470 likes | 598 Views
Calculation of Magnetic Anisotropies and Exchange Hamiltonians in Molecules with Density Functional Theory Mark R. Pederson Washington DC. Collaborators. Jens Kortus Max-Planck-Institute Stuttgart Noam Bernstein Naval Research Laboratory Tunna Baruah NRL/Georgetown/Howard
E N D
Calculation of Magnetic Anisotropies and Exchange Hamiltonians in Molecules with Density Functional Theory Mark R. Pederson Washington DC Collaborators Jens Kortus Max-Planck-Institute Stuttgart Noam Bernstein Naval Research Laboratory Tunna Baruah NRL/Georgetown/Howard Kyungwha Park NRL/Georgetown/Howard Stephen Hellberg Naval Research Laboratory Shiv Khanna Virginia Commonwealth University Supported by ONR, DOD HPCMP
NRLMOL + DFT FOR MOLECULAR MAGNETS Place Gaussians on each atom in molecule or crystal Yi(r) = Si Ci exp[-ai(r-Ri)2] |) DENSITY FUNCTIONAL FORMALISM USED (PBE GGA) Reduce Problem to Finding Expansion Coefficients
Exchange Coupling Parameters within Density Functional Theory? Small Anisotropy Example: V15 E1=-JS2 E2= JS2
Heisenberg Hamiltonian within DFT: [V15As6O42(H2O)] K6 [Kortus, Hellberg, Pederson PRL 86, 3400 (2001) ] J J’ J’’ Expt. DFT+ Heisenberg J2 J1 J1 J2 J’’ J J’ • Electronic Structure • Spin Ordering • Exchange Parameters Couple NRLMOL and many-spin Heisenberg Hamiltonian HF = lF Effective Moment vs Temperature Convergence? Diagonalize Many-Spin Hamiltonian for excitation spectra.
O C N H DFT Exchange Parameters: Mn12 is a classical Ferrimagnet (?!) Majority Spin Electrons Minority Spin Electrons = Mn Mn12O12(RCOO)16(H2O)4 1.5 nanometers • Interesting Deviations from classical bar magnet are due to: • “Large” Quantum Mechanical Effects (Defines Ideal Behavior) • “Small” Quantum Mechanical Effects (Complicates Behavior)
Mn3+ O2- Mn3+(S=2) Mn4+(S=3/2) Intramolecular exchange interactions:Mn12-acetate • S4 symmetry+3 inequivalent Mn sites : J1, J2, J3, J4 • Consider low-energy collinear spin-excitations • Use optimized ground-state geometry by DFT unknowns
Mn3+(S=2) Mn4+(S=3/2) DFT calculated exchange constants: Mn12-acetate • J1 , J2 : dominant, antiferromag. DFT: K. Park et al., PRB (2004) Ref.A: Regnault et al., PRB (2002). Qualitative Agreement with Experiment (50 %)
Spin excitation energetics • Heisenberg Hamiltonian DFT determined • Diagonalize H with DFT-determined Jij to find ground state & excited spin multiplets [ Lanczos method, Hellberg et al., JPSJ (1999). ] • Dimension of Hilbert space for Mn12-acetate: 108
S=9 33 K S=9 7 K S=9 41 K S=10 Calculated ground state & low-lying excited spin multiplets (Mn12-acetate) Park, Hellberg, Pederson, PRB (2004) Regnault et al., PRB (2002) Petukhov et al., PRB (2004) (2S+1)-fold degeneracy Experimental energy gap=35-40 K (Hill et al) “Classical” Ferrimagnet Configuration has amplitude of 0.6. Large but slightly smaller than that deduced from Godel et al
e- Energy Electric Field Electron Velocity e- Magnetic Field -S 0 +S Magnetic Anisotropy in Nanomagnets Electric Field Caused by Nuclei and Electronic Density Velocity Determined from Momentum Operator Computing Magnetic Effects due to Spin-Orbit Coupling Possible within Density-Functional Theory (Van Vleck 1937)
Energy/A <Sz> (M) Anisotropy HamiltonianPederson and Khanna PRB 1999 Effect on total energy due to spin-orbit L.S term Dependent on axis of spin quantization |1)= cos(/2) |) + eisin(/2) |) |2)=-eisin(/2) |) + cos(/2) |) To lowest order: (2nd order perturbation in L•S) Determine gab from DFT D2=Sabgab <Sa><Sb> DE2= -DSzSz - E(SxSx-SySy)
Q MAGNETOMOLECULAR ANISOTROPY ENERGYDE ~(1/4C4 ) M2 Spin Orbit Energy (Q) E= + DFT Energy ME’s (Q) = C4 (3.5x108)
NRLMOL DIRAC EXPT Kr 3d 1.282 1.303 Kr 3p 7.551 7.883 Kr 2p 50.97 53.43 Mn 2p 10.3 11-12 Ru 2p 121 125 Accuracy of Non-Relativistic DFT for Spin-Orbit? GAS PHASE Mn[N-(CN)2]2 Molecular Magnetic Material M.R. Pederson, A.Y. Liu, T. Baruah, E.Z. Kurmaev, A. Moewes, S Chiuzbidian, M. Neuman, C.R. Kmety, K.L. Stevenson and D.Ederer, Phys. Rev. B 2002
Density of States for Passivated Mn12O12 Magnet Minority Gap: 2.03 eV Majority Gap: 0.43 eV • No Mn (4s) • S=10 • Ferrimagnetic Expt: Gaps at 1.08 and 1.75 eV [Oppenheimer et al, PRB 65 05449 (2002)]
Second Order Molecular Magnetic Anisotropy BarrierTheory vs. Experiment All Electron GGA (NRLMOL): 55.7 K Expt. (Barra et al,Fort et al) : 55.6 K
How does Magnetization Barrier Depend on Addition of A single Electron?
Mn Mn Mn O2- Mn Mn Mn Projected Onsite Anisotropies? T. Baruah et al, CPL 360144 (2002) Eg T2g Eg ~-1 Kelvin T2g ~10 Kelvin Majority Spin Insertion should decrease anisotropy Local JT distortion
S=13.0 S=11.5 S=10.5 EF S=10.0 Barrier (55K) RBM: 10.3K SCF: 7.2K Mn8Fe4-Acetate: Additional Majority Spin Electrons Should Reduce Barrier? Rough Prediction for Mn12O12+14: U~31K Good Agreement with Experiment…but situation is more complicated (See Park et al, PRB 2004)
Spin Manifolds inMn12-Acetate Park, Pederson and Hellberg PRB 69 014416 (2004) Sz Energy (eV) MAE (K) 10 0.000 54.21 9-b 0.062 54.56 9-c 0.145 54.98 8 0.138 55.09 6-c 0.038 55.03 6-b 0.080 55.35 5-a 0.134 55.52 5-b 0.092 54.92 13 0.151 53.70 N.B. None of these states are eigenstates of S^2 ! Above states are NOT connected by one and two electron operators
DFT Prediction: Spin Ordering, Magnetic Anisotropies and Resonant Tunneling of Magnetization in Nanomagnets. 2nd-Order Spin Orbit Energy Depends on Quantization Axis D2=-DCOS2Q + ESIN2QCOS(2b) Fe4 GGA-DFT EXPT Mn12 (D) 55 55 Mn12 (E) 0 0 Fe8 (E) 5.4 5.5 Fe8 (D) >50 29 Cr (D) 5.6 6.0 Mn10 (D) 9.5 8.0 Fe4 (D) 14.0 14.3 Fe4 (E) 1.6 1.4 Co4 (D) 27 ~100 Blind Test Mn12 Mn10 Fe8 Cr Co4
How does Magnetization Tunneling Rate Depend on Chemical Environment?
D H= -DSzSz + E(SxSx-SySy) + O(SxSx SxSx + SySy SySy) +…. Higher Order Effects Lead to Tunnel Splittings E: Dislocations, Solvent Disorder, Spin Vibron O: Spin Vibron, Higher Spin-Orbit, NCM Continuous Range of E’s would explain experimental observation
Symmetry Breaking Due to Solvents Leads to Transverse 2nd Order Terms (Cornia model) Six isomers of Mn12-acetate DFT calculations give maximum value of E~0.016 K Accord with Expt Park et al Phys. Rev. B 69 144426 (2004).
Does the Correct Electronic Structure guarantee understanding of Magnetic Anisotropy?
O C N H Fe8: A problem case Theory and Experiment disagree on Anisotropy Hamiltonian by a factor of two Fe8-TACN
Electronic Structure of Fe8 Theory + Experiment: Good Agreement (with Baruah, Musfeldt, Dalal and coworkers)
Conclusions • Good Agreement with Experiment • Some Verified Predictions within DFT …Still Lots of Questions Remain
4th-order = + Total 2nd-order 4th-Order Anisotropy (responsible for tunnel splittings) • Higher order terms in L•S: • exact electronic (non-self-consistent) total energy with L•S • coupling of spins to vibrations DE4=G SzSzSzSz + H [SxSxSxSx +SySySySy] = A1(4)[S2(Sz2-S2/3)] + A2(4)[3S4+35Sz4+30S2Sz2] + B1(4)[Sx4+Sy4-6Sx2Sy2] + … } cubic harmonics Can have different angular dependence and different scaling with 1/[speed of light]
Vibrational Contribution to Magnetic Anisotropies. Pederson, Bernstein and Kortus PRL 89 097202 2002 • Spin Orbit Interaction Depends on Electric Fields and Kohn-Sham Orbitals • Electric Fields and Kohn-Sham Orbitals depend on Atomic Positions/ Vibrational Displacements • Zero Point Energy of a Vibrational Mode Changes as a function of Spin Projection due to spin-orbit-vibron coupling. • Lowest-Order effect is 1/[Speed of Light]8
SPIN-ORBIT VIBRON INTERACTION [P2+w2Q2]/2 + gzz Sz2 +QSab (dgab/dQ)SaSb |Y=|f |SM E=w/2 + gzz M2 - (A+BM2)2/(2w2) S(S+1)[d/dQ(gxx+gyy)]/2 d/dQ[gzz -(gxx+gyy)/2] Compute total energy, forces, gab for all atomic displacements Extract: vibrations (IR, Raman) from dynamical matrix vibration-spin coupling from d/dQ(gab) Difficult calculations (shortcuts): must compare to experiment!
Calculated Total and Infrared DOS - Mn12-Acetate Large Mn-crown contributions in Region identified as field-dependent in IR experiments. (Sushkov et al)
TOTAL AND PROJECTED RAMAN INTENSITIES Good agreement between predicted Raman and recent experimental measurements (North et al) Raman Intensity Vibrational Energy (1/cm)
EVOLUTION OF 4TH-ORDER ENHANCEMENT WITH COUPLING TO VIBRONS Total and Mn projected weight NRLMOL+GGA: 6K - 1K = 5K EXPERIMENT: 5 – 10K
Results from Spin-Orbit Vibron Calculation Barra et al, PRB 56 8192 (1997) • 4th-Order Angular Terms: S4sin4q [G cos(4f)+H sin(4f)] • Demonstrated Mechanism for Isotope Induced Tunnel Splittings Co4: 4th-order barrier is 3% of 2nd-order barrier
U = E S2COS(2f) + GS4(COS4f+d) Spin Vibron X 10 SPIN-VIBRON MAX AT: 50, 140 SOLVENT DISORDER MAX AT:-15,165 Classical Energy (eV) Solvent Disorder Angle in XY Plane (degrees) Incommensurate Principal Axes From Solvent Disorder and Spin Vibron Maxima out of phase by ~20 or ~70 degrees (+/-10) Qualitative Accord with Experiment [del Barco et al, PRL 047203 (2003)]
Total Spin Local Spins Interference? Anisotropy Mn12: S=10 S=3/2 and S=2 Constructive Large Mn10: S=12/13 S=5/2 Destructive Small Mn12 vs Mn10?
Energy/A <Sz> (M) NANOSCALE MOLECULAR MAGNETS LONGITUDINAL FIELD Classical Barrier Hopping vs. Resonant Tunneling of Magnetization } Barrier changes continuously with Bz field Yellow States Aligned with Blue States only if: Bz = [DN]D/2 Discontinuous changes of demagnetization rate at integer fields due to RTM Observable in Hysterisis curves DW = MB - DM2/2 AKA: Zero-Field Splittings in atomic physics/radical chemistry
Co4 Based Molecular Magnet Baruah and Pederson, CPL 2002 Magnetic moment:12 mB Local Co moment:3 mB Addition of 4 hydrogens reduced moment by4 mB Anisotropy varies strongly with molecular distortions (20-60K) Uniaxial alignment Global easy axis along Z Lowest-energy staggered structure Good Agreement with more recent experiments.
Co-d Minority Minority Spin Co(3d) has 2 electrons } Co -d Majority Majority Spin Co(3d) Full DOS (Arb. Units) Minority Majority Energy (eV) Spin Projected Total and Cobalt Density of States
N Soft O Co Cl Hard Hard Medium Hard Easy Barrier = 23K Local Magnetic Anisotropy axes Global Anisotropy axes Orthogonal alignment of local hard axes results in a uniaxial system.
3 Equivalent Expressons for Spin-Orbit Coupling Classical and Quantum Mechanical Most Common Most Straightforward (Kittel, Schiff?)
Angle of Applied Transverse Field Log( Tunnel Splitting) (M=5) Applied Magnetic Field Lower Symmetry Molecules (Fe8): Berry’s Phase Oscillations A. Garg: EPL 22, 205 (1993) Period in transverse tunneling Experiments (kB/gmB)2[2E(E+D)]1/2 H=-DSzSz - E(SxSx-SySy) +HxSx