPs–DI water solution. Accordingly, FESEM images of the obtained structures are shown in Figure A,B.

Table 1

Process Parameters Inspected to Optimize the NP Embedment into the BFTSA Structures

parameters	conditions
Loctite 5248	2%/5% (w/v)
concentration of HAp in DI water	0/1.5/5/20 mg/mL

Figure 6

FESEM images of the NP-decorated BFTSA structures at the 5% Loctite 5248–ethyl acetate solution and 20 mg/mL HAp/DI water. (A) Top-down view on the Ti6Al4V disc substrate. (B) Cross-sectional view on a glass substrate. (C,D) Top-down and the cross-sectional views of the BFTSA structure in the absence of NP decoration. The scale bars are set to 10 μm for (A–D) and to 1 μm for the magnified view in (B).

FESEM images of the Nclass="Chemical">P-decorated BFTSA structures at the 5% pan class="Chemical">Loctite 5248–ethyl acetate solution and 20 mg/mL HAp/DI water. (A) Top-down view on the Ti6Al4V disc substrate. (B) Cross-sectional view on a glass substrate. (C,D) Top-down and the cross-sectional views of the BFTSA structure in the absence of NP decoration. The scale bars are set to 10 μm for (A–D) and to 1 μm for the magnified view in (B).

As expected, class="Chemical">similar to Figure , the decorapan class="Chemical">tion mainly occurred in the inner part of the pores (Figure A) and at the three-phase (air–water–polymer) contact (cross-sectional view in Figure B); these results are also in accordance with the work of Sun et al.(24) and Yang et al.(31)

The effect of the increaclass="Chemical">sing amount of pan class="Chemical">HAp NPs on the pore morphology, regularity, and circularity was also investigated by comparing 5% Loctite 5248 samples with no HAp NPs and 1.5, 5, and 20 mg/mL NP concentrations. In Table S1 (see the Supporting Information), the statistical data of the pore areas and the circularity analysis are reported for the four samples. Furthermore, the equivalent diameter (ϕeq) is reported under the hypothesis of perfect circularity of the pores.

As given in Table S2 and Figure A, a decrease of the median value is noclass="Chemical">ticeable from the sample with no pan class="Chemical">HAp (2727 μm2 or ϕeq = 59 μm) to the one with 20 mg/mL HAp (1404 μm2 or ϕeq = 42 μm). The most relevant decrease is between the first two samples (no HAp vs 1.5 mg/mL), moving from 2727 (ϕeq = 59 μm) to 1748 μm2 (ϕeq = 47 μm), while it slows down with a further NP increase. On the contrary, for the circularity analysis, an opposite trend is observed: the higher the NP concentration, the higher the pore circularity, toward the asymptotic value of 1 (a perfect circle). A similar trend is observed for Q1, median, and Q3 values, which increase, respectively, from 0.71, 0.75, and 0.79 for the sample with no HAp to 0.80, 0.84, and 0.87 for the sample with 20 mg/mL HAp. Furthermore, a noticeable decrease of IQR with an increase of NP concentration is observed, which indicates that the distribution becomes narrower (Table S2 and Figure B). This is more evident from 5 mg/mL (IQR = 0.17) to 20 mg/mL (IQR = 0.07) because the IQR is almost half of the previous one. The circularity increase is however limited, and it is more pronounced for concentrations from 1.5 to 5 mg/mL, where the median values increase from 0.76 up to 0.80, than for concentrations from 5 to 20 mg/mL, where the median increases from 0.80 to 0.84. From these data, particle decoration for low concentrations is more effective in limiting the pore dimension, while for higher concentrations, NPs affect the pore circularity more.

Figure 7

(A) Pore area distribution histograms with log–normal fitting and the median value for the four tests at different HAp concentrations. (B) Box chart of the circularity statistical values related to the HAp concentration (mg/mL) for the same four tests reported in Table S1.

(A) class="Chemical">Pore area distribupan class="Chemical">tion histograms with log–normal fitting and the median value for the four tests at different HAp concentrations. (B) Box chart of the circularity statistical values related to the HAp concentration (mg/mL) for the same four tests reported in Table S1.

Ceramic Conversion

The class="Chemical">NH3 class="Chemical">pyrolypan class="Chemical">tic transformation of the decorated BFTSA preceramic structures was performed according to the procedures described by Carlomagno et al.,[20] which transforms a cross-linked Si-based polymer precursor into a SiON structure. The NH3 environment had been chosen in line with a further work of Carlomagno et al.,[21] where these structures, compared to SiO2 and SiOC coatings, are more suitable to promote cell adhesion, proliferation, and activity. The ceramic conversion has been evaluated by SEM, EDXS, and XPS analyses. SEM images of the SiON samples had been collected to evaluate the overall macrostructure. As shown in Figure , the surface porous structure has been preserved without any evident morphological changes as well as signs of significant crack propagation or delamination. A predictable shrinkage occurred because of the pyrolysis, as similarly reported in ref (20), which nevertheless does not significantly alter the morphological features of the parent polymeric BFTSA structures.

Figure 8

SEM image of a SiON BFTSA structure at 5% Loctite 5248–ethyl acetate. Few coating cracks visibly connect the pore rims. The scale bar is set to 100 μm.

SEM image of a class="Chemical">SiON BFTSA structure at 5% pan class="Chemical">Loctite 5248–ethyl acetate. Few coating cracks visibly connect the pore rims. The scale bar is set to 100 μm.

EDXS elemental maps were acquired to semiquanclass="Chemical">titapan class="Chemical">tively determine the spatial distribution of Si, Ca, and P in the inorganic coatings and to detect the incorporation of NPs into the structure. From Figure , where the Ca (Figure A), P (Figure B), and Si (Figure C) maps are reported, it is clear that the HAp NPs were decorated along the interface and the inner part of the pores, given by a higher intensity of both the Ca and P signals, while the Si signal is more homogeneous across the entire image, indicating that the matrix is mainly characterized by Si, as expected from the conversion into SiON.

Figure 9

EDXS elemental maps for calcium (A), phosphorous (B), and silicon (C). The SEM raw image of the mapped pore (D). The scale bars are 100 μm. (E) shows schematics of the pore decoration from a frontal view and a top view.

EDXS elemental maps for class="Chemical">calcium (A), pan class="Chemical">phosphorous (B), and silicon (C). The SEM raw image of the mapped pore (D). The scale bars are 100 μm. (E) shows schematics of the pore decoration from a frontal view and a top view.

Nevertheless, because of the limitaclass="Chemical">tions and the reliability of the EDXS technique to detect light elements such as pan class="Chemical">oxygen and nitrogen, XPS was performed to quantitatively determine the chemical composition and the chemical bonds generated during the organic-to-inorganic transformation. Two different specimens were analyzed: 5% Loctite 5248 with HAp NPs and 5% Loctite 5248 without HAp NPs.

The survey spectra of the two tested samples are shown in Figure A. As expected, in both spectra, class="Chemical">Si 2p, O 1s, and N 1s class="Chemical">peaks display the highest intenpan class="Chemical">sity, confirming the successful SiON conversion. The calculated atomic percentages for Si 2p, O 1s, and N 1s are, respectively, 27, 50, and 18% for the sample without HAp and 24, 54, and 16% for the heat-treated sample with HAp NPs. In both spectra, the C 1s peak is also visible at a binding energy of 285.10 eV and it has been attributed to an adventitious hydrocarbon contamination of the samples. The main difference between the two spectra is given by the Ca 2p peak revealed at 348.02 eV for the SiON 5%–HAp specimen, suggesting the HAp NP presence into the BFTSA structure. However, the low intensity of the Ca 2p spectrum observed can be attributed to a masking effect caused by the presence of particles of small dimensions onto the inner walls of the pores. Nevertheless, as shown in Figure F, the typical Ca 2p doublet is reported, with the two spin–orbit splitting component peaks at 351.34 eV (Ca 2p1/2) and at 348.00 eV (Ca 2p3/2). These binding energy values are compatible with Ca in HAp.[51−53] These values are also in agreement with other HAp XPS analysis results reported in the literature.[54−56] From the Si 2p deconvoluted spectra for both specimens, reported in Figure B,D, three components are present, SiO2, Si3N4, and SiON, of which the last is the prevailing component with a peak at 103.18 eV (Figure B) and 103.04 eV (Figure D). This trend is also confirmed by the deconvoluted spectra of N 1s in Figure C,E, where the main contribution is still given by SiON, reported at 398.30 and at 398.09 eV, respectively, while the second peak is related to Si3N4 (i.e., at 396.84 and 396.30 eV). The obtained values are in good agreement with the reported literature[57−60] and with the previous work of Carlomagno et al.,[20] conducted on the same polymeric precursor under identical pyrolysis conditions. Similarly, the wettability characteristics of flat and surface porous BFTSA structures were studied by means of static CA analysis. Nonpatterned surfaces evidenced a marked hydrophilic behavior (CA = 24 ± 2°), while surface porous BFTSA evidenced complete spreading of the water droplets. This behavior can be attributed to capillary phenomena deriving from small coating cracks (also visible in the SEM image, Figure ) produced by the pyrolysis conversion of the BFTSA specimens.

Figure 10

(A) XPS survey spectra for NH3-pyrolyzed BFTSA samples of a 5% Loctite 5248–ethyl acetate solution with (black spectrum) and without HAp NPs (red spectrum). (B) Si 2p and (C) N 1s spectra for the SiON 5%–no HAp sample; (D–F) Spectra of Si 2p, N 1s, and Ca 2p of the SiON 5%–HAp sample.

(A) Xclass="Chemical">PS survey spectra for pan class="Chemical">NH3-pyrolyzed BFTSA samples of a 5% Loctite 5248–ethyl acetate solution with (black spectrum) and without HAp NPs (red spectrum). (B) Si 2p and (C) N 1s spectra for the SiON 5%–no HAp sample; (D–F) Spectra of Si 2p, N 1s, and Ca 2p of the SiON 5%–HAp sample.

SEM, EDXS, and Xclass="Chemical">PS analyses confirmed the ceramic converpan class="Chemical">sion of the polymeric material as well as the preservation of the surface porous structure at a macroscale and a microscale. Furthermore, HAp NPs were detected and also characterized on the ceramic samples, stating, as expected, that the heat treatment did not provide any change in the structure characteristics.

Conclusions

Breath-figure surface porous structures are suitable for a wide range of applicaclass="Chemical">tions. pan class="Chemical">Particularly, it has been demonstrated that using a Si-based preceramic polymer as a precursor, it is possible to obtain inorganic coatings suitable for prosthetic coating applications.[21] Furthermore, by selectively decorating these pores with NPs, a further enhancement of osteointegration can be achieved. Nevertheless, the NP embedment into the BFTSA structures might require the need to surface-modify particles or the precursor by highly demanding, time-consuming, and expensive processes.[24,25,31]

In the present study, a novel, economic one-stage technique to combine the BF formaclass="Chemical">tion together with the Npan class="Chemical">P pore decoration is proposed. This is achieved by introducing an ultrasonic atomizer into the gas bubbler in order to directly deposit particles on the casted polymeric solution by mist transport. Adopting this new method, this study demonstrates the successful formation of coatings for prosthetic applications using Loctite 5248, a UV-cross-linkable PDC alkoxy silicone, and synthesized HAp NPs. NP decoration occurred at the three-phase interface and into the inner bottom area of the BFTSA pores. It has also been noticed that by increasing the HAp decoration, beneficial effects regarding the circularity of the patterned pores were achieved, in accordance with the combination of Pickering emulsion and BF mechanisms. Once the organic structures were characterized, a pyrolytic process, in an NH3 atmosphere, at 900 °C for 1 h was promoted. The inorganic transformation into SiON successfully occurred, together with successful decoration of HAp NPs, as confirmed by the XPS and EDXS measurements. As confirmed by SEM analysis, the overall morphological features of the coatings have not changed after the heat treatment, preserving the typical BFTSA structure.

Future developments, currently under invesclass="Chemical">tigapan class="Chemical">tion, can be related to embed other types of functional NPs, for example, drug-loaded mesoporous NPs, to further boost the osseointegration or to use the coating as a drug delivery scaffold. Further investigations concerning the biological response of these coatings are also underway.