StructBond(2006)117:1–264 DOI10.1007/b136907 © Springer-VerlagBerlinHeidelberg2005 Publishedonline:3December2005 MagneticParametersandMagneticFunctions inMononuclearComplexesBeyond theSpin-HamiltonianFormalism RomanBoˇca DepartmentofInorganicChemistry,SlovakTechnicalUniversity,SK-81237Bratislava, Slovakia [email protected] 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 MagneticParameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3 MagneticFunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2 EnergyLevelsofMultitermSystems. . . . . . . . . . . . . . . . . . . . . . 20 2.1 PathsforEvaluatingEnergyLevels . . . . . . . . . . . . . . . . . . . . . . . 20 2.2 MatrixElements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.3 ZeemanLevels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 3 ModelingtheMagneticParameters . . . . . . . . . . . . . . . . . . . . . . 41 3.1 Computer-AidedCalculationsofSpin-HamiltonianParameters . . . . . . . 41 3.2 PartitioningTechnique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.3 ManualCalculationsofSpinHamiltonianParameters . . . . . . . . . . . . 46 3.4 MagnetochemicalFormulae . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4 CalculationsofEnergyLevelsandMagneticParameters . . . . . . . . . . 62 4.1 d1-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 4.2 d2-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.3 d3-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.4 d4-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 4.5 d5-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.6 d6-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 4.7 d7-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 4.8 d8-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 4.9 d9-OctahedralReference . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 5 EmpiricalMagneticParameters . . . . . . . . . . . . . . . . . . . . . . . . 181 5.1 Jahn–TellerEffect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 5.2 Jahn–TellerEffectinCrystal-FieldModel . . . . . . . . . . . . . . . . . . . 186 5.3 PrincipalMagneticParameters . . . . . . . . . . . . . . . . . . . . . . . . . 193 5.4 OrbitalReductionFactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 A SpectroscopicConstants,Coefficients,andMatrixElements . . . . . . . . 206 B IrreducibleTensorsandTensorOperators . . . . . . . . . . . . . . . . . . 220 2 R.Boˇca C ClassificationofCrystal-FieldTermsandMultiplets. . . . . . . . . . . . . 233 D CalculatedEnergyLevelsandMagneticParameters . . . . . . . . . . . . . 244 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262 Abstract Usingthespin-Hamiltonianformalismthemagneticparametersareintroduced through the components of the Λ-tensor involving only the matrix elements of the an- gularmomentumoperator.Theenergylevelsforavarietyofspinsaregenerated,andthe modelingofthemagnetization,themagneticsusceptibility,andtheheatcapacityisdone. Theoretical formulae necessary for performing the energy level calculations for a mul- titerm system are prepared with the help of the irreducible tensor operator approach. Thegoaloftheprogrammingistoevaluatetheentirerelevantmatrixelement(electron repulsion, crystalfield, spin–orbit interaction, orbital-Zeeman,andspin-Zeeman opera- tors)inthebasissetoffree-atomterms.Themodelingofthezero-fieldsplittingisdone at three levels of sophistication: (i) the differences in the crystal-field multiplets when the high-dimensional matrix in the complete space is diagonalized; (ii) the differences in the energy levelsof a model subspace when the partitioning technique is applied in the first iteration; (iii) the differences in the energy levels as they are produced by the second-order perturbation theory for the spin Hamiltonian. The spin-Hamiltonian for- malismofferssimpleformulaeforthemagneticparameters(theg-factors,D-parameter, χTIP)byevaluatingthematrixelementsoftheangularmomentumoperatorinthebasis set of thecrystal-fieldterms. Themagneticfunctionsfor dn complexesaremodeledfor awiderangeofcrystal-fieldstrengths. Keywords Magnetism·Magneticparameters·Magneticsusceptibility· Multitermsystems·Transitionmetalcomplexes Abbreviations AT atomicterm CI configurationinteraction CF crystalfield CFM crystal-fieldmultiplet CFP coefficientoffractionalparentage CFT crystal-fieldterm CSC completespacecalculation IR irreduciblerepresentation ITO irreducibletensoroperator JT Jahn–Teller MAM magneticangularmomentum MO-LCAO molecularorbitals–linearcombinationofatomicorbitals MP magneticparameter SH spinHamiltonian TIP temperature-independentparamagnetism ZFS zero-fieldsplitting MagneticParametersandMagneticFunctions 3 1 Introduction 1.1 Motivation Magnetochemical/physical experiments are done at thermal equilibrium: at sufficiently stabilized temperatures the thermodynamic response functions are monitored, i.e., the temperature dependence of the magnetic suscepti- bility (or magnetization with the SQUID apparatus) at a constant magnetic field, the field dependence of the magnetization at constant temperatures, and eventually the temperature dependence of the heat capacity at a con- stant(orzero)magneticfield.Theaboveresponsefunctionsreflectthemag- netic energy levels (the Zeeman levels) whose separation is on the order of a fraction to several reciprocal centimeters. Complementary information couldbeobtainedfromspectroscopicmethods:electronspinresonance,far- infrared spectroscopy or its modern version called the frequency domain spectroscopy,magneticcirculardischroism,andinelasticneutronscattering. Eachofthesemethods,however, hasitsownlimitationssothatthebestway istocombinetheminordertohaveasmuchinformationaspossible. The monographs on magnetism incorporate theoretical aspects to a dif- ferent extent and at a different level of complexity (from almost nothing to a very broad presentation) [1–24]. However, most of the initial progress in this area was made in electron spin resonance (ESR) [25–33]. In treating zero-fieldsplitting(ZFS)systems,themagnetochemicalandESRsourcesuti- lizedextensivelytheoperatorequivalentapproachalongwiththecrystalfield (CF) of lowered symmetry. Suchan approach, however, is limited to ground electrontermsobeyingHund’srules.TheZFSparametersareconsideredpa- rametersthatarefixedbyfittingtheexperimentaldatatoeachsystemunder investigation. To date there has been no rational approach that could corre- lateZFSparameterstosomeother,morefundamentaldata(integrals).Sofar, tailoring and tuning of ZFS parameters has only been an unrealized dream. What wouldfacilitaterealizationofthisdreamisadeep theoreticalanalysis. This is the main goal of the present paper. However, a sufficient descrip- tion of the situation requires much more effort, and we will see later that there is a need for combining information from the electronic structure of atoms [34–42], crystal/ligandfieldtheory [43–51], fundamentals ofangular momentum [52–55], and, finally, the irreducible tensor operator (ITO) ap- proach [56–63] as a real working tool. Recent monographs utilize the ITO approachasacommonstandard[22,23,32,49]. The next paragraph introduces briefly the set of magnetic parameters (MPs) that determine the spacing of the lowest energy levels in metal com- plexes. Then the magnetic response functions can be reconstructed using statisticalthermodynamics. 4 R.Boˇca Section 2 prepares all the theoretical formulae necessary for performing energy level calculations for a multiterm system. The spectral parameters enteringtheformulaearetabulatedinAppendixA.Thegoaloftheprogram- ming is to evaluate the entire relevant matrix element (electron repulsion, CF, spin–orbit interaction, orbital-Zeeman, and spin-Zeeman operators) in the basis set offree-atom terms. The transition to crystal-fieldterms (CFTs) is considered a redundant operation that can be omitted as such a process represents a unitary transformation that leaves the eigenvalues invariant. In this way a difficult manipulation with the point-group coupling coeffi- cients(3Γ-,6Γ-,9Γ-,orV-,W-,andX-coefficients)isavoidedentirely.This partuses theITOapproachextensively. Sincethislanguageisavailable from alimitednumber ofsources,acomprehensive reviewisofferedtoreadersin AppendixB. ThemodelingoftheZFS(Sect.3)isdoneatthreelevelsofcomplexity: (a)The differences in the crystal-field multiplets (CFMs) when the high- dimensionalmatrixinthecompletespaceisdiagonalized; (b)The differences in the energy levels of a modelsubspace when the parti- tioningtechniqueisappliedinthefirstiteration; (c)The differences in the energy levels as they are produced by the second- orderperturbationtheoryforthespinHamiltonian(SH). The labeling oftheCFMsisbased onthedouble-groupconcept,whichisfar from common. Therefore, the most relevant items of the double groups are collected in Appendix C. The SH formalism offers simple formulae for the MPs (g-factors, D-parameter, χ ) by evaluating the matrix elements of the TIP angular momentum operator in the basis set of the CFTs. The values of the temperature-independent paramagnetism(TIP)areabyproductofthetheor- eticalmodeling. In Sect.4 detailed calculations are performed for d1 to d9 systems in the geometry of a regular octahedron and a compressed/elongated tetragonal bipyramid. Calculated energy levels and MPs are presented in Appendix D alongwiththeenergyleveldiagrams. Section 5 introduces the Jahn–Teller (JT) effect and its treatment within theCFmodel.Briefinsightintotheelectronic-structureparametersisgiven. Notationsused: (a)SI units are used consistently throughout: χ [SI] = 4π×106χ mol mol [cgs&emu]. (b)EnergyquantitiesE(likeε,J,D,E,a,F,etc.)arepresentedascorrespond- ingwavenumbers, i.e.,E/hc,andgiveninunitsofcm–1. (c)The angular momentum operators translate the reduced Planck constant (cid:1) duringoperationintoacorrespondingwavefunction(aket). (d)The fundamental physical constants (c, ε , µ , N , k=k , R, µ , e, h, 0 0 A B B (cid:1) ) adopt their usual meaning—they enter the reduced Curie constant C =N µ µ2/k. 0 A 0 B MagneticParametersandMagneticFunctions 5 1.2 MagneticParameters Transition-metal(dn)complexeswithopenshellsbelongtotheclassofpara- magneticmaterials:theirmagneticsusceptibilityispositive(thesampleisat- tractedtothemagneticfield)andistemperature dependent. Athighenough temperatures andinsmallfields,themolarmagneticsusceptibility normally obeystheCurielaw χ =C/T. (1) mol The Curie constant C can be considered a magnetic parameter (MP) asso- ciated with the sample. Theory, however, tells us that such a phenomeno- logical parameter could be made of fundamental physical constants and the magnetogyric-ratioparameterg inthefollowingway: (cid:1) (cid:2) C= N µ µ2/k g2S(S+1)/3=C g2S(S+1)/3, (2) A 0 B 0 where S is the spin of a Curie paramagnet. Thus we arrive at another MP— the g-factor—and a reduced Curie constant C . The hyperbolic decay of the 0 susceptibility with increasing temperature isbetter followedwhen the prod- uct function P=χ T/C (dimensionless) is plotted as a function of tem- mol 0 perature, and the perfect fulfillment of the Curie law is now represented by a straightline P vs. T withazeroslope. Inan olderrepresentation, the tem- perature dependence oftheeffective magneticmoment,µ giveninunitsof eff Bohrmagneton,isfollowedwhere (cid:3) (cid:1) (cid:2) √ µ /µ = 3k/ N µ µ2 χ T= 3P. (3) eff B A 0 B mol The Curie paramagnets [e.g., octahedral Fe(III) complexes] are rather rare, andthetemperaturedependenceofthemagneticsusceptibilityrequiresmore parameters, depending on the actual spacing of the low-lying energy levels. LetusenumeratetheMPsassociatedwiththeSHformalism: (a)g-tensorcomponents,g(cid:2) andg⊥,orgx,gy,gz; (b)AxialZFSparameterDandrhombicZFSparameterE; (c)TIPχ ; TIP (d)Eventually,biquadraticspin–spininteractionparametersaandF. Under the SH formalism it is understood that the energy levels are recon- structed by considering only the formal spin kets |S,M (cid:4) that are under the S action of the SH involving the spin–spin interaction and the spin-Zeeman term (cid:5) (cid:6) (cid:7)(cid:8) (cid:1) (cid:2) H(cid:4)S=(cid:1)–2 D (cid:4)S2–(cid:4)S2/3 +E (cid:4)S2–(cid:4)S2 +(cid:1)–1µ Bg (cid:4)S (4) a z x y B a a (a=x,y,z). Such a Hamiltonian is appropriate for ZFS systems. (When the spin–spin interaction term vanishes, the SH collapses to the pure spin- 6 R.Boˇca Zeeman term appropriate to the Curie paramagnet.) The MPs g , g , g , D, x y z andEareunderstoodasconstantsthatcharacterizethesampleunderstudy. The Hamiltonian that describes the interaction of the single magnetic center with the external magnetic field involves the spin-Zeeman term, the orbitalZeemanterm,andtheoperatorofthespin–orbitcoupling,i.e., H(cid:4)(cid:5)=(cid:1)–1µ g (S(cid:6)·B(cid:6))+(cid:1)–1µ (L(cid:6)·B(cid:6))+(cid:1)–2λ(L(cid:6)·S(cid:6)). (5) B e B The aboveHamiltonianactsasaperturbationoperator,sothatperturbation theoryyieldsthefirst-orderandsecond-ordercorrections (cid:9) (cid:10) (cid:10) (cid:11) H(cid:4)(1)= 0(cid:10)H(cid:4)(cid:5)(cid:10)0 , (6) (cid:9) (cid:10) (cid:10) (cid:11)(cid:9) (cid:10) (cid:10) (cid:11) (cid:12) (cid:10)(cid:4)(cid:5)(cid:10) (cid:10)(cid:4)(cid:5)(cid:10) 0 H K K H 0 H(cid:4)(2)=– . (7) E –E K(cid:7)=0 K 0 It is assumed that the state vectors are represented by orthonormal kets of spatial variables, viz., |K(cid:4)=|α,L,M (cid:4). The first-order correction can be L rewrittenasfollows: (cid:13) (cid:10) (cid:10) (cid:14) (cid:10) (cid:10) H(cid:4)(1)=(cid:1)–1µ g (B(cid:6)·S(cid:6))(cid:8)0|0(cid:4)+((cid:1)–1µ B(cid:6)+(cid:1)–2λS(cid:6))· 0(cid:10)L(cid:6)(cid:10)0 , (8) B e B wherethelasttermvanishesintheabsenceofthefirst-orderangularmomen- tum,andthuswearriveatthepurespin-Zeemaninteraction: H(cid:4)(1)=(cid:1)–1µ g (B(cid:6)·S(cid:6)). (9) B e Thesecond-ordercorrectioncontainstheterms (cid:9) (cid:10) (cid:10) (cid:11) (cid:13) (cid:10)(cid:10) (cid:10)(cid:10) (cid:14) 0(cid:10)H(cid:4)(cid:5)(cid:10)K =(cid:1)–1(µ B(cid:6)+(cid:1)–1λS(cid:6))· 0(cid:10)L(cid:6)(cid:10)K +(cid:1)–1µ g B(cid:6)·S(cid:6)(cid:8)0|K(cid:4) , (10) B B e where the last contribution vanishes owing to the orthogonalityof the state vectors.Thecompletesecond-ordercorrectionis (cid:5) (cid:13) (cid:10) (cid:10) (cid:14)(cid:8)(cid:5) (cid:13) (cid:10) (cid:10) (cid:14)(cid:8) (cid:10) (cid:10) (cid:10) (cid:10) (cid:12) (µ B(cid:6)+(cid:1)–1λS(cid:6))· 0(cid:10)L(cid:6)(cid:10)K (µ B(cid:6)+(cid:1)–1λS(cid:6))· K(cid:10)L(cid:6)(cid:10)0 B B H(cid:4)(2)=–(cid:1)–2 , E –E K(cid:7)=0 K 0 (11) andafterintroducingtheΛ-tensor (cid:9) (cid:10) (cid:10) (cid:11)(cid:9) (cid:10) (cid:10) (cid:11) (cid:12) (cid:10)(cid:4) (cid:10) (cid:10)(cid:4) (cid:10) 0 L K K L 0 Λ =–(cid:1)–2 a b , [energy–1], (12) ab E –E K(cid:7)=0 K 0 itcanberewrittenintheform H(cid:4)(2)=(µ B(cid:6)+(cid:1)–1λS(cid:6))·Λ·(µ B(cid:6)+(cid:1)–1λS(cid:6)). (13) B B MagneticParametersandMagneticFunctions 7 (OneshouldbecarefulinthedefinitionofthesignoftheΛ-tensorsinceone encounterstheoppositesignintheliterature.)Theoverallresultofperturba- tiontheoryuptothesecondordercanbewrittenasfollows: H(cid:4)S=– 1(B(cid:6)·κ·B(cid:6))+(cid:1)–1µ (B(cid:6)·g·S(cid:6))– 1(cid:1)–2(S(cid:6)·∆·S(cid:6)), (14) B 2 2 whereweintroducetheκ-tensor (reduced, temperature-independent param- agneticsusceptibilitytensor) 1 – κpara=µ2Λ , [energy×induction–2], (15) 2 ab B ab theg-tensor(magnetogyricratiotensor) g =g δ +2λΛ , [dimensionless], (16) ab e ab ab andtheD-tensor(spin–spin interactiontensor) – 1∆ =λ2Λ =D(cid:5) , [energy]. (17) 2 ab ab ab (Thenumericalprefactors,like–1/2,areamatterofconvention.) The SHH(cid:4)S actsonlyonthespin kets |S,M (cid:4)yielding eigenvalues that are S (cid:4)(cid:5) identical withthoseproduced bythe perturbationoperator H acting onthe fullsetofspin–orbit variables|α,L,M ,S,M (cid:4).Thissituationisexplained in L S Table1:thetruncatedSHmatrixinvolvesintegralsovertheangularmomen- tumviatheperturbationtheory. On concluding, MPsare comprised of three Cartesian tensors, each com- posed of the Λ-tensor and the spin–orbit splitting parameter λ. In this way, just the Λ-tensor is responsible for the shift of the g-components relative to thefree-electronvalueg ,fortheappearanceoftheZFStensor∆,andforthe e existenceoftheTIP. Table1 ExplanationofthemagneticHamiltonianandthespinHamiltonian 8 R.Boˇca The temperature-independent paramagnetic term is omitted hereafter (this can be included in the empirical correction of the experimental data, togetherwiththediamagneticterm)sothatwearriveattheZFSHamiltonian: (cid:5) H(cid:4)zfs=(cid:1)–2(S(cid:6)·D ·S(cid:6))+(cid:1)–1µ (B(cid:6)·g¯¯·S(cid:6)). (18) B As far as the D-tensor is concerned, it contains nine Cartesian components. Assumingthatthecoordinateaxesareidenticalwiththeprincipalaxesofthe D-tensor, onlythediagonalelements contribute. Byintroducingnewparam- eters D=(–D(cid:5) –D(cid:5) +2D(cid:5) )/2, (19) xx yy zz E=(D(cid:5) –D(cid:5) )/2, (20) xx yy J=(D(cid:5) +D(cid:5) +D(cid:5) )/3, (21) xx yy zz theequivalentformoftheZFSHamiltonianis H(cid:4)zfs=(cid:1)–2[D((cid:4)S2–(cid:4)S2/3)+E((cid:4)S2 –(cid:4)S2)+J(cid:4)S2]+(cid:1)–1µ (B(cid:6)·g¯¯·S(cid:6)). (22) z x y B The constant term J(cid:4)S2 can be removed from further consideration since it uniformly shifts all the energy levels. The omission of the constant term, in fact,correspondstothesubtractionononethirdofthetraceoftheD-tensor fromthediagonalelements,andinthiswaythe(diagonal)D-tensorbecomes traceless.NotethattheDandEparametersremainunchangedwhenthesame constant term is added (subtracted) to (from) the diagonal elements of the D-tensor.NormallyitisassumedthattheZFSparametersobeyarelationship |D|≥3E≥0. (23) One caninterchange the Cartesian axes (which cannot influence the proper- tiesofthesystem),buttheaboverelationshipwillstillhold. Theinvolvementoftheangularmomentumthroughtheperturbationthe- ory is not permitted when it is contained in the ground-stateket. The triply degenerateelectronterms2S+1Tbelongtosuchacaseofthemagneticangular momentum(MAM).ThentheinteractionHamiltonianinvolves(1)thespin– orbitinteraction,(2)theorbitalZeemanterm,(3)thespin-Zeemanterm,and (4)eventuallythesymmetry-loweringterm H(cid:4)=(cid:1)–2λsf(γsfAκL(cid:6) ·S(cid:6))+(cid:1)–1µ B(cid:6)·(γsfAκL(cid:6) +g S(cid:6)) p B p e +∆ [(cid:1)–2(cid:4)L2–L(L+1)/3] (24) ax z (whereL=1andγsf=–1shouldbeapplied).TheZeemantermcanbetaken asaperturbation,whereastheothertermsneedtobeincludedintheunper- (cid:4) turbedHamiltonianH .ThisapproachhasbeenoutlinedbyKotani[64]and 0 thengraduallyextendedbyGriffith[44],Figgis[45],andFiggisetal.[65–68]. In this case the only MPs are the spin–orbit splitting parameter λ, the or- bitalreductionfactorκ≤1,theFiggisCI-mixingparameter1≤A≤3/2,and MagneticParametersandMagneticFunctions 9 theaxialsplittingparameter∆ (whichispositivewhentheA-term,arising ax fromthesplittingoftheT-term,isthegroundstate). Wehaveseenthatmononuclearcomplexeswithorwithoutthefirst-order angularmomentumcanbetheoreticallytreatedinadifferentdegreeofcom- plexity. Thus we can speak of some magnetotheoretical hierarchy (Table2). Dependingonthebasicpostulateabouttheextentoftheactivespace(aspace of kets included in the zero-order Hamiltonian) various levels of the theory canbedistinguished. In the most complete treatment of this approach (such as the theory of König and Kremer [69–71]), the whole spectrum of the electron energy Table2 Magneto-theoreticalhierarchy Level Description Activespace Freeparameters Magnetic parameters 7 Notavailable Alltermsoffkdnsmpl ξ ,ξ,... d f 6a K¨onig& Alltermsofdn, B,C, ∆g=0 Kremer kets|vSLJΓγa(cid:4) Dq(Dt,Ds) κpara=0 [62,69–71] orF4 (F2,...), D=0 ξ ,κ d 6b Schilder& Alltermsofdnorfn, Lueken[78] kets—atomic microstates 6c [Thiswork] Alltermsofdn, kets|v,L,ML,S,MS(cid:4) 5a Figgisetal. LimitedCI, λ,κ,A,∆ax ∆g=0 [65–68] ketsoflowered (orv=∆ax/λ) κpara(cid:7)=0 symmetry 5b Weissbluth LimitedCIforspin [59] admixedstates kets|Γ,γ,S,MS(cid:4) 4 Figgis[45] LimitedCIfor λ,κ,A ∆g=0 2S+1T1 terms κpara(cid:7)=0 3 Kotani[64], Groundterm2S+1T, λ,κ ∆g=0 Griffith[44] ket(cid:10)s|L=(cid:11)1,ML,S,MS(cid:4) κpara(cid:7)=0 (cid:10) or J,MJ 2 Zero-field Groundterm2S+1A, gaa,D,E,(a,F), ∆g(cid:7)=0 splitting kets|S,MS(cid:4) χTIP,(orΛaa) κpara(cid:7)=0 D(cid:7)=0 1 Effectivespin LowestK(cid:10)ramers(cid:11) gaefaf ∆g(cid:7)=0 doublet(cid:10)(cid:15)S,±1/2 κpara(cid:7)=0 10 R.Boˇca levels is determined by the electronic-structure theory: the Racah parame- ters B and C, the CF parameters F (L) and F (L) for each ligand L (or the 4 2 conventional Dq, Ds, Dt, ... parameters), and the effective spin–orbit coup- ling constant ξ ; eventually the orbitalreduction factorκ canbeconsidered. d There is almost nothing to be included from the perturbation theory, ex- cept the Zeeman term. Therefore, the differential g-tensor, the temperature- independent paramagnetic tensor, and the spin–spin interaction tensor are blank:∆g=g–g 1=0,κpara=0,D=0.TheZFSparameterD(E,a)isdirectly e read off as the splitting of the lowest CFMs, and the g-factors are deter- mined bytheevolutionoftheZeeman levelsinthemagneticfield.Thereare nolongeranyphenomenologicalMPsasthemagnetism iscompletelydeter- minedbytheelectronicstructureandthemagneticfieldstrength(induction). By contrast, when the active space is restricted to the spin-only kets, the influence of all attainable excited states manifests itself in the filling of the MPs (tensors). In such a case the g-tensor deviates considerably from the free-electronvalue,theTIPappearssubstantial,andthespin–spininteraction tensortransformstohighvaluesoftheZFSparameters(DandE). 1.3 MagneticFunctions Having determined the magnetic energy levels ε (B) (as eigenvalues of the a,i interactionHamiltonian)wecanproceedwiththeapparatusofthestatistical thermodynamicbydefiningthe(magnetic)partitionfunction (cid:12)N Z (T,B)= exp[–ε (B)/kT] (25) a a,i i=1 andthefree(Helmholtz)energyperparticle F(T,M)=–kTlnZ. (26) Thentheobservablethermodynamicfunctionsaredeterminedasfollows: 1. Molarmagnetization (cid:16) (cid:17) (cid:16) (cid:17) (cid:16) (cid:17) ∂F ∂lnZ 1 ∂Z M =–N =–N kT =–RT ; (27) mol A ∂B A ∂B Z ∂B T T T 2. Molarisothermalmagneticsusceptibility (cid:16) (cid:17) (cid:16) (cid:17) ∂M ∂2F χ =µ =–µ ; (28) T 0 ∂B 0 ∂B2 T T 3. Molarisofieldheatcapacity(E–ethalpy,Q–heat,S–entropy) (cid:16) (cid:17) (cid:16) (cid:17) (cid:16) (cid:17) ∂Q ∂E(S,H) ∂S C = = =T , (29) H ∂T ∂T ∂T H H H