Direct observations of phase boundaries, invariant reactions and phase compositions. They are essential but may be sparse, internally inconsistent or inaccessible at extreme conditions.
CALCULATION OF PHASE DIAGRAMS · THERMOCHEMISTRY · DATABASE ASSESSMENT
CALPHAD as a method of computational thermodynamics
CALPHAD represents the Gibbs energy of every relevant phase with physically structured models, evaluates those models against heterogeneous evidence, and computes equilibrium by constrained Gibbs-energy minimisation. The phase diagram is an output of this framework—not its only purpose.
Originally “CALculation of PHAse Diagrams”; now a broader framework coupling phase equilibria, thermochemistry, physical models and computational minimisation.
01
WHY THE METHOD IS NEEDED
Experiments constrain a phase diagram; they rarely define the entire thermodynamic system.
Relative energies, defect energetics and vibrational contributions can constrain phases for which measurements are incomplete. Results depend on the electronic-structure and statistical-mechanical approximations used.
All defensible evidence is evaluated together to obtain a compact, self-consistent description of Gibbs energy as a function of temperature, pressure and composition.
Input
Thermochemical and phase-equilibrium data, crystal structure, first-principles results, uncertainty and expert judgement.
Internal representation
Reference states, endmembers, sublattices, configurational entropy and interaction parameters for each phase.
Output
Phase stability, phase fractions and compositions, driving forces, activities, chemical potentials and other modelled properties.
02
GIBBS-ENERGY REPRESENTATION
The model is a set of phase functions, not a digitised diagram.
Gmφ = ΣxiGi0,φ + RTΣxilnxi + Gmex,φ + Gmphys,φ
Reference term describes pure components or endmembers; ideal mixing follows configurational statistics; excess Gibbs energy represents non-ideal interactions; physical contributions may include magnetic, ordering, pressure or defect terms.
Binary regular-solution example
Adjust Ω and temperature to examine how the curvature of ΔGmix changes. This is an instructional model, not an assessed alloy system.
Model critical temperature962 K
At the selected point, the system is inside the spinodal region.
ΔGmix = RT[x ln x + (1−x) ln(1−x)] + Ωx(1−x)● selected composition
For Ω > 0, sufficiently low temperature can produce a double-well free-energy curve and a miscibility gap. Equilibrium phase compositions are found through chemical-potential equality, represented geometrically by a common tangent.
03
NICKEL-BASED SUPERALLOY CASE STUDY
CALPHAD connects alloy chemistry to the phase constitution that controls high-temperature performance.
Disordered matrix
A Ni-rich face-centred-cubic solid solution containing Cr, Co, Mo, W and other alloying elements. Its composition and solvus relations set the chemical environment for precipitation.
Coherent strengthening phase
An ordered Ni3(Al,Ti,Ta,Nb)-type precipitate. A compound-energy formalism with coupled sublattices represents site preference, ordering and continuity with the disordered phase.
σ · μ · P · R
Topologically close-packed phases can sequester refractory elements and degrade creep performance. Their complex structures generally require multi-sublattice descriptions and careful validation.
MC · M6C · M23C6
Carbide stability influences grain-boundary chemistry, heat-treatment response and long-term exposure. Carbon must be represented consistently with the metallic subsystem.
COMPOUND-ENERGY FORMALISM
(Ni, Al, Cr, …)3(Al, Ti, Ta, Nb, …)1
Constituents occupy crystallographically distinct sublattices. Endmember energies, ideal site mixing and interaction terms together describe ordering and substitution. The notation is a model definition—not a fixed stoichiometric formula for every γ′ composition.
Calculate γ′ solvus, equilibrium phase fractions and partitioning over solution and ageing temperatures.
Compare equilibrium and Scheil–Gulliver limits to assess segregation, incipient melting and terminal constituents.
Screen driving forces for TCP phases and carbides, then couple thermodynamics to mobility and precipitation models.
04
DATABASE DEVELOPMENT
Assessment is an iterative inference problem.
03
ACTIVE STAGE
Assessment
Model parameters are optimised against all selected evidence simultaneously. Residuals, data weights, parameter correlation and physical plausibility are examined rather than fitting one phase boundary in isolation.
THERMODYNAMIC DATABASE
A database is an integrated model collection.
It contains phase definitions, functions, parameters and metadata that share compatible reference states. Its reliability depends on assessment scope and evidence coverage, not on file size.
- Unary descriptionsReference Gibbs energies and heat-capacity functions for elements and structures.
- Binary assessmentsSolution interactions, compounds, ordering and invariant reactions.
- Ternary assessmentsTargeted corrections where binary extrapolation is insufficient.
- Higher-order predictionInterpolation from lower-order systems, with validation in the intended composition domain.
Parameters are commonly estimated by weighted nonlinear least squares. Data weights, covariance and parameter correlation matter; a low residual does not guarantee physical extrapolation.
Binary and ternary descriptions are combined using a stated geometric formalism, such as Muggianu interpolation, before higher-order corrections are introduced.
Bayesian or ensemble methods can propagate data and parameter uncertainty. Independent experiments remain necessary in the intended alloy and temperature domain.
05
EQUILIBRIUM, METASTABILITY AND KINETICS
What a calculation means depends on the imposed constraints.
GLOBAL MINIMUM
All admissible phases participate.
The calculation identifies the phase assemblage with the lowest total Gibbs energy under mass-balance constraints. It does not state how rapidly that state will be reached.
Global Gibbs-energy minimum under mass balance. Appropriate for limiting states, not transformation time.
Assumes complete liquid mixing, negligible solid diffusion and local interface equilibrium. It is a limiting solidification model.
DICTRA-type transport and KWN-type precipitation require mobility, nucleation and interfacial-property data in addition to thermodynamics.
REPRODUCIBLE COMPUTATION
Minimal pycalphad patterns
These examples expose the calculation conditions explicitly. They require a compatible thermodynamic database and phase names; replace placeholders with the identifiers defined in that database.
from pycalphad import Database, equilibrium, variables as v
db = Database("Ni_superalloy.tdb")
components = ["NI", "AL", "CR", "VA"]
phases = ["FCC_A1", "FCC_L12"]
conditions = {
v.P: 101325,
v.T: 1173,
v.X("AL"): 0.12,
v.X("CR"): 0.08,
}
result = equilibrium(db, components, phases, conditions)
print(result.NP, result.X)06
INTERPRETATION AND USE
Capabilities, limitations and responsible reporting
- Equilibrium and metastable phase diagrams
- Phase fraction and phase composition
- Activities, chemical potentials and driving forces
- Solidification paths under stated assumptions
- Inputs for diffusion, precipitation and phase-field models
- Treating a database as universally valid
- Equating equilibrium with transformation rate
- Ignoring metastable phases or imposed constraints
- Reporting extrapolation without validation
- Hiding parameter and data uncertainty
- Software and database version
- System components and selected phases
- Temperature, pressure and composition conditions
- Suspended phases and other constraints
- Assessment provenance and uncertainty domain
IMPLEMENTATIONS
One method, several software environments
Thermo-Calc, Pandat, JMatPro, MatCalc, OpenCalphad and pycalphad provide overlapping thermodynamic, solidification, diffusion or precipitation capabilities. ESPEI supports parameter assessment and uncertainty workflows. Interfaces differ; scientific meaning comes from phase models, databases, numerical conditions and constraints rather than the product name.