Molecular Magnetism Ilaria Ciofini and Claude Daul Institute of Inorganic and Analytical Chemistry University of Fribourg Switzerland History of magnetism Lodestone = magnetic core (Fe3O4) was mined in the province of magnesia: g in g greek writings g ((800 BC). ) Thales “the lodestone possesses a soul” (627-547BC). Descartes postulate the dichotomy of soul and body: modern science is born (1596-1650). Langevin g explains p p paramagnetism g and diamagnetism g classically y (1905). Bohr and Van Leeuwen (separately) proved a theorem excluding magnetism as a classical phenomenon. Heisenberg explained magnetism as a purely quantum mechanical effect:exchange. effect:exchange (1928). (1928) Magnetic Exchange Interactions: In insulators “the spin and magnetic moments whose alignments lead to magnetic effects are certainly localizable so that phenomenologically at least, they can be described by a spin hamiltonian which contains spin operators and exchange terms of H i Heisenberg b ttype.”” Philip W. Anderson Type of Interactions: STRONG WEAK diamagnetic ferro/antiferro magnetic What is “Molecular Magnetism”? “Molecular Magnetism deals with the magnetic properties of isolated molecules and assemblies of molecules.” molecules. O. Kahn in Molecular Magnetism Which are the possible applications? • Material Science • Molecular Devices • Bio-inorganic Bi i i S Systems t Molecular Magnetism and Computational Chemistry: QUALITATIVELY QUANTITAVELY Exchange Pathways analysis Energies of Spin Multiplets Semi-Empirical Rules Spin Density analysis PROBLEMS SMALL energy differences Does not give predictions Spin Contamination Computational Effort Illustration1 : ab initio calculation of singlet-triplet separation (in cm-1) in cupric acetate hydrate dimer 0th 2nd order I higher order orders exp II -244 I : JF = 2Kab II : JAF = -(εg- εu)2 / (Jaa - Jab) 1 P. de Loth, P. Cassoux, J.P. Daudey, J.P . Malrieu, J. Am. Chem. Soc. 103 (1981) 4007 -286 DFT: Heuristic approach X-ray diffraction dr ∫ ρ(r )dr Cusp ρ(r ) nuclear positions # electrons ⎛ ∂ρ ∂ (Rk )⎞ = −2Zk ρ(R0 ) ⎜ ⎟ ⎝ ∂Rk ⎠ R =R k ĤΨ H Ψ = EΨ 0 General Theory Exact energy expression 1 2 Eel = - + ∑ i ∑ ∫ A ∫ ZA → → | R A- r 1| → 1 2 → → → φ ( r 1 )∇2 φ ( r )d r i i 1 1 → ⌠ρ( r 1 )ρ( r 2) ⎮ → ⎮| → r ⌡ 1 r2 | → → ρ( r 1 ) d r 1 → → dr 1 dr 2 + Exc Parr R G ;Yang W : Density Functional Theory of Atoms A Parr,R.G.;Yang,W Molecules, Oxford University Press, New York 1989 The Kohn-Sham Equation hksφi = εiφi hKS = 1∇2 + 2 ∑ ZA → → A | R A- r | 1 → ⌠ ρ( r ) → ⎮ 2 + dr 2 → → ⎮| r - r | ⌡ 1 2 + VXC Approximate density functional theories for exchange and correlation Xα Local exchange LDA Local exchange + local correlation GGA Local exchange + local correlation + gradient corrections 3rd Generation of functionals Xα : Local exchange functional of the homogeneous electron gas LDA: Local exchange functional + local correlation functional of the homogeneous electron gas GGA: Same as LDA + “non-local” non local gradient corrections to exchange and correlation 3rd Generation of functionals: Same as GGA + instilation of “exact-exchange” and + 2nd derivatives of the density corrections Practical Implementation Solve Kohn KohnSham eqs. ⎡ 1 ⎤ ρ( r' ) d ' +VXC (ρ( r))⎥Ψi = ε i Ψi (r ) dr' ⎢− ∇ + v(r ) + ∫ r − r' ⎣ 2 ⎦ Features: LCAO expansion: expansion waves STO, GTO, numerical, plane Coulomb potential: potential set solve Poisson’s eq. or fit ρ(r) to a of one-center one center auxilliary functions Matrixelements: accurate numerical integration in the Matrixelements irreducible wedge of the molecule Methodology based on Approximate DFT MS-X α 1966 DV-X α 1970 ADF FRIMOL 1994- 1973 - development in progress DeMon 1976 - NUMOL 1982 MS-X α: Make use of partial-waves as basis 37). ( Relatively fast. Good for ionization potentials and excitation energies 1( 0). Total energies unreliable39). ( No geometry optimization. Full use of symmetry. H relativistic Has l ti i ti extension t i 53f). 5( 3f) M Make k use off muffin-tin ffi ti approximation (38). Developed by K.H. Johnson (37) . DV-Xα: Make use of numerical atomic orbitals or STO's. Avoids Muffin-tin Muffin tin approximation by fit of density 45a). ( Accurate total energies (76d). Relativistic extension (53e). Numerical integration of matrix elements by Diophantine integration (40). Developed by Ellis and Painter (40). Extensive improvements by Delley (D-MOL-program) including new integration scheme (46c) and geometry optimization. HFS-LCAO : Make use of STO's . Accurate potentials 41). ( Full use of symmetry. Relativistic extensions (53a,b). Highly vectorized (47). Accurate total energies (49). Geometry optimization (54c). Accurate numerical integration (46b). Many auxiliary property programs . Pseudo potentials (52a (52a,d). d) Embedding procedures (76h). Energy decomposition scheme (72). Developed by Baerends,Snijders,Ravenek,Vernooijs and te Velde (41,53,47,46d) LCGTO-LSD : Make use of GTO's. Fit of exchange-correlation and Coulomb potential43). ( Analytical calculation of matrix elements (48b). Accurate energies. Geometry optimization (54b,h). Strongly vectorized48b). ( First developed by Dunlap (43) as well as Sambe and Felton (42). Extensive improvements by Salahub and Andzelm (48b) (D-GAUSS-program) as well as Rösch (74a). Also work by Pederson (45e) and Painter (45d) NUMOL : Unique basis basis-set set free program 50a,e). ( Accurate numerical integration (46a). Efficient generation of Coulomb potential (50c). Geometry optimization. Developed by Becke (50 ). Methodology É Post-HF Calculations (as comparison) É DFT Broken Symmetry Single Determinant S i P Spin Projection j ti Techniques T h i Βroken Symmetry The electronic symmetry is artificially lowered Symmetry Equivalent Atoms Become Distinct BS= a+a' − a+ a’- Noodleman L. J. Chem. Phys. 74 (1981) 5737. E(S)= J/2[S(S+1)] + const. H=JS1S2 For WEAKLY Interacting systems For MEDIUM Interacting systems [E(S=1)-E(S=0)] =J E(HS) − E(BS) J= 2S1S2 E(HS) − E(BS) J= 2 2S1S2 + α (S)sab Single determinant method th d Multiplet wave function Ψi = Single determinant Ψi = α Γ m Γ S mS ∑ Aiμ i Φμ μ Φμ = χ1 χ2 χ3 ... E Energy off single i l determinant d i ( ) ( ) E Φ μ = E DFT Φμ tot COMPUTED Single Determinant Method: A X B Considering g just j the SOMO as active orbitals |b> |φ A> |φ B> Four multiplets arise: |a> | a2 1A> | b2 1A> | ab 1B> | ab 3B> |φ X> |φ 0> Energy of the Multiplet in Terms of Single Determinant |b2 1A> |ab 1B> Kab/[E(a2)E(b2)] E(b2) 2E(a↑b↓)-E(a↑b↑) Kab ||ab 3B> E(a ( ↑ b↑ ) |a2 1A A> E(a2) E(1B) − E(3B) = = E a↑ b ↑ − E a ↑b ↓ 2 ( Kab/[E(a2)E(b2)] where Kab is the exchange integral between the orbitals a and b ) ( ) Spin Projection Technique Ψunr,S = aΨS +bΨT 2 b = a2+b2 =1 1 2 ψ unr,S S ψ unr,S 2 E unr,S = ψ unr,S H ψ unr,S = a 2 ψ S H ψ S + b 2 ψ T H ψ T ES = E unr,S − b 2 E T 1− b2 Δ ST = E S − E T = E unr,S − E T 1− b 2 General Case: ψ unr,S = a S (S)ψ S + a S (S + 1)ψ S+1 + ... + aS (Smax )ψ S max B L = (S− )L (S + ) L L ψ unr,S BL ψ unr ,S = L! ∑ det[V ( j, j' )] j,j' =1 αβ β V( j, j' ) = δ j,j' − ∑ sαβ jk s j'k nk k L−1 ψ(S, M) B L ψ(S, M) = ∏ [S(S + 1) − (M + 1)( M + i + 1)] i=0 Defining: P−1 Tk (P,S) = ∏[S(S + 1) − (k + i)(k + i + 1)] i=0 Tk (0,S) ≡ 1 As (S) + AS (S + 1) + AS (S + 2) + ... + AS (S + L) = 1 TS (1,S +1) AS (S +1) + TS (1,S + 2)AS (S + 2) + ... + TS (1,S + L)AS (S + L) = B1 ... ... TS ( L,S + L) AS (S + L) = BL E uncor (S) − A S (S + 1)E corr (S + 1) − AS (S + 2)E corr (S + 2) − ... E corr (S) = 1 − AS (S + 1) − AS (S + 2) − AS (S + 3) − ... A. A. Ovchinnikov, J. K. Labanowski, Phys. Rev. A (1996) 53, 3946. H2 Dissociation Energy gy Curve: an Illustrative Example p H2 Dissociation Energy Curve: Triplet and Singlet 2.00 1.50 1.00 Singlet Triplet 0.50 0.00 -0.50 0 00 0.00 1 00 1.00 2 00 2.00 3 00 3.00 4 00 4.00 5 00 5.00 6 00 6.00 7 00 7.00 8 00 8.00 H2 Dissociation Energy Curve 0 50 0.50 0.40 0.30 0.20 0.10 0.00 -0.10 Full-CI RHF RDFT -0.20 -0.30 -0.40 0 00 0.00 1 00 1.00 2 00 2.00 3 00 3.00 4 00 4.00 5 00 5.00 6 00 6.00 7 00 7.00 Minimal Basis Set H2 Dissociation Energy Curve: long range behaviour Single g Determinant and Broken Symmetry y y Versus full-CI 0.30 0.20 0.10 0.00 0 00 -0.10 Full-CI RDFT Single Det UDFT (BS) -0.20 -0.30 0.30 -0.40 0 00 0.00 2 00 2.00 4 00 4.00 d(H-H) 6 00 6.00 8 00 8.00 0.20 0.15 0 10 0.10 0.05 0.00 -0 0.05 05 -0.10 -0.15 -0.20 -0.25 -0.30 -0.35 0.00 2.00 4.00 6.00 8.00 d(H-H) 1.00 0.80 0.60 0.40 0.20 0.00 0.00 2.00 4.00 d(H-H) 6.00 8.00 RDFT UDFT <S2>(UDFT) QuickTime™ and a Animation decompressor are needed to see this picture. H2 Dissociation Energy Curve: • DFT in the Local Approximation generally OVERBINDING • Restricted DFT and HF Approaches do not give the Correct Long Range Behaviour • The Single Determinant Method yields the correct behaviour at far distances • Allowing the spin α and β wave functions to have different spatial components (Unrestricted) yields a Localised (BS) state and includes non dynamical correlation (correct asymptotic behaviour) Application s • Model Systems: H-He-H • Organic g Radicals: BVD Nitroxides • Metal Complexes: Organic Radical and Dia/Para magnetic Metal Ions Binuclear Metal Complexes Singlet-triplet separation (cm -1)a computed for H −He−H at different H−He distances. -1 Singlet -Triplet separation (cm ) computed for H-He-H Method of calculation AB INITIO: CAS (2,2) OVB-MP2(2,2) CAS(4,3) OVB-MP2(4,3) QCISD(T) Res-BS Unres-BS S full CI DFT Xα BS Xα-BS LDA-BS BP-BS Xα-SD Xα SD LDA-SD BP-SD H−He Distances ( ) 1.25 1.625 2.00 4204 4358 4294 4530 4928 4298 4580 4860 476 420 484 492 790 526 554 544 48 26 46 48 202 60 60 50 6004 12432 10529 5631 9050 7799 646 760 1268 745 1553 1170 58 158 134 64 183 123 The reported values are E(3Σu)- E(3Σg). POSITIVE values indicate a SINGLET G a T he reported values are E(3Σu)-E(1Σg). Positive values indicate that the singlet is the ground state. H2NO-H NO H Singlet-triplet separation (cm-1)a computed for H2NO–H with an O-H distance of 1.9 Å. Method of Calculation E(3B1)-E(1B1) a The following ab initio: abbreviations are CAS(2,2)/ROHF -884 used here: RMP2 = RMP2 -2149 E(3B1) and E(1B1) CAS(6,12) -972 obtained from SOCI(6,4,8) -970 restricted open shell f ll CI full -962 962 MP2 calculations; DFT: SOCI(n,a,e) = -931 Xα-SD second order CI LDA SD LDA-SD -840 840 calculation on an BP-SD -668 active space of a active orbitals, e empty orbitals with n active electrons (this corresponds to a CAS(n,a) calculation followed by a CISD calculation including e virtual orbitals. [(acac)2Cu-ONH2] Ψ Si l t t i l t separation Singlet-triplet ti (cm ( -1) computed t da for f [(acac) [( )2Cu−ONH C ONH2]. ] Method of Calculation E(3A’’)-E(1A”) ab initio: CAS(2,2)/ROHF -52.4 SOCI(0,2,94) -48.5 -50.3 SOCI(8,2,8)b DFT: -180 Xα -SD LDA-SD -171 BP-SD -169 Experiment -10 ÷ -70 a ψ = 90° was used in all the calculations. b The 18 orbitals included in the SOCI calculations are the valence 3d orbitals on copper and the valence 2p orbitals of the oxygen atom of the radical. Singlet-triplet separation (cm-1) computed for [(acac)2Cu−ONH2] at different ψ angles. angles ψ (deg) Method of calculation ab initio: CAS(2 2)/ROHF CAS(2,2)/ROHF DFT: Xα-SD LDA-SD BP-SD 0° 45° -53.0 53 0 -49.2 49 2 -184 -174 -173 -190 -179 -178 [Cu2Cl6]2- Variation of ϕ : 0° : Antiferromagnetic 45°:: Ferromagnetic 45 70°: Antiferromagnetic Singlet - triplet separation for [Cu2Cl6]2- in cm-1 ϕ (deg) Method of Calculation 0° 45° 70° -36 58 -87 -122 -362 -256 41 -160 35 0 ÷ -40 40 309 246 200 437 374 402 80 ÷ 90 -72 -191 -109 106 -45 145 ---- 1 ab initio : DFT : Xα -BS LDA-BS BP-BS Xα -SD LDA-SD BP-SD E Experiment i t 1 O. O 1 Castell, Castell J. J Miralles, Miralles R. R Caballol, Caballol Chem. Chem Phys Phys. 179 (1994) 377 ϕ = 45 ° ϕ = 0° ϕ = 70° Ti(CatNSQ)2 Computed Multiplet Structure for Ti(CatNSQ)2 and Sn(CatNSQ)2. ΔE State 3A 2 1A 2 1A 1 1A 1 Ti 0. Sn 0. 57 [56(1)] 57. 58 [23(1)] 58. 9510. 11934. 9571 9571. 11991 11991. a The energies are computed as differences from the ground 3A2 state. The experimental values are in squ are brackets. -1) for Ti(CatNSQ)2 and Sn(CatNSQ)2 Electronic Transitions Energies (cm Ti Transition 8a 2 → 19b2 States 3 A2→3B2 Computed Experimental 8078 80 8 9800 A2→3A2 A2→ B1 13645 16990 14200 17300/18600 A2→ A2 26868 23900 A2→ B1 3 A2→3A2 31096 28200/30400 39772 35900 States Computed Experimental A2→ B1 5768 9900 18b2 → 19b2 3 7a 2 → 19b1 3 8a 2 → 9a2 3 8a 2 → 20b1 18b1 → 20b1 3 3 3 3 Sn Transition 9 2 → 20b1 9a 3 19b2 → 20b1 3 A2→ A2 11038 14800 8a 2 → 20b1 20b1 → 10a 10 2 A2→ B1 3 A2→3B1 17018 17900/19200 31968 26000 A2→3A2 35371 27400/30900 9a 2 → 10a2 3 3 3 3 3 35100 BVD (biverdazyl diradical) Experimental Data: • Ground State S=0 Anti ferromagnetic Coupling • ΔE(T-S)= 760 cm-1(solution); 887 cm-1(crystal) QuickTime™ and a p Animation decompressor are needed to see this picture. Free Rotor Model. Exchange Coupling Constant ( J=E(T)-E(S) ) J (cm-1) vs αJ 1600 1400 1200 1000 Jcry s. B3LYP(vacuo) 800 Jsol. 600 400 200 0 0 00 0.00 BLYP 20 00 20.00 40 00 40.00 60 00 60.00 80 00 80.00 100 00 100.00 B3LYP(sol) (E0 − E j )/ kT ] [ ∑ j ΔJ j e Averaging over the vibrational motion J T = Jmin + j (E0 − E j )/ kT ] [ ∑e j In Vacuo CHCl3 34.6 566 585 721 26.8 757 774 844 αmin Jmin <Jmin> (0 K) <Jmin> (298 K) Experimental: Js=760 cm-1; Jc=887cm-1 For all the calculation: UB3LYP/6-31G* J in cm-1; Angles in degree. Perspectives TD-DFT Linear Response Theory DMRG Density Matrix Renormalisation Group TD-DFT Linear Response Theory TD KS eqs. vext (r,t , ) = vstat ( r) + vt (r ) ⋅ f (t ) The noninteracting system is defined by ⎛ ∇2 ⎞ ∂ ⎜ ⎟ i ϕ i (r,t ) = − + v KS ( r,t ) ϕ i (r, t ) ∂t ⎝ 2 ⎠ v KS (r,t ) = vext (r,t ) + ∫ d 3 r' ρ (r' ,t ) δAXC [ρ ] + r − r'' δ (r,tt ) δρ In the adiabatic limit (low frequency): AXC [ρ ] = ∫ dt E XC [ρ ]ρ =ρ (r,t ) Time-Dependent Response ⎡ ⎤ ρ (1) (r'',t' ) δ 2 EXC 3 3 (1) ρ (r,t ) = ∫ d r dt χ KS(t,t' ,r,r' ) vt (r' ) f (t' ) + ∫ d r' ' + ∫ d r'' (1) ρ ( r'',t' ) ⎥⎦ ⎢⎣ r'−r'' δρ (r') δρ (1) (r'') (1) 3 (1) ⎡ ⎤ δ2EXC 3 ρ (r'',ω) 3 (1) ρ (r,ω) =∫ d r χKS(ω,r,r') vt(r') f (ω)+∫ d r'' +∫ d r'' (1) ρ (r'',ω) (1) ⎢⎣ ⎥⎦ r'−r'' ' '' δ (r'')δρ δρ δ (r'''') (1) 3 To find a self-consistent solution for ρ(1) (r,ω) we use the parametrisation: ρ(1)(r,ω) = ∑[Piaσ (ω)ϕ*aσ(r)ϕiσ (r)+Piaσ (ω)ϕaσ (r)ϕi*σ(r)] iaσ The kernel or generalized susceptibility χKS(ω,r,r') is given as (cf. Casida) χ KS,σ ,σ ' (ω , r,r' ) = δ σσ ' (vt )iaσ ⎡ϕ i*σ (r )ϕ aσ (r)ϕ i σ ( r' ) ϕ *aσ (r' ) ϕ iσ (r) ϕ a*σ (r )ϕ i*σ (r' )ϕ aσ ( r' ) ⎤ − ∑ ω − ε − ε ω + (ε aσ − ε iσ ) ia ⎢ ( ) ⎣ ⎦⎥ aσ iσ = ∫ d 3 r ϕ i*σ ( r) vt (r) ϕ aσ (r ) [δ στ δ ijδ ab (ε aσ − ε iσ + ω ) + Kiaσ ,jbτ ]Pjbτ + K iaσ ,bjτ Pbjτ = − (vt )iaσ [δ στ δ ijδ ab (ε aσ − ε iσ − ω ) + Kiaσ ,bjτ ]Pbjτ + K iaσ , jbτ Pjbτ = −(vt )iaσ K klσ ,nmτ ∂vklCoul ∂vklXCσ σ = + ∂Pnmτ ∂Pnmτ = ∫ d rd 3 3 r' ϕ *kσ ⎛ 1 ⎞ * δ 2 E XC ⎟ ϕ nτ ( r' )ϕ mτ ( r' ) + (1) (r )ϕ l σ ( r) ⎜ (1) r − r' δρ r' δρ r' ' ⎝ ( ) ( )⎠ Pseudo-Eigenvalue Problem: (M. Casida) Ω(ω ) FI = ω I2 FI ΩI: Excitation Energy FI: Intensity ( )( ) Ω ijσ ,klτ (ω ) = δ στ δ ik δ jl (ε l τ − ε kτ ) + 2 ni σ − n jσ ε jσ − ε i σ K ijσ ,klτ (ω ) (nlτ − nk τ )(ε lτ − ε k τ ) 2 K klσ ,nmτ ∂vklCoul ∂vklXCσ σ = + ∂Pnmτ ∂Pnmτ = ∫ d rd 3 3 r' ϕ *kσ ⎛ 1 ⎞ * δ 2 E XC ⎟ ϕ nτ ( r' )ϕ mτ ( r' ) + (1) (r )ϕ l σ ( r) ⎜ (1) r − r' δρ r' δρ r' ' ⎝ ( ) ( )⎠ DMRG Density Matrix Renormalisation Group H = ∑ Tijc iσ c jσ + i, j,σ 1 ∑ Vijkl c iσ c jσ' ckσ' c lσ 2 ijklσσ' DMRG Unoccupied LUMO HOMO Occupied Test Calculation: H2O Software: • ADF • Gaussian94 • Home o e made ade So Software t ae Acknowledgements: Prof. V. Barone (Uni. of Naples) Prof. V. Bencini (Uni. of Florence) Dr K. Dr. K Doclo (Uni. (Uni Fribourg / ETHZ) Prof. P. Fantucci (Uni. of Milan) F. Totti (Uni. of Florence) J L B J.-L. Barras (Uni. (U i off F Fribourg) ib ) R. Bruyndonckx (Uni. of Fribourg) M. Buchs (Uni. of Fribourg) F. Mariotti (Uni. of Fribourg) Swiss National Science Foundation and COST Action D9.
© Copyright 2026 Paperzz