We report transport measurements through a single-molecule magnet, the Mn 12 derivative Mn 12 O 12 O 2 C-C 6 H 4 -SAc 16 H 2 O 4 , in a single-molecule transistor geometry. Thiol groups connect the molecule to gold electrodes that are fabricated by electromigration. Striking observations are regions of complete current suppression and excitations of negative differential conductance on the energy scale of the anisotropy barrier of the molecule. Transport calculations, taking into account the high-spin ground state and magnetic excitations of the molecule, reveal a blocking mechanism of the current involving nondegenerate spin multiplets.

We consider transport through a single-molecule magnet strongly coupled to metallic electrodes. We demonstrate that for half-integer spin of the molecule electron-and spin-tunneling cooperate to produce both quantum tunneling of the magnetic moment and a Kondo effect in the linear conductance. The Kondo temperature depends sensitively on the ratio of the transverse and easyaxis anisotropies in a non-monotonic way. The magnetic symmetry of the transverse anisotropy imposes a selection rule on the total spin for the occurrence of the Kondo effect which deviates from the usual even-odd alternation.PACS numbers: 72.10. Fk, 75.10.Jm, 75.30.Gw, 75.60.Jk Introduction. Single-molecule magnets (SMMs) such as Mn 12 or Fe 8 have been the focus of intense experimental and theoretical investigation [1]. These molecules are characterized by a large spin (S > 1/2), easy-axis and transverse anisotropies, and weak intermolecular interaction. Molecular-crystal properties are due to an ensemble of single molecules and exhibit quantum tunneling of magnetization (QTM) on a mesoscopic scale. Recently, a single molecule magnet (Mn 12 ) was trapped in a nanogap [2,3] and fingerprints of the molecular spin were observed in electron transport. Furthermore, transport fingerprints of QTM were predicted [4] when the individual excitations can be resolved by the temperature. Using easy-axis anisotropy for magnetic device operation was also proposed [5]. These works focused on the regime where single electrons charge and discharge the molecule through weak tunneling. In this Letter we investigate linear transport through a half-integer spin SMM deep inside the blockade regime [6] where the charge on the molecule remains fixed. A strong tunnel-coupling to the metallic electrodes induces spin fluctuations and allows the magnetic moment to tunnel. This is remarkable, since for an isolated SMM with half-integer S this is forbidden by time-reversal (TR) symmetry. At the same time, the resonant spinscattering allows electrons to pass through the SMM: the Kondo effect for transport [7,8] results in a zero bias conductance anomaly that has been studied experimentally in many systems with small spins (e.g. quantum dots [9,10,11,12,13] and single molecules [14,15]). Such an effect is unexpected in SMMs because the S > 1/2 underscreened Kondo effect is suppressed by the easyaxis anisotropy barrier which freezes the spin along the easy axis. However, we find that even a weak transverse anisotropy induces a pseudo-spin-1/2 Kondo effect. The corresponding Kondo temperature is experimentally accessible due to a compensation by the large value of the physical spin S. We perform a scaling analysis [16] for the effective pseudo-spin-1/2 model and verify the results by a non-perturbative numerical renormalization group (NRG) calculation [17,18] for the full large-spin Hamiltonian.Model. We consider SMMs which can be described by the following minimal model in the limit of strong tunnel-

We demonstrate that in a single molecule magnet (SMM) strongly coupled to electrodes the Kondo effect involves all magnetic excitations. This Kondo effect is induced by the quantum tunneling of the magnetic moment (QTM). Importantly, the Kondo temperature TK can be much larger than the magnetic splittings. We find a strong modulation of the Kondo effect as function of the transverse anisotropy parameter or a longitudinal magnetic field. For both integer and half-integer spin this can be used for an accurate transport spectroscopy of the magnetic states in low magnetic fields on the order of the easy-axis anisotropy parameter. We set up a relationship between the Kondo effects for successive integer and half-integer spins.

we investigated the involvement of excited states in the Kondo effect in half-integer spin single-molecule magnets (SMM) in the limit of strong exchange interaction J. Unfortunately, the numerical results of this Letter are not correct. In particular, in the exchange interaction JS Á s, coupling the SMM to the first site on the Wilson chain in the NRG algorithm, the term JS z s z was inadvertently implemented incorrectly. As a result we did not account for matrix elements of the z component of the SMM spin operator hijS z jji with i Þ j, while accounting for those with i ¼ j. Here jii, jji denote eigenstates of the SMM part of our HamiltonianThus, the NRG algorithm was applied correctly but using incorrect matrix elements, thereby not capturing the transitions into SMM excited states. This only leads to incorrect results when two conditions are met: (i) the excited states of the SMM are relevant; i.e., J is large enough compared to the largest anisotropy splitting Á ¼ ð2S À 1ÞD, and additionally (ii) there is transverse anisotropy, as in the Letter in question (B 2 ), or a transverse magnetic field, as in [2]. Both conditions (i) and (ii) have to apply at the same time for quantitative errors to occur. Importantly, in the limit where J is sufficiently weak compared to Á, where the Kondo effect in SMM was first studied [3], condition (i) does not apply and all results reported there are correct. This we checked by explicit recalculation of the results presented in [3] with the corrected code. Also, for large magnetic field H z ) D, B 2 the anisotropy becomes unimportant and the eigenstates approach spin eigenstates and the correct suppressed and split Kondo peaks are found. For the above reasons the problem could not be detected in the numerous checks we performed against known results for the Kondo effect for various spin S in a magnetic field but without magnetic anisotropy.Recalculation of the results of [1] confirm one of the central conclusions of that Letter, namely, that the excited states are involved in the Kondo effect. In Fig. 1 we show the dependence of the Kondo temperature on the transverse anisotropy B 2 , correcting Fig. 2 in Ref. [1]. As the exchange interaction J is increased T K grows. However, the correct order of magnitude of T K is much smaller than previously reported [1]. The discrepancy with the correct results in fact shows that for the parameters chosen the excited states are important even though the Kondo scale T K does not exceed the excitation energy Á. This implies that the two level picture which was valid in [3] does not apply for these parameters. Moreover, Fig. 1 shows no oscillatory dependence of T K on the magnetic anisotropy parameter B 2 as shown in Ref. [1]. We note that the cross check on Fig. 2(a) reported in Fig. 2(b) of Ref.[1] suffers from the same problem since the incorrect coupling matrix elements are used there as well (there was no problem with the NRG algorithm).In Fig. 2 we show the dependence of the spectral function on the external longitudinal magnetic field ...

We analyze a model for a metal-organic complex with redox orbitals centered at both the constituent metal ions and ligands. We focus on the case where electrons added to the molecule go onto the ligands and the charge fluctuations on the metal ions remain small due to the relatively strong Coulomb repulsion. Importantly, if a nonzero spin is present on each metal ion it couples to the intramolecular transfer of the excess electrons between ligand orbitals. We find that around special electron fillings, addition of a single electron switches the total spin S tot = 0 to the maximal value supported by electrons added to the ligands, S tot =3/2 or even S tot =7/2 for metal ions with spin 1/2. This charge sensitivity of the molecular spin is a strong correlation effect due to the Nagaoka mechanism. Fingerprints of the maximal spin states, as either ground states or low-lying excitations, can be experimentally observed in transport spectroscopy as spin blockade at low bias voltage and negative differential conductance and complete current suppression at finite bias, respectively.