Creating thermodynamic databases: unary data, model compatibility, naming of phases, validation of databases, nano-materials. Examples using databases: Sigma-Phase Formation in Ni-bases Anticorrosion Superalloys, Intermetallic Phases in Lead-Free Soldering, Equilibria with Laves Phases for Aircraft Engines CT – 15: Creating thermodynamic databases, Examples using databases Creating thermodynamic databases Purpose of an assessment is to provide building blocks for multicomponent thermodynamic database. Database based on binary asssessments and a limited number of ternary assessments, all centered around a „base“ element (Fe, Al, Sn,…), can give reliable extrapolations to multicomponent alloys based on that element. Such databases are valuable tool for new alloy development, because experimenal investigations of multicomponent system are very expensive. Some ternary assessment must be included in such databases! There are ternary compounds that must be present in database and ternary assessment may reveal that some binary assessments require modification. Unary data The data for the pure elements must be the same in all assessments in which the element appears in order to make it possible to merge them into a database. International agreement on pure-element data (e.g. metastable modifications Cr fcc, Au bcc, etc,.) – lattice stabilities – collected by A.T.Dinsdale: CALPHAD 15 (1991) 317, and it is updated e.g. on http://www.sgte.org Database dictates keeping the unary data unchanged even if it is evident, that these need to be improved. (Example: data for Laves phases as Hydrogen storage, which used value of G of fictitious end members equal to 5 kJ.mol-1) It is desirable to develop an automatic and continuous reassessment procedure. Model compatibility nPhases that may form a continuous solutions from one binary system to another should be described by the same model which make it possible to combine these to give a single Gibbs energy description. nExample: Fe-Cr-C system: model (Fe, Cr)1(C, Va)1 will create new fictitious end member belonging to Cr-C system. It is not enough to work with A1 and B1 structures of Fe-Cr and Fe-C systems. n nOnly phases with the same structure may be treated as he same. nExample: Cr-Si, Fe-Si, and Mn-Si has structure of general stoichiometry M3Si, but Cr3Si has another structure and it should not be combined with them. n Structurbericht notation is a great help. It is important in extrapolation to higher-order systems. n nPartitioning of Gibbs energy in modeling of order-disorder transition is help in modeling in ternary systems nExample: B2 phase Al-Ni (ordered) should be combined with A2 phase Cr-Ni (A2 parts can be added, combined) nModels should be selected so, that they are compatible with the systems with which the assessment should be combined later Experimental database nAssessment is working with the file containig experimental data (in Thermo-calc „*.pop“ – file): n In this file all experimantal information should be referenced correctly and all transformations or corrections made from original publication documented. Also theoretical information should be referenced there. n (It is „temporary“ database.) n nExperimental datafile should be kept for the future use (similarly as assessment logbook). Naming of phases nPhases that can extend into a multicomponent system have to have a unique name. nExample: Ca-Mg-Fe-O system: CaO (lime), MgO (periclase), FeO (wustite) form continuous solution having the NaCl (B1) structure type. n In a database, the parameters for these three phases must be stored together with a single name (at present „halite“ has been selected). nExample: in iron database can be used name „ferrite“ for BCC structure and „austenite“ for FCC structure for complex alloys. nIt is important to agree to use the same name for the same phase for many applications. nStrukturbericht notation – has no relation to the actual structure (A – pure elements, B – binary systems 1:1, C – binary systems 2:1 etc.). It is popular. n„Structure family“, e.g. ordered structures (A1, L12, L10) based on simple disordered lattice (FCC), (B2, D03, L21) based on BCC lattice, and (D019 and B19) on HCP lattice. 4 sublattice model is used for L12, L10 ordering, but for is D022 – 8 sublattices is needed - separate phase. nSome compounds and intermetallic phases have traditional names (sigma, mju) nLiquid phase is usually treated as a single phase, aqueous solutions, polymers as separate phases (they do not mix with metallic and oxide liquids). Consistent database - summary nThermodynamic database cannot be simple collection of thermodynamic assessments published by various authors for relevant binary and higher-order systems in spite of beeing based on sound and proper models. nCondition of consistency of database: n- The models used have to represent Gibbs energy of a phase (unary and interactions). Violating this may lead to that e.g. order (sequence) of elements in parameter in various assessments (LA:B vs.LB:A) may require the change of sign in odd RK parameters – therefore, alphabetical order of element is recommended. Other (different) polynomials used, (e.g.Thermodynamically Adapted Polynomial – TAP), may create problems. n- The models and names used for description of the same phase existing in different systems have to be the same (sublattice model used in CEF must be the same, e.g. the number of sublattices, site occupancies etc.) n- The thermodynamic data for Gibbs energies of the same element in related systems (Lattice Stability) have to be the same - difference in unary data between SGTE1 and SGTE4 in fictitious structures (continuous revision). It is an opportunity to involve the ab initio modeling in applied research. From assessment to database nA database is (not only) a merged set od thermodynamic assessments of binary, ternary and higher-order systems. n nIndividual assessments is advantageous to keep in separate files for check (of unary data, models used, parameter values) – first task is to reproduce the original publication! n nFurther, phase names should be unified (software exists for it). nTransformation softwares for transformation of format of databases exist (e.g MTDATA to Thermocalc and vice versa). n Unassessed parameters nThe specific format of databases – there are programs for their conversion n nExample: Laves phases (A,B)2(A,B)1 and (A,C)2(A,C)1 are in database and ternary Laves phase exists, modeled by (A,B,C)2(A,B,C)1. n Parameters oGB:C and oGC:B are „unassessed“ and in most softwares are treated as zero. It might not be a good assumption! nAll unassessed parameters can be listed by program usually (e.g. in GES module of Thermocalc) and estimated value can be set to them. n Clear reference must be given for each estimated parameter n n Missing parameters nExample: Cu-Fe system has no stable HCP_A3 phase. Adding Zn to data-base in systems Cu-Zn and Fe-Zn, the ternary HCP phase (in Cu-Fe-Zn) can be modeled succesfully taking Cu-Fe HCP phase as ideal (interaction equal zero). nIn the Figure, it is evident that missing parameter must be assessed. LFS - CT Missing parameters – cont. nHCP phase will appear in Cu-Fe system even when HCP_A3 phase is explicitly suspended. n nSome hints in the cases of no parameters: -HCP can have the same parameter as FCC -HCP and FCC parameters can be set equal to the values for Liquid or BCC -BCC parameter can be set equal to parameter for Liquid (for unstable BCC) - Validation of the database n- Checking that the assessed system can be recalculated from the database n- Checking that the extrapolations from assessed systems are reasonable (valid composition range for each component in database) n nExample: Cr, Ni, Mo steel: all binaries and ternaries assesed n Problem: austenite/ferrite equilibria at 1150 oC were not n reproduced (duplex stainless steel) n Solution: introducing the same positive interaction in the FCC_A1 n phase in the Cr-Mo system as for the BCC_A2 phase n and reassessing the Fe-Cr-Mo system nExample: Nicrofer and Avesta superaustenitic steel: equilibrium state after n long-term annealing used for validation of database n Database management and updating nConstant updating of database is necessary. n nUpdating: adding new assessments n replacing existing assessments n- Individual assessments used to create the database must be kept in separate files all the time. n- Estimated parameters which was added to database should be also kept in special file. n- Extrapolation to higher-order systems must be checked after each updating n nAdding a binary data that was missing from the original database may not improve the database. (Addition of estimations of missing parameters may disturb). Documentation of a database nDocumentation is necessary for the management of database (change of the manager). n nDocumentation: systems included, test points used for extrapolation, n list of references of all assessed systems in database, n each parameter should be referenced to paper or other source n (in „TDB format“ at the end of statement) Existing thermodynamic databases nSeveral databases are available – classified in terms of their main component (e.g. Al database, Ni database etc.) or physical property n (semiconductor database) or special application (solder database). n More: CALPHAD 26 (2002) 141-312. n n(Our databases available: Steel-ex.tdb, solder.tdb) n nReference in publication: database name, suplier, year and version n It is important as an instrument for measurement! Website should be added. n nNew commercial databases are coded (not readable)!!! n n n Complementing database – Mobility database nComputational thermodynamics is most important for simulating phase transformations. n nThermodynamic description of the phase – source of chemical potentials of the components and the thermodynamic factor of diffusion. n nIn addition, mobility of the elements in the various phases have to be available. Source of that data: self-diffusion (or tracer diffusion) experiments. Mobility is often closely related to bulk modulus, which can be easily calculated using ab initio technique. It is possibility to estimate mobilities. n nThere is special version of PARROT module of Thermocalc program for assessment of mobilities using diffusion data or concentration profiles. nThermodynamic description is the base for it (calculating thermodynamic factor and chemical potential) nA few commercial mobility databases exist (e.g.Thermocalc). Nano-materials nPolycrystalline materials (grain size > 100 nm) have isotropic mechanical properties (monocrystals are used for a few applications only – e.g.turbine blades). n n„Bulk“ polycrystalline material: surface properties several order less important than bulk properties. n nInterface region between two different crystalline grains (grain boundaries) have larger density of defects than inner regions of the grains and the atoms on the interface can have different coordination number from those in bulk. nTherefore, interface properties are different from those of the „perfect“ crystal by a factor up to 30%. n nReducing the size of the grains to nanometers make the number of atoms in the grain comparable to the number of atoms at the interface – „nanomaterials“ – metastable equilibrium state. n Nano-materials - example nNanoparticles have different melting temperature than bulk material. n Gsurf = S.n.s = (3Mr s /r) (1/r) (spherical particles, n = (rM/r)3) T = Tm - 3Mr (Tm/ DHm) (-1/rsolid) [(s /r) liquid - (s /r) solid ] at Tm: Gliq = G sol and melting temperature T is given by: nDick K. et al., JACS 124 (10), 2312-2317 (2002), Au nCrosses: Calculation with estimation for Au (Buffat, Borel): n (s /r) liquid = 0,74/17300=4,28.10-5 n (s /r) solid = 0,90/19000=4,74.10-5 sejmout002 Nano-materials - example Nano-materials - example Nano-materials - example Influence of substrate on the melting temperature of nanoparticles Lee07 Melting temperature of gold nanoparticles (r>5nm) a)Graphite substrate b) b) b)Tungsten substrate Role of substrate only when it shows good wettability 31 (2007) 106 - 111 Nano-materials - example nNanoparticles of tin before heating Nano_Sn3 100 nm J.Vrestal: WG1 COST MP0602 meeting, Genova 2007, Sn-stand DSC - bulk Sn, atmosphere Ar, 5N sn-nano DSC - nano Sn, atmosphere Ar, 5N P.Brož: CALPHAD XXXVII, Saariselka,2008 Distribution of particle size before heating N particles Diameter of particles / nm Diameter of particles / nm V particles / .10-3 nm3 J.Vrestal: WG1 COST MP0602 meeting, Genova 2007, Surface in materials nGrain boundary can move and at increasing temperature grains can grow, minimizing its surface energy. nMechanical properties are usually better for materials with small grains nEffort to stabilizing grains against changing temperature was devoted – e.g.pinning the grain boundaries by placement of particles. n nRelative stability of different phases at the phase interface is the most important factor for determining the interface movement. nCrystal lattice and the composition change across a phase interface and thus bulk diffusions are also needed for the interface to move. n„Local equilibrium“ assumption at the phase interface: compositions at the n interface are given by an equilibrium tie-line in the phase diagram. n For very different mobilities of elements (e.g. C in Fe) – „para-equilibrium“ n assumption is valid. nAll of these models for phase transformations needs good description of the bulk thermodynamics. Surface tension of liquids - complementing database nThermodynamic database is needed also for calculation of surface tension of various systems (salts, oxides and liquid alloys) based on Butler equation: n ns = s1 + (1/A1)(G1E,s - G1E) + (RT/A1) ln ((1-x2s)/(1-x2)) s ns = s2 + (1/A2)(G2E,s - G2E) + (RT/A2) ln ((x2s)/(x2)) n nSolving these equations we get values of s and x2s for every value of x2 nThe superficial areas of pure liquid components A1, A2, the surface tension nof pure liquid components s1, s2 and partial excess Gibbs energies in the nbulk G1E, G2E, and in the surface G1E,s, G2E,s have to be known. nA1 = 1.091 NA1/3 V12/3 and similarly A2 = 1.091 NA1/3 V22/3 (N is Avogadro‘s number and V molar volume). nIt is suppposed that G1E,s/G1E = G2E,s/G2E = 0.83 and then thermodynamic database for G1E, G2E make us possible to calculate s and x2s by simple n computer program (http://www.gsp.ipm.cz/kroupa/) having complementary database for density and surface tension of pure components. Surface tension of liquids - example nR.Picha, J.Vrestal, A.Kroupa: CALPHAD 28 (2004) 141-146 n Nucleation in materials nFor the formation of the new phase in nucleation stage, the surface energy of the new phase is even more important than that for bulk phase. nClasical nucleation theory gives for critical radius of spherical particle of the new phase: n rc = 2 s Vm / DGm n (for r > rc: spontaneous growth of nucleus) nWhere s is the surface energy, Vm is the molar volume of the new phase and DGm is the difference in Gibbs energy between the new and old phases („driving force“) n nFor solid phase, it is difficult to calculate DGm exactly. Only approximative calculations are available. Examples using databases n Multicomponent phase diagram calculations nHigh-speed steel, 4% Cr, 9% Mo, 1.5% W, 1%V, 8% Co (rest Fe), C-varying. nIsopleth - reliability of lines is comparable with experimental determination – nmust be verified, but it helps to alloy design substantially LFS - CT Examples using databases – cont. nA.Kroupa, J.Havránková, M.Coufalová, M.Svoboda, J.Vřešťál: Journal of Phase Equilibria 22 (2001) 312 – 323 – Improvement of thermodynamic ndescription of carbides – important for long-term exploitation of that material (improved database for Fe-Cr-Mo-V-C system) Examples using databases – cont. nSimulation of phase transformation by DICTRA: one dimensional diffusion problem solved using thermodynamic calculations of driving force for nucleation and thermodynamic factor for diffusion outside the stability range Dissolution of spherical cementite particle in an Fe-Cr-C austenite matrix during heating. Assumption of local equilibrium at the interface (cementite surface) Chromium has an effect of slowing down the transformation – dissolution of cementite particle in Fe-C system takes less than 1 sec. LFS - CT Examples using databases – cont. n Dissolution of cementite in Fe-Cr-C austenite n Independent thermodynamic assessment of Fe-Cr-C system and mobilities of Cr and C in austenite and cementite were needed LFS - CT Examples using databases – cont. nScheil (1942) - Gulliver (1913) solidification model: nNo diffusion in solid phase – liquid is assumed to be homogeneous LFS - CT Examples using databases – cont nPhase field method: simulation of phase transformation in two or three dimensons. Grid is imposed and the amounts of phases at each grid are calculated based on thermodynamics and kinetic data assumed not „sharp“, but „diffuse“ interface (MICRESS phase-field software – (http://www.micress.de), TQ Thermocalc interface (http://www.thermocalc.com), and COST-507 light-alloy database. (Al-5.5Mg-0.4Mn) LFS - CT Examples using databases – cont. nSigma phase in corrosion-resistant steel: Fe–20Cr–18Ni-6Mo-0.5Mn-0.7Cu-0.3Si-0.2N p. 1025-1030 M.Svoboda, A.Kroupa, J.Sopoušek, J.Vřešťál, P.Miodownik Examples using databases – cont. nSolder database – Ag-Cu-Sn (SAC) solder – phase relations A.Dinsdale, A.Watson, A.Kroupa, J.Vrestal, A.Zemanova, J.Vizdal: Atlas of Phase Diagrams for Lead-Free Soldering. COST office 2008 Examples using databases – cont. nSolder database – Ag-Cu-Sn (SAC) solder – phase relations nA.Dinsdale, A.Watson, A.Kroupa, J.Vrestal, A.Zemanova, J.Vizdal: nAtlas of Phase Diagrams for Lead-Free Soldering. COST office 2008 Examples using databases – cont. nSolder database – prediction of ternary diagram based on binary data and verification of it by experiment, Bi-Sb-Sn:Manasijevič D.,Vřešťál J.,Minič D.,Kroupa A.,Živkovič D.,Živkovič Ž.: Journal of Alloys and Compounds 438 (2007)150-157 Examples using databases – cont. nLaves phases database for aircraft engines. System Cr-Zr. (Special separate database) Examples using databases – cont. CrTi CrHf Experiments: original sources in (Zhuang 2000) (Carlson 1968) – squares (Svechnikov 1965) – triangles W. Zhuang, J. Shen, Y. Liu, L. Ling, S. Shang, Y. Du,J.C. Schuster, Z. Metallkde 91 (2000) 121 O.N. Carlson, D.G. Alexander, J. Less-Common Metals 15 (1968) 361 V.N. Svechnikov, A.K. Shurin, G.P. Dmitrijeva, Prevrashchen Faz., AN Ukr.SSR 1965 159 Cr-Ti Cr-Hf (Vrestal J.: CALPHAD XXXVIII Prague 2009, Book of abstracts) (Special separate database) Thermodynamic models of phases nBCC: 2 \ 1 3 nHCP: 2 \ 1 0.5 nLiquid: 1 \ 1 n nLaves_C14: 2 \ 2 1 or 3 \ 4 6 2 nLaves_C15: 2 \ 2 1 nLaves_C36: 2 \ 2 1 or 3 \ 8 12 4 (simplified) n n Assignment of lattice sites in models nSublattice 1 2 3 nLaves_C14 4f 6h 2a (Cr) nLaves_C15 8a 16d (two sublattice model) nLaves_C36 4e, 4f 6g, 6h 4f (Cr) n Ab initio values (NEW) for DHo(T=0K) for Cr-Ti and Cr-Hf Laves phases C14, C15, C36 vs. SER phases (all combinations) – not changed in assessment – thermodynamic data! Parameter values of solution phases: Cr-Ti, Cr-Hf (in J/mol of atoms) – adjusted to experimental data Cr-Ti Cr-Hf [Zhuang 2000] [Yang 2008] BCC Lo = 11824 Lo = 42847.5 – 12*T L1 = 5012 L1 = 12064 HCP Lo = 25500 Lo = 43774 + 0.64729*T L1 = 15000 Liquid Lo = - 992 * Lo = -30000 + 8*T L1 = 1811 * L1 = 3800 * this work Liquid phase Cr-Hf: Miedema‘s guess: Lo,H = - 9000 (J/mol of atoms), Tanaka‘s rule: LH/LS = 1150 K nContribution of vibrational heat capacity and vibrational entropy to the Gibbs energy – influence of temperature G = DHo +0òDCp.dT - T0ò(DCp /T).dT DHo = EoL - EoSER (0 K) DCp = CpL - CpSER = a + bT G =DHo + T(a - b - a lnT) + (b/2)T2 (assume: b = 0, DCp = a) G @ DHo + Ta(1- lnT) New: ( DCp(T – TlnT) – adjusted) How to calculate vibrational contribution to heat capacity? nNew: Contribution of heat capacity n- DCpL‑SER is more positive for less stable C14 than for more stable (at 0 K) C15 n- Term DCpL‑SER (T-TlnT) at higher temperature stabilises less stable structures. n- The origin of this heat capacity is more probably vibrational (not configurational) – (it may be calculated from vibrational data – phonon spectra) n nProgrames calculating phonon spectra: nABINIT – NL – free nPHONON – PL - commercial nMEDEA – US - commercial n Parameter values of Laves phases (in J/mol of comp.) Cr-Ti system - 3 sublattices for C14 and C36 Laves phases (adjusted values are in red - correspond to the contribution of vibrational entropy and vibrational heat capacity) ab initio calculated values are in blue – not changed during optimization C14 G(Cr:Cr:Cr): 343598 + 12*GHSERCR G(Cr:Ti:Cr): -101605 -3.15*T + 3.15*T*ln(T) + 8*GHSERCR + 4*GHSERTI G(Cr:Cr:Ti): 425552 + 6*GHSERTI + 6*GHSERCR G(Cr:Ti:Ti): 151040 + 2*GHSERCR + 10*GHSERTI L(Cr:Ti:Cr,Ti;0): -20000 C15 G(Cr:Cr): 81877 + 3*GHSERCR G(Cr:Ti): -30486 -1.412*T + 1.412*T*ln(T) + 2*GHSERCR + GHSERTI G(Ti:Cr): 171806 + 2*GHSERTI + GHSERCR G(Ti:Ti): 96780 + 3*GHSERTI L(Cr:Cr,Ti;0): -57900 C36 G(Cr:Cr:Cr): 333020 + 12*GHSERCR G(Cr:Ti:Cr): -114138 -4.427*T+ 4.427*T*ln(T) + 8*GHSERCR + 4*GHSERTI G(Cr:Cr:Ti): 444722 + 6*GHSERTI + 6*GHSERCR G(Cr:Ti:Ti): 161657 + 2*GHSERCR +10*GHSERTI L(Cr:Ti:Cr,Ti;0): -32200 Parameter values of Laves phases (in J/mol of comp.) Cr-Hf system - 3 sublattices for C14 and C36 Laves phases (adjusted values are in red - correspond to the contribution of vibrational entropy and vibrational heat capacity) ab initio calculated values are in blue – not changed during optimization C14 G(Cr:Cr:Cr): 343598 + 12*GHSERCR G(Cr:Hf:Cr): -104379 -2.70*T + 2.70*T*ln(T) + 8*GHSERCR + 4*GHSERHF G(Cr:Cr:Hf): 566979 + 6*GHSERHF + 6*GHSERCR G(Cr:Hf:Hf): 231990 + 2*GHSERCR + 10*GHSERHF L(Cr:Hf:Cr,Hf;0): -10000 C15 G(Cr:Cr): 81877 + 3*GHSERCR G(Cr:Hf): -31130 -0.185*T + 0.185*T*ln(T) + 2*GHSERCR + GHSERHF G(Hf:Cr): 310380 + 2*GHSERHF + GHSERCR G(Hf:Hf): 114000 + 3*GHSERHF L(Cr:Cr,Hf;0): -21000 C36 G(Cr:Cr:Cr): 333020 + 12*GHSERCR (prediction) G(Cr:Hf:Cr): -116208 -1.72*T+ 1.72*T*ln(T) + 8*GHSERCR + 4*GHSERHF G(Cr:Cr:Hf): 597242 + 6*GHSERHF + 6*GHSERCR G(Cr:Hf:Hf): 234827 + 2*GHSERCR +10*GHSERHF L(Cr:Hf:Cr,Hf;0): -15000 Assessment results – Cr-rich region Cr-Ti Cr-Hf CrTiLaves CrHfLav Experiments: (Chen 1995) triangles up, diamond – 2 phase field (Carlson 1968) – squares Other symbols – single phase fields (Svechnikov 1965) – triangles (Zhuang 2000) points for xCr=0.6 and 0.7 Chen, K.C., Allen, S.M., Livingstone, J.D.: Mater. Res. Symp. Proc. 364 (1995) 1401 Conclusion nCalphad method n is physically based semiempirical method of calculations of phase diagram on thermodynamic bases. n It represents valuable tool for prediction of stable state of materials which enable us to do prediction of their properties and scientific materials design. Questions for learning n1. Explain the term „compatibility“ of database n n2. Explain the rules which have to be kept in preparing the database n n3. Explain the use of thermodynamic database for calculation of diffusion phenomena and in phase field method n n4. Explain the use of thermodynamic database for calculation of surface tension of binary and higher-order systems, for nucleation processes and for stability of nanomaterials n n5. Describe process of management and updating of databases