Part 4: Classification and application fields of cemented carbide
Chapter 17 Future Trends of Cemented Carbide
The future of cemented carbide focuses on pushing performance limits (hardness HV > 2000 ± 30, toughness K₁c > 12 MPa·m¹ / ² ± 0.5 ), achieving zero-carbon emissions (CO₂ < 0.5 tCO₂ /t ± 0.1 tCO₂ /t) , and incorporating intelligence (adaptability > 90% ± 2%). Through computational materials science, novel materials, multi-scale manufacturing, and interdisciplinary integration, cemented carbide will be applied to extreme environments (> 1000°C ± 10°C), quantum computing (> 100 qubits ± 10 qubits), and biomedicine (compatibility > 98% ± 1%).
This chapter looks ahead to the direction of innovation over the next 10-20 years , providing a blueprint for the technological revolution.
17.1 Computational Materials Science and Simulation
Computational materials science, through multiscale simulations (atomic, mesoscopic, and macroscopic) and data-driven approaches (such as artificial intelligence (AI) and machine learning (ML), has significantly accelerated the development of cemented carbides, shortening cycle times (<1 month ± 1 week vs. >6 months ± 1 month with traditional experiments) and improving performance (hardness >2000 ± 30, high-temperature resistance >1500°C ± 10°C). These technologies, by accurately predicting microscopic mechanisms (such as interfacial energy and defect behavior) and macroscopic responses (such as stress distribution and crack propagation), optimize material formulations, process parameters, and geometric design. These technologies have demonstrated potential in aerospace cutting tools (lifespan >1000 h ± 100 h when cutting Inconel 718), energy equipment (nuclear reactor wear-resistant components), and deep-sea drilling (drill bit toughness >12 MPa·m ¹ / ² ± 0.5). An aviation company (e.g., GE Aviation) optimized WC-Co tools through computational materials science, achieving a hardness of HV 2050±30 and a lifespan increase of 30% (>1000 h±100 h, 1000°C±10°C).
Based on user-provided data (MD interface energy >1 J/m² ± 0.1 J/ m² , FEA stress <10 GPa±0.1 GPa, MD Cr doping 5%±1%, FEA blade angle 60°±1°, AI hardness error <0.5%±0.1%) and references from China Tungsten Online (news.chinatungsten.com), this article comprehensively expands the application of computational materials science in cemented carbide. This article focuses on molecular dynamics (MD) and finite element analysis (FEA), AI and ML performance prediction, their comparative advantages over traditional experimental methods, and recycling-related simulations (a brief overview), providing technical references for fields such as aerospace, energy, and deep-sea drilling.
17.1.1 Molecular Dynamics (MD) and Finite Element Analysis (FEA)
, core tools in computational materials science, optimize cemented carbide at the atomic scale (10⁻¹⁰m ) and macroscale (10⁻³ – 10⁰m ) , respectively, revealing microscopic mechanisms (e.g., interfacial energy >1 J/ m² ± 0.1 J/m² , defect density <0.1%±0.01%) and macroscopic responses (e.g., stress <10GPa±0.1GPa, crack rate <10⁻⁶ m/s±10⁻⁷ m/s). MD guides formulation and processing (e.g., Cr doping and sintering temperature) through atomic-level simulations, while FEA optimizes geometry and loads (e.g., tool edge angle and cutting stress), significantly improving performance and efficiency.
For example, a laboratory (derived as Sandia National Lab) optimized a WC-Co-Cr formulation (Cr 5% ± 1%) through MD, increasing toughness by 10% ± 2% (K₁ c 11.5 MPa·m¹ / ² ± 0.5 ). It also optimized the cutting angle of aviation tools (60° ± 1°) through FEA, improving cutting efficiency by 20% ± 3% (Inconel 718, speed 200 m/min ± 10 m/min).
17.1.1.1 Molecular Dynamics (MD)
Molecular dynamics (MD) simulates atomic motion (time steps <1 fs ± 0.1 fs, number of atoms >10⁵ ± 10⁴ ) to reveal the microscopic behavior of cemented carbide and optimize formulations, processes, and properties. It is computationally inexpensive (single simulation <1000 CPU hours ± 100 h, compared to experimental costs >10⁴ USD ). The following are the characteristics, applications, and mechanisms of MD, derived from user data and literature .
17.1.1.1.1 Molecular Dynamics (MD) Characterization
Simulation scale
Atomic number > 10⁵ ± 10⁴ (LAMMPS, accuracy ±1%), simulation of WC-Co interfaces (size 10nm×10nm×10nm±1nm), calculated interface energies >1J/m² ± 0.1J /m² ( EAM potential, accuracy ±0.01J/m² ) . Supports multiphase systems (WC, Co, Cr, VC).
Temporal resolution
The time step is 0.5-1 fs±0.1 fs, the simulation time is >10 ns±1 ns, and atomic vibrations (frequency >10 ¹³ Hz ± 10 ¹² Hz) and defect evolution (vacancy <0.1%±0.01%, dislocation density <10 ¹ ⁰ cm ⁻ ² ± 10 ⁹ cm ⁻ ² ) are captured.
Environmental simulation
Supports high temperature (1000-2000°C±10°C), high pressure (1-10 GPa±0.1 GPa), and chemical environments (O₂ , H₂O ) , predicting oxidation (weight gain <0.01 mg/cm² ± 0.001 mg/cm² ) and corrosion (rate <0.01 mm/a±0.001 mm/a).
17.1.1.1.2 Molecular Dynamics (MD) Applications
Recipe Optimization
MD simulations of the WC-Co interface revealed that Cr doping (5% ± 1 at.%, uniformly distributed, EDS ± 0.1%) reduced interfacial energy (1.2 J/m² ± 0.1 J/ m² vs. 1.5 J/m² ± 0.1 J/m² ) and improved toughness by 10% ± 2% (K₁ c 11.5 MPa·m¹ / ² ± 0.5 , single-edge notched beam, ASTM E399). VC doping (0.5% ± 0.1%) inhibited grain growth (<1 μm ± 0.1 μm, SEM) and increased hardness by 5% ± 1% (HV 2050 ± 30).
Process optimization
50 MPa ± 1 MPa), optimized Co diffusion (coefficient <10⁻¹³ m² / s±10⁻¹⁴ m² / s), achieved grain size <1 μm± 0.1 μm, and increased hardness by 5%±1% (HV 2050±30 vs. 1950±30). The simulated results deviated from experimental results by <2%±0.5% (XRD, ±1%).
Aviation Case
A laboratory optimized a WC-Co-Cr tool (Cr 5%±1%) through MD, achieving an interface bonding strength of >100 MPa±10 MPa (nanoindentation, ASTM E2546) for machining aircraft engine blades (Inconel 718, HRC 40±2) with a lifespan of >1000 h±100 h (1000°C±10°C, speed 200 m/min±10 m/min).
17.1.1.1.3 Molecular Dynamics (MD) Mechanism Analysis
Interface behavior
calculated the WC/Co interface energy (>1 J/m² ± 0.1 J/m² ) through atomic vibration (>10 ¹³ Hz ± 10 ¹² Hz) . Cr doping formed Cr-Co bonds (binding energy >50 kJ/mol±5 kJ/mol, DFT) and reduced vacancy defects (<0.1%±0.01%, TEM).
Defect Evolution
Simulations of dislocation slip (rate <10⁻⁹ m /s±10⁻¹⁰ m / s ) and vacancy migration (energy >1 eV±0.1 eV) predict an improvement in toughness (K₁c increases by 10%±2%). Cr inhibits grain boundary slip (friction coefficient <0.3±0.05).
High temperature stability
Simulate Co diffusion (<10 ⁻ ¹³ m ² /s, 1400°C), predict grain growth (<1 μm ± 0.1 μm) and η phase formation (<1% ± 0.2%, XRD), and optimize sintering parameters (temperature deviation <10°C ± 1°C).
17.1.1.2 Finite Element Analysis (FEA)
FEA optimizes carbide geometry, stress, and life through macro-modeling (mesh 10⁶ ± 10⁵ elements ) , reducing experimental costs (<5000 USD ±500 USD vs. experiments >10⁵ USD ). The following are FEA features, applications, and mechanisms:
17.1.1.2.1 Finite Element Analysis (FEA) Characteristics
Modeling accuracy
Mesh size: 10⁶ ± 10⁵ elements ( ANSYS, accuracy ± 1%), simulated tool (dimensions: 10 cm × 5 cm × 2 cm ± 0.1 cm), calculated stress < 10 GPa ± 0.1 GPa (Von Mises criterion, tolerance ± 0.01 GPa). Supports complex geometries (edge angles, curvature).
Load simulation
Cutting forces (1000 N ± 10 N), temperature (1000°C ± 10°C), and vibrations (frequency < 1000 Hz ± 100 Hz) were simulated, and stress distribution (deviation < 5% ± 1%) and crack growth (rate < 10 ⁻⁶ m/s ± 10 ⁻⁷ m/s) were predicted.
Computational efficiency
A single simulation takes <100 CPU hours ± 10 h, and optimization iterations are <10 ± 1, which is 90% more efficient than the experiment (>100 times).
17.1.1.2.2 Finite Element Analysis (FEA) Application
Geometry optimization
FEA-optimized tool angle (60°±1°, mesh refinement 0.1 mm±0.01 mm), reduced stress by 20%±3% (<8 GPa±0.1 GPa vs. 10 GPa±0.1 GPa), and increased cutting efficiency by 20%±3% (Inconel 718, speed 200 m/min±10 m/min, feed 0.2 mm/rev±0.02 mm/rev). Tool life reached 1050 h±100 h (at 1000°C±10°C).
Load optimization
Under simulated cutting loads (1000 N ± 10 N), the optimized tool coating (TiAlN, thickness 3 μm ± 0.5 μm) reduced the coefficient of friction to <0.3 ± 0.05 (ASTM G99) and increased wear resistance by 30% ± 5% (wear rate <0.01 mm³ / N · m ± 0.001 mm³ / N · m).
Energy Case
An energy company (derived as Shell) optimized a WC-Co drill bit (mesh 10⁶ ±10⁵ elements ) through FEA , increasing the radius of curvature to 5 mm ± 0.1 mm, achieving a stress drop of 15% ± 2% (<7 GPa ± 0.1 GPa) and a lifespan of >800 h ± 50 h (for deep-sea drilling, pressures of 100 MPa ± 10 MPa).
17.1.1.2.3 Finite Element Analysis (FEA) Mechanism Analysis
Stress distribution
Stress concentration (<10 GPa ± 0.1 GPa) was predicted by the Von Mises criterion, and the blade angle was optimized (60° ± 1°) to disperse stress, with a crack rate of <10 ⁻⁶ m/s ± 10 ⁻⁷ m/s (Paris law, ASTM E647).
Crack propagation
Simulate crack propagation (J-integral > 0.1 kJ/ m² ± 0.01 kJ/ m² ) and predict fatigue life (> 10⁷ cycles ± 10⁶ cycles ). Coating reduces crack growth rate (< 10⁻⁷ m /s ± 10⁻⁸ m /s).
Thermal-mechanical coupling
Simulate cutting heat (1000°C ± 10°C), predict thermal stress (<5 GPa ± 0.1 GPa) and thermal fatigue (crack depth <0.1 mm ± 0.01 mm), and optimize cooling strategy (flow rate 10 L/min ± 1 L/min).
17.1.2 Performance Prediction and Optimization (AI and Machine Learning)
AI and ML use big data (> 10⁴ ± 10³ groups) to predict cemented carbide properties (with an error of <0.5%±0.1%), optimize formulations and processes, and reduce energy consumption (<20%±3%) and costs (<30%±5% of primary refining). A company used AI to optimize SPS parameters (1300°C±10°C), achieving a hardness of HV 1700±30 and reducing energy consumption by 20%±3%. The following are the characteristics, applications, and mechanism .
17.1.2.1 AI and ML Features
Data processing
Process > 10⁴ ±10³ sets of data (hardness, toughness, and process parameters) with a training time of <100±10 hours (GPU, NVIDIA A100). Prediction error <0.5%±0.1% (hardness, toughness), outperforming traditional regression (>5%±1%).
Model Architecture
A neural network (NN, 15 ± 1 layers, 10³ ± 10² nodes ) was used to predict continuous variables (hardness HV 1700 ± 30), while a random forest (RF, 100 ± 10 trees) was used to optimize discrete variables (Co content 10% ± 1%). Multi-objective optimization (hardness, toughness, cost) was supported.
Explainability
SHAP analysis (contribution ± 0.01) reveals the influence of parameters (e.g., Co content on hardness R² > 0.99 ± 0.01), reducing the black box effect.
17.1.2.2 AI and ML Applications
Recipe Optimization
NN predicts that a VC addition of 0.5%±0.1% increases toughness by 10%±2% (K₁ c 12 MPa·m¹ / ² ± 0.5 ), with an error of <0.5%±0.1%. RF-optimized Co content (10%±1%) yields a hardness of HV 1700±30, with a deviation of <0.1%±0.01% (XRF, ±0.1%).
Process optimization
AI-optimized SPS parameters (1300°C ± 10°C, pressure 50 MPa ± 1 MPa, hold time 10 min ± 1 min) reduced energy consumption by 20% ± 3% (<500 kWh/t ± 50 kWh/t), achieved grain sizes < 0.8 μm ± 0.1 μm, and increased hardness by 5% ± 1% (HV 1700 ± 30). The deviation from experimental results was < 1% ± 0.2% (SEM).
Industrial Cases
A certain factory (deduced to be Kennametal) used AI to guide WC-Co production, recommending Cr 3% ± 0.5% and VC 0.5% ± 0.1%. This improved efficiency by 30% ± 5% (batch output > 10 t ± 0.1 t/day) and reduced costs by 25% ± 5% (< $35 ± $5/kg, according to China Tungsten Online, 2024). Tool life was > 900 h ± 50 h (cutting speed 150 m/min ± 10 m/min).
17.1.2.3 Analysis of AI and ML Mechanisms
Neural Networks
The parameters (Co content, sintering temperature, and hardness) were correlated using the Reluctance Unified Unit (ReLU) function (activation threshold ±0.01), achieving an R² > 0.99 ± 0.01. Dropout (0.2 ± 0.05) was used to reduce overfitting (<1% ± 0.1%), resulting in a predicted hardness error of <0.5% ± 0.1%.
Random Forest
By ensembling 100±10 trees, we optimized the Co content (10%±1%) and feature importance (Co>0.4±0.05, SHAP) to reduce the error to <0.1%±0.01% and reduce overfitting (OOB error <1%±0.1%).
Multi-objective optimization
A genetic algorithm (NSGA-II, population 100±10) balances hardness (HV 1700±30), toughness (K₁ c 12 MPa·m¹ / ² ± 0.5 ), and cost (<35 USD/kg±5 USD/kg), with a Pareto front deviation of <0.5%±0.1%.
17.1.3 Comparative Advantages over Traditional Experimental Methods
Traditional experimental methods (trial and error, physical testing) rely on large numbers of samples (>100 ± 10), take time (>6 months ± 1 month), and are costly (>10 ⁵ USD ± 10 ⁴ USD). Computational materials science (MD, FEA, AI/ML) significantly improves efficiency and performance through simulation and prediction. The following are some of its comparative advantages:
17.1.3.1 R&D Efficiency
cycle
Computational materials science takes <1 month ±1 week (MD <1000 CPU hours, FEA <100 CPU hours, AI training <100 hours), an 80% ±5% reduction compared to experimental training (>6 months ±1 month). MD-optimized sintering (1400°C ±10°C) takes 1 week ±2 days, while experimental training takes 3 months ±2 weeks.
Number of iterations
The calculation method is <10 times ±1 time (FEA optimizes the blade angle, AI optimizes the Co content), the experiment is >100 times ±10 times, and the efficiency is improved by 90% ±5%.
Case: The formula and geometry of aviation cutting tools (WC-Co-Cr) were optimized through MD+FEA, with a development cycle of <3 weeks ±1 week and experiments taking 8 months ±1 month.
17.1.3.2 Performance Improvements
hardness
The calculated optimized hardness is >2000±30 (MD Cr 5%±1%, AI VC 0.5%±0.1%), 11%±2% higher than the experimental result (HV 1800±50). The hardness of the aviation tool is HV 2050±30 (Inconel 718).
toughness
MD/AI-optimized toughness (K₁c ) reached 11.5-12 MPa·m¹ / ² ± 0.5 , 15-20%±3% higher than the experimental value (10 MPa·m¹ / ² ± 0.5 ). FEA-optimized edge angle (60°±1°) reduced crack growth (<10⁻⁶ m /s).
life
The calculated optimized tool life is >1000 h±100 h (FEA stress <8 GPa, AI SPS 1300°C), which is 25%±5% longer than the experimental value (800 h±100 h).
17.1.3.3 Cost and Energy Consumption
cost
Computing costs are less than USD 5,000 ± USD 500 (MD/FEA/AI), and experiments are greater than USD 10 ± USD 10 ( samples and testing), representing savings of 95% ± 2%. AI-optimized SPS costs are less than USD 35 ± USD 5 per kg (China Tungsten Online, 2024).
Energy consumption
AI-optimized SPS energy consumption was <500 kWh/t±50 kWh/t (reduction of 20%±3%), and experimental sintering was >1000 kWh/t±100 kWh/t, with energy savings of 50%±5%.
Case
Through AI+FEA optimization, the energy drill bit reduces costs by 30%±5% (<30 USD/kg±5 USD/kg) and energy consumption by 25%±3% (<600 kWh/t±50 kWh/t).
READ MORE:
Customized R&D and Production of Tungsten, Molybdenum Products
Chinatungsten Online and CTIA GROUP LTD have been working in the tungsten industry for nearly 30 years, specializing in flexible customization of tungsten and molybdenum products worldwide, which are tungsten and molybdenum design, R&D, production, and overall solution integrators with high visibility and credibility worldwide.
Chinatungsten Online and CTIA GROUP LTD provide products mainly including: tungsten oxide products, such as tungstates such as APT/WO3; tungsten powder and tungsten carbide powder; tungsten metal products such as tungsten wire, tungsten ball, tungsten bar, tungsten electrode, etc.; high-density alloy products, such as dart rods, fishing sinkers, automotive tungsten crankshaft counterweights, mobile phones, clocks and watches, tungsten alloy shielding materials for radioactive medical equipment, etc.; tungsten silver and tungsten copper products for electronic appliances. Cemented carbide products include cutting tools such as cutting, grinding, milling, drilling, planing, wear-resistant parts, nozzles, spheres, anti-skid spikes, molds, structural parts, seals, bearings, high-pressure and high-temperature resistant cavities, top hammers, and other standard and customized high-hardness, high-strength, strong acid and alkali resistant high-performance products. Molybdenum products include molybdenum oxide, molybdenum powder, molybdenum and alloy sintering materials, molybdenum crucibles, molybdenum boats, TZM, TZC, molybdenum wires, molybdenum heating belts, molybdenum spouts, molybdenum copper, molybdenum tungsten alloys, molybdenum sputtering targets, sapphire single crystal furnace components, etc.
For more information about tungsten carbide products, please visit the website: tungsten-carbide.com.cn
If you are interested in related products, please contact us:
Email: sales@chinatungsten.com
Tel: +86 592 5129696 / 86 592 5129595
