Boron nitride nanosheets as an effective strategy against the slow crack growth and hydrothermal ageing in zirconia composites

Boron nitride nanosheets as an effective strategy against the slow crack growth and hydrothermal ageing in zirconia composites C. Mu˜ noz-Ferreiro a,b,c , A. Morales-Rodríguez a , H. Reveron b , E. Guisado-Arenas a , S. Cottrino b , P. Moreno d , J. Prada-Rodrigo d , J. Chevalier b , ´ A. Gallardo-L´ opez a , R. Poyato c,* a Dpto. de Física de la Materia Condensada, ICMS, CSIC-Universidad de Sevilla, Apdo. 1065, 41080 Sevilla, Spain b CNRS, INSA Lyon, Universit´ e Claude Bernard Lyon 1, MATEIS, UMR5510, 69621 Villeurbanne, France c Instituto de Ciencia de Materiales de Sevilla, ICMS, CSIC-Universidad de Sevilla, Avda. Am´ erico Vespucio 49, 41092 Sevilla, Spain d Grupo de Aplicaciones del L´ aser y Fot´ onica (ALF-USAL), Unidad de Excelencia en Luz y Materia Estructuradas (LUMES), Universidad de Salamanca, Pl. de la Merced, s/n, 37008 Salamanca, Spain ARTICLE INFO Keywords: Boron nitride nanosheets Yttria stabilized zirconia Low temperature degradation ABSTRACT This paper explores the effectiveness of boron nitride nanosheets in preventing the premature failure of yttriastabilized tetragonal zirconia ceramics, particularly in humid environments. A simple, low-cost and scalable technique -shear exfoliation in a kitchen blenderwas used to prepare BNNS, and pure zirconia and composites with 1, 2.5 and 5 vol. % BNNS were spark plasma sintered. Accelerated hydrothermal ageing experiments in autoclave revealed a remarkable improvement of low temperature degradation resistance in all the composites. Fracture toughness and slow crack growth of the composites with 1 and 2.5 vol. % BNNS were evaluated by bending tests performed in notched specimens. Although the composites presented fracture toughness values similar to those of the reference zirconia, an increase of ~18 % on crack-tip toughness was achieved. Similar Rcurves evaluated in air and in oil-impregnated 2.5 vol. % BNNS composites revealed a limitation of stress-assisted corrosion by water in zirconia, thanks to the BNNS incorporation. 1. Introduction Yttria-stabilized tetragonal zirconia polycrystals (YTZP) are technical ceramics widely used in structural and biomedical applications due to their superior fracture resistance resulting from a phase transformation toughening mechanism. In particular, the best known TZP bioceramic contains 3 mol. % of yttria (Y 2 O 3 ; 3YTZP) and leads to strength values above 1000 MPa, excellent biocompatibility, high hardness and wear resistance. However, in YTZP ceramics, mechanical properties can deteriorate in humid or aqueous environments. The interaction of water molecules with strained Zr-O-Zr bonds at any preexisting cracks initiates its propagation under tensile loading at a stress intensity (K I ) value below the critical stress intensity factor or fracture toughness (K IC ). This process deteriorating the bonding at the crack tip, referred to as “subcritical crack growth” or “slow crack growth” (SCG), results in a delayed failure that significantly limits the long-term performance of zirconia components [1,2]. On the other hand, the water molecules can also promote the process of spontaneous nucleation and growth of monoclinic grains, from the surface to the bulk, resulting in grain pull-out, microcracks and surface roughening. This spontaneous transformation from the tetragonal (t) to the monoclinic (m) phase, which occurs at low temperature, in the absence of stress but in the presence of water has been referred to as low-temperature hydrothermal degradation (LTD) or ageing [3]. Significant research efforts are currently underway to develop tougher and stronger zirconia-based composites to ensure the long-term reliability of these oxide ceramics. Zirconia-toughened alumina composites, as well as ceriaand magnesia-doped zirconia, are the most promising choices for overcoming ageing/SCG issues discussed above and related to water-biased fracturing processes in YTZP [4]. The microstructural refinement of highly transformable Ce-TZP-based composites through the addition of different oxides resulted in an improvement of their resistance to SCG [5]. The addition of La 2 O 3 was beneficial in improving the ageing stability of 2YTZP (i.e. 2 mol. % Y 2 O 3 ) matrix alumina toughened zirconia (ATZ) composites, with a superior balance between ageing and crack growth resistance compared to conventional 3YTZP and ATZ composites without La 2 O 3 doping [2]. Since metal transition carbides and nitrides are almost insensitive to * Corresponding author. E-mail address: [email protected] (R. Poyato). Contents lists available at ScienceDirect Open Ceramics journal homepage: www.sciencedirect.com/journal/open-ceramics https://doi.org/10.1016/j.oceram.2025.100816 Received 23 April 2025; Received in revised form 10 June 2025; Accepted 10 June 2025 Open Ceramics 23 (2025) 100816 Available online 10 June 2025 2666-5395/© 2025 The Authors. Published by Elsevier Ltd on behalf of European Ceramic Society. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ). SCG, Zhang et al. [1] have shown that the reinforcement of zirconia with covalent ceramics also resulted in composites with promising resistance to subcritical crack propagation assisted by water molecules. In the last decade, preventing the water-induced fracture of zirconia composites by incorporating low-dimensional carbon nanostructures has received increasing attention [6–11]. The incorporation of a low volume fraction of carbon nanotubes (CNT) advantageously balances the ageing and crack propagation resistance of 3YTZP composites thanks to the small grain size of the sintered zirconia matrix, which impedes the t-m phase transformation that characterizes the ageing, and the incorporation of new toughening mechanisms related to CNT [6]. The reduced micro–cracking and the substantially increased resistance to ageing in 3YTZP composites with 0.5–2.5 vol. % CNT have been related to the modification of the ceramic grain boundaries due to the nanotubes incorporation and to the ability of CNT to absorb the elastic energy related to the martensitic t-m transformation, by means of nanotube bending and inter–bundle sliding [8]. Graphene-based nanomaterials (GBN) have also been successfully used to prevent hydrothermal ageing of zirconia-based materials [7,9]. Although some authors [7] have related the inhibition of the t-m phase transformation in zirconia composites containing graphene platelets (GPL) to the grain refinement promoted by the introduction of GPL, more recent studies [9] have also correlated the enhanced resistance to hydrothermal ageing in zirconia-GBN composites to the graphene’s impermeability to moisture [12]. In recent years, several studies have reported the enhanced mechanical properties and tailored thermal and electrical properties of ceramic or polymer composites when incorporating inorganic nanostructured fillers, including hybrid or modified fillers as boron nitride (BN) or alumina [10,11,13–15]. These composites are proposed for their use in structural and functional applications. Specifically, a few works [10,11] have studied the incorporation of 2D boron nitride (BN) to zirconia to prevent the degradation of mechanical properties in humid environments. This 2D nanomaterial has properties as exceptional as those of graphene –high mechanical strength, excellent thermal conductivity and impermeability to moisture and other gases [16]– and also overcomes some of graphene’s disadvantages, such as the risk of cytotoxicity, which could hamper the use of ceramic composites with graphene as biomaterials [10]. Previous studies have revealed the high biocompatibility of BN [17,18] and 3YTZP composites with boron nitride nanoplatelets (BNNP) have been proposed for application as dental materials [10]. In this sense, the study of the low-temperature degradation (LTD) of these composites takes an enormous relevance. The only work available up to date on LTD of ceramic composites with boron nitride 2D nanomaterials showed noticeable enhancement of the resistance to LTD in ATZ nanocomposites. The increased ageing resistance was attributed to the homogeneously dispersed, randomly oriented BN nanoplatelets, which prevented moisture penetration at the surface of the polycrystalline ceramic material [10]. Recently, Mu˜ noz-Ferreiro et al. [11] showed that the incorporation of 2.5 vol. % hydroxylated BN nanosheets (BNNS) into 3YTZP composites successfully inhibited slow crack growth, increasing the crack growth resistance of the composite by ~30 %. The nanosheets were distributed throughout the 3YTZP matrix with the ab plane perpendicular to the crack front, and the authors related the enhancement in slow crack growth resistance to the blocking by the BNNS of any water molecule attempting to reach deep in the composite. However, despite these promising results, to the best of our knowledge, to date there are no publications on the hydrothermal degradation of BNNS/YTZP composites. The common synthesis approaches to obtain BNNS include bottomup techniques such as chemical vapor deposition or solid-phase reaction and top-down routines such as liquid phase exfoliation of hexagonal h-BN powder by ultrasonication or ball milling [11,13,19,20]. However, the search for environmentally friendly procedures with high production rate and easiness to scale up that use common solvents and minimize pollutant chemical agents is still ongoing. It has been demonstrated that defect-free few-layer graphene (FLG) can be obtained from graphite flakes using a kitchen blender [21,22], even with household detergent as surfactant [23]. The dominating lateral forces in the mixer lead to simple self-exfoliation of the particles. Furthermore, Varrla et al. [23] characterized the scaling of both FLG concentration and production rate with the mixing parameters - mixing time, initial graphite concentration, rotor speed and liquid volumeand found that the FLG production rate increased with the blended volume, meaning that it can be scaled up to industry levels, contrary to ultrasonic agitation methods in which the production rate stuck at a certain volume. Coleman et al. [24] had already optimized the liquid exfoliation of various layered materials, including h-BN, when Paton et al. [22] managed to exfoliate large quantities of defect-free few-layer graphene with a kitchen blender as a proof of concept, suggesting that this method could be applied to exfoliate BN, MoS 2 and a range of other layered crystals. In this study, the scaling behavior of the graphene production rate was fully characterized, revealing that exfoliation could be achieved in liquid volumes from hundreds of milliliters up to hundreds of liters and beyond. The objective of this study is to evaluate the effectiveness of BNNS in preventing the premature failure of zirconia ceramics in humid environments. To that end, BNNS were exfoliated with a simple, low-cost and scalable technique and incorporated into a 3YTZP matrix. The hydrothermal ageing response of zirconia composites containing 1, 2.5, and 5 vol. % BNNS was investigated by performing accelerated autoclave ageing tests. The fracture toughness (K IC ) and slow crack growth (SCG) behavior of the composites with 1 and 2.5 vol % BNNS were evaluated in bending tests and compared with the reference zirconia ceramic to clarify the effect of BNNS incorporation on the fracture behavior and slow crack growth sensitivity to water environment. 2. Experimental procedure 2.1. Materials processing 2.1.1. Synthesis and characterization of the BN nanosheets BNNS were produced by the shear exfoliation of a commercial hexagonal h-BN powder (<44 µm particle size, 99.5 % purity; Ref. 11,078.18; Alfa Aesar, Kandel, Germany) using a kitchen blender (KB). This method was selected due to its high production rate, relatively low-cost, and easiness to scale up [21–23]. The liquid shear exfoliation of BNNS was performed using a Silvercrest kitchen blender (Lidl, Germany) with a maximum power of 600 W and a maximum capacity of 1.75 L. The rotational speed of the blades could not be controlled, as only five fixed velocities (named from 1 to 5) were available. The optimal speed was selected by comparing the resulting BNNS after their exfoliation at low (1), medium (3) and high (5) velocities. For each procedure, 5 g of h-BN powder were added to the blender jug filled with 500 ml of isopropyl alcohol (iPrOH), and the blender operated for 30 min. The suspension concentration and exfoliation time were chosen based on the studies from Varrla et al. [23] and Coleman et al. [24]. The mixer was turned off for a few minutes every 3 min of exfoliation until the rotor and suspension had cooled down. The suspension was then collected, allowed to settle for one day and centrifuged at a relative centrifugal force RCF of 360 g (with g=Earth’s gravitational field) for 45 min in either a Rotofix 32 A or an Allegra X-12R equipment (Beckman Coulter, USA; at Centro de Investigaci´ on, Tecnología e Innovaci´ on de la Universidad de Sevilla, CITIUS). Finally, the upper three quarters of the supernatant were collected to avoid collecting possible agglomerates located towards the bottom of the tube. This suspension of BNNS in i PrOH was used directly as the raw material to prepare the composite zirconia powders. The particle size distributions after the exfoliation processes were quantified by laser granulometry (LG, Mastersizer 2000, Malvern Panalytical, UK; at CITIUS) and dynamic light scattering (DLS, Zetasizer Nano Z, Malvern Panalytical, UK; at CITIUS). The BNNS suspension obtained after centrifugation was added dropwise to the Hydro 2000S C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 2 accessory (Malvern Panalytical, UK) filled with i PrOH. The Fraunhofer model, which assumes a spherical shape for particles in suspension, was used for the LG calculations. Therefore, when analyzing 2D nanomaterials, such as the BNNS in the present study, the particle size determined by this technique does not represent the real particle size distribution and should be considered only for qualitative comparison between similar materials. The LG particle size detection ranges from 0.02 to 2000 μ m while DLS size measurements are limited from 0.6 nm to 6 µm. The BNNS suspension was diluted to a concentration of 1:10 in i PrOH to prevent decantation during measurement acquisition. Both resulting particle size distributions were averaged from at least three measurements. In this work, the absorbance spectrum in the entire visible range of BNNS synthesized in the kitchen blender was recorded with a Cary 5000 spectrophotometer (Agilent, USA; at Instituto de Ciencia de Materiales de Sevilla, ICMS). The BN nanosheet suspension was diluted tenfold and placed in a 1 cm side glass cuvette for analysis. UV–Vis spectroscopy results were then used to determine the concentration of the BNNS in the suspensions. The absorbance at 300 nm was measured (Pharo 300 spectrophotometer, Merck, Germany; at CITIUS) and the extinction coefficient was taken as ε 300 =2367 ml mg -1 m -1 , according to Coleman et al. [24]. The morphology and structural integrity of the exfoliated BNNS were characterized by transmission electron microscopy (TEM), highresolution transmission electron microscopy (HR-TEM) and Raman spectroscopy. To that end, a few droplets of the sonicated BNNS suspension in i PrOH were deposited on a Cu transmission grid with C coating (for TEM and HR-TEM inspection) and on a glass slide (for Raman spectroscopy). The Raman spectra were acquired using a dispersive microscope (Horiba Jobin Yvon LabRam HR800, Kyoto, Japan; at ICMS), equipped with a He-Ne green laser (532.14 nm) at 20 mW. The microscope used a 100x objective and a confocal pinhole of 100 μ m. The Raman spectrometer was calibrated using a silicon wafer. TEM and HR-TEM images were acquired with a FEI Talos S200 transmission electron microscope (FEI, Oregon, USA; at CITIUS). 2.1.2. Processing and sintering of 3YTZP-BNNS composites A commercially available 3YTZP powder with 40 nm particle size was used as matrix (TZ-3YB-E, Tosoh Corporations, Tokyo, Japan). The as-received 3YTZP powder was annealed at 850 ◦C for 30 min to remove organic additives. Then, this powder was added directly to the BNNS suspension obtained from the shear exfoliation process in the kitchen blender, and the two phases were mixed by strong magnetic stirring. Composite suspensions containing 1, 2.5, and 5 vol. % BNNS were prepared. After drying on a hot plate, the resulting powders were homogenized in an agate mortar before sintering. The composite powders were spark plasma sintered at 1250 ◦C for 5 min under vacuum, using heating and cooling ramps of 100 and 50 ◦C/ min, in a SPS model 515 S (Dr. Sinter, Inc., Kanagawa, Japan; at CITIUS) or in an HPD 25 model (FCT Systeme GmH, Germany; at Laboratoire Mat´ eriaux Ing´ enierie et Sciences, MATEIS) to prepare disk-shaped samples of 15-mm or 40-mm diameter, respectively. A 75 MPa pressure was applied from 700 ◦C when heating and it was maintained during the sintering process until its removal at the same temperature while cooling down. The temperature was monitored using an optical pyrometer focused on a hole drilled in the center of the graphite die. A graphite foil (0.35 mm thick) was placed between the powders and the die/punches and along the inner wall of the mold to ensure electrical, mechanical, and thermal contact and for easy sample removal. The resulting sintered composite disks (3 – 4 mm thick) were manually ground to remove the graphite paper from the SPS molding system. 2.2. Microstructural characterization The bulk density of the sintered samples was measured by the Archimedes’ method, using distilled water as impregnation and immersion liquid. The immersion protocol described in ASTM C373–18 was followed. The specimens were dried in an oven at ~ 100 ◦C for at least 2 h before being weighed in air (m a ) using a precision balance (Mettler AM100). The samples were then degassed by placing them under vacuum in a desiccator for 2 h 30 min. The immersion process was completed by bringing the desiccator to atmospheric pressure and leaving the samples immersed for 24 h to ensure complete impregnation of the open porosity. After this procedure, the impregnated samples were weighed in air (m ia ) and in distilled water (m iw ) using a precision balance (Mettler AM100) and the experimental density ( ρ exp ) calculated as follows: ρ exp =ma mia −miw ρ w where ρ w is the water density at the test temperature. For each material, the experimental densities were averaged from at least three different measurements. The rule of mixtures was used to calculate the theoretical densities of the composites, considering the density values of 6.05 g cm -3 for 3YTZP and 2.1 g cm -3 for h-BN, provided by the material suppliers. Semi-quantitative analysis of the crystallographic phases present in the sintered composites was performed by X-ray diffraction (XRD, model D8 Advance A25, Bruker Co. Massachusetts, USA; at CITIUS) in BraggBrentano configuration, equipped with a Lynxeye PSD detector (Bruker, Germany), with copper K α radiation. Low-magnification scanning electron microscopy (SEM, FEI-Teneo, FEI, Oregon USA; at CITIUS) using backscattered electrons (BSE) for imaging was used to characterize the BNNS distribution in the zirconia matrix. To that end, cross section (c.s.) surfaces, i.e. surfaces parallel to the SPS pressing axis, were polished with diamond paste down to 1 μ m. SEM was used to examine the fracture surfaces of the composites as well as to characterize the BNNS morphology and ceramic grain size, quantifying the mean equivalent plane diameter and its corresponding standard deviation. ImageJ and OriginLab softwares were used to analyze the grain size distributions by measuring more than 300 grains for each material and fitting to a lognormal distribution. 2.3. LTD experiments – accelerated ageing tests Autoclave ageing tests were performed to evaluate the sensitivity of the composites to hydrothermal degradation. Samples with in-plane sections polished down to 1-micron were placed in an autoclave (SANOclav, model TKL-MCS-5, Germany) with distilled water at 134 ◦C and 0.2 MPa for controlled periods of time according to ISO standard 13356. In-plane sections of the samples were selected for the analysis since previous studies revealed that hydrothermal ageing resistance on these composites was decreased when analyzed on the cross-sections [25]. The evolution of the transformed monoclinic phase was quantitatively followed on the polished surface by XRD after each steam exposure period. The XRD data were acquired over a short 2θ range (from 27◦to 33◦), with a step size of 0.015◦and a time per step of 0.5 s. The monoclinic zirconia volume fraction (V m ) was estimated as described by Toraya [26]: Vm=1.311Xm 1+0.311Xm The monoclinic/tetragonal zirconia content ratio, Xm, was determined by measuring the area under the monoclinic and tetragonal principal diffraction peaks registered in the XRD patterns, using the Garvie and Nicholson equation [27]. The transformed monoclinic fraction as a function of the ageing time was analyzed assuming nucleation and growth processes in the framework of the Kolgomorov-Jhonson-Mehl-Avrami (KJMA) formalism: X=1−exp{ − [b⋅(t−t0)]n}, where X is the monoclinic content rescaled to the saturation value, X= C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 3 Vm Vmax, b is an Arrhenius thermally activation term, t0 is the induction time, and n is the Avrami exponent related to the nucleation and growth conditions of the monoclinic regions [28]. The onset of transformation t0 was accurately estimated after [29] as t0=tslope − α (tinf −tslope), where tinf is the time at which the transformation rate is maximum (obtained using dX dt curves), tslope =tinf −X(tinf ) dX dt (tinf )and α = − 0.014,corresponding to n=1 (which is consistent with the best-fit results that were obtained in the KJMA-plots of our study). The ageing parameters n and b were estimated by linear fitting of the KJMA-plots, i.e. ln(− ln(1−X)) versus ln(t−t0), in the interval corresponding to X =0.1 and X =0.9. 2.4. Mechanical characterization Fracture toughness (K IC ) and crack growth resistance curves (Rcurves) were evaluated using the single edge V-notched beam (SEVNB) technique in bars machined from 40-mm disk-shaped sintered composites. The straight-through-thickness V notch was made in the center of 3 mm x 3.8 mm x 25 mm (b x w x L) rectangular specimens. The 3 mm wide side received the starting notch (diamond saw) that was finished by ultrashort pulsed laser ablation (UPLA). The notch-length-to-width ratio, which is the total notch length, was 0.3 for the K IC measurements and 0.5 for the R-curve analyses. Experimental details of the notch laser ablation were described elsewhere [11]. The notch was performed parallel to the SPS pressing axis. In a previous study [30] we assessed R-curve analysis on anisotropic 3YTZP composites with reduced graphene oxide (rGO) as filler in which the notch was performed perpendicular or parallel to the SPS pressing axis, reporting that the highest crack resistance was achieved when the planar crack front was perpendicularly oriented to the rGO main ab plane. Thus, in the present study this notch configuration was selected to perform the bending tests. K IC was measured in a four-point bending configuration, with support outer and inner spans of 21 mm and 10 mm, using a universal testing system INSTRON 8500 machine (Norwood, USA; at MATEIS). Specimens were loaded in air at a constant displacement-controlled rate of 0.5 mm/min and force-deflection curves were recorded. The fracture toughness values were calculated as the stress intensity factor (K I ) for the maximum applied load, F max , as described in [11]. Resistance to stable crack growth (R-curve) was studied both in air and olive oil in order to evaluate the effect of slow crack growth in the composites. R-curve was estimated using the compliance method on a three-point bending device with a support span of 20 mm following the procedure recently validated for similar composites [11,30]. Tests were conducted at a constant displacement rate of 10 μ m/min to achieve slow crack propagation during loading. All tests were unloaded at 100 μ m/min prior to complete failure to ensure no further crack propagation. For each material, the K IC value and K I0 and ΔK I,th parameters were averaged from at least three measurements, and the dispersion of the results was evaluated as the demi-dispersion of the individual values. Olive oil was chosen among other oils because of its lower water content. The impregnation protocol was established following the international standard for density measurements through liquid immersion (ASTM C373–18 [31]) and the methodology presented by Chevalier et al. [32]. The specimen was placed in a desiccator under a vacuum of 10 –2 mbar for 2.5 h, after properly drying it in stove at 105 ◦C for more than 2 h, in order to remove moisture and isolate the crack path from the air humidity. The immersion liquid was also placed inside the desiccator to minimize air in the oil. Finally, the specimen was immersed in olive oil maintaining the vacuum for another 2 h. To guarantee no contact with the environmental humidity, normal pressure was recovered just before the testing. The fracture surfaces and crack paths before failure were analyzed by SEM (Zeiss SUPRA VP55, Switzerland; at MATEIS) to investigate the possible strengthening mechanisms. The stress-induced transformed monoclinic fraction less than 2 µm below the fracture surface was estimated by grazing incidence X-ray diffraction (GIXRD, D8 Advance Bruker AXS diffractometer, Billerica, USA; at MATEIS). A Weibull mirror was coupled to a fixed 0.2◦slit in the incident beam of the diffractometer. Scans were taken in the crack plane after fracture, from 27◦to 33◦, with a step size of 0.02◦and a time per step of 30 s. The V m was calculated from the diffractograms as described in section §2.3. 3. Results and discussion 3.1. Analysis of the BNNS obtained by shear exfoliation in kitchen blender The liquid shear exfoliation of h-BN particles was performed at three different velocities –low, medium and high– of the kitchen blender, as detailed in section §2.1.1. The obtained BNNS suspensions are evaluated in this section to determine the optimal rotor speed in terms of exfoliation degree and production rates. The particle size distribution of the obtained suspensions was jointly evaluated by laser granulometry (Fig. 1a) and dynamic light scattering (Fig. 1b) due to their technical limitations for measuring fine and large particles, respectively. As shown by the BNNS particle size distribution curves, the majority of the nanosheets exhibited particle sizes below 1 µm, regardless of the exfoliation velocity, although a minor proportion were found in the 1–10 µm range. This is in agreement with previously reported size distributions of BNNS obtained by mechanical exfoliation in liquid media, since wide range of nanosheets sizes are usually found, commonly between 100 nm and 1 µm [24,33,34]. The larger particles may emerge from either stacked BNNS or poorly exfoliated h-BN nanoparticles that were not properly removed during centrifugation. Anyhow, these techniques estimate a spherical diameter, thus the numerical values here obtained should only be considered as an estimation of the order of magnitude of the particle size, as previously mentioned in section §2.1.1. When comparing the particle size distribution of the suspensions obtained at the different speeds, it can be established that increasing the kitchen blender velocity produces larger fractions of smaller BNNS. The main peak from the laser granulometry analysis in Fig. 1a (centered at ~ 0.1 µm) intensifies for higher rotor speeds, whereas a second minor band (~ 2 µm) reduces its intensity. Furthermore, higher velocities of the kitchen blender rotor also reduce the size of large particles, as a second peak at ~ 0.7 µm develops simultaneously to the diminution of the 2 µm band. Evaluation of the dynamic light scattering results (Fig. 1b), provides similar conclusions. The BNNS suspension obtained at the lowest speed presents a particle size distribution centered at ~ 0.5 µm. For the other two speeds (i.e. medium and high) the distributions are centered at ~ 0.3 µm, so an increase on the rotor velocity decreases the fraction of larger BNNS. Both techniques show a bimodal distribution of BNNS and both point to smaller nanosheets sizes when using the kitchen blender at its medium and maximum power (speeds 3 and 5). However, dynamic light scattering estimates slightly larger particle sizes than laser granulometry due to the distinct physical phenomena employed for estimating particle sizes in each technique. Fig. 1c illustrates the optical absorption behavior of the exfoliated BNNS suspensions obtained at the three kitchen blender velocities over the near UV and the visible light spectrum. All three spectra show similar profiles, indicating scattering of the BNNS in solution near the UV range, as previously reported [24,34]. The concentration of the BNNS suspensions evaluated by determining the absorbance of the diluted suspensions at 300 nm ranges from 0.07 to 0.22 mg ml -1 at speed 5 to ~ 0.05 mg ml -1 at speeds 1 and 3. The suspended BNNS obtained at the highest rotational speed (5) promote a greater scattering, indicating that the amount of material retained was maximized for this exfoliation condition. C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 4 Even at the higher blender speed, the concentrations of BNNS in suspension are not outstanding, however, the easiness and rapidity of this shear exfoliation technique, together with the large suspension volumes that can be prepared, make it an adequate routine for the production of BNNS for use as fillers in composite materials. Since the highest rotor speed (5) of the kitchen blender increased the BNNS production rate and provided a higher proportion of small-sized nanosheets, velocity 5 of the kitchen blender was selected as the optimal condition for the BNNS production. The structural integrity of the nanosheets was confirmed to be intact by Raman spectroscopy (Fig. 1d). The E 2g band of the exfoliated BNNS is centered at ~ 1370 cm -1 , presenting a blue-shift of ~ 4 cm -1 relative to that of bulk h-BN (E 2g at 1366 cm -1 ) [34–36]. Previous studies indicated that an upshift of 2–4 cm -1 reveals the presence of monolayer h-BN predominantly, in agreement with the phonon mode theoretically found for this 2D material [34,35]. Therefore, the Raman analysis suggests the obtaining of primarily monolayer h-BN in the BNNS suspended after exfoliation with the kitchen blender at speed 5. TEM micrographs of these nanosheets (Fig. 2) allowed estimating the real planar size of the BNNS. Numerous stacked nanosheets are shown in Fig. 2a, illustrating the wide particle size distribution (100–800 nm). Evaluation of more than 30 individual nanosheets revealed that BNNS with sizes between 100 and 400 nm were predominant, with the sporadic observation of nanosheets larger than 1 µm (Fig. 2b). These observations are in agreement with the LG and DLS analyses, which indicated a significant proportion of small nanosheets (sizes below 1 µm). The BNNS lateral sizes after exfoliation in the kitchen blender are significantly lower than those reported for nanosheets obtained by other exfoliation techniques, such as planetary ball milling exfoliation (sizes up to ~3 μ m) [11,33]. The morphology of the BNNS can also be described from these images, which show thin rounded laminates for those with diameter below 400 nm. The larger nanosheets are also thin, as shown by their transparency and flexibility (Fig. 2b) and exhibited a more irregular shape with rounded edges. Finally, no holes or structural damage were observed in the BN nanosheets, and the 6-fold symmetry was corroborated by high-resolution TEM (inset in Fig. 2b). Fig. 1. a), b) Analysis of the particle size distribution of the BNNS suspensions obtained at low (1), medium (3) and high (5) blender rotor velocities by a) laser granulometry and b) dynamic light scattering; c) optical absorbance spectra of the suspensions by UV–vis spectroscopy, and d) Raman spectrum of the BNNS exfoliated at speed 5. Fig. 2. Transmission electron microscopy images of as-exfoliated BNNS. High-resolution TEM image is presented in b) as an inset. C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 5 3.2. Microstructural characterization of the sintered composites Fully dense composites (Table 1) with reduced tetragonal zirconia (ZrO1.95, JCPDS 01–081–1544) as the main crystallographic phase were obtained (XRD patterns not shown). The zirconia matrix exhibited submicrometric grain sizes (0.17–0.32 μ m) regardless the BNNS concentration. All the composites presented a homogeneous distribution of the BNNS throughout the zirconia matrix. As an example, a representative BSE-SEM image from the composite with 2.5 vol. % BNNS is presented in Fig. 3a. Moreover, a marked anisotropy is obtained, with the basal plane of the BN nanostructures (horizontal direction) oriented perpendicular to the SPS pressing axis (vertical direction). As previously discussed [11, 13], this preferential orientation is caused by the synergic effect of the uniaxial pressure exerted during the SPS and the 2D structure morphology of the BNNS. The fracture surface of the composite with 2.5 vol. % BNNS presented in Fig. 3b corroborates the smaller dimension of the BNNS exfoliated in kitchen blender in comparison with those coming from other techniques such as planetary ball milling exfoliation [11,33]. The thin BN nanosheets are located between the zirconia ceramic grains, individually or in small groups, and they maintain the rigid character previously reported for BNNS [11,13,37]. 3.3. Low temperature degradation of the 3YTZP-BNNS composites A sigmoidal increase in the hydrothermal degradation is observed in the ageing curves of the zirconia ceramic and the composites displayed in Fig. 4a, which clearly show that the fraction of monoclinic zirconia, Vm,accelerates more rapidly over ageing time in the zirconia reference ceramic compared to the BNNS composites. In addition to the improvement of the LTD resistance observed for all the composites, since their ageing curves are systematically lower than that of the monolithic zirconia, a more remarkable enhancement with increasing BNNS content was also observed. Considering that the zirconia grain size in the composite with 1 vol. % BNNS is similar to that of the reference 3YTZP (see Table 1), the delay in LTD seems to be related to the BNNS incorporation. Contrary to the expected behavior that zirconia with a coarser grain size would undergo faster LTD degradation [38], the composite with 2.5 vol. % BNNS demonstrates an ageing rate comparable to that of its counterpart with 1 vol. %, notwithstanding its coarsened matrix microstructure. Furthermore, the composites with 2.5 and 5 vol. % BNNS, both with similar coarsened matrices show a significant improvement in LTD resistance compared to the monolithic zirconia, especially at the highest BNNS content. Finally, the monoclinic fraction for 3YTZP (reference) and all composites containing BNNS remains below 10 % following 5 h of exposure to water vapor at 134 ◦C, which makes them suitable for biomedical applications in accordance with ISO 13356 standards [39]. In the present study, the incorporation of BNNS has been shown to have a beneficial intrinsic effect on the retardation of zirconia LTD. No secondary interphases at BNNS/zirconia interfaces are expected based on the exhaustive HR-TEM study previously performed in partly hydroxylated BN nanosheets reinforced zirconia composites [11], exhibiting mostly abrupt interfaces without intermediate phases between the BNNS and the zirconia grains. Thus, the improvement in LTD resistance would not appear to be related to the presence of other secondary phases at the grain boundaries. Nevertheless, the O 2 and H 2 O barrier properties of boron nitride nanosheets have been pointed out to be responsible for the prevention of moisture penetration into different materials [10,11, 33,40]. The effectiveness of boron nitride nanoplatelets in improving the water vapor barrier properties in polymer composites has also been previously demonstrated [33,40]. Regarding a zirconia matrix, the addition of a low quantity of boron nitride nanosheets (1.5 vol. %) demonstrated a significant inhibitory effect on degradation, as observed in accelerated low-temperature tests conducted over a five-hour period [10]. Following these results, Lee et al. [10] proposed a potential solution to the issue of low-temperature degradation taking advantage of the water barrier properties of two-dimensional hexagonal boron nitride fillers, but no further studies were conducted on the evaluation of LTD after different time periods in an autoclave or on the microstructural effects in relation to the presence/absence of BN nanofillers (platelets or sheets) in the grain boundary areas. In the composites analyzed in the present study, the BN nanosheets are distributed throughout the matrix with their ab plane preferentially oriented perpendicular to the pressing axis during SPS, i.e. parallel to the sample in-plane. Thus, any water molecule attempting to reach into the matrix is blocked by the nanosheets, which act as moisture barrier and prevent the propagation of the t-m transformation into the composite bulk in a very effective way, resulting in an improvement of the ageing resistance. As illustrated in Fig. 4b, the best linear regression to plots ln(-ln(1-X) versus ln (t-t 0 ) demonstrate comparable slopes for both the reference ceramic and the composites. Table 1 shows the kinetics parameters n and b in the KJMA equation that are essential for predicting the ageing kinetics of composites under any length of steam exposure at 134 ◦C. Regardless the BNNS content, a similar n value (n≅1.2 ±0.2) is obtained, which is consistent with previous observations for 3YTZP composites containing multilayered graphene [9]. Wei and Gremillard [38] have recently reviewed the literature on n values for zirconia ceramics doped with yttria, which spread over a range from 0.5 to 4. These values were achieved by applying various forms of KJMA equation to fit the t-m transformation curves without considering the induction time in their expressions. Their findings indicate that these values are not affected by grain size, yttrium content in the tetragonal phase, or the fraction of the cubic phase. In line with this review, our previous outcomes also suggest the independence of the nature of the incorporated impermeable nanosheets. The values of the b-parameter slightly decreased with increasing BNNS content in the composites, from b =0.028 ±0.09 h -1 for the reference zirconia value to 0.008 ±0.002 h -1 for the composite with 5 vol. % BNNS, in agreement to the trend of the retarded onset of ageing [38]. As already reported in 3YTZP ceramics, the n values around 1 indicate that the nucleation of the monoclinic phase is the dominant ageing mechanism. 3.4. Assessment of the stable crack propagation in the composites The fracture toughness and R-curve behavior were only evaluated on the composites with 1 and 2.5 vol. % BNNS, since previous studies have shown a decrease in the mechanical performance of zirconia composites when incorporating GBN contents higher than 2.5 vol. % [30,41]. Table 2 summarizes the toughness and the R-curve parameters for the 3YTZP (reference) and each 3YTZP-BNNS composite. The fracture toughness of the composites attained similar values to those of the reference zirconia, i.e. K IC ~ 4.2 MPa⋅m 1/2 (Table 2). The limited fracture toughness of the materials developed in this work is related to the refined zirconia grains produced by the relatively low SPS temperatures needed for densification. This grain refinement limits the Table 1 Relative density ( ρ rel ), zirconia mean grain size (d) and standard deviation (s.d.) and values of the ageing parameters (n and b) of the studied materials, including the r 2 of the KJMA linear plot fitting. BNNS vol % ρ rel (%) d (µm) σ d (µm) n b (h -1 ) r 2 0 100 0.21 0.13 1.35 ± 0.07 0.028 ± 0.009 0.98849 1 99.7 0.17 0.07 1.13 ± 0.03 0.012 ± 0.002 0.99171 2.5 99.0 0.27 0.19 1.32 ± 0.07 0.012 ± 0.004 0.98313 5 100 0.32 0.15 0.93 ± 0.04 0.008 ± 0.002 0.97729 C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 6 stress-induced t-m transformation toughening mechanism. Previous works on several ceramic matrices incorporating either BNNS or BNNP have reported toughness enhancement. Lee et al. [10] measured a 20 % increase in SENB K IC on alumina-TZP composites with the addition of 1 vol. % BNNP. Reinforcing Si 3 N 4 with 2 vol. % BNNP also promoted tougher materials when compared to the monolithic ceramic [37]. Similarly, Sun et al. [42] observed superior toughness on fused silica composites when incorporating up to ~ 5.5 vol. % BNNS. On the contrary, in our previous study [11] we have reported no toughness enhancement for 3YTZP composites with 2.5 vol. % BNNS exfoliated by hydroxide assisted milling. Thus, the intrinsic toughness of 3YTZP composites containing BNNS and sintered by SPS remained unchanged, regardless of the content (1 to 2.5 vol. %) and the manufacturing method (blender exfoliation or hydroxyde-assisted planetary ball milling) of the second phase. Fig. 5 illustrates the R-curves obtained by the compliance method for the two composites (1 and 2.5 vol. %) and the reference zirconia (3YTZP) to evaluate the effect of the BNNS content. Table 2 gathers the R-curve parameters for each material averaged from at least three tests. All the studied materials present almost flat curves, with an increase of the crack growth resistance from the crack-tip toughness to the plateau of up to ΔK IR,th ~ 0.4 MPa⋅m 1/2 (Table 2), behaving as ideally fragile materials. The absence of a rising R-curve for the composites implies that the incorporation of these BN nanosheets as filler does not promote significant increasing resistance to stable crack growth. Among the extrinsic mechanisms that could be taking part in the toughening of zirconia-based composites, the stress induced t-m zirconia transformation and crack bridging by second phase usually induce the greatest reinforcements [1]. The monoclinic content on the fracture surface of the composite containing 2.5 vol. % BNNS after failure of the SEVNB specimen, measured by grazing-incidence XRD, revealed that no transformation occurred in the material (Table 2). Fig. 6 shows SEM micrographs of the fracture surfaces of the composites after the mechanical tests, revealing that the addition of BNNS increased the rugosity of the fracture surface due to crack deflection, the increase in rugosity being more remarkable for the highest BNNS content (Fig. 6d), due to a greater deflection of the crack during its propagation. Indeed, the crack plane rugosity only accounts for the crack deflection during its propagation, a less reinforcing mechanism than phase transformation or bridging [1], and which may be also occurring as a way for the crack to pursue lower-energy paths through weaker interfaces [43,44] Recent studies on 3YTZP composites incorporating high aspect-ratio rGO (nanosheets of reduced graphene oxide) or BNNS have related the enhanced resistance to stable crack growth to the low stacking order and large surface area of the nanosheets [11,30]. On the contrary, the low lateral dimensions of exfoliated graphene nanoplatelets (e-GNP) have been pointed out as the cause of the absence of reinforcement on 3YTZP Fig. 3. SEM micrographs of the composite with 2.5 vol. % BNNS: a) polished cross-section surface and b) fracture surface. Fig. 4. a) Ageing curves of the 3YTZP-BNNS composites plotted with the monolithic 3YTZP zirconia as reference; b) KJMA-plot linear fits of the experimental data for determination of the ageing parameters. Table 2 Mechanical properties of the composites with BNNS and the reference 3YTZP: SEVNB fracture toughness (K IC ), crack tip toughness (K I0 ), R-curve toughening (ΔK IR,th ) and monoclinic volume fraction on fracture surface (V m,f ). BNNS vol. % K IC (MPa⋅m 1/ 2 ) K I0 (MPa⋅m 1/ 2 ) ΔK IR,th (MPa⋅m 1/2 ) V m,f (vol. %) 0 4.2 ±0.1 3.3 ±0.1 0.3 0.8 1 4.2 ±0.2 3.9 ±0.2 0.1 Not measured 2.5 4.2 ±0.2 3.9 ±0.2 0.4 0.4 C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 7 composites [30]. Thus, the lack of enhanced resistance to stable crack growth in the composites studied in the present work is consequence of the small dimensions of the BNNS exfoliated in the kitchen blender (100–800 nm), as described in section §3.1. Although BNNS in this study do not promote rising R-curves, the Rcurve of both composites is above that of the monolithic zirconia, with an increase of about 18 % on the crack-tip toughness. The K I0 increase in the studied composites (~ 3.9 MPa m 1/2 ) with respect to the monolithic zirconia (3.3 MPa m 1/2 ) cannot be related to toughening by grain growth since all the materials present similar grain sizes (Table 1) and K IC values during unstable crack propagation (Table 2). Thus, it is concluded that it is a direct effect of the BNNS incorporation. According to literature, the differences in K I0 with respect to K IC are a sign of stress-assisted corrosion by water in zirconia [11]. Moreover, it has been reported that for a given zirconia material, the crack growth resistance (K I0 ) can change from ~ 3 MPa m 1/2 in water to ~ 5 MPa m 1/2 under vacuum conditions [45]. Thus, resistance to stable crack growth was studied in the present work also on oil-impregnated SEVNB specimens in order to guarantee no contact with the environmental humidity. The R-curve of the composite with 2.5 vol. % BNNS assessed on oil-impregnated specimens revealed an almost identical curve to the analogous in air (Fig. 7). Surprisingly, the isolation of the crack tip from the environmental humidity did not change the resulting curves, which showed equal toughness at the onset of crack propagation than their analogous in air. The shape of the Fig. 5. Crack growth resistance curves for the 3YTZP composites with 1 and 2.5 vol. % BNNS. The reference monolithic zirconia is included for comparison purposes. Fig. 6. SEM images of the fracture surfaces after bending tests at different magnifications of: a) the monolithic 3YTZP zirconia, and the composites with b) 1 vol. % BNNS and c) and d) 2.5 vol. % BNNS. In all these images, the crack propagated from top to bottom, parallel to the SPS pressing axis. C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 8 R-curves also remained unchanged, revealing that the BN nanosheets were not deteriorated by the presence of water. Thus, it can be concluded that, as the crack front advances and the crack tip line encounters the BNNS ab plane, the nanosheets hinder the water molecules from reaching the matrix in the same way that we have mentioned in Section §3.3. Since the BN nanostructures are very finely dispersed, the proportion of these submicrometric-sized nanosheets between the zirconia grains is much higher than that achieved with nitride or carbide particles considering similar vol. % addition [1]. Consequently, the BNNS advantageously reduce the water contact into zirconia grains at the front of cracks, hence, restraining the zirconia corrosion by slow crack growth. 4. Conclusions Boron nitride nanosheets mostly having planar diameters in the range of 100 - 400 nm were successfully exfoliated through an easy and environmentally friendly procedure using a low-cost kitchen blender. The incorporation of BNNS contents up to 5 vol. % in 3YTZP zirconia resulted in almost fully dense composites after SPS sintering of submicrometric zirconia grain size, with a homogeneous anisotropic distribution of the BN nanosheets throughout the matrix. A remarkable improvement of the ageing resistance was achieved in all the 3YTZP-BNNS composites, thanks to the blocking of the water molecules attempting to reach into the matrix by the nanosheets, which act as moisture barrier and prevent the propagation of the spontaneous t-m transformation. The long-term reliability of the zirconia-BNNS composites in terms of ageing degradation was demonstrated. Although the composites presented a fragile performance with fracture toughness values similar to those of the reference zirconia, an increase of ~18 % on the crack-tip toughness was achieved. Almost identical Rcurves were assessed in air and in oil-impregnated specimens for the composite with 2.5 vol. % BNNS, revealing a restrained stress-assisted corrosion by water in zirconia, thanks to the BNNS incorporation. The homogeneous distribution of BNNS at the zirconia grain boundaries represents a promising approach to enhance the hydrothermal ageing stability of 3YTZP ceramics while simultaneously increasing their intrinsic crack resistance. CRediT authorship contribution statement C. Mu˜ noz-Ferreiro: Writing – original draft, Methodology, Investigation. A. Morales-Rodríguez: Writing – review & editing, Writing – original draft, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. H. Reveron: Writing – review & editing, Visualization, Validation, Supervision, Methodology, Investigation, Data curation, Conceptualization. E. Guisado-Arenas: Methodology, Investigation. S. Cottrino: Methodology, Investigation. P. Moreno: Methodology, Investigation. J. Prada-Rodrigo: Methodology, Investigation. J. Chevalier: Writing – review & editing, Visualization, Validation, Supervision, Methodology, Investigation, Data curation, Conceptualization. ´ A. Gallardo-L´ opez: Writing – review & editing, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. R. Poyato: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This research was supported by projects PID2022–140191NB-I00 and PID2020–119003GB-I00 funded by MCIN/AEI/ 10.13039/ 501100011033 and by FEDER, UE. C. Mu˜ noz-Ferreiro acknowledges the financial support of a VI PPIT-US (Plan Propio Universidad de Sevilla, Spain) fellowship through the contract USE-18740-H. The authors are grateful to the JECS Trust for funding the visit of C. Mu˜ noz-Ferreiro to MATEIS Laboratory (INSA Lyon) (Contract No. 2020248). The authors thank Prof. J.S. Bl´ azquez (University of Seville) for his advice and assistance regarding the KJMA model. Fig. 7. Crack growth resistance curves for the composite with 2.5 vol. % BNNS measured in air (continuous line) and in olive oil (symbols). C. Mu˜ noz-Ferreiro et al. Open Ceramics 23 (2025) 100816 9