The materials utilized in this study included phosphoric acid (85%), zirconium (IV) oxide chloride 8-hydrate (ZrOCl₂·8H₂O), absolute ethanol was obtained from Sigma-Aldrich. Commercial epoxy resin (Bisphenol A-epichlorohydrin prepolymer CAS 25068-38-6-3 units of bisphenol-A, molar mass 853 g/mol) and the hardener (cycloaliphatic amine) were purchase from Siligel. For mould construction, commercial silicon resin and hardener were bought from Siligel. All reagents were used as received.

Nano-zirconium phosphate (ZrP) was synthesized using P:Zr ratio equal 18; a mixture of phosphoric acid (H₃PO₄) and zirconium (IV) oxide chloride 8-hydrate (ZrOCl₂·8H₂O) was kept under reflux, for 48 hours; sequentially, the precipitated was centrifuged, washed with distilled water until pH of 6. Finally, the product was dried in an oven, at 80˚C, until constant weight [26] .

Mixing of epoxy resin and hardener at different mass ratios (1:0.25; 1:0.5 and 1:1 wt./wt.%) was used as obtain polymer matrix. Zirconium phosphate (ZrP) nanoparticle was applied as filler at different wt.% (0; 0.25; 0.5 and 1) related to resin:hardner weight ratio. To accommodate the post, a mould was prepared by mixing of silicon resin + hardener (1:0.05 wt./wt.%) as specified in the label. The composites were prepared at different weight ratios of resin/hardener, filler amount, reaction time and temperature as shown in Table 1 . To better understand, the steps of the mould construction as well as the prepared pin were highlighted ( Figure 1 ).

*Weight ratio.

Diffraction analysis was carried out in a Rigaku Ultima IV diffractometer using 40 kV, 20 mA, step of 0.05, 2θ angle ranging from 2˚ to 40˚.

The infrared evaluation was performed in Perkin-Elmer equipment, model Frontier MIR/FIR, within the range of 4000 - 400 cm⁻¹. The spectra were obtained by attenuated total reflectance (ATR), using 60 scans and a resolution of 4 cm⁻¹. Before and after curing, the epoxy resin degree of conversion was evaluated by ratio between absorptions at 914 cm⁻¹ (epoxy ring group, variable band) and 1581 cm⁻¹ (C=C aromatic ring, invariable band). Each ratio value was divided by the same ratio determined using the uncured resin. The difference from 100% was considered the degree of conversion [27] [28] .

The insoluble content was determined by immersing the specimen in the absolute ethanol. Two specimens were taken in ethanol for 24 hours. After that, the specimens were removed from the liquid and left in an oven for 24 hours. The specimen weight was weighted before and after its immersion in the liquid. The weight difference was taken and related to the initial specimen weight and accepted as extracted matter. The insoluble matter was calculated considering the initial mass of the specimen as 100% and decreasing the value of the extracted.

TGA data was acquired throughout TA analyzer model Q500, between 30˚C - 700˚C, at 10˚C·min⁻¹, under a nitrogen atmosphere. T_onset, T_max and the temperatures where the mass loss were 10, 25, 50, 75 wt.%—T₁₀, T₂₅, T₅₀, T₇₅—were registered.

The flexural properties were performed adapting the ISO 4049 Standard in an Emic DL2000, load cell 20 kN, speed of 1 mm/min. Seven specimens were tested being the flexural modulus and flexural strength evaluated. The median was considered. The fractured surface was evaluated to the confocal and AFM equipments.

The analysis was conducted in OLS4100 Olympus equipment in the specimens after the flexural test. Images of the transversal surface topograph were taken when possible.

AFM analysis was conducted in Park Systems XE7 equipment. For each sample, a disk with 2 mm was prepared, cutting it through Isomet machine and polished with sandpapers with different grain sizes.

The transverse section SEM images were captured using a Tescan field emission microscope, model MIRA 4 LMU (LowVac Mode UniVac^TM) equipment, voltage of 10 kV). Elemental analysis was performed with an EDS detector equipped with a 30 mm² Si₃N₄ window, with a resolution lower than 129 eV for the MnKα emission line.

All samples revealed very similar X-ray diffraction patterns. For this analysis, the posts contained 0.25 wt.% of ZrP were chosen as representative varying the ratio of epoxy resin:hardener 1:0.25, 1:0.50 and 1:1 ( Figure 2 ). Diffraction angles were recorded at 12.1˚, 20.1˚, 25.3˚ and 34.4˚ which correspond to the ZrP crystallographic planes d₀₀₂, d₁₁₀, d₁₁₂ and d₀₀₉, respectively [29] [30] . As amorphous polymer, the epoxy resin did not exhibit crystalline planes.

Due to their similarity, Figure 3 displays the representative posts’ infrared spectra of each Group. The absorptions at 3393 cm⁻¹ (N-H stretching); 2958 cm⁻¹ (C-H stretching of CH₃); 2914 cm⁻¹ (C-H stretching of CH₂), 2870 cm⁻¹ (C-H stretching of aldehyde); 1607 cm⁻¹ (N-H bending); 1581 cm⁻¹ (C=C aromatic ring); 1454 cm⁻¹ (CH₂ deformation mode) 1362 cm⁻¹ (C-O stretching); 1295 cm⁻¹ (C-C-H aromatic ring); 1181 cm⁻¹ (C-O and C-N deformation mode); 1034, 1018 and 961 cm⁻¹ (P-O-H and PO deformation mode); 914 cm⁻¹ (epoxide ring vibration); 828 cm⁻¹ (C-O-O epoxy ether and C-H out of plane deformation mode) were endorsed with the articles of Sabu et al., Sukanto et al., Ullah et al. and Mendes et al. [27] [28] [31] [32] . Table 2 presents the epoxy resin degree of conversion arranged according to the ratio of epoxy resin:hardener as function of ZrP content, time and temperature. The pins without ZrP showed a high degree of conversion (70-80%). Increasing reaction time and temperature, the degree of conversion was not affected. The addition of ZrP showed significant influence on the degree of conversion. Posts of the Group II and III showed a significant decrease in the degree of conversion. The greatest decrease was observed in the posts of the Group III (0.5 wt.% of ZrP). When the time reaction was increased there was an

*Weight ratio; **24 h; ***90˚C; ^#no post formation.

increment of 34% in the degree of conversion (1:0.5:0.5, Group) while the increase of the reaction temperature did not reveal any improvement. The variation of the posts’ degree of conversion with the addition of ZrP can be attributed to the restriction of mobility of the reactive groups of the epoxy resin and the hardener. The phosphate structure promoted a type of physical block, reducing the chances of collision between the epoxide groups and the amine groups, which resulted in a lower degree of conversion.

To evaluate the effect of epoxy resin:hardener ratio on the insoluble matter, the posts contained 0.25 wt.% of ZrP were chosen with ratios of 1:0.25, 1:0.50 and 1:1. As seen in Table 3 , the insoluble content ranged between 97% - 99%. It was understood that the time and temperarure used in the posts’ preparation were enough to maintain the molecules of epoxy resin and hardener in an interconnected network independent on the degree of conversion attained for each one. The behavior coud be extended to the posts embedded with 0.5 wt.% of ZrP.

*Weight ratio.

Figure 4 shows the loss mass and derivative curves of the posts 1:0.25:0, 1:0.25:0.5, 1:0.5:0.5, 1:1:0.5, 1:0.5:0.5, 1:0.5:0.5 (Groups are described in parentheses). All samples revealed two steps of degradation. In general, the first one occurred around 100˚C - 300˚C while the second one appeared at 300˚C - 500˚C. Table 4 displays the maximum degradation temperature and onset temperature of each post. T_onset of the posts ranged from 320˚C to 340˚C. The values found for the posts obtained at 24 hours and 90˚C were slightly upper than those build at 4 hours and 70˚C. Similar behavior was observed for T_max1 and T_max2. Zhang et al. developed composites based on epoxy resin containing ZrP viewing application as flame retardant. The presence of the phosphate decreased the thermal stability. Only one step of degradation was detected and T_max varied between 370˚C - 380˚C [33] . Guo et al. prepared a composite of epoxy resin compounded with mixing of aluminum oxide/aluminum oxide modified with magnetite (Al₂O₃/Al₂O₃@Fe₃O₄) [20] . Only one degradation step was found and there was no improvement in the thermal stability. The two degradation steps may be related to the reaction conditions (time and temperature) that were not enough to promote complete curing of the resin. Regardless of the presence of filler, all posts showed two steps of thermal degradation. From insoluble content, it was possible supposed that even with incomplete curing, there is a three-dimensional network formed by the reaction between epoxy and amine groups meaning that practically there were not free molecules of resin and hardener into the posts. Thus, the two degradation steps represent domains with unven crosslinking density.

*No post formation.

Figures 5(a)-(e) shows the representative curves of the flexural test. For the posts of the Group I, that one without ZrP (1:0.25:0) the mechanical profile seemed to be a hard and soft material. Regions of yielding, cold drawing and breaking. The curve of the post (1:0.25:0.25) presented britlle behavior. For the Group II, the post without ZrP had similar mechanical mode to that Group I without breaking. The posts (1:0.5:0.25) and (1:0.5:0.5) broken as a brittle material. For the posts of the Group III, the curves with and without ZrP resembled a plasticizing material without yielding region. All of them did not break. For the posts of the Group IV and V, the mechanical behavior of the posts (1:0.5:0) and (1:0.5:0.5) were resembled to their counterparts in Group II. Table 5 displays the flexural modulus and strength. The results did not show trend. For the posts of Group I, the presence of ZrP in the post (1:0.25:0.25) induced an increase of 54% and 18% of the flexural modulus and flexural strength, respectively. Group II registered an increment of the flexural modulus (4.9 %) for the post (1:0.5:0.5) but the flexural strength of (1:0.5:0.25) and (1:0.5:0.5) posts were significantly decreased. For all posts of the Group III, flexural modulus and strength attained the worst values. When time (Group IV) and temperature (Group V) were increased no improvement of the mechanical properties was noticed. The flexural properties of six commercial endodontic fiber posts were evaluated by Plotino and collaborators. The flexural modulus ranged from 24.4 ± 3.8 GPa and 108.6 ± 10.7 GPa for silica fiber and stainless posts, respectively. The flexural strength fluctuated from 879.1 ± 66.2 MPa and 1545.3 ± 135.9 MPa for silica fiber and cast gold fiber posts, respectively.

*Weight ratio; **median value; ***no post formation; ****no break.

The values were upper than those observed for flexural modulus and flexural strength of human dentin, 17.5 ± 3.8 GPa and 212.9 ± 41.9 MPa, respectively [34] . Taskiran et al. compounded a dental formulation based on composite containing mixing of monomers—ethoxylated bisphenol A glycol dimethacrylate (Bis-EMA) plus urethane dimetacrylate (UDMA) plus triethylene glycol dimethacrylate (TEGDMA) which was embedded with barium glass, colloidal and fumed silica, zirconia (ZrO₂) and hydroxyapatite (HA) nanoparticles [35] . According to the authors, flexural strength (70 - 80 MPa) and flexural modulus (3.5 - 4.0 GPa) were superior than that for control sample. Zirconium dioxide (ZrO₂) stabilized with yttrium oxide (Y₂O₃) has been used as dental prosthetics material. Mustafa et al. prepared composites of ZrO₂ and Y₂O₃ (0.5 to 3.0 wt.%) in matrix of epoxy resin. The flexural strength varied between 90 - 110 MPa and for both composites the best behavior was achieved to 1 wt.% of each oxide [36] . Considering the reaction parameters—resin:hardener ratio, ZrP content, time and temperature reaction, it was possible to verify that each Group presents a specific behavior. For Groups I and II there was some improvement in the modulus independent of the resin:hardener ratio, but the increase in the hardener content causes a sharp decline in the flexural strength. The best mechanical properties were achieved for the post 1:0.25:0.25. In group III, the modulus and flexural strength were the worst while the variation of time and temperature did not produce the expected results. The poor results could be associated with the low resin degree of conversion, agglomeration of ZrP nanoparticles and also their distribution and dispersion in the epoxy resin matrix.

Confocal images ( Figure 6(a) ) and three-dimensional topology ( Figure 6(b) ) of the fractured surface from flexural test of the following posts: 1:0.25:0 and 1:0.25:0.25 (Group I); 1:0.5:0.25 and 1:0.5:0.5 (Group II); 1:0.5:0 and 1:0.5:0.5 (Group IV); 1:0.5:0 and 1:0.5:0.5 (Group V) were evaluated. In Figure 6(a) , Group I posts surfaces presented as continuous slots which were better seen in the post without ZrP. The presence of the ZrP rendered the surface rougher. About epoxy resin-ZrP adhesion, it was believed that it was high, although some voids were noticed which could be associated with the localized poor dispersion of ZrP in the epoxy matrix. The surface images of the Group II posts showed great similarity to that one of Group I with ZrP. As can be seen, the post with the higher ZrP content showed the largest number of voids. This irregularity could be attributed to a lower dispersibility and non-homogeneity in size of the ZrP into epoxy matrix. The posts of Group IV are related to the evaluation of the reaction time effect. The surface topology of the post without ZrP was similar to that its counterpart in Group I. The post with ZrP (50 wt.%) showed heterogeneity in the size of the ZrP domains, high adhesion between of matrix-ZrP and some voids also attributed to the dispersibility of ZrP in the polymer matrix. The images of Group V posts resembled to those of Group IV. In Figure 6(b) , the images possess a range of color which represents variation in the height of each surface. The purple color is related to the lowest height while the highest is the red one. The images endorse the details taken from the black and white images regarding the size heterogeneity of ZrP, its dispersion and distribution in the epoxy matrix and the matrix-ZrP adhesion. Qian et al. studied the use of metal cations viewing the formation of passivation of coating layer [37] . Graphene oxide was modified with cerium dioxide after being incorporated into epoxy matrix. The coating was immersed in salt solution for 14 days. After that, laser scanning confocal microscope was used to evaluate the coating corrosion. The best result was attained by the sample enriched with CeO₂@rGO/EP-0.50. Nanosheets of graphene oxide (F-GO) was chemically modified with zinc oxide quantum dots (ZnO QDs) forming nanohybrid filler labeled as F-GO@ZnO QDs [38] . Nanocomposite of waterborne epoxy coating was formulated with 1 wt.% of nanohybrid filler. By laser scanning confocal microscopy, the authors pointed out that s the roughest surface was attained with F-GO@ZnO QDs nanohybrid filler. These findings are in consonance with those of flexural analysis.

AFM images of the 1:0.25:0.25 (Group I); 1:0.5:0.5 (Group II); 1:0.5:0.5 (Group III); 1:0.5:0.5 (Group IV); 1:0.5:0.5 (Group V) posts are presented in Figure 7 . Observing the topology, some similarities among the posts can be highlighted such as the lamellar nature and nanometric size of ZrP, heterogeneity of size, dispersion and distribution of ZrP domains due to cluster formation. ZrP increased the posts surface roughness. Through AFM images, Khan et al. prepared nanocomposites based on epoxy resin embedded with alumina nanoparticles [39] . The authors associated the increase of nanocomposite breakdown strength with high surface roughness. The images showed excellent wettability of ZrP by the resin and regardless of the conditions imposed for the development of each group of posts, it was possible to conclude that high adhesion between resin-ZrP was achieved.

between epoxy resin-ZrP, a rough surface can be observed. The gray to black colored tiny regions were associated with the epoxy resin which was capable to conduct the repairation of the previous damage occurred on the fragmented surface of the ZrP particles. A brittle fracture was seen. Post formulated as 1:0.5:0 (Group IV) revealed a smooth and homogenous surface, low shearing deformation, uninterrupted crack propagation path and brittle fracture. Post designated as 1:0.5:0.5 (Group IV) showed good dispersion and distribution of ZrP, rough surface and brittle fracture. Smooth and uniform surface was noticed for the post 1:0.5:0 (Group V). The post 1:0.5:0.5 (Group V) displays high adhesion

between matrix and ZrP, good dispersion and distribution of ZrP, rough surface and brittle fracture. EDS evaluation was performed in order to describe the profle of the ZrP dispersion and distribution on the fractured surfaces ( Figure 9 ). Generally speaking, the ZrP particles were better accommodated in a matrix developed from epoxy resin:ZrP ratio of 1:0.5 but their posts showed the worst mechanical results. Additionally, as the amount of ZrP increased the greatest is the roughness promoting heterogeneity in the stress distribution. Also, high content of ZrP tends to create difficulties in the collision between the reactive groups of the epoxy resin and the hardener, the degree of conversion and the crosslinking density are drastically reduced. Through dual functionalization of ZrP, Zhu et al. prepared exfoliated nanocomposites based on epoxy resin [40] . Although the roughness has increased at high content of modified ZrP (16 wt.%) high elastic modulus was achieved. Composites of epoxy resin embedded with epoxy-functionalized POSS (E-POSS) and glass fiber (GF) were studied by Jiang and collaborators [41] . SEM images indicated that the epoxy resin was well accommodated in the composite with 10 wt.% of E-POSS but those ones with 10 wt.% of GF and mixing of E-POSS and GF (26 wt.%) revealed voids on account of GF pull out. Herein, the 1:0.25:0.25 post presented the best mechanical result in both modulus and flexural strength which could suggest that epoxy:resin ratio of 1:0.25 was the most suitable.