ORBITALLY DEPENDENT SUPEREXCHANGE IN MIXED-VALENCE CYANO-BRIDGED Mn(III)-Mn(II) DIMER. A NEW PERSPECTIVE FOR SINGLE MOLECULE MAGNETS A.V.Palii Institute of Applied Physics, Academy of Sciences of Moldova, Academy Str. 5, MD-2028 Kishinev, Moldova e-mail: andrew.palii@uv.es The model of the orbitally dependent magnetic exchange in the mixed-valence bioctahedral Mn(III)-CN-Mn(III) dimer is developed. The kinetic exchange mechanism involves the electron transfer from the single occupied t2 orbitals of the Mn(II) ion ( 6 A1 t 23 e 2 ground state) to the single occupied t2 orbitals of the Mn(III) 3 ground state) resulting in the charge transfer ion ( T 1 t 4 2 5 T2 t 22 e 2 Mn(III) 2 T2 t 25 Mn(II) state of the pair. The deduced effective exchange Hamiltonian leads to an essentially non-Heisenberg energy pattern. The energy levels are shown to be dependent on both spin and orbital quantum numbers providing thus the direct information about the magnetic anisotropy of the system. Along with the magnetic exchange the model includes the axial component of the crystal field and the spin orbit coupling operating within the ground 3T1(t24) cubic term of the Mn(III) ion. We have shown that under some conditions the interplay between these three interactions leads to the appearance of the barrier for the reversal of magnetization, so the results obtained can be regarded as a first step in the explanation of the magnetic bistability exhibiting by the recently synthesized trigonal bipyramidal cyanide cluster {[MnII(tmphen)2]3[MnIII(CN)6]2} (tmphen = 4,5,7,8-tetramethyl-1,10-phenantroline). 1. INTRODUCTION Molecules that exhibit magnetic bistability, commonly referred to as SingleMolecule Magnets (SMM), are of high interest due to their unusual physical properties and potential importance for high-density data storage and quantum computing (1). To date, almost all the molecules firmly established as displaying SMM behavior incorporate oxide-based bridging ligands that mediate the magnetic exchange coupling between metal centers. A remarkable feature of these systems is that in them all orbital angular momenta are quenched by the local low-symmetry crystal fields, so the oxobridged SMMs can be referred to as pure spin systems. Such molecules possess a large total spin ground state (S) formed by the isotropic Heisenberg magnetic exchange, which, when combined with a negative axial zero-field splitting DS S Z2 (DS <0), leads to the appearance of the energy barrier for spin reversal. Recently in the interest of producing clusters with larger spin reversal barriers, the trigonal bipyramidal cyano-bridged cluster [MnIII(CN)6]2[MnII(tmphen)2]3 (tmphen = 4, 5, 7, 8 – tetramethyl–1, 10–phenanthroline) was synthesized and characterized, ref. (2). The observed ac-susceptibility signal indicates that this cluster (hereunder abbreviated as Mn5-cyanide cluster) represents a new SMM. The Mn(III) ions in the Mn5-cyanide cluster occupy almost perfect octahedral sites. As a result the strong cubic crystal field produced by six carbon ions leads to the 3 - 95 orbitally degenerate 3T1 t 24 ground term of the Mn(III) ion. Insofar as this state possesses the unquenched orbital angular momentum the system under consideration is drastically different from the classical oxo-bridged SMMs consisting of orbitally non-degenerate ions. The first difference is that the conventional Heisenberg-DiracVan-Vleck Hamiltonian fails when one deals with the Mn5-cyanide cluster containing the Mn(III) ions with unquenched orbital angular momenta. In fact, the exchange Hamiltonian of the Mn5-cyanide cluster should involve not only spin but also orbital operators, i. e., such magnetic exchange proves to be orbitally dependent. The most important feature of the orbitally dependent exchange is that it is highly anisotropic, refs. (3-5). The second difference is a significant (first order) single ion anisotropy that can be expected in the Mn5-cyanide cluster. Such anisotropy represents a first order effect with respect to the spin-orbit coupling and axial component of the crystal field acting on each Mn(III) ion, i. e., this single ion anisotropy can not give rise to a global anisotropy described by the second order Hamiltonian DS S Z2 . One can expect that both exchange anisotropy and single ion anisotropy contributes to the global magnetic anisotropy responsible for the formation of the barrier for the reversal of magnetization in the Mn5-cyanide cluster. In this paper we endeavor to develop a model that would be able to qualitatively explain the existence of the barrier for the reversal of magnetization in the Mn5cyanide cluster. In order to avoid complications implied by the consideration of polynuclear the cluster entire and to make the results more clear and transparent we will restrict ourselves by considering the Mn(III)-CN-Mn(II) pair, that seems to retain the main peculiarities inherent in the entire Mn5-cyanide cluster. In fact, the analysis of the structure of the Mn5-cyanide cluster shows that only the superexchange interaction between Mn(II) and Mn(III) ions through the cyanide bridges can be significant, meanwhile the interactions between two Mn(II) ions and two Mn(III) ions are negligible due to the large intermetallic distances. The model includes the orbitally dependent superexchange mediated by cyanide bridge, as well as the spin-orbit coupling and axial crystal field operating within the ground 3T1 t 24 state of the Mn(III) ion. The results obtained in the framework of this model can be considered as the first step in the understanding of the magnetic behavior of the Mn5-cyanide cluster. 2. HAMILTONIAN OF THE KINETIC ORBITALLY DEGENERATE IONS EXCHANGE BETWEEN Let us consider the kinetic exchange between two octahedrally coordinated transition metal ions A and B assuming that both ions are in the ground states 2 s 1 2 s 1 ( A g A g d n A and B g B g d nB terms). We focus on the particular case when one or both ground terms are orbitally degenerate. Kinetic exchange appears as a second order contribution with respect to the following intercenter one-electron transfer operator playing a role of perturbation: V t Aγ A B γB γ , B γ B CB γBσ C Aγ Aσ h.c. , A A where the operator CB γBσ C Aγ Aσ (1) σ creates (annihilates) electron on the orbital φ B γB φ Aγ A of the ion B (A) with spin projection σ , A , B t 2 or e, γ A , γ B label the 3 - 96 one-electron basis, and t is the hopping integral. Hereunder we will consider a corner shared bioctahedral dimer and use the real one-electron cubic basis related to the cubic local coordinate frames, with zA (zB) axes being directed along C4 axis of the pair. This means that A and B will run over ξ yz , η xz, ζ xy (t2basis) and u 3z 2 r 2 , υ 3 x 2 y 2 (e-basis). The operator, eq. (1), connects the ground state with the excited charge transfer (CT) states arising from the electronic configurations d n A 1 d nB 1 in which one electron is transferred from the site A(B) to the site B(A). The general approach to the problem of the magnetic exchange between orbitally degenerate ions is outlined in our recent papers, refs. (3)-(5) and will not be repeated here. Application of this approach leads to the following general expression for the kinetic exchange Hamiltonian operating within the ground 2 s A g 1 2 s 1 A g d nA B g B g d nB manifold of the pair: H ex A, B 2 t γ A γ A A γ A A B γ A γ B γA γB γ γ γ , B γ B t B γ B , A γ A A A γ B γ B B γ B OAγ OBγ (2) F0 A , B F1 A , B s A s B . In eq. (2) γ A γ A A γA are the Clebsch-Gordan coefficients for the Oh group, Oi γ are the cubic irreducible tensor operators acting within the orbital i g i g manifold of the center i (i = A, B ) and s i are the single ion spin operators. The operators Oi γ are defined in such a way that their reduced matrix elements ig Oi ig ig 2 , 1 where i g is the dimension of i g , so the matrix elements of these operators coincide with the Clebsch-Gordan coefficients appearing in the Wigner-Eckart theorem , ref. (12). Finally, the parameters F0 and F1 are expressed in terms of the energies of the CT states, the general formulas for these parameters are given in ref. (4). 3. EXCHANGE MODEL FOR A Mn(III)-CN-Mn(II) PAIR Hereunder we will apply the general formalism outlined in the previous Section to the cyano-bridged Mn(III)-CN-Mn(II) pair in which the Mn(III) ion is surrounded by six carbon ions, and the Mn(II) ion is surrounded by six nitrogen ones. As a first step we assume that both metal ions are in a perfect octahedral ligand fields. The ground term of the Mn(III) ion in a strong cubic field produced by the carbon atoms is expected to be the low-spin orbital triplet 3T1(t24). On the contrary, weak crystal field induced by the nitrogen atoms gives rise to a ground high-spin orbital singlet 6 A1 t 23 e 2 of the Mn(II) ion, so the ground state of the bioctahedral corner-shared dimer (overall C4v symmetry) will be 6 A1 t 23e 2 Mn II 3T1 t 24 Mn III . In order to apply the general formula for the exchange Hamiltonian, eq.(2), to the Mn(III)Mn(II) pair we will assign the indices A and B to the Mn(II) and Mn(III) 3 - 97 ions, respectively, so that S Ag Ag 6A1 t23 4 A2 e 2 3 A2 6A1 t23e 2 and SB g B g 3T1 t24 . Insofar as the transfer of the electron from the site B to the site A leads to the CT states with very high excitation energies we will neglect such processes and consider only the A B electron transfer. There are two possibilities for the A B electron transfer, namely, the transfer from the single occupied t 2 orbitals of the Mn(II) ion to the single occupied t2 orbitals of the Mn(III) ion through the bonding and antibonding * orbitals of the cyanide ion, and the transfer from the single occupied e orbitals of Mn(II) to the empty e orbitals of Mn(III) through the cyanide -orbitals (the hopping parameters corresponding to the t 2 e transfer are expected to be negligible due to the orthogonality of t2 and e orbitals). At the same time recent density functional theory calculations of the exchange parameters in cyano-bridged species (6) demonstrated that the interaction through the the cyanide -orbitals was significantly smaller compared to the interaction through the and * orbitals (see also (7) and refs. therein). That is why only t 2A t 2B transfer processes are assumed to be important and will be taken into account in our exchange model. It is easy to see that the overlap between -type t2 orbitals of Mn(II) and Mn(III) through the and * orbitals of cyanide bridge is strong, and the same overlap takes place between orbitals. So there are two equivalent hopping parameters t t t associated to these overlaps (Fig.1). At the same time the integral t can be omitted because there is no effective overlap between orbitals. Note that the t 2A t 2B transfer can not affect X XB A ZB ZA Mn(I I) Mn(II I) Fig.1. Scheme of overlap between t2 orbitals of Mn(III) and Mn(II) through π orbitals of cyanide bridge the e 2 subshell of the ion A ( 3 A2 e 2 -state). At the same time this transfer decreases the spin of the ion A by 1 2 . The analysis of the Tanabe-Sugano diagrams (8) for d 5 and d 4 ions shows that the only appropriate state for the oxidized t 22 e 2 configuration of the ion A is the state 5T2 t 22 3T1 e 2 3 A2 5T2 t 22 e 2 . Analogously, the reduced t 25 configuration of the ion B gives rise to the only state 2T2 t 25 . We thus arrive at the conclusion that the t 2A t 2B transfer results in the only CT state 5T2 t 22 e 2 A 2 T2 t 25 B . It is remarkable that the single-ion states involved in this CT state are the “pure” 3 - 98 states, each resulting from the only electronic configuration. For this reason, no complications implied by the Coulomb mixing of different electronic configuration can appear in the kinetic exchange problem under consideration. The orbital schemes for the [Mn(II)]A[Mn(III)]B pair (ground state) and [Mn(III)]A[Mn(II)]B pair (CT state), and the electron transfer process connecting these states are shown in Fig.2. It is Excited (CT) to be noted that each orbital scheme depicts only one Slater determinant (microstate) of the B many-electron open shell waveu υ A e function. For example, the only u υ e determinant ξ η ζ ζ involved t2 ζ t2 η ξ ζ 5 T2 t22e2 2 in the two-determinant wave3 T1 t 24 , γ , ms 0 function η ξ T2 t25 1 Ground t2 ζ 6 υ υ u e B u e η ξ A1 t25e2 t2 ζ ξ 3 T1 t24 Mn(II) Mn(III) Fig.2. Image representation Gordan coefficients of the kinetic exchange mechanism for the Mn(III)-CN-Mn(II) pair ξηζ ζ of η is represented by the only microstate, so the corresponding orbital scheme in this case shows the full wavefunction of the high-spin Mn(II) ion. The jumping electron does not change its spin projection and selects the initial and final microstates as exemplified in Fig.2. Now it is a straightforward work to adapt the general expression for the kinetic exchange Hamiltonian, eq.(2), to the Mn(III)-CNMn(II) pair under consideration. Substituting the relevant values of the Clebschinto eq.(2) one finds: 4 H ex A, B t 2 FA10A1 t 2 ,t 2 OAB1 1 2 FA10E t 2 ,t 2 OEBu 3 4 t 2 FA11A1 t 2 ,t 2 OAB1 1 2 FA11E t 2 ,t 2 OEBu s A s B . 3 ξ η ζ ζ the low-spin Mn(III) ion is shown in Fig.2. On the contrary, the state 6 3 2 A2 t 2 e , ms 5 2 ξ η ζ u υ Mn(III) Mn(II) A 2 3 - 99 (3) While deducing eq. (3) it has been taken into account that A1 A1 A1 1 and A1 A1 γ 0 for all A1 , so O γ O 1 (only spin operators act within the A A A1 orbitally non-degenerate ground state of the Mn(II) ion). The orbital operators OAB1 and OEAu are represented by the following matrices in the cubic T1 basis α ,β ,γ : 1 0 0 OAB1 0 1 0 , 0 0 1 OEBu 12 0 0 0 0 . 1 0 12 0 (4) The parameters Fk in eq. (3) can be found with the aid of the approach developed in (3-5); the results are the following: FA10A1 t 2 ,t 2 F A1 E t 2 ,t 2 1 0 where ε A B FA11A1 t 2 ,t 2 1 , 2 ε A B 2 2 ε A B , F A1 E t 2 ,t 2 1 , 5 ε A B 1 1 is the energy of the 5 2 ε A B 6 A1 t 23e 2 A (5) , 3T1 t 24 B 5T2 t 22 e 2 A 2 T2 t 25 B excitation. The T-P-isomorphism makes it possible to consider the ground 3T1 t 24 term of the Mn(III) ion (ion B) as a state possessing fictitious orbital angular momentum lB 1 (8). This allows us to express the cubic irreducible tensor OEBu in eq. (3) in terms of the orbital angular momentum operator l BZ acting within the lB 1, m l B 0, 1 basis ( m l B is the projection of the fictitious orbital angular momentum) as follows: 2 O EBu 1 3 2 l ZB . (6) Substituting eqs. (5) and (6) into eq. (3) we arrive at the following final formula for the kinetic exchange Hamiltonian of the Mn(III)-CN-Mn(II) pair: t2 5 2 s A s B 2 3 l ZB2 . H ex A, B 30 ε A B (7) This Hamiltonian is essentially non-Heisenberg and includes both spin and orbital angular momenta operators (orbitally-dependent exchange). 4. RESULTS AND DISCUSSION The Hamiltonian, eq. (7), proves to be isotropic in the spin subspace and axially symmetric in the orbital subspace, so that S , M S (total spin of the pair and its 3 - 100 projection) and ml B are the good quantum numbers. The eigenvalues of H ex A, B are calculated as follows: E S , ml B 0 E S ,| ml B t2 63 4 S S 1 , 15 ε A B (8) t2 63 4 S S 1 . | 1 6 ε A B The energy pattern (formed by the magnetic exchange) of the Mn(III)-CNMn(II) pair contains two superimposed groups of the energy levels with ml B 0 and | ml B | 1 (see Fig.3). The total spin S of the pair takes the values S 3 2 , 5 2 , 7 2 , and the energy levels within each group obey the Lande’s rule. The exchange splitting of both ml B 0 and | ml B | 1 multiplet proves to be antiferromagnetic S gr 3 2 . The conclusion about the antiferromagnetic exchange splitting in each group of the energy levels is in agreement with the underlying ideas of Anderson, ref. (9) and Goodenough and Kanamori (see ref. (10) and refs. therein). In fact, these authors indicated that the electron hopping between the half-occupied orbitals should result in the antiferromagnetic exchange coupling. 0.5 t E 2 ε AB S 7 2 , ml B 0 , 1 0 S 5 / 2, ml B 0 0.5 S 3 / 2 , ml B 0 1 S 5 / 2, ml B 1 1.5 S 3 / 2, ml B 1 2 0.02 0.04 0.06 0.08 0.1 Fig. 3. Energy pattern formed by the magnetic exchange It is to be underlined that the Lande’s rule is not valid for the whole energy pattern, particularly, the non-monotonic alternation of the levels with S 3 2 and S 5 2 takes place. Another important result is that the energy levels depend not only on the total spin quantum number S but also on | ml B | . This leads to the interesting peculiarities in the magnetic behavior of the system. In the magnetic field applied parallel to the C4 3 - 101 axis of the bioctahedron the orbital contribution to the Zeeman splitting of the ground level is significant (first order effect) because the operator κ β l Z B H Z possesses the following nonvanishing matrix elements within the ground level: S 3 2 , M S , ml B 1 κ β l Z B H Z l Z B S 3 2 , M S , ml B 1 κ β H Z (9) In eq. (9) κ is the orbital reduction factor, sign “minus“ appears due to the fact that the matrix elements of l B within T1 and P bases are of the opposite signs, ref. (8). On the contrary, the orbital contribution to the Zeeman splitting of the ground level in a perpendicular field is much smaller because it appears as a second order effect due to the mixing of the ground level with the second excited level ( S 3 / 2 , ml B 0 ) by the operator κ β l X B H X lY B H Y (Van Vleck paramagnetism). Therefore, as distinguished from the Heisenberg magnetic exchange, the orbitally dependent exchange interaction described by the Hamiltonian, eq. (7), produces the strong magnetic anisotropy of the pair. Now in order to make our consideration more realistic we will take into account the fact that the nearest surrounding of the Mn(III) ion is axially distorted. In this case the operator of the axial crystal axial crystal field H ax acting on the Mn(III) ion should be added to the magnetic exchange Hamiltonian. The operator H ax can be defined as follows: H ax l Z2 B 1 , (10) where is the parameter of the axial field. This interaction splits the ground 3T1 term of the Mn(III) ion into the orbital doublet 3 E (orbital basis ml B 1 ) and the orbital singlet 3 B2 ( ml B 0 ) in such a way that the orbital singlet (orbital doublet) becomes the ground state providing 0 0 . Finally, one should also take into account the spin orbit (SO) coupling acting within the 3T1 term of the Mn(III) ion, the corresponding operator is given by H SO κ λ sB l B , (11) where λ 180 cm 1 is the many-electron SO coupling parameter for the 3T1 t24 term , ref. (11). We will consider the most typical situation when the axial field is strong significantly exceeding both the magnetic exchange and the SO coupling ( | | t 2 ε A B , κ | λ | ). Provided that 0 the strong axial crystal field totally removes the orbital degeneracy giving rise to the pure spin system with the Heisenberg-type pattern of the low-lying energy levels. We will not consider this trivial situation and focus on the analysis of more interesting 3 - 102 (from the point of view of the barrier for the reversal of magnetization) case when 0 . Strong negative axial field that results in a strong destabilization (by the value | | ) of the subset of the energy levels (formed by the magnetic exchange) with ml B 0 . The SO coupling splits the levels belonging to the low-lying subset with ml B 1 and mixes the levels with different S values belonging to this subset. In addition the SO coupling mixes the | MJ | ↓ 9/2 7/2 5/2 E κ| λ | 3/2 1/2 0 1/2 3/2 5/2 2 7/2 5/2 3/2 1/2 4 1/2 3/2 1/2 1/2 6 3/2 5/2 0.1 0.2 0.3 0.4 0.5 t 2 6 ε A B κ λ Fig.4. The low-lying energy levels of the Mn(III)-CN-Mn(II) pair in the limit of strong negative axial crystal field subsets of the levels with ml B 1 and ml B 0 , but in the strong axial crystal field limit such mixing can be neglected, and the effective SO coupling operator acting within the low-lying group of levels with ml B 1 becomes axial ( H SO κ λ s ZB l ZB ). The low-lying energy levels as the functions of the magnetic exchange are shown in Fig. 4. One can see that providing t 2 ε A B 0 (exchange interaction is switched off) the energy pattern contains three equidistant levels with the κ | λ | (eigenvalues of the operator κ λ sZB lZB ). When energies κ | λ | , 0 and 3 - 103 the exchange interaction is switched on ( t 2 ε A B 0 ) the energies of the levels become dependent on M J , where M J ms A ms B ml B is the projection of the total angular momentum ( ms A and ms B are the spin projections of the ions A and B ). It should be emphasized that the energies of three low-lying levels (with M J 5 2 , M J 3 2 and M J 1 2 ) monotonically increases with the decrease of M J , so these three levels form the barrier for the reversal of magnetization. The magnitude of this barrier monotonically increases with the increase of the exchange interaction. These results qualitatively explain the formation of the potential barrier in the entire Mn5-cyanide SMM. The application of the ideas of this paper to a quantitative description of the SMM properties of the Mn5-cyanide cluster will be considered in the forthcoming publications. ACKNOWLEDGMENTS The author thanks Professor B. S. Tsukerblat for many useful discussions on the subject of the paper. REFERENCES 1. Gatteschi D., Sessoli R.: Quantum Tunneling of Magnetization and Related Phenomena in Molecular Materials. Angew. Chem. Int. Ed. 2003 42, 269-297. 2. Berlinguette C. P., Vaughn D., Cañada-Vilalta C., Galán-Mascarós J.-R., Dunbar K. R.: A Trigonal-Bipyramidal Cyanide Cluster with Single-Molecule-Magnet Behavior: Synthesis, Structure, and Magnetic Properties of II III {[Mn (tmphen)2]3[Mn (CN)6]2}. Angew. Chem. Int. Ed. 2003 42, 1523-1526. 3. Borras-Almenar J. J, Clemente-Juan J. M., Coronado E., Palii A. V., Tsukerblat B. S.: Magnetic exchange interaction in a pair of orbitally degenerate ions: Magnetic anisotropy of [Ti2Cl9]3-. J. Chem. Phys. 2001 114 (3), 1148-1164. 4. Borras-Almenar J. J, Clemente-Juan J. M., Coronado E., Palii A. V., Tsukerblat B. S.: Magnetic exchange interaction in clusters of orbitally degenerate ions. I. Effective Hamiltonian. Chem. Phys. 2001 274, 131-144. 5. Borras-Almenar J. J, Clemente-Juan J. M., Coronado E., Palii A. V., Tsukerblat B. S.: Magnetic exchange interaction in clusters of orbitally degenerate ions. II. Application of the irreducible tensor operator technique. Chem. Phys. 2001 274, 145-163. 6. Nishino M., Yoshioka Y., Yamaguchi K.: Effective exchange interactions and magnetic phase transition temperatures in Prussian blue analogs: a study by density functional theory. Chem. Phys. Lett. 1998 297, 51-59. 7. Weihe H., Güdel H.U.: Magnetic exchange across the cyanide bridge. Comments Inorg. Chem. 2000 22 (1-2), 75-103. 8. Sugano S., Tanabe Y., Kamimura H.: Multiplets of transition-metal ions in crystals. New York and London, Academic Press, 1970. 9. Anderson P.W., in: Magnetism, G. T. Rado, H. Suhl, Eds. (New York, Academic Press, 1963). 10. Goodenough J. B.: Magnetism and Chemical Bond. New York, Interscience, 1963. 11. Abragam A., Bleaney B.: Electron Paramagnetic Resonance of Transition Ions. Clarendon, Oxford, England, 1970. 3 - 104