Abstract
This study presents a sustainable, performance-oriented approach for developing tribologically efficient polymer composites by reinforcing ultra-high molecular weight polyethylene (UHMWPE) with calcium oxide (CaO) synthesized from waste eggshells. CaO particles were prepared through calcination at 850°C and modified using silane to enhance compatibility with the UHMWPE matrix. Composites containing 0.5–1.5 wt% CaO were fabricated via compression molding and evaluated under varying loads and sliding speeds in simulated body fluid using a central composite design. An artificial neural network (ANN) model accurately predicted wear rate (R2 = 0.9792) and coefficient of friction (COF) (R2 = 0.9007). Multi-objective optimization using the Non-dominated Sorting Genetic Algorithm III (NSGA-III) produced 30 Pareto-optimal solutions, with a hypervolume of 0.78 and generational distances of 2.65 mg/km (wear rate) and 0.014 (COF). The experimentally validated knee point (110 N, 0.2 m/s, 1 wt% CaO) showed less than 5% deviation from predicted results. SEM analysis of worn surfaces revealed dominant wear mechanisms such as micro-ploughing, particle pullout, and delamination. This work demonstrates the potential of combining waste valorization and AI-based optimization to engineer high-performance, eco-friendly composites suitable for biomedical applications.
Keywords
Introduction
The global emphasis on sustainable materials has intensified research into converting biogenic and agricultural waste into high-performance, environmentally conscious materials, particularly in engineering applications. This movement has been especially significant in the development of polymer composites such as ultra-high-molecular-weight polyethylene (UHMWPE) and other thermoplastics reinforced with waste-derived or eco-friendly fillers. Researchers have explored a variety of additive strategies to enhance mechanical, tribological, and structural properties while maintaining or improving sustainability metrics.
In this context, Abdul Samad Mohammed et al. demonstrated that incorporating 10 wt% crumb rubber into UHMWPE coatings resulted in a 94% increase in hardness and an 86% increase in elastic modulus. The composite sustained a 15 N load, surpassing the performance of conventional fillers. 1 Complementarily, Ashish Soni et al. engineered sand-filled LDPE, HDPE, and PET composites, achieving a minimum abrasive wear of 0.02498 cm3 and compressive strengths up to 46.20 N/mm2 under varying operational loads and speeds. 2 Furthering this, Chaobao Wang et al. synthesized PAAm hydrogel microsphere/UHMWPE composites, which exhibited improved tribological behavior via hydration lubrication—a promising route for water-lubricated bearing systems under high load and low-speed conditions. 3
Continuing the exploration of reinforcement mechanisms, Besma Sidia et al. evaluated MS-reinforced UHMWPE biocomposites under hygrothermal aging. Their findings indicated increased wear resistance and elastic modulus with filler content, dependent on MS loading levels. 4 In contrast, T. J. Joyce et al. assessed the role of hyaluronic acid (Ostenil®) in orthopedic applications. While no significant wear reduction was observed in bovine serum, the addition to water reduced wear but resulted in non-biological transfer films. 5 Q. Wang et al. introduced titanium particles into UHMWPE, achieving up to 50% wear reduction in simulated body fluid with 20 wt% loading. However, wear debris size increased beyond 12 wt% inclusion, indicating a trade-off. 6
To further enhance multifunctionality, Sandeep Kumar Pandey et al. developed UHMWPE composites reinforced with MoS2 microcapsules, glass fibers, and Al2O3. These composites exhibited a 45.28% wear rate reduction, a 54.47% decrease in the coefficient of friction (COF), and approximately 40% enhancement in both flexural and tensile strength. 7 In a similar vein, Sakhayana N. Danilova et al. employed waste-derived wollastonite and 2-mercaptobenzothiazole as binary fillers. The result was improved interfacial bonding and superior mechanical and wear performance in UHMWPE composites. 8 Leonardo Roberto da Silva et al. validated the potential of incorporating grinding waste into UHMWPE, observing increased crystallinity (32–45%) and a notable decrease in friction coefficient. 9
The trend of natural fiber and bio-waste utilization was also evident in S. Prashanth et al.’s work on rHDPE composites. They reported peak tensile strength (155.7 MPa) and modulus (6.6 GPa) at 1.5 wt% treated bamboo fiber loading, with reduced wear loss. However, performance declined at higher fiber content due to agglomeration effects. 10 Meanwhile, M. Saravana Kumar et al. investigated Al8011 composites with goat dung ash and Si3N4, resulting in a 56.26% wear rate reduction, 50% grain size refinement, and a 23% increase in brake wear resistance compared to Al6061–10% SiC. 11
Building on wear optimization strategies, Saravanan et al. investigated gamma-irradiated UHMWPE composites, relevant for biomedical implants. These materials exhibited enhanced cross-linking and wear resistance, where adhesion and ploughing emerged as dominant wear modes. 12 Likewise, Chang et al. reported that 10 wt% ZnO-reinforced UHMWPE composites exhibited optimal wear resistance, reduced friction, superior mechanical strength, and notable antibacterial properties against E. coli and S. aureus. 13 Extending to biodegradable waste-derived materials, Tertaraw M. Baye et al. achieved 110.85 MPa tensile strength and 48.57 J impact energy in recycled HDPE composites with 10 wt% medium n-HAp. 14 Venkata Ramana Murty Yerubandi et al. also noted that 10 wt% nano ZnO in UHMWPE reduced wear loss from 5.1 mg to 1.1 mg in wet conditions due to ZnO’s hydrophobic behavior. 15
Recent advances have also leveraged machine learning (ML) and optimization algorithms to predict and enhance composite behavior. Vipin Kumar et al. employed ML models for UHMWPE wear prediction, with Random Forest achieving remarkable accuracy: 0.06 MAE, 0.17 RMSE, and 0.96 R2, reducing reliance on physical testing. 16 Abdul Jawad Mohammed et al. demonstrated that SVM models best predicted UHMWPE-SiC tribology, achieving R2 values of 0.9999 and 0.9998 for COF and specific wear rate, respectively, outperforming four alternative models. 17
To enhance material design, Parijat Srivastava et al. utilized entropy-COCOSO optimization and identified optimal parameters (5% graphene, 200 N load) to attain a COF of 0.04 and a wear rate of 4.87 × 10−4 mm3/Nm in UHMWPE-Al2O3-graphene systems. 18 A. Vinoth et al. integrated artificial neural networks (ANN) and genetic algorithms (GA) to optimize UHMWPE composites for orthopedic implants, predicting enhanced wear and mechanical properties using hybrid reinforcements such as MWCNT, graphene, and carbon fibers.19,20 Similarly, Huihui Feng et al. achieved a low 1.04% prediction error in COF optimization of textured UHMWPE using orthogonal array modeling (OAM), multiple linear regression (MLR), and particle swarm optimization (PSO) from a minimal dataset. 21
Complementing predictive models, A. Borjali et al. developed ML models for PoD test-based wear prediction in polyethylene, enabling accurate forecasting from input parameters while quantifying their influence on wear response. 22 Ahmed Fouly et al. applied adaptive neuro-fuzzy inference systems (ANFIS) to predict HDPE-graphene composite behavior, determining 2 wt% graphene as optimal for mechanical properties and 0.5 wt% for tribological performance, consistent with simulation and experimental validation. 23 Halil Ibrahim Kurt et al. used a feedforward ANN model to assess UHMWPE wear behavior, identifying load and speed as primary influencers and enabling effective design of reinforcement schemes involving CNT, CF, GO, and wollastonite. 24
Closing the discussion with numerical simulation methods, Nishant Verma et al. developed a multiscale finite element (FE) model using a representative volume element (RVE) to predict adhesive wear in nano-hydroxyapatite (nHA)-reinforced UHMWPE, demonstrating strong agreement with experimental data and confirming biocompatibility through cytotoxicity testing. 25 Rajesh Ghosh et al. utilized finite element analysis and GA to optimize acetabular design, concluding that while UHMWPE minimized bone loss, it exhibited higher wear, whereas CFR-PEEK provided an effective balance, emerging as the most promising alternative. 26
Despite extensive research into UHMWPE composites reinforced with various fillers, ranging from metal oxides and carbon-based nanofillers to bio-waste and synthetic additives, the incorporation of waste-derived calcium-based reinforcements, particularly CaO synthesized from eggshells, remains largely unexplored for biomedical tribological applications. Most existing studies focus on improving wear and mechanical properties using conventional or commercially sourced fillers, overlooking the potential of biogenic CaO, which is both abundant and inherently biocompatible. Furthermore, while predictive modeling and optimization techniques such as ANN, SVM, and PSO have been applied to UHMWPE systems, there is a clear lack of studies that systematically integrate ANN-based surrogate modeling with multi-objective optimization (such as NSGA-III) to tailor performance for orthopedic components like tibial inserts. Addressing this gap, the present study introduces a sustainable and functional composite material by synthesizing CaO from waste eggshells and incorporating it into UHMWPE to develop wear-resistant, low-friction composites suitable for load-bearing biomedical applications. The work involves experimental evaluation of tribological properties, surface wear characterization via SEM, ANN-based modeling of COF and wear rate responses, and NSGA-III optimization of process parameters.
Materials and methods
Materials
The primary materials used in this study were ultra-high molecular weight polyethylene (UHMWPE) and calcium oxide (CaO) synthesized from waste eggshells. UHMWPE powder was procured from local supplier in Chennai, India, with a purity of 99.5%. The ultra-high molecular weight polyethylene (UHMWPE) employed in this study has a weight-average molecular weight (Mw) of approximately 4.5 × 106 g/mol, consistent with standard medical-grade UHMWPE ranges typically between 3 × 106 and 6 × 106 g/mol. This high molecular weight ensures superior wear resistance and mechanical durability, making it suitable for tribological applications in biomedical components. Based on the weight-average molecular weight (Mw) of approximately 4.5 × 106 g/mol and the molar mass of the ethylene repeating unit (28.05 g/mol), the average polymer chain length in the UHMWPE used for this study is calculated to be around 160,450 monomer units. This high chain length contributes to the material’s superior entanglement density and excellent resistance to wear and mechanical deformation under tribological loading conditions.
The FTIR spectrum of the as-received UHMWPE powder (Figure 1) reveals distinct vibrational features that allow reliable assessment of its crystalline structure. The prominent –CH2 rocking vibration at 730 cm−1, characteristic of orthorhombic crystalline domains, appears significantly more intense than the adjacent amorphous –CH2 rocking mode at 719 cm−1, indicating a predominantly crystalline morphology. The well-defined peaks at ∼1471 cm−1 (CH2 scissoring) and ∼1376 cm−1 (CH3 symmetric bending) further support the presence of dense chain packing and conformational order in the raw polymer. FTIR results of UHWPE powder.
The additional presence of C–C skeletal stretching at 1063 cm−1 and the absence of carbonyl absorption near 1740 cm−1 confirm minimal oxidative degradation, indicating the preservation of chain integrity during synthesis and storage. Moreover, the sharp absorption bands at 2920 cm−1 (asymmetric CH2 stretching) and 2851 cm−1 (symmetric CH2 stretching) reflect a high degree of molecular orientation with limited amorphous relaxation. Based on the intensity ratio of the crystalline (730 cm−1) to amorphous (719 cm−1) bands, the degree of crystallinity in the UHMWPE powder is estimated at ∼66–68%. This high initial crystallinity establishes a robust structural baseline for subsequent composite fabrication and mechanical property development.
The particle size distribution of the UHMWPE powder used in this study was determined by dynamic light scattering (DLS) and is presented in Figure 2. The distribution exhibits a unimodal profile, with differential intensity peaking at approximately 6.7 µm and a total size span ranging from 5 to 10 µm. Notably, over 90% of the particle population lies below 8 µm, as indicated by the cumulative intensity curve. This narrow particle size range is advantageous for powder blending and uniform dispersion with submicron CaO reinforcements, ensuring a homogeneous composite microstructure. Moreover, the predominance of particles within the 6–8 µm range aligns well with the matrix–filler scale compatibility, promoting effective interfacial contact, minimal agglomeration, and reproducible compression molding behavior. Particle size distribution of UHMWPE powder showing a unimodal profile with over 90% of particles below 8 µm.
Composition of simulated body fluid (SBF) used in tribological testing.
Synthesis CaO from waste eggshell
The synthesis of calcium oxide (CaO) from waste eggshells, as illustrated in Figure 3, exemplifies a sustainable and innovative approach to tackling both environmental and material science challenges. Waste eggshells, a byproduct of the food industry, residence, are often discarded as waste, contributing to landfill burden and environmental pollution. However, they are an abundant source of calcium carbonate (CaCO3), making them an excellent precursor for CaO production. Repurposing eggshells not only reduces waste but also eliminates the need for mining natural limestone, conserving natural resources and reducing carbon emissions associated with its extraction and processing. This process aligns with green manufacturing principles, supports the circular economy by transforming waste into value-added materials, and contributes to the development of eco-friendly composites with enhanced mechanical properties. Through this synthesis, sustainability is integrated at every step, creating a model for reducing waste and supporting sustainable material development. Synthesis of CaO particles from waste eggshells.
The process begins with the collection of waste eggshells from local food sources. These shells are washed thoroughly to remove organic residues and membranes that could contaminate the final product. After cleaning, the shells are dried at a controlled temperature of 80°C for 15 minutes, ensuring complete moisture removal, which is critical for efficient downstream processing. Later, the dried eggshells are manually crushed using a mortar and pestle to reduce their size. This initial size reduction is essential for preparing the material for fine grinding in subsequent stages. The coarse powder obtained is then subjected to ball milling, a mechanical process that ensures the particles are reduced to submicron sizes. Ball milling parameters were carefully optimized to achieve uniform particle size distribution, with a rotation speed of 250 rpm, a ball-to-powder weight ratio of 10:1, and a milling duration of 2 hours. These settings were chosen to maximize energy transfer during milling and enhance the reactivity of the CaCO3 particles during calcination.
The calcining the fine powder was done in a muffle furnace at a temperature of 850°C for 3 hours. During this high-temperature process, the calcium carbonate undergoes thermal decomposition, releasing carbon dioxide gas and converting into calcium oxide according to the reaction represented in equation (1).
This step is critical to obtaining high-purity CaO with optimal physical and chemical properties for use as a reinforcement material. Post-calcination, the powder is cooled naturally in a desiccator to prevent moisture absorption, which could compromise its stability and reactivity. The final stage of the process involves surface functionalization of the CaO particles to improve their compatibility with the UHMWPE matrix. This is achieved using a silane coupling agent, 3-aminopropyltriethoxysilane. The CaO particles are dispersed in an ethanol solution containing the silane agent, and the mixture is ultrasonicated for 30 minutes to ensure uniform coating. Functionalization enhances the interfacial bonding between the CaO particles and the polymer, leading to improved mechanical properties of the composite material.
The adhesion mechanisms between CaO and UHMWPE in the presence of 3-aminopropyltriethoxysilane involve both chemical and physical interactions at the filler-matrix interface as shown in Figure 4. Upon hydrolysis, the ethoxy groups of the silane coupling agent generate silanol (Si–OH) groups, which can condense with hydroxyl groups present on the CaO particle surface to form stable Si–O–Ca bonds. This creates a covalently anchored silane layer on the CaO particles. The terminal amino groups of the silane molecule, although not reactive with the nonpolar UHMWPE chains through direct covalent bonding, facilitate adhesion through secondary interactions. These include hydrogen bonding with micro-polar regions of the UHMWPE crystalline lamellae, van der Waals forces, and enhanced mechanical interlocking due to increased surface roughness imparted by the silane layer. Furthermore, the silane treatment improves the dispersion stability of CaO particles within the UHMWPE melt during blending and compression molding by reducing particle agglomeration and lowering interfacial tension. Together, these mechanisms contribute to a more uniform stress transfer at the interface, resulting in enhanced mechanical integrity and tribological performance of the UHMWPE/CaO composites. Schematic of APTES functionalization of CaO particles and interfacial interaction with UHMWPE.
The synthesis of CaO from waste eggshells demonstrates a practical and sustainable approach to addressing environmental and resource conservation challenges. By integrating technical innovation with sustainability principles, this process contributes to the development of eco-friendly composites with enhanced performance, showcasing the potential of circular economy models in material science.
Characterization of CaO
The synthesized calcium oxide (CaO) particles were comprehensively characterized using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS) to evaluate their morphological architecture, microstructural attributes, and elemental composition. Figure 5(a)–(d) present SEM micrographs at progressive magnifications, providing detailed insights into the hierarchical surface topography of CaO, while Figure 5(e) illustrates the elemental profile via EDS. (a)–(d) SEM images of CaO & (e) EDS results of CaO.
At low magnification in Figure 5(a), the CaO particles exhibit dense agglomerated grains composed of irregularly distributed clusters. These agglomerations arise from the sintering effect during the calcination of CaCO3 at 850°C, leading to thermally fused particulate domains. In Figure 5(b), the morphology transitions into a more defined worm-like architecture, where elongated and entangled CaO structures form a quasi-continuous network. This is accompanied by porous bridges inter-particle connections that likely result from local material flow and diffusion during high-temperature decomposition, enhancing the internal porosity and active surface area.
At higher magnification, Figure 5(c) and (d) reveal nanoscale structural features critical to the functionality of CaO. Nano-scale pores and branched CaO necks in Figure 5(c) represent partially fused regions formed due to incomplete grain separation, likely influenced by the controlled ball milling and thermal treatment synergy. These interconnected necks aid in load transfer and dispersion when embedded in a polymeric matrix. In Figure 5(d), dome-shaped CaO nodules are visible hemispherical surface protrusions potentially arising from localized recrystallization phenomena. Additionally, the presence of nano surface dots, observed as bright spherical features across the surfaces, may correspond to residual CaO nuclei or secondary growth structures formed during calcination cooling cycles.
Detailed morphological examination of the synthesized CaO particles revealed a heterogeneous size distribution, with individual particle dimensions ranging from approximately 35 nm to 591 nm, reflecting the combined effects of mechanical ball milling and thermal decomposition processes. Such a wide size span ensures the coexistence of nanoscale domains that enhance surface reactivity and larger submicron structures capable of bearing mechanical loads within the UHMWPE matrix. Morphologically, the particles exhibit predominantly irregular geometries, including worm-like elongated formations, porous interconnected bridges, and dome-shaped nodules formed due to localized recrystallization during calcination at 850°C. These surface features produce a high degree of texturing and porosity, which are advantageous for mechanical interlocking and interfacial adhesion within polymer composites, mitigating the risk of particle debonding under tribological stress. The presence of such diverse shapes and hierarchical structures is consistent with prior observations for biogenic CaO systems and is integral to the observed improvements in wear resistance and frictional behavior, as it promotes uniform load transfer and enhances the stability of the composite under dynamic sliding conditions.
The EDS spectrum in Figure 5(e) confirms that the synthesized particles predominantly comprise calcium (57.3 wt%) and oxygen (39.7 wt%), consistent with the theoretical stoichiometry of CaO. Minor traces of magnesium (1.2 wt%), silicon (1.0 wt%), and aluminum (0.8 wt%) are also detected, which could stem from inherent impurities in the eggshell precursor or process-induced contamination. Notably, the strong Ca Kα and O Kα peaks indicate complete conversion of CaCO3 to CaO, while the absence of any significant carbon peaks supports the effective thermal decomposition and removal of carbonate species during calcination. Collectively, these structural and compositional features highlight the efficacy of the adopted synthesis route in producing high purity, nanoengineered CaO particles from biogenic waste, tailored for high-performance composite reinforcement applications.
The particle size distribution of the synthesized CaO particles was analyzed using dynamic light scattering (DLS), and the results are presented in Figure 6. The differential intensity profile exhibits a unimodal distribution, peaking at approximately 420 nm, with the overall particle size range extending from ∼340 nm to 530 nm. The cumulative intensity curve demonstrates that over 90% of the particles fall below 500 nm, indicating a relatively narrow distribution and low polydispersity. This confirms that the combined ball milling and calcination approach effectively minimized agglomeration and produced well-dispersed submicron particles suitable for composite reinforcement. Particle size distribution of synthesized CaO particles showing differential and cumulative intensity curves.
Such a particle size distribution is critical in achieving uniform dispersion within the UHMWPE matrix, promoting efficient stress transfer and minimizing localized stress concentrations during sliding contact. Particles in this submicron range offer a favorable balance between surface area and mechanical stability, enhancing interfacial adhesion without compromising matrix toughness. Additionally, the absence of secondary peaks confirms that the synthesized CaO is free from large residual aggregates or unprocessed precursor material.
Based on the features visible in Figure 5(d), which more closely reflect the overall dimensions of individual CaO particles, the estimated major and minor axes range between approximately 533.6 nm–591.1 nm and 342.1 nm–354.2 nm, respectively. These dimensions correspond to aspect ratios in the range of 1.5 to 1.7, indicating that the particles generally exhibit mildly elongated, ellipsoidal geometries. Compared to more spherical or highly anisotropic shapes, this moderate aspect ratio is favorable in composite systems where uniform stress transfer and isotropic reinforcement are desired.
Such aspect ratios may help suppress localized stress concentrations while still offering sufficient anisotropy to enhance mechanical interlocking within the UHMWPE matrix. This balance supports improved stress distribution across the filler–matrix interface during sliding, which is particularly relevant in tribological environments. Moreover, the consistency between the aspect ratio observed in SEM and the particle size range reported by DLS further reinforces the appropriateness of this geometry for stable and effective composite reinforcement.
The XRD pattern of the calcined eggshell powder is presented in Figure 7, clearly demonstrating the successful thermal decomposition of CaCO3 to CaO. The dominant diffraction peak at 2θ ≈ 29.4°, indexed to the (111) plane, along with well-resolved reflections at 32.2° (200), 36.0° (220), 42.6° (222), 47.3° (311), 49.4° (213), 54.0° (204), 58.5° (420), 62.3° (331), and 72.3° (422), are in excellent agreement with the standard face-centered cubic (FCC) phase of calcium oxide (CaO). These peaks confirm the formation of phase-pure, highly crystalline CaO. XRD results of synthesized CaO particles.
Notably, a minor peak at ∼23.0° corresponding to the (104) plane of calcite (CaCO3) is observed, which may be attributed to trace amounts of unconverted CaCO3 or slight re-carbonation of CaO upon exposure to ambient air. However, the absence of other characteristic CaCO3 reflections and the dominance of high-intensity CaO peaks indicate that the overall phase transformation is effectively complete. The sharpness and narrow width of the CaO reflections further confirm high crystallinity and structural integrity of the synthesized filler.
Manufacturing of UHMWPE/eggshell CaO composites
The manufacturing of ultra-high molecular weight polyethylene (UHMWPE)/eggshell calcium oxide (CaO) composites involves a three-stage process as showed in Figure 8. The process was carefully carried out to ensure uniform dispersion of silane-treated CaO particles within the UHMWPE matrix, thereby enhancing the mechanical and functional properties of the resulting composite. In the first stage, composite blend preparation, UHMWPE powder and silane-treated CaO particles were blended in the required proportions as per the experimental plan. Silane-treated CaO, derived from waste eggshells, serves as the reinforcing material. The silane treatment modified the surface of CaO particles, improving their compatibility with the hydrophobic UHMWPE matrix and promoting strong interfacial bonding. The blending process begun with weighing the powders in the desired ratio as per the experimental plan. The powders were then homogenized using a planetary ball mill at 200 rpm for 30 minutes, which ensures uniform particle dispersion, reduces agglomeration, and enhances the composite’s structural consistency. Ball milling also increases the surface area of the CaO particles, further promoting adhesion with the matrix during the molding process. Manufacturing procedure of UHMWPE/eggshell CaO composites.
In the second stage, compression molding, the composite blend was consolidated into a solid structure using a hydraulic compression molding setup. The prepared blend was transferred into a mold cavity, ensuring even distribution to minimize defects during molding. The mold was heated to 160°C, a temperature chosen to soften the UHMWPE matrix without causing thermal degradation, enabling it to flow and encapsulate the CaO particles effectively. The system applied a pressure of 5 MPa through a hydraulic piston, ensuring compaction of the composite blend, elimination of voids, and improved particle-matrix contact. The pressure enhances interfacial adhesion and increases the composite’s density, while a molding time of 15 minutes ensures complete fusion of the UHMWPE matrix and uniform embedding of CaO particles.
The final stage involves finishing the molded composite to achieve the desired structural and surface quality. Following compression molding at 160°C under 5 MPa, the composites were subjected to controlled in-mold cooling at a rate of approximately 5–7°C/min, as regulated by passive air convection in a semi-closed mold environment. This intermediate cooling regime was intentionally selected to suppress abrupt thermal gradients while enabling gradual lamellar rearrangement, thus promoting uniform crystallization and minimizing residual stress buildup.
X-ray diffraction analysis of the fabricated UHMWPE/CaO composites, as shown in Figure 9, reveals distinct and sharp peaks at 2θ ∼ 18.6°, 21.3°, and 23.6°, corresponding to the (100), (110), and (200) crystallographic planes of orthorhombic UHMWPE, respectively. The intensity and sharpness of the (110) and (200) reflections, particularly the dominant peak at 21.3°, confirm a highly ordered lamellar structure and a well-developed crystalline phase. Additionally, the presence of a pronounced (210) reflection at ∼36.1° suggests crystallographic refinement and extended chain alignment due to the synergistic effects of CaO particle nucleation and controlled thermal relaxation. XRD results of UHMWPE/CaO composites.
The diffraction pattern exhibits a low amorphous background with minimal halo intensity, further affirming high crystallinity. Quantitative estimation using the peak area method yields a final degree of crystallinity of approximately 67–70%, which exceeds typical values for compression-molded UHMWPE processed under uncontrolled cooling. This enhancement is attributed to the dual influence of moderate cooling rate and CaO-induced heterogeneous nucleation, which collectively facilitate lamellar thickening and crystalline growth during thermal quenching.
The successful retention of narrow and phase-pure CaO peaks, particularly (111), (200), (220), and (311), confirms that the filler retained its crystallographic integrity during thermal processing and contributed to interfacial templating. The overall XRD signature thus substantiates both the thermal stability of the composite system and the enhanced crystalline architecture, which are critical to the improved tribomechanical response discussed in the subsequent sections.
To confirm the successful incorporation and spatial distribution of CaO particles within the UHMWPE matrix, cross-sectional SEM imaging and EDS analysis were conducted. As shown in Figure 10(a), the low-magnification (500×) SEM micrograph reveals a uniform dispersion of fine CaO particles without observable agglomeration, indicating effective mixing and distribution during compression molding. The high-magnification image in Figure 10(b) further demonstrates well-embedded individual CaO particles with sharp boundaries and no interfacial voids, highlighting strong physical compatibility and interfacial adhesion between the filler and polymer matrix. Complementary EDS point analysis (Figure 10(c)) confirms the elemental composition at a selected location, with significant concentrations of calcium (18.92 wt%) and oxygen (45.76 wt%), characteristic of CaO, alongside carbon (30.43 wt%) from the UHMWPE phase. Minor signals of Na, Mg, and P originate from trace elements in the natural eggshell precursor. The combined morphological and compositional evidence affirms the homogeneous incorporation of CaO particles and supports the structure–property enhancements observed in the composite system. (a) and (b) SEM images (c) EDS results of UHMWPE/CaO composites.
Density of neat UHMWPE and UHMWPE/CaO composites.
Mechanical properties of neat UHMWPE and UHMWPE composites.
However, as typical for ceramic-reinforced polymer composites, a decrease in elongation at break was observed, with ductility declining by roughly 5–15% as filler content increased. This reduction is linked to the presence of rigid filler particles creating localized zones of stress concentration and impeding large-scale chain deformation, thereby diminishing the material’s capacity for plastic flow. Despite this decrease in ductility, the combination of enhanced hardness, tensile strength, and stiffness aligns well with the observed reductions in wear rate and coefficient of friction, substantiating the suitability of these composites for load-bearing biomedical applications where both tribological performance and mechanical integrity are critical.
Design of experiments
Factors and levels in central composite design.
Sliding speed, varying between 0.1 m/s and 0.3 m/s, was chosen for its impact on frictional heating, wear rates, and material interactions. These levels simulate different movement speeds, from slow walking to brisk running, replicating operational conditions for joint replacements. Higher speeds may induce thermal effects and transition wear mechanisms, such as melting or chemical reactivity, making it a critical parameter for analysis. The CaO content (0.5 wt% to 1.5 wt%) was included to evaluate the reinforcement’s effect on wear resistance and mechanical properties. While increased CaO enhances hardness and reduces wear, excessive amounts may cause agglomeration or weak matrix-particle bonding, negatively impacting composite performance.
Design of experiment using CCD.
Testing methods
The testing of UHMWPE/CaO composites for their suitability in biomedical applications, particularly as tibial inserts in knee joint prosthetics, was performed using a pin-on-disc tribometer DUCOM TR-20LE as shown in Figure 11. This testing method is widely employed for evaluating wear and frictional properties under controlled conditions, simulating the interaction between a UHMWPE/CaO composite pin and its counterface disc in physiological environments. The composite samples were prepared according to ASTM G99 standards, which specify guidelines for wear testing using pin-on-disc setups. The samples were cylindrical pins with dimensions of 10 × 10 mm in cross section and 30 mm in height, ensuring a consistent contact area with the disc. Pin-on-disc test setup for wear test of composites.
To replicate the physiological conditions of a knee joint, a simulated body fluid (SBF) like human plasma was applied at the interface during testing to mimic the joint environment. This simulated fluid not only replicates lubrication that could occur in vivo, providing a realistic evaluation of the composite’s performance in biomedical applications. The counterface material was selected as medical-grade Stainless Steel 316 L alloy, which is commonly used in knee joint prosthetics due to its excellent wear resistance, biocompatibility, and low friction properties. The disc was polished to a surface finish of Ra 1 µm, which matches the smoothness required in joint replacements to minimize wear and ensure long-term performance.
The test conditions, including normal load, sliding speed, and environmental parameters, were selected based on a Design of Experiments (DOE) approach to comprehensively evaluate the composite’s performance under simulated human activity. Normal loads were chosen in the range of 50, 100, and 150 N, while sliding speeds were set at 0.1, 0.2, and 0.3 m/s, representing the forces and relative movements experienced during activities such as walking and running. These parameters were carefully selected to capture the biomechanical demands of knee joint prosthetics under realistic conditions, with a sliding distance maintained at 1 km for consistent and thorough evaluation. The normal load is designed to simulate the varying forces exerted on the tibial insert during these activities, reflecting the dynamic stress experienced in real-world conditions. Similarly, the sliding speed is chosen based on the relative movement between the knee components of prosthetic implants during these activities, ensuring the testing closely mimics the actual kinematics of a knee joint. The test was conducted at a temperature of 37°C to replicate body temperature, further enhancing the physiological relevance of the results. The wear rate of the composite was calculated by measuring the mass loss of the pin before and after the test using a high-precision analytical balance. Frictional forces were recorded continuously throughout the test to calculate the coefficient of friction.
Results and discussion
Influence of normal load, sliding speed and Cao content on responses
Pin-on-disc test results of UHMWPE/CaO composites.
In contrast, run 3 (170 N, 0.3 m/s, 0.5 wt% CaO) recorded the highest wear rate (75.74 mg/km) and COF (0.179), indicating a regime dominated by abrasive and fatigue wear mechanisms. The elevated contact stress and frictional heating under such severe tribological conditions likely caused matrix softening, CaO debonding, and third-body wear, diminishing the reinforcement efficacy. Intermediate values observed in Runs 2, 6, 13, and 18 suggest a nuanced balance between reinforcement dispersion and tribo-mechanical stability, where moderate CaO content under controlled sliding conditions enabled the formation of a protective tribo-film that attenuated wear propagation.
Table 6 collectively highlights that wear resistance is maximized when the reinforcing phase is homogeneously distributed and securely bonded within the matrix under mild-to-moderate loading regimes. Overloading or under-reinforcement disrupts this balance, leading to escalated wear and friction. These findings reinforce the necessity of integrated optimization of filler content and operational parameters to exploit the full potential of UHMWPE/CaO composites for biomedical tribological applications.
To contextualize the tribological enhancement imparted by eggshell-derived CaO reinforcement, a direct comparative analysis was performed between the wear and friction performance of the fabricated UHMWPE/CaO composites and that of neat UHMWPE under identical testing conditions. Unreinforced UHMWPE specimens were subjected to pin-on-disc testing under a standard load of 50 N and sliding speed of 0.1 m/s in simulated body fluid (SBF), yielding a baseline wear rate of 28.63 mg/km and an average coefficient of friction (COF) of 0.117.
When compared to the lowest wear configuration among the composites (Run 5: 50 N, 0.1 m/s, 1.5 wt% CaO), the wear rate was reduced from 28.63 mg/km to 1.33 mg/km, representing a remarkable reduction of approximately 95.36%. Likewise, the COF decreased from 0.117 for neat UHMWPE to 0.054 in the composite, corresponding to a reduction of approximately 53.85%. This significant reduction is attributed to the dual contribution of silane-functionalized CaO particles, which improve mechanical interlocking within the UHMWPE matrix and simultaneously promote the formation of a stable tribo-film, effectively mitigating direct polymer-counterface contact and reducing interfacial shear stresses.
Even at moderate reinforcement levels (e.g., 1 wt% CaO in Run 4 under the knee-point conditions), the wear rate declined to 45.12 mg/km relative to neat UHMWPE. Although this represents a lower percentage reduction (∼57.45%) than at maximum CaO loading, it underscores the potential for significant performance improvements without risk of filler agglomeration or matrix embrittlement. Notably, under higher loads and speeds, the presence of CaO reinforcement continued to yield measurable reductions in wear and friction relative to neat UHMWPE, though to a lesser extent due to the increased severity of tribological stresses.
These quantitative comparisons firmly establish the efficacy of eggshell-derived CaO as a reinforcing phase in UHMWPE, delivering substantial tribological benefits crucial for biomedical applications such as tibial inserts, where wear debris minimization and friction reduction are essential for implant longevity and biocompatibility. The magnitude of these improvements confirms that sustainable, bio-derived fillers can achieve performance levels comparable to, or exceeding, those of conventional reinforcements while simultaneously contributing to environmental sustainability objectives.
Beyond tribological evaluation, microhardness measurements were performed to assess the fundamental mechanical response of the composites to localized deformation. Vickers microhardness was measured according to ASTM E384-17 under a load of 50 g and a dwell time of 10 s. The hardness test was performed in 3 different locations of the sample, and the average was considered. The neat UHMWPE exhibited an average hardness of 4.2 HV, consistent with its highly ductile molecular structure and low crystalline modulus. Incorporation of CaO particles led to a progressive increase in hardness, with measured values rising to 5.1 HV, 6.4 HV, and 6.7 HV for composites containing 0.5 wt%, 1.0 wt%, and 1.5 wt% CaO, respectively.
This incremental enhancement in hardness aligns well with the tribological improvements observed in the composite systems. The inclusion of CaO particles, characterized by high stiffness and hierarchical surface features, restricts molecular chain mobility in the UHMWPE matrix and promotes effective load transfer under indentation. The porous and irregular morphology of the CaO further contributes to mechanical interlocking, enhancing resistance to plastic deformation. This mechanical stiffening effect corresponds directly to the reduced wear rates and lower coefficient of friction reported in the tribological tests, as the harder composite surface is better able to withstand micro-cutting and ploughing actions during sliding.
Nonetheless, while increased hardness improves surface durability, excessive filler content may elevate stiffness to a level where the composite loses some of the intrinsic toughness and ductility crucial for load-bearing biomedical applications. The observed data suggest an optimal balance in the range of 1.0–1.5 wt% CaO, where significant hardness enhancement is achieved without compromising the mechanical resilience required for functional performance in demanding physiological environments. Thus, the hardness measurements substantiate the tribological findings and confirm the beneficial role of CaO reinforcement in elevating both surface integrity and overall composite durability.
Artificial neural network (ANN) model
The artificial neural network (ANN) model developed in this study provides a predictive framework for evaluating the tribological performance of UHMWPE composites reinforced with eggshell-derived CaO. Designed to capture the nonlinear interactions between normal load, sliding speed, and CaO content, the model consists of an input layer with three neurons, a hidden layer with 20 neurons employing the ReLU activation function, and an output layer predicting wear rate and coefficient of friction. The choice of architecture was guided by the complex, multivariate nature of tribological behavior, where synergistic effects between input parameters influence interfacial stress, heat generation, and material removal mechanisms. As illustrated in Figure 12, training was conducted using the Adam optimizer over 500 epochs, with mean squared error (MSE) as the loss metric and an 80:20 train/test split to ensure generalization and avoid overfitting. This modeling approach enhances the interpretability and prediction accuracy of composite behavior, supporting data-driven design optimization for high-performance biomedical applications. Architecture of artificial neural network model.
Performance metrics of ANN model.
The performance of the ANN model in predicting tribological responses is visually validated in Figure 13 through parity plots and training-validation loss curves. The parity plots for wear rate (Figure 13(a))) and coefficient of friction (Figure 13(b))) demonstrate excellent predictive agreement, with R2 values of 0.9792 and 0.9007, respectively, confirming the model’s high fidelity in learning the input-output mapping governed by normal load, sliding speed, and CaO content. The loss curves in Figure 13(c) and (d) exhibit rapid convergence and minimal divergence between training and validation losses, indicative of an optimal model with negligible overfitting. The use of second-degree polynomial features and standard scaling, coupled with L2 regularization and the Adam optimizer, facilitated effective generalization. These results validate the model’s capacity to accurately estimate wear behavior in UHMWPE/CaO composites, where nonlinear tribo-mechanical interactions are influenced by filler dispersion, matrix reinforcement synergy, and process-induced stress localization. Parity plot (a) wear rate; (b) coefficient of friction; Training versus validation loss curves (c) wear rate; (d) coefficient of friction.
The correlation matrix and SHAP-based feature importance analysis, shown in Figure 14(a) and (b), offer critical insights into parameter influence within the UHMWPE/CaO composite system. SHAP analysis is a machine learning method that explains how much each input factor contributes to the model’s predictions, helping interpret the influence of variables such as load, speed, or filler content on wear and friction outcomes. CaO content exhibited the highest predictive relevance for wear rate, supported by its negative correlation with wear (r = −0.46), reflecting its reinforcing effect through load transfer and microstructural stability. Normal load showed strong positive correlation with both wear rate (r = 0.66) and COF (r = 0.79), indicating its dominant role in stress-induced degradation. Speed had a moderate but consistent influence across both outputs. Error histograms (Figure 14(c) and (d)) revealed tightly clustered residuals near zero, confirming model stability and low bias across test samples. These patterns affirm the ANN model’s precision and reinforce that tribological performance is chiefly governed by optimized CaO dispersion and applied contact mechanics. The data-driven interpretation enhances the mechanistic understanding of particle–matrix interactions in these bio-sourced composite systems. (a) Correlation heatmap; (b) feature importance plot; error histogram (c) wear rate; (d) coefficient of friction.
Permutation importance results presented in Figure 15(a) indicate that normal load is the most influential factor governing wear rate, followed by sliding speed and CaO content. This hierarchy reflects the predominance of contact-induced stresses in initiating wear mechanisms in UHMWPE/CaO systems. In contrast, Figure 15(b) highlights sliding speed as the most significant contributor to variations in coefficient of friction (COF), with CaO content and load showing secondary but non-negligible influence, emphasizing the role of interfacial shear dynamics and thermal effects. The learning behavior in Figure 15(c) shows that while training R2 remains consistently high, the validation R2 improves steadily as training data size increases, suggesting effective model generalization with added data. Similarly, Figure 15(d) exhibits early convergence of training and validation R2 for COF, stabilizing after a critical sample size is reached. These results collectively confirm that the ANN model, supported by L2 regularization and polynomial-transformed inputs, captures essential nonlinearities inherent to tribological interactions in this bio-reinforced composite system. Permutation feature importance plot (a) wear rate; (b) coefficient of friction; learning curve of (c) wear rate; (d) coefficient of friction.
Non-dominated sorting genetic algorithm III (NSGA-III) optimization
Multi-objective optimization using Non-dominated Sorting Genetic Algorithm III (NSGA-III) was implemented to identify non-dominated solutions that minimize both wear rate and coefficient of friction (COF) for UHMWPE/CaO composites. Given the inherent trade-offs between these responses, where conditions favorable for one may degrade the other, NSGA-III provides a robust evolutionary framework tailored for handling such conflicting objectives with improved convergence and diversity preservation. Unlike its predecessors, NSGA-III introduces reference-point-based selection and enhanced niche preservation, which are particularly valuable for this study’s high-dimensional design space defined by normal load, sliding speed, and CaO content.
The procedure, depicted in Figure 16, begins with the integration of CCD experimental results to train surrogate models for wear rate and COF. These surrogates, developed from polynomial-transformed features and validated ANN models, allow rapid fitness evaluations without additional physical experimentation. Following model training, NSGA-III parameters were initialized with a population size (μ) of 100, offspring size (λ) of 100, and 50 generations (G). The algorithm then generates an initial random population within bounded decision spaces: normal load (50–170 N), sliding speed (0.1–0.3 m/s), and CaO content (0.5–1.5 wt%). Procedure flowchart of NSGA-III algorithm.
Each generation involves evaluating the population and offspring using the trained surrogate models, with selection based on reference-point sorting and crowding distance. Performance metrics, hypervolume, generational distance (GD), spread, spacing, and epsilon indicators, are computed to assess solution diversity and proximity to the Pareto front. Spread indicates how well solutions cover the range of objectives in optimization. Spacing reflects how evenly distributed solutions are on the Pareto front. Epsilon indicator (ε) measures how close solutions are to dominating a reference front. Pareto front represents a set of solutions in which improving one objective would worsen another: useful for choosing trade-offs in multi-objective problems like minimizing wear and friction simultaneously. After 50 iterations, the model yields 30 Pareto-optimal configurations, offering a spectrum of process conditions that balance tribological performance. This data-driven approach enables precision design of bio-reinforced composites under complex operational constraints.
Performance metrics of NSGA-III optimization.
Hypervolume is a metric used in multi-objective optimization to measure how much of the objective space is covered by the optimal solutions; higher values indicate better solutions. A hypervolume (HV) of 0.78, computed with respect to a reference point of 80 mg/km (wear rate) and 0.20 (COF), indicates substantial coverage of the optimal search space. Generational distance (GD) indicates how close the optimization solutions are to the ideal (true) Pareto front; lower values mean better accuracy. The generational distance (GD) values, 2.65 mg/km for wear rate and 0.014 for COF demonstrate proximity to the true Pareto front. The moderate spread (Δ = 0.27) and low spacing (S = 0.014) confirm solution uniformity and effective diversity preservation. A low epsilon indicator (ε = 0.05) further signifies that only a minimal shift is required to weakly dominate the approximated front, validating convergence. Collectively, these metrics underscore the robustness of NSGA-III in guiding optimal parameter selection for enhanced tribological behavior in sustainable composite systems.
Pareto optimal solutions from NSGA-III.
Comparison of the Pareto front with CCD-based experimental solutions in Figure 17(a) demonstrates that the NSGA-III optimization significantly outperformed traditional design points in both wear rate and average coefficient of friction (COF). The clustering of Pareto-optimal points along the lower-left corner of the trade-off surface confirms the superiority of optimized configurations derived from the surrogate-driven evolutionary model. Convergence behavior, as shown in Figure 17(b), reveals a rapid increase in hypervolume across early generations, stabilizing near 0.78 by generation 45. This indicates successful exploration and exploitation of the objective space, maximizing solution coverage relative to the defined reference point. (a) Pareto front versus CCD solutions; (b) Hypervolume convergence curve; (c) Generational distance and spread of NSGA-III; (d) Spacing metric distribution of NSGA-III.
Stability in generational distance (GD) and spread (Δ), as depicted in Figure 17(c), further validates convergence toward a well-formed Pareto front. GD values remained consistently low across iterations, while the spread metric preserved solution diversity, avoiding premature convergence. The spacing distribution in Figure 17(d) shows a tight clustering around a mean value of ∼0.015, affirming uniform distribution of non-dominated solutions without excessive redundancy. Collectively, these results underscore the effectiveness of the NSGA-III framework in capturing nonlinear and multi-objective dependencies between tribological outputs and input parameters, normal load, sliding speed, and CaO content, for engineering high-performance, waste-derived UHMWPE composite systems.
Knee point of Pareto optimal solutions.
The knee point reported in Table 10 (normal load 110 N, sliding speed 0.2 m/s, and 1 wt% CaO) represents an optimal balance between minimizing wear rate and coefficient of friction (COF) under moderate tribological conditions. However, this knee-point condition is inherently influenced by the operational regime, and the optimal parameters may shift depending on the application’s load and speed requirements. Analysis of the NSGA-III Pareto front suggests that at higher loads beyond 110 N, wear rate and COF tend to increase due to elevated contact stresses and localized thermal effects, which can promote mechanisms such as plastic deformation, filler-matrix debonding, and micro-ploughing. Under such intensified conditions, a slightly higher CaO content, such as 1.5 wt%, may become advantageous, as it provides increased surface hardness and distributes mechanical loads more effectively within the composite matrix.
Conversely, at lower loads, for example around 50 N, or at reduced sliding speeds like 0.1 m/s, the wear rate and COF generally decrease because of lower contact stresses and reduced frictional heating. Under these milder conditions, the data indicate that lower CaO contents (0.5–1.0 wt%) are sufficient to maintain excellent tribological performance, as the polymer matrix experiences less deformation and the risk of filler pullout or localized surface damage diminishes. The interplay between filler content, load, and speed is thus critical, enabling tailored material formulations depending on the specific biomechanical demands of the intended application.
These insights allow for practical translation of the optimization results to various biomedical contexts. For instance, in knee prostheses, different patient activities—from slow walking to high-impact movements, involve distinct combinations of load and sliding speed. Lower-load, low-speed conditions would permit the composite to be formulated with a reduced filler content, preserving ductility and toughness, while higher-load, higher-speed applications would benefit from elevated CaO levels to reinforce the surface and enhance wear resistance. It should be recognized that material formulation for biomedical applications is inherently multifactorial, and although this study focuses on normal load, sliding speed, and CaO content, other factors such as counterface material, lubrication environment, long-term fatigue performance, and biocompatibility remain outside the current experimental scope. Nonetheless, within the parameters studied here, the NSGA-III optimization provides a valuable framework for guiding the development of UHMWPE composites tailored to diverse mechanical and physiological requirements in biomedical devices.
Validation of optimized parameters and wear mechanism
Results of predicted and experimental values with variation.
Worn surface analysis of the UHMWPE composite reinforced with 1 wt% waste-derived CaO, tested under optimized conditions (110 N, 0.2 m/s), reveals critical tribological features that elucidate the operative wear mechanisms, as shown in Figure 18(a)–(f). The image in Figure 18(a) exhibits delaminated debris and micro-ploughing grooves, which are indicative of mild abrasive wear resulting from interfacial shearing at asperity contacts. Under the applied load, stress concentrations at filler-rich zones initiate subsurface cracking, eventually leading to delamination and the formation of loose wear fragments. The orientation and depth of plough lines suggest that although UHMWPE’s ductility accommodates shear strain, localized reinforcement-induced stiffness gradients amplify wear track roughness. SEM images of worn out surfaces of samples undergone for validation test.
Figure 18(b) highlights the formation of a transfer film along with visible CaO particle pullout, pointing to partial interfacial debonding despite silane surface treatment. This behavior is attributable to repeated tangential forces at the interface, which, over time, overcome the interfacial adhesion strength. While the transfer film may aid in friction stabilization by forming a protective layer, the presence of particle detachment signifies fatigue-induced weakening of the matrix-filler interface, a phenomenon exacerbated by cyclic micro-sliding.
Plastic deformation zones and mild surface cracks in Figure 18(c) further validate viscoelastic behavior under the test conditions. The plastic smear across the surface confirms that UHMWPE accommodates localized stress without catastrophic failure, preserving structural continuity. In contrast, Figure 18(d) displays deep furrows and debris clusters, evidence of third-body abrasion where dislodged wear particles re-enter the contact zone and intensify wear through gouging and ploughing.
The compressed flakes and plough ridges in Figure 18(e) point to severe plastic flow and material pile-up under directional sliding, where entrapped particles exacerbate surface roughening. Figure 18(f) reveals fractured ridges and pluck zones, typically formed due to localized tensile failure at filler-matrix interfaces. These features confirm microstructural damage resulting from mismatch in mechanical properties and stress concentrations during repeated sliding. The wear mechanisms observed—abrasion, delamination, particle pullout, and surface fatigue are strongly influenced by the synergistic interaction between applied loading parameters and the optimized CaO content. These microstructural insights reinforce the effectiveness of the optimized configuration in achieving durable and predictable tribological performance in UHMWPE/CaO composites, even under moderate operating stresses.
The worn surface morphologies obtained under the optimized knee-point parameters (110 N, 0.2 m/s, 1 wt% CaO) exhibit distinct features that can be directly correlated to the applied tribo-loading conditions and filler–matrix interactions. In Figure 18(a), the presence of continuous but shallow plough lines indicates that the applied Hertzian contact pressure was sufficient to induce localized plastic shear without causing severe subsurface rupture of the UHMWPE lamellae. This is consistent with a regime where the matrix retains its structural integrity while accommodating interfacial shear through viscoelastic flow, aided by the uniform dispersion of submicron CaO particles that act as load-bearing micro-asperities.
Delamination observed in the form of thin, plate-like debris in Figure 18(a) and (d) originates from subsurface crack initiation at filler–matrix interfaces where repeated tangential stresses concentrate. The extent of delamination under these conditions remains limited, suggesting that the silane-mediated interfacial adhesion between CaO and UHMWPE effectively delays crack propagation. Instances of particle pullout, visible in Figure 18(b), correspond to localized interfacial decohesion, where cyclic loading eventually exceeds the adhesion strength; however, the frequency and size of pullout sites are minimal, indicating a high degree of mechanical interlocking.
Areas showing compacted wear debris and smeared transfer film in Figure 18(b) and (e) demonstrate that the tribo-contact facilitated partial re-deposition of detached material, forming a protective layer that stabilizes the coefficient of friction. The simultaneous observation of micro-ploughing (Figure 18(a)), constrained delamination (Figure 18(d)), and low-density pullout (Figure 18(b)) confirms that, at the selected process parameters, wear proceeds through a mixed mechanism in which abrasive and adhesive contributions are moderated by the reinforcing phase. This direct microstructural evidence substantiates that the optimized loading and speed conditions balance interfacial stability with controlled material removal, thereby achieving the wear and COF values predicted by the ANN–NSGA-III optimization.
Comparative benchmarking of tribological performance with reported UHMWPE composites
Benchmarking of tribological performance of UHMWPE composites from literature.
Hydrogel-filled UHMWPE 3 under water lubrication at substantially lower loads (10 N) and moderate speeds (0.03–0.12 m/s) exhibited a markedly lower COF range (0.04–0.14) and wear coefficients on the order of 10−4 mm3/N·m, reflecting the low Hertzian stresses and effective hydration lubrication afforded by the polymeric hydrogel phase. In contrast, the present work’s values are obtained under an order-of-magnitude higher load, with SBF lubrication where ionic content, protein analogues, and biofouling tendencies introduce boundary lubrication effects, thereby amplifying adhesive and ploughing contributions to wear.
When compared with high-load simulator studies such as UHMWPE + Ti, 6 which at 784 N in SBF-9 achieved volumetric wear rates as low as 4.98 × 10−7 mm3/N·cycle, the present work operates under intermediate load severity and sliding velocity. The Ti-filled systems benefit from metallic particle stiffness and load distribution under conformal contact in oscillatory motion, whereas the present CaO-filled system is optimized for pin-on-disc-like articulation with more severe shear components, making its retention of COF in the ∼0.11 range under SBF notable.
Dry-sliding literature systems, such as wollastonite-filled UHMWPE 8 and SiC-filled UHMWPE, 17 report wear coefficients in the 10−6–10−5 mm3/N·m range with higher COFs (0.17–0.42), despite operation at lower nominal contact pressures or loads than the present study. The disparity underscores the combined role of lubrication chemistry and particle–matrix interfacial compatibility; the hydrophilic, bioactive CaO phase here promotes a stable interface in SBF, whereas inert ceramic phases in dry contact may exacerbate third-body abrasion and increase COF.
Biofunctional hybrid systems such as UHMWPE + Al2O3 + HA + chitosan 12 achieve extremely low COFs (down to 0.02) but at very low loads (2–6 N) and under dry sliding, where reduced contact severity and polymeric lubricity from chitosan dominate. The present work’s ability to sustain low wear rates and moderate COF under realistic bio-lubrication and high load offers a more clinically representative measure of long-term in vivo performance.
The benchmarking confirms that although the present CaO-reinforced UHMWPE operates under harsher tribo-mechanical conditions than many literature-reported composites, it maintains wear and COF values competitive with or superior to systems tested under milder regimes. The unique combination of bio-derived filler morphology, silane-mediated interfacial bonding, and ion-rich lubrication environment facilitates a mixed wear mechanism with controlled ploughing and limited delamination, positioning the material as a high-performance, sustainable alternative for load-bearing biomedical applications.
Biomedical implications of composite wear
The long-term suitability of the developed UHMWPE/CaO composites for tibial insert applications is intrinsically linked to their in vivo biocompatibility and the biological response to any generated wear debris. The CaO reinforcement derived from waste eggshells possesses an inherently bioactive and bioresorbable chemistry, which, under physiological conditions, can hydrolyze to Ca(OH)2 and gradually release Ca2+ ions. This ion release is beneficial within controlled limits, as Ca2+ can contribute to periprosthetic mineralization and support osteoblastic activity without provoking adverse inflammatory cascades. The silane functionalization applied in this study not only enhanced interfacial adhesion and load transfer in the tribo-mechanical regime but also minimized the likelihood of early particle detachment, thereby reducing the burden of free particulates in the joint space.
Wear debris from UHMWPE is widely recognized as a potential initiator of macrophage-mediated osteolysis, with submicron particles (<1 µm) posing the highest risk due to their cellular uptake efficiency. The present composite system demonstrated a significant reduction in wear rate, indicating a proportionally lower volumetric release of UHMWPE-derived debris over the expected service life of an implant. Furthermore, any detached CaO particulates, owing to their submicron size distribution (∼340–530 nm) and moderate aspect ratio (1.5–1.7), are likely to undergo gradual dissolution in synovial fluid, transforming into calcium phosphate-like phases via interaction with phosphate ions in the joint environment. This dissolution pathway mitigates long-term particle persistence and, unlike inert ceramic debris, offers potential osteoconductive benefits.
From a mechanical standpoint, the increased hardness (up to ∼60%) and tensile strength (∼21%) achieved in the composites contribute to maintaining articular surface geometry under cyclic load, delaying the onset of pitting, delamination, and fatigue cracking that could otherwise exacerbate debris generation. The sustained low COF (∼0.115 at the optimized knee-point condition) under simulated body fluid lubrication further indicates that the composite can maintain boundary-lubricated contact with reduced interfacial shear, which correlates with lower rates of matrix micro-fracturing and subsurface crack nucleation. These attributes are critical for maintaining joint kinematics and preventing accelerated wear in vivo.
In combination, the tribological improvements, bioresorbable nature of the inorganic phase, and absence of toxic metallic or ceramic inclusions present a compelling case for the long-term biocompatibility of the composite in tibial insert applications. The observed material behavior under SBF lubrication at physiologically relevant loads suggests that the system can significantly reduce biologically active debris generation and ensure that any inorganic fraction present in the debris is benign and potentially beneficial to the peri-implant environment. Extended joint simulator testing and standardized cytotoxicity assays would be the logical next steps to confirm these inferences over multi-million-cycle wear regimes.
Conclusion
This research presents the sustainable development, predictive modeling, and multi-objective optimization of ultra-high molecular weight polyethylene (UHMWPE) composites reinforced with calcium oxide (CaO) synthesized from waste eggshells. The study successfully integrates green material processing with advanced data-driven techniques to evaluate and enhance the tribological performance of bio-derived composites under simulated physiological conditions. A comprehensive experimental design was formulated using Central Composite Design (CCD), followed by artificial neural network (ANN) modeling and NSGA-III-based multi-objective optimization to balance wear rate and coefficient of friction (COF). The optimal solution was experimentally validated and further supported through SEM-based mechanistic analysis. • Synthesized CaO particles exhibited hierarchical porosity and nanoscale features (35–591 nm) that enhanced mechanical interlocking within the UHMWPE matrix. • The lowest wear rate (1.33 mg/km) and COF (0.054) were achieved at 50 N, 0.1 m/s, and 1.5 wt% CaO, attributed to strong interfacial bonding and minimal matrix degradation. • ANN models achieved R2 values of 0.9792 (wear rate) and 0.9007 (COF), confirming high predictive reliability using polynomial-transformed features and L2 regularization. • Permutation feature importance and SHAP analysis confirmed that wear rate was most influenced by normal load, followed by CaO content, while sliding speed primarily impacted COF. • NSGA-III yielded 30 Pareto-optimal solutions with a hypervolume of 0.78 and low generational distance (WR: 2.65 mg/km; COF: 0.014), highlighting convergence and diversity. • The experimentally validated knee point (110 N, 0.2 m/s, 1 wt% CaO) exhibited deviations <5% from predicted values, confirming the model’s real-world reliability. • SEM analysis of validated samples revealed wear mechanisms such as micro-ploughing, delamination, and third-body abrasion, linked directly to the optimized tribo-parameters.
The current framework can be extended to study fatigue and corrosion behavior of UHMWPE/CaO composites and adapted for other waste-derived reinforcements targeting broader biomedical and structural applications.
Footnotes
Author contributions
Hariharasakthisudhan Ponnarengan: Conceptualization, Methodology, Supervision, Writing – Original Draft, Project Administration, Funding Acquisition. Sathickbasha Katharbasha: Data Curation, Investigation, Software, Validation, Formal Analysis. Logesh Kamaraj: Resources, Visualization, Writing – Review & Editing. Sathish Kannan: Software, Methodology, Validation, Writing – Review & Editing.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Ethical approval
There are no human participants in this article and informed consent is not required.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.