Bioactive-Enhanced Polyetheretherketone Dental Implant Materials: Mechanical Characterization and Cellular Responses

Today, the practice of dental implantology is a scientifically proven and well-documented treatment option for total or partial edentulism and is considered to be the best alternative to natural teeth.^1–3  Titanium (Ti) and its alloys (principally Ti6Al4V) are the biomaterials of choice for dental implants,^4–6  based on their biocompatibility, mechanical characteristics, and ability to facilitate osseointegration.^3,7,8  However, the incidence of undesirable allergic reactions, the gradient difference in the elastic moduli of a titanium implant and its surrounding bone, cellular sensitization, galvanic current formation, and an unesthetic gray hue have heightened the demand for improved implant materials.^5–9 

Biocompatible polymers and polymer-based composite materials have been considered as an alternative to the systematic use of metallic alloys for a number of biomedical applications, including dental, orthopedic, and cardiovascular devices.^10,11  Polyetheretherketone (PEEK) is commonly described as one of the possible available polymers for the production of medical devices because of its high biocompatibility, chemical resistance to biodegradation, and esthetic properties.^5,9,12  At the same time, this material has similar mechanical properties to human bone, which has been widely employed in orthopedics, traumatology, and spinal implants to minimize stress shielding and bone resorption.^11,13–15  It thus presents a promising possibility for dental implants. Early research has been controversial in terms of biological behavior, and different strategies have been used to improve the bioactive properties of PEEK implants^16–21 ; however, based on the literature, there are currently no published reports of the PEEK dental implant body as applied to the mandible or maxilla.

Currently, a series of surface modifications including grit blasting, acid etching, calcium phosphate coatings, or different combinations of these techniques has been developed and applied to commercial implants to improve the implant surface and osseointegration.^8,22,23  According to the literature, some modifications of the implant surface composition with calcium phosphate coatings, most commonly hydroxyapatite (HA) and beta-tricalcium phosphate (βTCP), showed good results because of the osteoblasts' compatibility^3,24 ; however, the layer tends to delaminate, leading to implant failure in mid- and long-term term studies.^{4,11,19,23,25} 

Furthermore, it is crucial to state that any surface modifications described will lead to alterations in surface chemistry, roughness, hydrophilicity, charge, or surface energy. These parameters play a role in the implant-tissue interaction.²⁴  Today, it is known that these represent key parameters in the cell behavior of the interface material and can influence the speed and strength of osseointegration.^8,26,27  The current literature regarding surface features effects on osteoblast proliferation suggests that, similarly to what is observed with titanium substrates, porous PEEK surfaces are associated with higher osteoblast proliferation and differentiation as compared with smooth surfaces.^21,28 

In this context, based on the potential of PEEK as a dental implant material overcoming the current limitations of titanium, and calcium phosphate bioactive coatings for future clinical applications, our group developed a material design by creating a functionally graded material (FGM) by means of a hot pressing technique.^29–31  Our strategy involves a groundbreaking FGM approach for dental implants manufactured with a low percentage of bioactive compounds on the outer layer of the implant structure embedded in the implant matrix, with the implant core composed of PEEK. This study was developed to perform both a mechanical characterization (shear strength and hardness) and a biological response evaluation taking into consideration the behavior of gingival and bone cells in contact with these materials to preliminarily validate its potential use as a dental implant material. The samples tested were produced under appropriate conditions to achieve controlled and equivalent surface features between samples to compare with Ti6Al4V (gold standard).

Thus, the aim of this study was to characterize the mechanical properties of PEEK-based materials with hydroxyapatite (HA) or embedded β-tricalcium phosphate (βTCP) for dental implants produced with an innovative manufacturing approach and to compare the in vitro biological behavior of human osteoblasts and gingival fibroblasts on these materials.

The manufacturing approach created a functionally gradated material (FGM) samples using a previously described hot-pressing technology.^29–31  PEEK powders (particles of 65 μm and a density of 1.32 g/cm³; Evonik Industries, Essen, Germany) were mechanically mixed with HA or βTCP (95 wt% PEEK and 5 wt% HA or βTCP) in a stainless-steel jar containing steel milling balls at 25 rpm for 21 hours. A total of 8 samples per group with a diameter of 8 mm and height of 3 mm were processed according to each group: PEEK, PEEK-HA, and PEEK-βTCP. The powder mixture was placed into graphite molds. The mold was placed inside the chamber, where the sample was compressed at 2 bar and then heated until reaching 380 °C, which is higher than the melting point of PEEK (345°C) using a heating rate of 80°/min, following which the temperature was decreased to 300°C. At this stage, a pressure of 4 bar was applied and maintained for 4 seconds, and then the power was turned off and the system cooled down to room temperature under a vacuum. Similar samples of Ti6Al4V were produced (n = 8) as the control. Ti6Al4V powder (SLM Solutions, Lübeck, Germany) was dehydrated at 110°C for 1 hour and placed into graphite molds. The mold was placed inside the chamber and produced by hot pressing, in a primary vacuum. After this step, the samples were compressed at 2 bar and then heated up to 1200°C at 31°C/min. At 1100°C, the pressure on the sample was raised to 20 MPa and then maintained for 30 minutes.

All test samples were wet ground on SiC papers down to 4000 mesh and then polished to a near-mirror finish using aluminum oxide suspension (1 μm). Samples were then cleaned ultrasonically. All samples were etched with 5% HNO₃, 10% HF, and 85% distilled water. Finally, all samples were grit blasted with 100 μm alumina particles under the appropriate conditions to achieve the most common equivalent roughness values (Ra between 1.5 and 2 μm)³²  used on actual dental implant surfaces. Following this step, samples were cleaned ultrasonically in 100% ethanol and autoclave sterilized.

Before the mechanical and biological tests, one sample of each group was observed under ultra-high-resolution field emission gun-scanning electron microscopy (FEG-SEM; FEI NOVA 200 Nano SEM, FEI, Hillsboro, Ore). Secondary electron images were performed at 150× magnification at an acceleration voltage of 10 kV. Atomic contrast images were obtained using a backscattering electron detector, at an acceleration voltage of 15 kV.

To confirm similar surface features, the roughness of the disc samples was evaluated, taking into account roughness parameters as the arithmetical mean roughness values (Ra) and the mean roughness depth (Rz). Ra consists of the arithmetic mean value between the peak and valley height values in the effective roughness profile. Rz was calculated by measuring the vertical distance from the highest peak to the lowest valley within 5 sampling lengths, then averaging these distances. Ra and Rz were recorded in 5 different areas on each material (n = 5) using an optical profilometer (NewView, 7300, Zygo, Middlefield, Conn). The measurement length was 0.7 mm, with a cutoff at 0.25 mm for 3 seconds.

The surface hydrophilicity of the specimens was assessed by water contact angle measurements performed in a digital goniometer (OCA 20, Data Physics, Filderstadt, Germany). A volume of 5 μL water was coated onto different areas of the sample surfaces for hydrophobicity measurements at room temperature and humidity (n = 5).

Shear bond strength was evaluated at room temperature using a universal testing machine (Instron 8874, Norwood, Mass), with a load cell of 25 kN capacity and under a crosshead speed of 0.5 mm/s. Shear bond strength (MPa) was calculated by dividing the highest recorded fracture force (N) by the cross-sectional area of the specimens (n = 3). Hardness tests were performed using a Vickers micro hardness tester (DuraScan, EMCO-TEST, Traunstein, Germany) on a load of 4.9 N (500 g) for 15 seconds. Calculation of the average hardness values was obtained from 5 indentations on each of 5 different samples.

Double blinding was ensured through the labeling of the sample using a concealed code that was broken at the end of the study.

Human fetal osteoblasts (hFOB 1.19) were sourced from ATCC (CRL-11372TM; American Culture Collection, Manassas, Va). These are conditionally immortalized cells with the ability to differentiate into mature osteoblasts expressing the normal osteoblast phenotype. Cells were cultured in an atmosphere of 5% CO₂ and 100% humidity in a culture medium composed of a mixture (1:1 v/v) of Ham's F12 medium (Sigma-Aldrich 51651C, Hampshire, UK) and Dulbecco's modified eagle medium (DMEM; Biowhittaker, Lonza, Walkersville, Md) supplemented with 0.3 mg/mL G418 (Roche, Indianapolis, Ind) and 10% fetal bovine serum (Biowest, Nuaillé, France) in 75-cm² flasks (Corning, Corning, NY) until reaching 80% confluence. Cells were detached using trypsin-EDTA (Lonza, Veners, Belgium), centrifuged at 800 rpm and resuspended at a density of 1 × 10⁴ cells/well and cultured at 37°C for all biological assays. All experiments were conducted using a fourth passage.

Immortalized human gingival fibroblasts (Applied Biological Materials Inc, Richmond, BC, Canada) were used. These are conditionally immortalized cells from human gingiva of normal tissue. Cells were cultured at 37°C in an atmosphere of 5% CO₂ and 100% humidity in a culture medium composed of DMEM (Biowhittaker, Lonza) supplemented with 10% fetal bovine serum (Biowest) in 75-cm² flasks (Corning) until reaching 80% confluence. Cell passaging was performed as previously described for osteoblasts and cultured at 37°C for all biological assays. All experiments were conducted using a fourth passage.

Four groups were considered: PEEK, PEEK-HA, PEEK-βTCP, and Ti6Al4V as a control. Sample discs (n = 8) were distributed in 48-well culture plates (Corning) under sterile conditions. Cell viability and proliferation were evaluated on the osteoblast and fibroblasts cells using a resazurin-based viability assay (Cell-Titer Blue reagent, Promega, Madison, Wis) according to the manufacturer's protocol. The conversion rate was measured as fluorescence intensity in arbitrary fluorescence units after 1, 3, 7, and 14 days of culture. Fluorescence intensity was detected at excitation/emission wavelengths of 560/590 nm using a luminescence spectrometer (PerkinElmer LS 50B, Waltham, Mass).

Osteoblasts and fibroblasts were cultured on samples for 1 and 7 days (5% CO₂, 37°C). After being washed, a fixative solution (1.5% glutaraldehyde solution for 10 minutes) was added to the culture wells. A dehydration process took place via serial dilutions of EtOH. Samples were covered with an ultrathin film (15 nm) of Au-Pd (80–20 weight %) using a high-resolution sputter coater (208HR Cressington Company, Watford, UK), coupled to a MTM-20 Cressington high-resolution thickness controller.

Samples were observed under FEG-SEM (FEI NOVA 200 Nano SEM, FEI). Secondary electron images (ie, topographic images) were taken at different magnifications (500×, 1000×, 2000×, and 5000×), at an acceleration voltage of 10 kV. Atomic contrast images were obtained using a backscattering electron detector, at an acceleration voltage of 15 kV. Image analysis was performed by 2 calibrated researchers, focusing on cell morphology, spreading, and the establishment of early contact with materials.

Alkaline phosphatase activity (ALP) was measured at 7 and 14 days under osteoblast culture using a fluorometric enzymatic assay (ab83371 ALP assay Fluorometric, Abcam, Cambridge, UK) following manufacturer instructions. A standard curve was performed at each measurement to calculate enzymatic activity. Standards and samples were measured by fluorescent intensity at excitation/emission wavelengths of 360/440 nm using a fluorescence spectrometer (PerkinElmer LS 45), and values were converted to mU/μL of enzyme (ALP) based on the standard regression equation.

For immunofluorescent staining of nucleic acids, discs cultured with fibroblasts were removed after 3 days of incubation. Cell culture staining was evaluated by means of DAPI (Sigma-Aldrich D9542, Hampshire, UK). Following this, cell fixation was performed using 1.5% glutaraldehyde solution (10 minutes). Images were obtained at 460-nm wavelengths using a Leica TCS SP5 confocal microscope (Leica Microsystems, Buffalo Grove, Ill) coupled to LAS-AF LITE v2.0 software (Leica Microsystems) on 3 replicas.

Statistical analyses were performed using IBM SPSS 24.0 statistics software for Mac (SPSS, Chicago, Ill). Data were tested for normality (Shapiro-Wilk test). Comparisons between groups for roughness values, surface hydrophilicity, shear bond strength, Vickers micro hardness test, proliferation, and ALP activity were carried out using a one-way analysis of variance (ANOVA) and repeated-measures ANOVA for cell viability. A Tukey post hoc test was used to identify groups with significant differences. The significance level was set at P < .05. All data are presented as mean ± standard deviation (SD). All of the methodology was reviewed by an independent statistician.

FEG-SEM micrographs were obtained for all tested groups (Ti6Al4V, PEEK, PEEK-HA, PEEK-βTCP) before cell cultures were undertaken (Figure 1). FEG-SEM micrographs presented similar surface features in all groups with regard to general pattern and sample roughness.

Figure 2 shows the roughness values of Ra (μm) and Rz (μm) as well as surface hydrophilicity by water contact angle (°) of the specimens, presented as mean and SD. The results showed similar roughness values between all groups, with no significant differences (P > .05). However, statistical differences in water contact angle were observed between study groups (P < .05). Moreover, these significant differences were observed in all pairwise comparisons (P < .05). The Ti6Al4V group showed the highest contact angle, followed by PEEK-βTCP, then PEEK and PEEK-HA with the lowest value (P < .05, for all pairs).

Mean values of shear bond strength (Figure 3a) in the Ti6Al4V group were significantly higher when compared with other groups (P < .05). The addition of calcium phosphate bioactive compounds to PEEK implant materials revealed no statistical differences in shear strength between PEEK, PEEK-HA, and PEEK-βTCP (P > .05). Similarly, mean values of Vickers hardness (Figure 3b) in the Ti6Al4V group were significantly higher when compared with other groups (P < .05), whereas no statistically significant differences were observed between the PEEK, PEEK-HA, and PEEK-βTCP groups (P > .05).

Cell viability and proliferation results were obtained for 1, 3, 7, and 14 days in 2 different cellular cultures: osteoblasts and fibroblasts (Figure 4). Osteoblast viability was statistically lower in Ti6Al4V when compared with PEEK and PEEK-βTCP specimens for all time points (P < .05). After 3 days in culture, osteoblasts on PEEK surfaces revealed a higher viability as compared with all other groups (P < .05).

The results showed that early osteoblast proliferation was less expressive in PEEK-HA specimens, with regard to the first week proliferation rate (P < .05). After 14 days in culture, the PEEK group presented the highest ratio of proliferation and PEEK-HA the lowest (P < .05).

Fibroblast behavior showed a different pattern: at up to 7 days of culture, no statistically significance differences between study groups were observed, but after 14 days of culture, fibroblast viability was statistically higher on PEEK as compared with other groups (P < .05). At this time point, PEEK-βTCP also presented statistically higher viability when compared with Ti6Al4V (P < .05). PEEK specimens showed a higher rate of proliferation from 1 to 14 days in cultures when compared with Ti6Al4V and PEEK-HA groups (P < .05). The 7-day proliferation rates did not present statistical differences in any of the groups (P > .05).

FEG-SEM obtained on all group samples after 1 day and 7 days of osteoblast (Figure 5a) and fibroblast (Figure 5b) cultures are shown with corresponding magnifications. FEG-SEM micrographs showed adherent osteoblast cells in all groups after 1 day of culture with similar morphology and typical elongated shape. However, these observations were more evident on PEEK specimens, with increased membrane spreading and more expressive filopodia formations. At 7 days of culture, no apparent differences were observed between groups.

On FEG-SEM with fibroblasts, it was possible to observe in the PEEK, PEEK-HA, and PEEK-βTCP specimens a higher number of adherent cells after 1 day of culture when compared with Ti6Al4V. At the same time point, PEEK surfaces showed a typical fibroblast phenotype, with flattened cell bodies and numerous cellular extensions, whereas in the Ti6Al4V images, the cell bodies had a round shaped. After 7 days of culture, Ti6Al4V specimens showed a lower cell number, density, and poor cell attachment as compared with other groups. In particular, PEEK-based discs revealed a uniform cellular monolayer that almost reached complete confluence over these surfaces.

The ALP activity of osteoblast cell suspension at 7 and 14 days of culture is shown in Figure 6. For the 2 measurements, the PEEK-βTCP group showed the highest ALP activity in osteoblasts when compared with PEEK and Ti6Al4V after 7 days of culture (P < .05) and as compared with all groups after 14 days of culture (P < .05). Differences in ALP activity between 7 and 14 days in culture were not observed (P > .05).

As illustrated in Figure 7, the immunofluorescent staining of fibroblast nucleic acids cultured on discs at 3 days of culture in the PEEK group suggested a higher number of fibroblasts on PEEK surfaces, which is in agreement with the previously described results of initial adhesion. Ti6Al4V surfaces showed lower staining of all groups, followed by PEEK-HA.

PEEK is usually described as an efficient alternative to the systematic use of metallic alloys in orthopedic therapies, as it has both biocompatibility and suitable chemical resistance.²⁰  PEEK is used in dentistry as a material for implant infrastructures, abutments, and fixtures, in cases of bruxism and allergic reactions. However, it is bioinert and does not have any intrinsic osteoconductive properties. In an attempt to increase its bioactivity and, consequently, the ability to increase predictability and rapid osseointegration, bioactive coatings are used. Very limited studies have been done using PEEK as a dental implant, and although it is a promising material, there is currently insufficient evidence to support PEEK as an alternative to titanium for dental implants.

The main goals of our group strategy to produce dental implants with PEEK and bioactive compounds in the outer layer are both to improve biological behavior while maintaining mechanical properties and to overcome the coating detachment problem, thus presenting a major breakthrough in this field with several potential clinical applications.^29,31 

Our results show Ti6Al4V shear bond strength and hardness values as the highest as compared with other groups, which is in accordance with the intrinsic properties of these materials.^9,23,30,33  On PEEK specimens, the addition of 5% of bioactive calcium phosphate compounds did not produce differences in shear bond strength and Vickers hardness when compared with pure PEEK. Within the limitations of these tests, they suggest that this contemporary material design with the integration of bioactive particles might avoid the common outcome of delamination observed in bioactive coatings.³⁴ 

Moreover, as described by several authors, PEEK may be an excellent choice for implant material based on similar mechanical properties of bone and the modulus of elasticity that ensures a uniform distribution of stress at implant, minimizing the relative movement at the implant-bone interface and bone loss attributed to stress shielding.^5,9,12,14,35 

There are several published in vitro toxicity tests that have been performed with nonhuman and human malignant cell lines,^16–18,36  and because of the malignancy of the cell lines described, the results may not be clinical relevant.

For that purpose, our study was developed with immortalized human fetal osteoblasts and fibroblasts as primary cells undergoing a developmental sequence of proliferation and differentiation similar to in vivo behavior. These cells appear to be an excellent model system for the study of in vitro osteoblast biology,³⁷  since they have the ability to differentiate into mature cells, expressing the normal phenotype maintaining characteristics that enable standardization of the assays, thus being important in testing the “biocompatibility” or toxicity of new dental materials.

Our results are in line with previous studies that compared PEEK to titanium alloy (Ti6Al4V) for implant devices.^9,38–41  In our study, the group with a pure PEEK surface showed the strongest viability and proliferation of osteoblasts, with increased cell spreading and numerous filopodia formations when compared with the other groups, which is in accordance with the results reported by Petrovic et al in 2006.³⁵  However, other authors have voiced concerns about the ability of PEEK as a biomaterial for dental implants.^18,36,42 

As widely stated in the literature, bioactive calcium phosphate compounds are optimal components for stimulating bone formation with an affinity to osteoblasts.^3,24  However, our results of PEEK with added bioactive components (HA and βTCP) were not in line with this assumption. Osteoblast adhesion, viability, and proliferation were higher on pure PEEK surfaces, whereas PEEK-HA samples promoted the lowest cell attachment and proliferation values. These results are distinct from a previous study, in which the authors concluded that the effect of HA coating resulted in early bone formation, suggesting that HA-coated PEEK implants could provide improved clinical responses.⁴³  It is noteworthy, however, that the authors do not state that the surface roughness could be an interfering factor. Moreover, given the fact that most of the published studies were realized using coatings of HA or βTCP on titanium implants, the comparison of results from this study with the previous literature is limited,⁷  and we have to consider that our technique of production can modify the structure of bioactive compounds, not reflecting on the same osteoblast results behavior.

Interestingly, although osteoblast viability and proliferation were higher on the pure PEEK group, PEEK-βTCP surfaces showed higher ALP activity results, which corresponded to increased cellular differentiation. Based on PEEK properties, as a bioinert material,^9,15  we may speculate that these results could be associated with the bioactive compounds integrated on the PEEK-βTCP implant matrix specimens. Our results suggest that PEEK-βTCP embedded surfaces may improve osteoblast differentiation and therefore stimulate osseointegration in vivo.^19,33,44,45 

Although fibroblast results were not as expressive, the behavior of these cells was similar to osteoblasts, with a better behavior in PEEK followed by PEEK-βTCP. Under the limitations of this study, the current results support the hypothesis that PEEK may provide an optimal surface for osteoblasts and fibroblasts, thus favoring peri-implant hard- and soft-tissue cell behavior. A potential approach to improve soft-tissue behavior could be the incorporation of amorphous silica fibers as described by Monich et al.⁴⁶  This strategy could provide a possible improvement in the biological properties of pure PEEK in contact with gingival fibroblasts.⁴⁶  This is a desired goal for the long-term stability of dental implants, with optimal surface interaction on the initial wound healing through osteointegration but also in the event of peri-implantitis, enabling favorable gingival and bone cell adhesion.^47,48 

Most of the published results of cell behavior are based on surfaces prepared for a minimum change of surface key parameters,^8,24,26  being that most compared the behavior of cells on different surface roughness. However, other parameters were not assessed.⁴⁹  This study was carried out to understand the behavior of gingival and bone cells represented by fibroblasts and osteoblasts on the contact with these materials produced by hot pressing. For this, all specimens were prepared in each and between groups with similar roughness values after surface treatment, as confirmed by the results of the roughness and FEG-SEM.

However, when surface hydrophilicity was tested, the Ti6Al4V group revealed the highest water contact angle and therefore the lowest hydrophilicity, with the worst cell behavior observed in this group. Although PEEK-HA presented the highest hydrophilicity, cell proliferation was significantly worse as compared with PEEK and PEEK-βTCP. The comparison of our results with other topographical studies is limited based on the few available reports, which refer roughness, wetting, or surface energy data on PEEK dental implant material.⁸ 

These results may question the well-accepted hypothesis that wettability alone can influence the adsorption of proteins and cellular adhesion. Probably, and as proposed previously by others, early protein adsorption is determined by a combination of different parameters including wettability, surface energy, chemistry composition, and others.^8,26,50  In this study, the surface chemistry of pure PEEK dental implants presented better osteoblast and fibroblast viability and proliferation, whereas PEEK-βTCP induced the highest osteoblast differentiation. Although we used the equivalent roughness of the samples, the cell behavior was not improved by the hydrophilicity of surface materials. Based on our data, it may be speculated that cell behavior is probably more dependent on surface composition (ie, surface chemistry) than on wetness itself.

The exact role of chemistry and topography in the early stages of bone integration are still poorly understood. This study presents the major limitation of being an in vitro study, implying that in vivo studies will be required to confirm these findings. Further studies will also be needed to understand bacterial adhesion and growth on these new materials, based on the etiology of Peri-implantitis, the major concern for long-term dental implant success. However, these results set an important basis for the optimization of PEEK as a future alternative material for dental implants.

The results of this study suggest that the addition of 5% HA or βTCP by FGM did not alter the mechanical properties of these new materials. However, they led to a poorer cell response in terms of viability as compared with pure PEEK. PEEK-βTCP showed the highest osteoblast differentiation, which may provide a promising option to improve osseointegration on PEEK surfaces. Our data suggest that cell behavior was not improved by the hydrophilicity of surface materials but is probably more dependent on surface composition (ie, surface chemistry). Further in vitro and in vivo studies are needed to fully understand the determining factors of a successful biological response on PEEK dental implant surfaces for further clinical applications.

Abbreviations

ALP:

alkaline phosphatase activity

βTCP:

β-tricalcium phosphate

CLSM:

confocal laser scanning microscopy

FEG-SEM:

field emission gun-scanning electron microscopy

FGM:

functionally graded material

HA:

hydroxyapatite

PEEK:

polyetheretherketone

SD:

standard deviation

Ti:

titanium

Ti6Al4V:

titanium alloy

The statistical methodology and analysis were reviewed by an independent statistician (Joana Fialho, PhD, adjunct professor, Escola Superior de Tecnologia e Gestão de Viseu, Centro de Estudos em Educação, Tecnologias e Saúde, Viseu, Portugal). The support of Professor Rui Malhó (Faculty of Science, Universidade de Lisboa) in obtaining confocal laser microscopy images is highly appreciated. We would also like to thank Professor Helena Raposo Fernandes and Professor Pedro Gomes (Faculty of Dental Medicine, Universidade do Porto) for their invaluable support in setting up cell experiments. This work was supported by FCT project NORTE-01-0145-FEDER-000018 – Portugal, by FCT reference projects PTDC/EME-EME/30498/2017 and UID/EEA/04436/2013, by FEDER funds through the COMPETE 2020 – Programa Operacional Competitividade e Internacionalização (POCI) with the reference project POCI-01-0145-FEDER-006941, and by Coordination of Improvement of Higher Level Personnel – CAPES - Brazil, PDSE 99999.006407/2015-03 Process.

The authors declare no other potential conflicts of interest with respect to the authorship and/or publication of this article.