Magnetic Anisotropy of Transition Metal Complexes Nicholas F. Chilton
A thesis submitted to e University of Manchester for the degree of Doctor of Philosophy in the Faculty of Engineering and Physical Sciences
2015 School of Chemistry e University of Manchester
2
Contents
Contents ...... 3
List of abbreviations ...... 5
Abstract ...... 7
Declaration ...... 9
Copyright statement ...... 11
Acknowledgements ...... 13
1. Preface ...... 15
2. Introduction ...... 17
Rationale for the alternative format and organization of thesis ...... 17
Brief of included works and roles of authors ...... 17
Magnetically anisotropic materials ...... 21
3. The electronic and magnetic properties of molecules ...... 23
Magnetic properties of the electron ...... 23
Quantum mechanics and the wavefunction ...... 24
A single electron: the hydrogen atom ...... 26
Multi-electron atoms ...... 28
Magnetic and relativistic effects: the Dirac equation ...... 29
Energy scales ...... 30
Linear algebra ...... 32
Perturbation theory ...... 34
The spherical harmonics ...... 36
More than one way to skin a cat ...... 37
3
The parametric approach: angular momentum in quantum mechanics ...... 38
The ab initio approach ...... 42
Connection of the ab initio and parametric approaches ...... 45
Experimental observables ...... 47
Conclusion ...... 48
References ...... 48
4. Paper one: “On the Possibility of Magneto-Structural Correlations: Detailed Studies of
Dinickel Carboxylate Complexes” ...... 51
5. Paper two: “Large Zero-Field Splittings of the Ground Spin State Arising from
Antisymmetric Exchange Effects in Heterometallic Triangles” ...... 65
6. Paper three: “An electrostatic model for the determination of magnetic anisotropy in dysprosium complexes” ...... 79
7. Paper four: “The first near-linear bis(amide) f-block complex: a blueprint for a high temperature single molecule magnet” ...... 111
8. Paper five: “Design Criteria for High-Temperature Single-Molecule Magnets” ...... 131
9. Paper six: “Direct measurement of dysprosium(III)···dysprosium(III) interactions in a single-molecule magnet” ...... 149
10. Conclusion ...... 185
11. Outlook ...... 187
12. Appendix: Further Ph.D. publications ...... 189
Total word count: 55,000
4
List of abbreviations
AO Atomic Orbital
BO Born-Oppenheimer
CASPT2 Complete Active Space Second Order Perturbation Theory
CASSCF Complete Active Space Self-Consistent Field
CF Crystal Field
CI Configuration Interaction
EPR Electron Paramagnetic Resonance
HF Hartree-Fock
LCAO Linear Combination of Atomic Orbitals
MCSCF Multi-Configurational Self-Consistent Field
MF Mean Field
MO Molecular Orbital
NEVPT2 N-Electron Valence State Perturbation Theory
PT Perturbation Theory
QM Quantum Mechanics
RASSCF Restricted Active Space Self-Consistent Field
RASSI Restricted Active Space State Interaction
RS Russell-Saunders
SCF Self-Consistent Field
SMM Single Molecule Magnet
SOC Spin-Orbit Coupling
ZFS Zero Field Splitting
5
6
Abstract
Magnetic Anisotropy of Transition Metal Complexes: a thesis submitted to the University of Manchester for the degree of Doctor of Philosophy in the Faculty of Engineering and Physical Sciences.
The study of magnetic anisotropy in molecular systems permeates the physical sciences and finds application in areas as diverse as biomedical imaging and quantum information processing. The ability to understand and subsequently to design improved agents requires a detailed knowledge of their fundamental operation. This work outlines the background theory of the electronic structure of magnetic molecules and provides examples, for elements across the Periodic Table, of how it may be employed to aid in the understanding of magnetically anisotropic molecules.
II III II The magnetic anisotropies of a series of dimetallic Ni 2 complexes and a Ru 2Mn triangle are determined through multi-frequency Electron Paramagnetic Resonance (EPR) spectroscopy and ab initio calculations. The magnetic anisotropy of the former is found to be on the same order of magnitude as the isotropic exchange interactions, while that of the latter is found to be caused by large antisymmetric exchange interactions involving the RuIII ions.
An intuitive electrostatic strategy for the prediction of the magnetic anisotropy of DyIII complexes is presented, allowing facile determination of magnetic anisotropy for low symmetry molecules.
Through the presentation of the first near-linear pseudo-two-coordinate 4f-block complex, a new family of DyIII complexes with unprecedented Single Molecule Magnet (SMM) properties is proposed. Design criteria for such species are elucidated and show that in general any two-coordinate complex of DyIII is an attractive synthetic target.
The exchange interaction between two DyIII ions is directly measured with multi- frequency EPR spectroscopy, explaining the quenching of the slow magnetic relaxation in the pure species compared to the SMM properties of the diluted form. The interpretation of this complex system was achieved with supporting ab initio calculations.
Nicholas F. Chilton
February 2015
7
8
Declaration
No portion of the work credited to the author in this thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.
9
10
Copyright statement
The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The
University of Manchester certain rights to use such Copyright, including for administrative purposes.
Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made.
The ownership of certain Copyright, patents, designs, trade marks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such
Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual
Property and/or Reproductions.
Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or
Reproductions described in it may take place is available in the University IP Policy (see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=487), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.manchester.ac.uk/library/aboutus/regulations) and in The
University’s policy on Presentation of Theses.
11
12
Acknowledgements
Firstly I would like to thank The University of Manchester for a President’s Doctoral
Scholarship, which has given me the opportunity to perform this work. Furthermore, none of this would have been possible without the support, assistance and advice from my supervisors Prof. Eric McInnes and Prof. Richard Winpenny. I am also extremely grateful for all the sage wisdom provided by Prof. David Collison.
A large debt of thanks is owed all my co-authors and all those in the Molecular
Magnetism group, but specifically to Dr. Andrew Kerridge, Dr Richard Layfield, Dr.
Samantha Magee, Dr. David Mills, Dr. Eufemio Moreno Pineda, Dr. Thomas Pugh, Dr.
Alessandro Soncini, Dr. Floriana Tuna and Dr. James Walsh for all their hard work and advice.
I would also like to acknowledge Prof. Keith Murray, without whom I would never have been here in the first place.
Enormous thanks are due to my friends and family for all their love, support and encouragement over the years. Finally and especially, all my love and thanks to Rachael for always being by my side and for understanding.
13
14
1. Preface
The development of parametric models during the 1950’s to the 1970’s for the interpretation of magnetic and spectroscopic data of high symmetry transition metal sites relied on a sound understanding of Quantum Mechanics (QM) and group theory.
These models, largely based around Crystal Field (CF) theory, were very successful in describing the experimental results in a rigorous theoretical framework. The renaissance of 4f-based complexes in molecular magnetism in the new millennium brought with it the use of such parametric CF approaches to model the magnetic data and to understand the electronic structure. However, many of these works were riddled with oversimplifications based on ‘pseudo-symmetric’ environments which were in fact devoid of any symmetry elements. These inappropriate symmetry restrictions resulted in qualitatively incorrect conclusions in many cases, contributing no insight into the origin of the magnetic anisotropy. It was not until the arrival of sophisticated ab initio computational approaches that the treatment of low symmetry lanthanide complexes was remedied. However, unsurprisingly, these complex calculations provided complex answers which were unable to be simply understood.
Despite the plethora of dubious magnetic interpretations, much progress was made in enhancing the barrier to magnetic reversal in lanthanide-based SMMs. However, after the initial flood of new and interesting results, there was no clear direction for how more substantial progress could be made. This stemmed from a lack of understanding of the origin of the magnetic anisotropy in these systems and how it could be controlled.
While it was largely understood that rigorous high symmetry would define and allow prediction of magnetic anisotropy, there was no real grasp on how to engineer the magnetic anisotropy in the much more common cases of low symmetry.
The work presented herein provides some answers to the challenges posed above and begins to suggest how the magnetic anisotropy can be controlled in low symmetry environments.
15
16
2. Introduction
Rationale for the alternative format and organization of thesis
The work undertaken during the course of this Ph.D. was of a topical nature which required rapid dissemination in peer-reviewed literature. Therefore all the research contained in this thesis is presented as published peer-reviewed articles.
Following a brief introduction to magnetic anisotropy in molecular magnetism, a detailed, but far from comprehensive, text on the electronic and magnetic properties of molecules is presented. Both the parametric and ab initio approaches to electronic structure are expounded and this text could be considered a primer for students starting a course of study in theoretical molecular magnetism.
The research studies of this Ph.D. have involved numerous research collaborations and investigations in different aspects of molecular magnetism. In order to tell a coherent and compelling story, six key works highlighting the study of magnetic anisotropy across the Periodic Table are presented in a non-chronological manner. Other publications containing work performed during this Ph.D. can be found as citations in the appendix and it is anticipated that further publications resulting from original research during this Ph.D. will be forthcoming.
Brief of included works and roles of authors
Paper one: “On the Possibility of Magneto-Structural Correlations: Detailed Studies of
Dinickel Carboxylate Complexes”.
J. P. S. Walsh, S. Sproules, N. F. Chilton, A.-L. Barra, G. A. Timco, D. Collison, E. J. L.
McInnes and R. E. P. Winpenny, Inorg. Chem., 2014, 53, 8464.
This paper highlights the difficulty of determining the magnetic parameters of even seemingly simple complexes. The study of a closely related family of dimetallic complexes finds that the Zero Field Splitting (ZFS) of the = 1 NiII ions is on the same
푆 17 order of magnitude as the magnetic interactions between them, greatly complicating the analysis. However, magneto-structural correlations between the Ni-O(H2)-Ni bridging angle and the magnetic exchange interaction are revealed.
JPSW synthesized all the complexes (with assistance from GT) and performed the magnetic and EPR measurements (with assistance from SS). High frequency EPR measurements were performed by ALB (with assistance from JPSW). NFC estimated the single-ion magnetic parameters for each dimer, through the use of ab initio calculations, providing a good starting point for the spectral simulations. JPSW and NFC modelled the EPR spectra and the magnetic data. The manuscript was written by JPSW with assistance from the other authors.
Paper two: “Large Zero-Field Splittings of the Ground Spin State Arising from
Antisymmetric Exchange Effects in Heterometallic Triangles”.
S. A. Magee, S. Sproules, A.-L. Barra, G. A. Timco, N. F. Chilton, D. Collison, R. E. P.
Winpenny and E. J. L. McInnes, Angew. Chem. Int. Ed., 2014, 53, 5310.
This paper elucidates the anomalous magnetic anisotropy in the ground state of a
III carboxylate bridged, oxo-centred Ru2Mn triangle. As the Ru ions are strongly antiferromagnetically coupled, the spin ground state of the molecule is owed to the
= 5 2 state of MnII. However, the magnetic anisotropy associated with the spin ground푆 ⁄ state is much too large to originate from manganese alone and was determined to arise through significant antisymmetric exchange interactions with the RuIII ions.
SAM synthesized the complex (with assistance from GT) and performed the magnetic and EPR measurements (with assistance from SS). High frequency EPR measurements were performed by ALB (with assistance from SAM). NFC developed an analytical antisymmetric exchange model and applied it to the coupled spin eigenstates of the isotropic exchange Hamiltonian, deriving an expression for the ground state axial ZFS
18 parameter with second-order perturbation theory. The manuscript was written by
SAM, SS, NFC퐷 and EJLM.
Paper three: “An electrostatic model for the determination of magnetic anisotropy in dysprosium complexes”.
N. F. Chilton, D. Collison, E. J. L. McInnes, R. E. P. Winpenny and A. Soncini, Nat.
Commun., 2013, 4, 2551.
This paper develops an intuitive and simple model to explain and to predict the magnetic anisotropy of low-symmetry DyIII complexes. The aspherical electron density
III distribution of the = ± 15 2 state of Dy , derived from the / spin-orbit 6 multiplet, is employed푚 퐽in conjunction⁄ with a minimal valence bond model퐻15 for2 the partial charges of the ligands and determines the orientation of the main magnetic axis by electrostatic optimization.
NFC developed the idea for this approach in concert with AS. NFC performed all ab initio calculations, wrote a program for the electrostatic analysis and subsequently performed all electrostatic calculations. The manuscript was written by NFC with assistance from the other authors.
Paper four: “The first near-linear bis(amide) f-block complex: a blueprint for a high temperature single molecule magnet”.
N. F. Chilton, C. A. P. Goodwin, D. P. Mills and R. E. P. Winpenny, Chem. Commun.,
2015, 51, 101.
This paper presents the first near-linear pseudo-two-coordinate f-block complex. The near-linear N-Sm-N coordination mode is the ideal environment to exploit the oblate electron density distribution of the large angular momentum states of DyIII and the
i + proposed complex [Dy{N(Si Pr3)2}2] is predicted to have a barrier to magnetization
19 reversal of 1800 . It is predicted that such a complex would display −1 magnetic hysteresis푈푒푓푓 ≈ at temperatures푐푚 above that of liquid nitrogen.
CAPG synthesized the complex (with assistance from DPM) and performed all characterizations. NFC identified the potential magnetic properties of the linear motif and performed all ab initio and CF calculations. The manuscript was written by NFC with assistance from DPM and REPW.
Paper five: “Design Criteria for High-Temperature Single-Molecule Magnets”.
N. F. Chilton, Inorg. Chem., 2015, 54, 2097.
This paper elucidates the requirements for two-coordinate DyIII complexes to show large barriers to magnetization reversal. The work reveals that near-linearity is not a requirement and that any two-coordinate complex of DyIII should show a large barrier to magnetization reversal. Furthermore, the effect of coordinating solvent is shown to be a major detriment to the relaxation barrier.
The entire paper was the work of NFC.
Paper six: “Direct measurement of dysprosium(III)···dysprosium(III) interactions in a single-molecule magnet”.
E. Moreno Pineda, N. F. Chilton, R. Marx, M. Dörfel, D. O. Sells, P. Neugebauer, S.-D.
Jiang, D. Collison, J. van Slageren, E. J. L. McInnes and R. E. P. Winpenny, Nat.
Commun., 2014, 5, 5243.
This paper is the first direct spectroscopic determination of a dysprosium-dysprosium exchange interaction in a coordination complex. It is shown for this asymmetric dimetallic complex that one pocket behaves as a SMM, while the other does not, and that the interaction between the two metals is responsible for quenching the SMM behaviour in the pure complex.
20
The synthesis, crystallographic characterization and magnetic data collection and processing for all complexes were performed by EMP. Low-frequency EPR data were collected jointly by DOS and EMP. FIR and HF-EPR measurements were performed by
EMP, RM, MD, PN, S-DJ and JvS. NFC performed ab initio calculations for the two dysprosium sites, providing a starting point for the model. EMP and NFC performed fits of the magnetic data as well as simulations of the EPR spectra. The paper was written by EMP, NFC, EJLM and REPW with input from the other authors.
Magnetically anisotropic materials
The study of magnetic anisotropy in transition metal complexes is pertinent to many areas of science, with applications reaching from biomedical imaging to quantum information processing. The magnetic anisotropy of molecular complexes is distinct from the collective anisotropy of ordered ferromagnets and is an intrinsic feature of the molecule. Magnetic anisotropy arises in molecules which possess unquenched orbital angular momentum and therefore large anisotropy is most commonly associated with open shell transition metal complexes. Even if such metal complexes have orbitally non- degenerate ground states, mixing of excited orbitally degenerate configurations into formally spin-only ground states via Spin-Orbit Coupling (SOC) can produce non- negligible magnetic anisotropy in the ground state.
The origin of the magnetic anisotropy in open shell transition metal complexes is the interaction of the coordinating ligands with the orbital magnetic moment, and is therefore directly connected to the spatial arrangement and symmetry of the coordination environment. The magnetic anisotropy results in a physical localization of the magnetic moment and is most apparent in complexes with degenerate orbitals, such as complexes of the d- and f-block. While not directly influenced by the environment, the spin magnetic moment is coupled to the orbital magnetic moment via SOC and therefore the magnetic anisotropy can be enhanced when this relativistic effect is large.
The success of inorganic coordination chemistry in being able to design molecular
21 architectures around metal ions provides a direct handle for tailoring the magnetic anisotropy, thus facilitating the wide range of applications which often require very different magnetic properties.
The difficulty in determining the nature of the magnetic anisotropy varies for different blocks of the Periodic Table. For example, the magnetic anisotropy and exchange interactions can often be of the same order of magnitude for compounds of 3d ions, therefore untangling the two becomes difficult, while for complexes of the 4f ions, the magnetic anisotropy is much more sensitive to the environment. Compounds employing the 4d, 5d and 5f ions often possess magnetic anisotropy and magnetic exchange interactions that are of the same order as other electronic interactions, making the determination of the magnetic properties even more challenging.
The unravelling of the magnetic properties and the understanding of the magnetic anisotropy, while at times a very challenging task, is a requirement for the in-depth understanding of magnetic phenomena. It is only with a complementary suite of experimental techniques and some theoretical insight that the true nature of complex magnetic problems can be revealed. The connection between experiment and theory can be elegantly broached by interpretation of the magnetic properties with a simple parametric model. It will be the subject of this text to outline the basic theory behind the electronic properties of magnetic molecules and present some selected research highlights to show the application of such approaches to complexes of the 3d ions, those of mixed 3d/4d ions, and some of 4f ions. This work shows the great promise held by such approaches to further the fundamental understanding of magnetic interactions in molecules, and provides some insight for how more interesting magnetic molecules may be designed.
22
3. The electronic and magnetic properties of molecules
Magnetic properties of the electron
e traditional picture of electrons orbiting a positively charged nucleus is an amazingly intuitive picture of the atom, Figure 1. Although the innitely localizable formalism of classical mechanics fails at length scales such as those of the atom or molecule, this classical picture is a useful point of departure for QM. e rst step into the quantum realm is the acknowledgement that this classical picture is inherently awed; radial acceleration due to the circular motion of such an orbit would cause the electron to emit radiation, thus losing energy and eventually spiralling in towards the nucleus, leading to the collapse of matter. ankfully, this catastrophe is prevented by QM due to the quantization of the orbital states, allowing only certain congurations and transitions.
Indeed, the classical concept of simultaneous knowledge of position and momentum is juxtaposed by QM, which is inherently probabilistic in nature. In place of accurate position and momentum, QM dictates dened quantum states which have certain probabilities of possessing given momentum or position.*
l ⇀
+ r ⇀- p ⇀
Figure 1 | Traditional circular electron orbit around the nucleus.
* When measured however, particles can be found to have a given position and momentum simultaneously, which, according to the Copenhagen interpretation of QM, is a deterministic result due to the collapse of the wavefunction under the inuence of the measurement device. 23
Suppose for a moment, however, that accurate knowledge of the position, , and momentum, , of an electron as it orbits the nucleus are available. With this 푟⃑can be associated a classical푝⃑ angular momentum, , as described by (1), where the cross product indicates that the angular momentum 푙⃑vector is perpendicular to both the radial direction and the direction of orbital motion of the electron, Figure 1.
= × (1)
The orbital motion of the electron could푙⃑ also푟⃑ 푝be⃑ considered as a current loop, which generates a magnetic field owing to the Biot-Savart law. Indeed, this magnetic field is also perpendicular to the plane in which the electron orbits and defines a deep connection between angular momentum and magnetism. In the context of a single electron, this is referred to as the orbital magnetic moment which arises from the physical orbital angular momentum of the electron in three-dimensional space. There exists an entirely separate,* additional, form of angular momentum associated with the electron, which is referred to as the ‘spin’ angular momentum. It is a purely quantum property with no classical analogue and resides in an isolated ‘spin space’. Electrons are
= 1 2 particles which may possess either the = ± 1 2 quantization. One should
푠disregard⁄ any association between the spin degree푚푠 of freedom⁄ and a physical spinning motion – the electron cannot be thought to have any axis around which to spin.
The reader should rest assured that the existence and quantization of both the orbital and spin angular momentum are experimentally verified and in fact belong to a group of core observations that led to the inception of QM.
Quantum mechanics and the wavefunction
The wavefunction is an entity that encapsulates all the information about a given particle or group of훹 particles. It is generally a complex valued function of the spatial and spin coordinates of each particle, whose squared magnitude = | | is associated ∗ 2 훹 훹 훹 * The spin is separate in its native state. Under the influence of spin-orbit coupling these two forms of angular momentum cannot be separated. See Magnetic and relativistic effects: the Dirac equation. 24 with the probability distribution of the particle(s).* Given that there is certain probability of finding the particle(s) somewhere in space, the integral of the probability distribution over all coordinates is defined as unity, (2). A key concept is that of an operator, denoted by , which is the mathematical representation of a measurement in the framework of QM.� Wavefunctions that are not altered by a particular operator are said to be eigenfunctions of that operator, and the action of the operator is to reveal a characteristic value of the wavefunction, known as an eigenvalue. These kinds of equations are known as eigenvalue equations and, while not limited to QM, the fundamental eigenvalue equation of QM is Schrödinger’s equation, (3),1 where the
Hamiltonian operator measures the total energy of the wavefunction. To calculate the expectation value 퐻(eigenvalue)� of any operator,퐸 (3) can be pre-multiplied by the complex conjugate and integrated over all coordinates, (4). While the integral ∗ notation of Schrödinger훹 is more traditional, this text will employ the more compact ‘bra-ket’ notation of Dirac, where the ket | is equivalent to the wavefunction and the bra | is its complex conjugate – integration훹⟩ over all variables is implied훹 in the ∗ combination⟨훹 of bra and ket.2 훹
ö : = 1 : | = 1 (2) ∗ 푆푐ℎ푟 푑푖푛푔푒푟 � 훹 훹 푑휏 퐷푖푟푎푐 ⟨훹 훹⟩ 휏 ö : = : | = | (3)
푆푐ℎ푟 푑푖푛푔푒푟 퐻�훹 퐸훹 퐷푖푟푎푐 퐻� 훹⟩ 퐸 훹⟩
ö : = : = (4) ∗ 푆푐ℎ푟 푑푖푛푔푒푟 � 훹 퐻�훹푑휏 퐸 퐷푖푟푎푐 �훹�퐻��훹� 퐸 The absolute positions, velocities휏 or time evolution of the electrons are not of concern for molecular systems. Rather, the interest lies in the bound states of the molecule where the electrons are in a steady-state and the wavefunction can be considered as a standing wave. In such cases the Hamiltonian takes the form (5),3,4 which is constructed to
* Recall that for any complex number = + , the complex conjugate is = and that = ( )( + ) = + = | | . ∗ ∗ 2 2 2푧 푎 푖푏 푧 푎 − 푖푏 푧 푧 푎 − 푖푏 푎 푖푏 푎 푏 푧 25 contain all the interactions involved in the system, comprising the kinetic and potential energy of each particle, and , respectively.*
푃�푖 푉�푖
= 푁 + = 푁 + (5) 2 2 4 2 −ℏ 2 푒 푞푖푞푗 푖 푖 �푖 퐻� ��푃� 푉� � � � 푖 훻 � 0 푖푗 � 푖=1 푖=1 푚 푗≠푖∈푁 휋휀 푟
| = | | (6)
푒 푛 Although the discussion thus far has 훹focussed⟩ 훹 ⟩on훹 electrons,⟩ nuclei are also described by the Hamiltonian (5) and the nuclear wavefunction does indeed exist. However as the rest mass of nucleons are three orders of magnitude larger than that of the electron
( 1840 ) their kinetic energy and magnetic moment are three orders of magnitude푚푝 ≈ 푚푛 smaller.≈ Therefore푚푒 the total wavefunction of an atom or molecule is assumed to be separable into an electronic and nuclear part, (6); this is known as this is the Born-
Oppenheimer (BO) approximation.5 This separation allows the solution for the electronic wavefunction only, where the nuclei are considered as fixed point charges.
A single electron: the hydrogen atom
In view of seeking a general solution to the electronic wavefunction, the special case of the hydrogen atom is considered briefly. In this case, the many-body wavefunction | is replaced with the single-electron wavefunction | , the nucleus defined as the origin훹⟩ and the position vector of the electron is given in휓 spherical푒⟩ coordinates, = ( , , ).
Then, the potential energy term is simply the spherically symmetric Coulomb푟⃑ potential푟 휃 휙 for the electron’s attraction to the nucleus, and the Hamiltonian reduces to (7).
= (7) 2 2 4 2 −ℏ 2 푒 퐻� 훻� − 푚푒 휋휀0푟
* is the number of particles, is the reduced Planck constant, is the rest mass of particle ,
= + + is the Laplacian operator, is the elementary charge,푖 is the vacuum permittivity, 푁 2 2 2 ℏ 푚 푖 2 휕 휕 휕 2 2 2 훻� is the휕푥 charge휕푦 of particle휕푧 and is the distance푒 between particles and . 휀0 푞푖 푖 푟푖푗 푖 푗 26
When (7) is written in spherical coordinates and | inserted, the radial and angular parts of the equation are separable and may be 휓solve푒⟩ d independently. The angular solutions are the spherical harmonics, (8),* which are the origin of the familiar Atomic
Orbitals (AOs) and are chosen specifically to have well defined orbital angular momentum , (9) and (10), where is the projection of along the z-axis.† An
푙 analytical solution푙 to the radial wavefunction푚 , is not of concern푙 here as there is no general solution and is dependent on the exact�푅푛 problem,푙� ‡ however it is in solving for
, that the principal quantum number, , is introduced and that only one radial
�푅function푛 푙� for each pair of allowed quantum푛 numbers , is required. Finally, a spin function is included in an ad-hoc manner which accounts푛 푙 for the two possible quantizations of the = 1 2 electron spin, (11) and (12). This yields the final single- electron wavefunction푠 (13)⁄, where represents the electron position and its set of unique quantum numbers ( , ,훼, , , ).§
훼 ≡ 푟⃑ 푛 푙 푚푙 푠 푚푠
( , ) | , (8) 푙 푙 �푌푚 휃 휙 � ≡ 푙 푚푙⟩ | , = ( + 1)| , (9) 2 2 푙̂ 푙 푚푙⟩ ℏ 푙 푙 푙 푚푙⟩ | , = | , (10)
푙̂푧 푙 푚푙⟩ ℏ푚푙 푙 푚푙⟩ | , = ( + 1)| , (11) 2 2 푠̂ 푠 푚푠⟩ ℏ 푠 푠 푠 푚푠⟩ | , = | , (12)
푠̂푧 푠 푚푠⟩ ℏ푚푠 푠 푚푠⟩ | ( ) = , | , | , (13)
휓푒 훼 ⟩ �푅푛 푙� 푙 푚푙⟩ 푠 푚푠⟩
* See The spherical harmonics for more information. † Note that the choice of the z-axis for quantization is arbitrary and is simply convention. ‡ Generally, however, the radial function decays exponentially as and is often a highly nodal polynomial function. 푟 → ∞ § Recall the allowed values for the quantum numbers are = 1, 2, 3, …; = 0, 1, 2, … , 1; = , + 1, … , 1, ; = 1 2; and = ± 1 2; and that the Pauli exclusion principle prohibits more 푛 푙 푛 − 푚푙 than one electron possessing the same set of quantum numbers. −푙 −푙 푙 − 푙 푠 ⁄ 푚푠 ⁄ 27
Multi-electron atoms
With more than one electron the Coulomb repulsion between the electrons is an important part of the Hamiltonian. However, as this term is a function of the coordinates of two particles, there are no analytical solutions for the many-body wavefunction. If the electron-electron Coulomb interaction is approximated by assuming each electron moves in a spherically symmetric field that is the average effect of the remaining electrons, the two-electron operator can be omitted and the potential term replaced with a modified single-electron potential , resulting in the Mean Field
(MF) Hamiltonian, (14).* This approximation results in 푈the푖 electrons not sensing each other correctly, and is known as the lack of dynamic correlation.†
= 푁푒 + ( ) (14) 2 2 −ℏ 2 푀퐹 �푖 푖 푖 퐻� � � 푒 훻 푈 푟 � As the MF Hamiltonian only contains푖= single1 푚-electron operators, the approximate many- body wavefunction can be decomposed into single-electron wavefunctions and each solved independently.‡ These single-electron wavefunctions have the same angular form as the solution for the hydrogen atom, however the radial wavefunctions differ owing to the modified potential. However, a simple product of these single-electron wavefunctions does not directly satisfy the Pauli exclusion principle; the wavefunction must be anti-symmetric with respect to a simultaneous position and spin permutation and therefore a special linear combination of the single-electron functions is required.
This can be elegantly achieved with a Slater determinant, (15), which represents a unique orbital population of electrons. Note that the Slater determinant does not necessarily represent an entire electron configuration – cases with orbital degeneracy require multiple determinants to describe the many-body wavefunction, an effect known as static correlation.
* is the number of electrons. † See The ab initio approach for further discussion. 푁푒 ‡ Subject to the constraint of orthogonality. 28
( ) ( ) ( )
1 �휓푒1(훼1)� �휓푒2(훼1)� ⋯ �휓푒푁푒(훼1)� | = (15) ! � � 푒1 2 푒2 2 푒푁푒 2 푒⟩ �휓 훼 � �휓 훼 � ⋯ �휓 훼 � 훹 푒 �푁 � ⋮ ⋮ ⋮ ⋮ � �휓푒1�훼푁푒�� �휓푒2�훼푁푒 �� ⋯ �휓푒푁푒 �훼푁푒�� Magnetic and relativistic effects: the Dirac equation
That astute reader will note that the Schrödinger treatment neglects interactions of magnetic origin and that Einstein’s theory of Special Relativity is also curiously absent.
For = 1 2 particles such as electrons, it is the Dirac equation which is built to include spin,푠 relativity⁄ and electro-magnetic fields, and provides a much more rigorous description of the electron.6 Although it will not be discussed in detail, it is worth mentioning that this remarkable equation does indeed reduce to the time-independent
Schrödinger equation in the limit of non-relativistic electrons. For the purposes of this text, the most important result from the Dirac treatment is that electrons moving in potentials like that of an atomic nucleus experience a coupling of their spin and orbital angular momentum. The so-called Spin-Orbit Coupling (SOC) may have been intuitively expected – the magnetic moment generated by the orbital angular momentum can interact with the intrinsic spin magnetic moment of the electron. SOC combines the orbital and spin angular momenta into a total angular momentum of the electron, = + . Importantly, this renders the individual projections and poor quantum 횥⃑numbers,푙⃑ 푠⃑ replaced by the projection of the total angular momentum푚푙 푚푠 .
Although the Dirac treatment of the electronic wavefunction has many advantages,푚 푗it proves to be a much more challenging problem which has no general solutions. Luckily, it turns out that relativistic effects only become important for elements with large atomic number where the core electrons move at a significant fraction of the speed of light.7,* For this reason, the Schrödinger treatment is usually a good starting point and
* The definition of ‘significant’ depends on the calculation, but here implies compounds containing elements of the fourth period, 19, and beyond.
푍 ≥ 29 the SOC can be included as a perturbation with the SOC Hamiltonian, (16), which is based on the result of the Dirac equation.3,4,*
( ) = 푁푒 (16) 2 2 ℏ 푑푈푖 푟푖 푆푂퐶 2 2 ⃑̂푖 ⃑푖 퐻� � � 푒 푖 푖 푙 ∙ 푠̂ � 푖=1 푚 푐 푟 푑푟 Energy scales
A number of approximations have been introduced when determining the electronic wavefunction, without much justification. The Hamiltonian is a recipe for calculating the energy of a quantum system and therefore any approximations should be judged on their contribution to the total energy. Take for example the BO approximation; it was justified that as the masses of the proton and neutron are three orders of magnitude larger than that of the electron, the kinetic energy of the nuclei are at least three orders of magnitude smaller than that of the electrons. It is these vastly differing energy scales that make ignoring the nuclear wavefunction justifiable. In the case of multi-electron systems the explicit electron-electron Coulomb repulsion was ignored and replaced with an average single-electron potential; the electron-electron repulsion energy is usually at least an order of magnitude smaller than the difference between subsequent electron configurations, see for example Figure 2, and therefore this is often a reasonable approximation. The exclusion of relativistic effects is again justifiable on the basis of energy scales; even the largest relativistic effect, the SOC, is usually an order of magnitude weaker than the electron-electron repulsion. The result of these vastly different energy scales is well visualized in the electronic structure of a multi-electron atom or ion, Figure 2, where the successive inclusion of each correction is a minor departure compared to the preceding stage.
* is the speed of light.
푐 30
1 G0
3 P2 3 P0
1 D2
3 F4 3 F3 3 F2
Figure 2 | Lowest electronic energy levels of a free VIII ion. e le hand levels correspond to the electron congurations, the central levels are the terms split by electronic repulsion and the levels on the right are the resultant SOC states. Magnied region shows SOC states for the lowest energy levels. SOC state labels are omitted for all but the lowest levels.
According to the energy scales outlined above, the electron-electro n Coulomb repulsion acts to remove the degeneracy of the electron congurations into terms that are characterized by the total spin and orbital momenta, commonly given in spectroscopic notation (17).* ese total angular momenta are sums of the individual electron moments, (18) and (19), and this is referred to as the Russell-Saunders (RS) or
LS coupling scheme. e ground term of the ground electron conguration is given by the empirical Hund’s rules8 which state:
1) the ground term will have the maximal spin multiplicity; and
2) for that multiplicity it will have the maxima l orbital momentum.
* Total ( ) or single particle ( ) orbital angular momentum are assigned codes where 0, 1, 2, 3, 4, and are subsequently alphabetic. 31
Within a manifold of degenerate orbitals, the ground term can easily be determined by assigning and quantum numbers to each electron, commencing with =
+1/2; 푚=푠 + , followed푚푙 by = +1/2; = + 1 etc. 푚푠1
푚푙1 푙 푚푠2 푚푙2 푙 − For light elements where the SOC is relatively weak, the LS coupling scheme is not a bad approximation and the SOC can be considered to couple total spin and orbital moments of each term leading to a further splitting into multiplets characterized by a total angular momentum = + .* In cases where the SOC is stronger than the Coulomb repulsion, the so-called퐽⃑ jj coupling퐿�⃑ 푆⃑ scheme is more appropriate, where the SOC is first applied to each electron individually to give a total one-electron angular momentum = + , followed by the inter-electronic Coulomb repulsion. 횥⃑ 푙⃑ 푠⃑
(17) 2푆+1 퐿퐽
= 푁푒 (18)
푆 � 푚푠푖 푖=1
= 푁푒 (19)
퐿 � 푚푙푖 푖=1 Linear algebra
The method most commonly employed for the solution to the Schrödinger wave equation is the equivalent matrix formalism of Heisenberg.3 With a defined basis,† the required integrals, given in Dirac notation as = , form a matrix and are therefore also known as matrix elements. The basis퐻푖푗 in�훹 which푖�퐻��훹 the푗� Hamiltonian matrix is constructed is a complex vector space known as a Hilbert space, which contains all the possible states for the wavefunction.‡ For example, the basis for a particle with total
* The coupling = + is a vector sum and therefore the allowed values are given by = | |, | | + 1, … , + 1, + . 퐽⃑ 퐿�⃑ 푆⃑ 퐽 퐽 퐿 − 푆 퐿 − † Explicitly and generally, | | , | . 푆 퐿 푆 − 퐿 푆 ‡ While Hilbert spaces can in fact be of infinite dimension, only spaces of finite dimension are considered Ψ푖⟩ ∈ 푎⟩ �Ψ푗� ∈ 푎⟩ here. 32 angular momentum (ignoring spatial coordinates for the time being) is given by all possible projections of푗 the angular momentum, (20), which has the dimensionality (21). All elements of the basis are orthonormal* and are said to span the Hilbert space.†
, {| , , | , + 1 , … , | , 1 , | , } (20)
�푗 푚푗� ∈ 푗 −푗⟩ 푗 −푗 ⟩ 푗 푗 − ⟩ 푗 푗⟩ = 2 + 1 (21)
푑푖푚푗 푗 = (22) −1 The language of QM describes the action퐻 푃퐷of the푃 Hamiltonian ‘connecting’ elements of the basis (i.e. the matrix element between two basis states is non-zero) as a ‘mixing’ of the two states. Once the Hamiltonian matrix is constructed, there always exists a unitary‡ transformation which can bring the퐻 matrix to a diagonal form (22).§ This diagonalization is in fact 푃the solution to a special case of the eigenvalue problem where the Hamiltonian matrix must be Hermitian** such that the energy eigenvalues are real- valued observables. The diagonal entries of are the eigenvalues whose corresponding eigenvectors are the columns of . The eigenvectors퐷 must necessarily be orthogonal, leading to the connection of the푃 diagonalization, and therefore solution, of the Hamiltonian matrix as a rotation of the basis.†† Accordingly, such a transformation is also known as a change of basis, where the rotated basis described by is the eigenbasis of the Hamiltonian. 푃
* That is, each element does not share any commonalities with another and that they are normalized to unit magnitude. † A set of basis kets | span the Hilbert space by definition of the closure relation | | = 1 ‡ Unitary matrices are those with the property ( ) = ( ) = , where is the identity matrix. 푎⟩ ∑푎 푎⟩⟨푎 § Or equivalently, = . ∗ 푇 ∗ 푇 퐴 퐴 퐴 퐴 퐼 퐼 ** Hermitian matrices−1 are equal to their conjugate transpose, = = ( ) . 푃 퐻푃 퐷 †† Assuming the initial basis vectors are not already the eigenbasis of† the Hamiltonian.∗ 푇 As the initial basis 퐴 퐴 퐴 vectors span the Hilbert space, the eigenvectors (linear combinations of these basis vectors) representing the eigenbasis of the Hamiltonian must also span the Hilbert space and are therefore also all orthogonal. As the two sets of vectors span the same Hilbert space, all elements of each set are orthogonal to one another and the two sets are not coincident, the eigenbasis is simply a rotation of the initial basis in the Hilbert space. 33
A commonly employed change of basis is the coupling of two angular momenta to form one total angular momentum, having application to, for example, the SOC of a single electron or the coupling of two spins. The coupling coefficients, given for the coupling of two arbitrary angular momenta , and , to form , , , ,* are
1 푗 2 푗 1 2 퐽 known as the Clebsch-Gordan coefficients,�푗 푚 1� (23), �푗which푚 2form� an orthonormal�푗 푗 퐽 푚 �set of eigenvectors spanning the new coupled Hilbert space. It is important to realize that while the basis has been changed, the dimensionality and closure of the Hilbert space remains; merely the matrix representation of the Hamiltonian is now different.
, , ,
�푗1 푗2 퐽 푚퐽� (23) = 푗1 푗2 , , , , , , , , ,
� � �푗1 푚푗1 푗2 푚푗2�푗1 푗2 퐽 푚퐽� �푗1 푚푗1 푗2 푚푗2� 푚푗1=−푗1 푚푗2=−푗2 Perturbation theory
While of course in reality all interactions should enter the Hamiltonian on an equal footing, sometimes it may be desirable to determine how a small perturbation influences the wavefunction of a more important Hamiltonian. With knowledge of the unperturbed or zeroth order wavefunction, which is an eigenfunction of the zeroth order Hamiltonian, and the assumption that the perturbed wavefunction is not too dissimilar (i.e. the perturbation is small), Perturbation Theory (PT) provides an approach to calculate the corrections to both the energies and wavefunctions to arbitrary order. For well-behaved perturbations the series is convergent and at the limit of infinite order the solution for the perturbed states is exact. The beautiful formalism of
PT is well explained in other texts9,10 and will not be explicitly derived here, save for the final results. The zeroth order (unperturbed) Hamiltonian is denoted ( ), with 0 퐻�
* Two independent Hilbert spaces are formally connected with the Kronecker or Tensor product , , , , , , , with dimension = (2 + 1)(2 + 1).
The coupling1 =2 + is 1a vector2 sum and1 therefore2 the allowed values1 2are given by = | �푗1 푚푗 � ⊗ �푗2 푚푗 � ≡ �푗1 푚푗 푗2 푚푗 � ≡ �푚푗 푚푗 � 푑푖푚푗 ⊗푗 푗1 푗2 |, | | + 1, … , + 1, + . The new coupled Hilbert space, spanned by the , , , basis 퐽⃑ 횥⃑1 횥⃑2 퐽 퐽 푗1 − vectors, has exactly the same dimension and contains exactly the same information as the uncoupled 푗2 푗1 − 푗2 푗1 푗2 − 푗1 푗2 �푗1 푗2 퐽 푚퐽� , , , basis.
�푗1 푚푗1 푗2 푚푗2� 34
( ) ( ) eigenvectors and eigenvalues , and the perturbing Hamiltonian denoted as 0 0 푖 푖 ( ). The first�Ψ order� correction to the퐸 energies is given as (24), which is simply the 1 퐻expectation� value (matrix element) for the perturbation in the unperturbed eigenbasis. The second order corrections to the energies are given by (25), which involve the matrix elements that mix the unperturbed functions. The first order correction to the wavefunction is very similar to the second order correction to the energies, as this must of course involve the mixing of the unperturbed functions, (26).
( ) ( ) ( ) = ( ) (24) 1 0 1 0 퐸푖 �훹푖 �퐻� �훹푖 � ( ) ( ) ( ) ( ) 2 0 0 (25) = ( ) 1 ( ) 2 ��훹푗 �퐻� �훹푖 �� 퐸푖 � 0 0 푗≠푖∈푎 푖 푗 ( )퐸 ( −) 퐸( ) ( ) ( ) 0 0 (26) = ( ) 1 ( ) 1 �훹푗 �퐻� �훹푖 � 0 �훹푖 � � 0 0 �훹푗 � Of course the expressions (25) and푗 ≠(26)푖∈푎 become퐸푖 − undefined퐸푗 in the presence of degenerate eigenstates, in which case the numerator is required to be exactly zero such that a singular expression is avoided. Unfortunately this can only be achieved by taking the appropriate linear combination of the unperturbed eigenstates that span the degenerate subspace, which is equivalent to diagonalizing the block of the Hamiltonian matrix containing the degenerate states. As with the case of degeneracy, in the case of near- degeneracy the corrections will be very large due to small denominators and therefore
PT is not valid, so the near-degenerate block must also be diagonalized. Only when the energy separation between states is large will PT yield good approximations to the perturbed states.
Suddenly the picture becomes clear; diagonalization of the matrix is in fact infinite order
PT within the subspace of our chosen basis. For example, the basis space including all electronic configurations is infinite and therefore by restricting the basis to the ground configuration only, the mixing of excited configurations into the ground one by action
35 of the Hamiltonian are ignored.* is has been previously justied, but now it can be seen that diagonalizing the Hamiltonian matrix in the subspace of a single conguration is simply innite order PT in this restricted basis. Any further truncation of the basis, such as only considering the lowest energy term, follows exactly the same protocol,
Figure 3.
Figure 3 | Schematic of the Hamiltonian matrix for the VIII problem. Largest grid shows basis containing all states from the 3d2, 3d14s1 and 3d14p1 congurations, smaller grid shows all terms arising from the 3d2 conguration and black box represents the 21 spin- orbit states of the ground 3F term.
e spherical harmonics
e spherical harmonics are a versatile set of angular functions, dened as the angular solutions to the Laplace Equation in spherical coordinates and formally given by (27) and (28).3,11,† ey are a complete set of orthonormal functions, (29) and (30), and are therefore capable of representing any scalar function on a sphere via a generalized
Fourier Transform, (31) and (32). Note that the Kronecker delta function appearing in
(29) is dened as (33).
* is is either justied by large energy gaps to excited states creating large denominators, small mixing coecients creating small numerators, or both. † e functions are the associated Legendre Polynomials. 36
(2 + 1)( )! ( , ) = ( 1) ( ) | , (27) 4 ( + )! 푙 푚푙 푙 푙 푖푚푙휙 푚푙 푙 푙 − 푚 푚푙 푙 푌 휃 휙 − � 푙 푃 푐표푠 휃 푒 ≡ 푙 푚 ⟩ 휋 푙 푚 ( , ) = ( 1) ( , ) (28) 푙 ∗ 푚푙 푙 푌푚푙 휃 휙 − 푌−푚푙 휃 휙
, | , = 휋 휋 ( , ) ( , ) = , , (29) ′ ∗ ′ ′ 푙 ′ 푙 ′ ′ ⟨푙 푚푙 푙 푚푙⟩ � � 푌푚푙 휃 휙 푌푚푙 휃 휙 푠푖푛 휃 푑휃 푑휙 훿푙 푙훿푚푙 푚푙 휃=0 휙=−휋
휋 휋 ( , ) = 1 (30) 푙 2 � � �푌푚푙 휃 휙 � 푠푖푛 휃 푑휃 푑휙 휃=0 휙=−휋
= 휋 휋 ( , ) ( , ) (31) 푙 푙 ∗ 푓푚푙 � � 푓 휃 휙 푌푚푙 휃 휙 푠푖푛 휃 푑휃 푑휙 휃=0 휙=−휋
( , ) = ∞ 푙 ( , ) (32) 푙 푙 푓 휃 휙 � � 푓푚푙 푌푚푙 휃 휙 푙 푙=0 1푚 = −푙 = = , 0 (33) 푖푓 훼 훽 훿훼 훽 � 푖푓 훼 ≠ 훽 More than one way to skin a cat
Under the MF approximation discussed earlier, the single-electron wavefunctions for a multi-electron atom have the same angular form as the electron in hydrogen, which is capable of exact analytical evaluation. However, the unknown form of radial function defies a general analytical solution and historically this problem has been approached in two distinct ways. Experimentalists have preferred a parametric approach where the radial integrals of the Hamiltonian are treated as parameters to be determined by experiment, while theorists have preferred an ab initio approach and sought a numerical approximation to the radial function. Both approaches have their distinct advantages and disadvantages, and it will be the subject of the remainder of this text to describe the two techniques and how they can be employed in concert to gain useful insight into the electronic and magnetic properties of molecular systems.
37
The parametric approach: angular momentum in quantum mechanics
The great usefulness of treating the unknown radial integrals as parameters while exactly solving the angular part is that calculations are much more facile and can be easily compared with experiment. To illustrate the power of the parametric approach, the SOC in the free ion configuration [Ar]3d1 will be examined. Owing to the Pauli exclusion principle, all angular momentum within closed shells will cancel* and only electrons in open shells contribute to the magnetic properties. Furthermore, it is only the energy differences to the excited states which are of importance for the magnetic and spectroscopic properties and therefore these paired electrons, which only contribute to the absolute energy of the ground state, can be neglected.† Because the free ion has spherical symmetry, all five 3d orbitals are degenerate and the electron is free to occupy any of these orbitals, leading to 10 possible states for the electron once the spin degree of freedom is accounted for. These states, where = 2 and = 1 2 and labelled | , for brevity, form a basis for the angular part of 푙the wavefunction푠 ⁄.‡ Due to the separation푚푙 푚푠⟩ of variables, the matrix elements (integrals) that must be computed when evaluating the
SOC Hamiltonian are of the type (34). With a reminder of the implication of Dirac notation, (35),§ the radial integral can be replaced with a single parameter , the single- electron SOC parameter for the 3d electron, leading to the simplified SOC 휁Hamiltonian (36) which can be expressed in an angular momentum only basis, (37). It is important to note that in the evaluation of matrix elements such as those in the final expansion of
(37), the orbital operators only act on the orbital part of the wavefunction and the spin operators only on the spin part. This is perhaps best expressed in the full Kronecker product expansion of the operator acting on the basis kets, (38).
* i.e. the sums (18) and (19) are zero when each state of angular momentum is doubly occupied. † The zero of the energy scale can be arbitrarily set and thus only the orbital manifold with unpaired 푚푙 푙 electrons needs to be considered. ‡ Recall that this combined Hilbert space is actually the Kronecker product of the two independent spaces for the orbital and spin angular momentum. The dimension of the Hilbert space is the product of the dimensions of the constituent spaces, which in this case is 5 × 2 = 10. § Note that the infinitesimal volume element for integration, commonly referred to as the Jacobian, in spherical coordinates is = sin . 2 푑푉 푟 푑푟 휃 푑휃 푑휙 38
( ) ( ) ′ 1푒 ( ) 푆푂퐶 푒 (34) = �휓 훼 �퐻� �휓 훼, � , 2 2 , , ℏ 푑푈 푟 ′ ′ 2 2 �푅푛 푙� �푅푛 푙� �푚푙 푚푠 �푙⃑̂ ∙ 푠⃑̂�푚푙 푚푠� 푚푒 푐 푟 푑푟 1 ( ) 1 ( ) , , = ∞ , = (35) 2 2 2 2 ℏ 푑푈 푟 ℏ 2 푑푈 푟 2 2 2 푛 푙 푛 푙 2 2 푛 푙 푒 �푅 � �푅 � 푒 � 푅 푟 푑푟 휁 푚 푐 푟 푑푟 푚 푐 푟=0 푟 푑푟 = (36)
퐻�푆푂퐶 휁푙⃑̂ ∙ 푠⃑̂ ( ) ( ) = , , ′ ′ ′ (37) �휓푒 =훼 �퐻�푆푂퐶, �휓푒 훼 � + 휁 �푚푙+ 푚푠 �푙⃑̂ ∙ 푠⃑,̂�푚푙 푚푠� ′ ′ 휁�푚푙 푚푠 �푙̂푥푠̂푥 푙̂푦푠̂푦 푙̂푧푠̂푧�푚푙 푚푠� | , = [| | ] = | [ | ] (38)
To determine 푙̂훼the푠̂훼 푚matrix푙 푚푠⟩ elements푙̂훼⨂푠̂훼 of푚 푙⟩the⨂ 푚Hamiltonian푠⟩ �푙̂훼 푚 푙⟩matrix,�⨂ 푠̂훼 푚the푠⟩ action of the operators on the basis kets must be determined; these rules are given for a general angular momentum , but are equally applicable for any orbital angular momentum or spin angular momentum푗 . The first two relations (39) and (40) are no more than 푙the original quantization rules푠 for the definition of the angular momentum and therefore the basis kets are unsurprisingly eigenstates of and . Due to the Heisenberg 2 uncertainty principle, definite values of all components횥̂ of 횥̂the푧 angular momentum are not available simultaneously and therefore the basis kets are not eigenstates of the and
operators. These two operators are most simply expressed as functions of the r횥̂aising푥
횥̂and푦 lowering operators, (41) – (44). These operators either add or subtract a quantum of angular momentum from the state on which they act. Accordingly then, no more angular momentum can be added to or subtracted from states already possessing either the maximal = or minimal = , respectively, (45).
푚푗 푗 푚푗 −푗
, = ( + 1) , (39) 2 2 횥̂ �푗 푚푗� ℏ 푗 푗 �푗 푚푗� , = , (40)
푧 푗 퐽 푗 횥̂ �푗 푚 �1 ℏ푚 �푗 푚 � = ( + ) (41) 2 횥푥̂ 횥+̂ 횥−̂ 39
1 = ( ) (42) 2 푦 + − 횥̂ 횥̂ − 횥̂ , = ( + 1)푖 + 1 , + 1 (43)
+ 푗 푗 푗 푗 횥̂ �푗 푚 � ℏ�푗 푗 − 푚 �푚 ��푗 푚 � , = ( + 1) 1 , 1 (44)
횥−̂ �푗 푚푗� ℏ�푗 푗 − 푚푗�푚푗 − ��푗 푚푗 − � | , = | , = 0 (45)
Therefore, the matrix elements of횥+̂ (37)푗 푗⟩ can횥 −̂be푗 further−푗⟩ expanded and evaluated as (46) –
(49). The only remaining components then are the integrals of the type
, | , which, due to the orthonormality of the spherical harmonics* and the ′ ′ spin⟨푚푙 functions,푚푠 푚푙 푚 simplify푠⟩ to Kronecker deltas, (50).
1 , , = , + + , (46) 2 ′ ′ ′ ′ �푚푙 푚푠 �푙⃑̂ ∙ 푠⃑̂�푚푙 푚푠� �푚푙 푚푠 � �푙̂+푠̂− 푙̂−푠̂+� 푙̂푧푠̂푧�푚푙 푚푠� , ,
= ( + 1) ′( ′+ 1) ( + 1) ( 1) (47) �푚푙 푚푠 �푙̂+푠̂−�푚푙 푚푠� 2 × , | + 1, 1 푙 푙 푠 푠 ℏ �푙 푙 − 푚′ 푚 ′ �푠 푠 − 푚 푚 − ⟨푚푙 푚푠 푚푙 푚푠 − ⟩ , ,
= ( + 1) ′( ′ 1) ( + 1) ( + 1) (48) �푚푙 푚푠 �푙̂−푠̂+�푚푙 푚푠� 2 × , | 1, + 1 푙 푙 푠 푠 ℏ �푙 푙 − 푚′ 푚 ′− �푠 푠 − 푚 푚 푙 푠 푙 푠 ⟨푚 푚 푚 − 푚 ⟩ , , = , | , (49) ′ ′ 2 ′ ′ �푚푙 푚푠 �푙̂푧푠̂푧�푚푙 푚푠� ℏ 푚푙푚푠⟨푚푙 푚푠 푚푙 푚푠⟩ , | , = , , (50) ′ ′ ′ ′ 푙 푙 푠 푠 Therefore the SOC Hamiltonian⟨푚푙 푚 푠 푚푙 푚푠 expressed⟩ 훿푚 푚 in훿 푚our푚 chosen basis takes the form 2 (51), where the bras and kets are labelled퐻�푆푂퐶⁄ℏ as , | and | , , and the labels are simplified as + + 1 2 and 1 2. ⟨Once푚푙 푚 푠diagonalized,푚푙 푚푠 ⟩the SOC 푚Hamiltonian푠 reveals two characteristic≡ ⁄ eigenvalues− ≡ − of ⁄ 3 /2 and which are four- and six-fold 2 2 degenerate, respectively, corresponding −to ℏthe휁 two spinℏ -휁orbit manifolds = 3/2 and
= 5/2 of the term of the [Ar]3d1 configuration. In fact, the coupled푗 , , , 2 푗 퐷 �푠 푙 푗 푚푗�
* See The spherical harmonics. 40 basis expressed by the Clebsch-Gordan coefficients (23) is exactly the basis which diagonalizes the single particle SOC Hamiltonian matrix and indeed the Hamiltonian could have been constructed in this basis to begin with, leading to immediate identification of the eigenvalues.
| 2, | 1, |0, |+1, |+2, | 2, + | 1, + |0, + |+1, + |+2, + 2, | 0 0 0 0 0 0 0 0 0 − −⟩ − −⟩ −⟩ −⟩ −⟩ − ⟩ − ⟩ ⟩ ⟩ ⟩ 1, | 0 0 0 0 0 0 0 0 ⟨− − 휁 2 휁 ⟨− − 휁 3 0, | 0 0 0 0 0 0 0 0 0 2 ⟨ − � 휁 3 +1, | 0 0 0 0 0 0 0 0 2 2 휁 ⟨+2, −| 0 0 0− 0 0 0� 0 휁 0 (51) 2, +| 0 0 0 0 0 0 0 0 ⟨ − −휁 휁 3 ⟨−1, +| 0 0휁 0 0−휁 0 0 0 0 2 2 휁 ⟨− � 휁 − 3 0, +| 0 0 0 0 0 0 0 0 0 2 ⟨ � 휁 +1, +| 0 0 0 0 0 0 0 0 2 +2, +| 0 0 0 0 0 0 0 0 0휁 ⟨ 휁 More generally,⟨ the flexibility of the parametric approach allows the treatment휁 of one or more metal ions in molecular environments, by considering the loss of spherical symmetry for the individual ions as well as any magnetic interactions between them. A generalized angular momentum Hamiltonian accounting for some common features of molecular systems can be written in the basis of the ground Russell-Saunders terms,
(52), accounting for the SOC, CF, spin exchange and Zeeman interactions, respectively.*
= 푁푆 + 푁푆 푘 2 , , 푞 푞 푖 푖 푖 푖 푖푗 푗 퐻� � 휆 퐿�⃑ ∙ 푆⃑̂ � � � 퐵푘 푖푂�푘 푖 − � 푆⃑̂ ∙ 퐽 ∙ 푆⃑̂ (52) 푖=1 푖=1 푘=2 4 6 푞=−푘 푖<푗∈푁푆 + 푁퐼 +
휇퐵 � �퐿�⃑푖 ∙ 퐼 푆⃑̂푖 ∙ 푔푖� ∙ 퐵�⃑ Although the parameters appearing푖= 1in such a general Hamiltonian have well defined origins, they should be considered purely phenomenological. For example, the CF
* = ± is the multi-electron SOC constant, where is the total spin of the ground term and the 휁푖 negative휆푖 2푆sign푖 is for shells that are more than half full,푆 푖 are the CF operators in Stevens operator equivalent notation, is the electronic Bohr magneton, is푞 the identity matrix and is the g-matrix. 푂�푘 푖 휇퐵 퐼 푔푖 41 parameters have a clear physical interpretation – they represent the loss of 푞 degeneracy of퐵 푘the metal orbitals upon complexation – however they cannot be derived. Likewise, the spin exchange terms result from a host of physical interactions such as the dipole and superexchange mecha퐽푖푗nisms, yet are not simply separable.
The ab initio approach
The aim of the ab initio approach is a direct numerical approximation for the elusive form of the radial function. The simplest and perhaps most widely used ab initio approach is the Hartree-Fock (HF) method. With the many-body wavefunction decomposed into simpler single-electron functions, the HF method constructs the wavefunctions for each Molecular Orbital (MO) from a combination of atom-centred
AOs; this is the Linear Combination of Atomic Orbitals (LCAO) approach.* In a further computational approximation, the radial parts of the atomic basis functions are composed of multiple Gaussian functions.† These basis sets can contain only a few or a very large number of AOs, which provide increased flexibility in the MOs they seek to describe and therefore with a larger basis set the MOs approach the true HF spatial wavefunctions. The HF method seeks a numerical solution by iteratively minimizing the total energy with respect to the MO expansion coefficients; the variational principle provides that the HF wavefunction is given at the stationary point. This procedure, known as the Self-Consistent Field (SCF), usually converges well for molecules that are aptly described by HF theory, providing the initial trial MOs are of ample quality.
Despite these numerical approximations, and not forgetting the previous approximations, HF theory can provide useful results for closed shell molecules at equilibrium geometry in the electronic ground state. The approximations contained in the HF solution to the non-relativistic electronic many-body wavefunction, excluding the necessity of a finite basis set, are the static and dynamic correlation effects. More
* This is only one approach to construct the wavefunction; any complete set of functions, such as plane waves for example, can also be used. † Sometimes other functions, such as Slater functions, are also used. 42 advanced ab initio calculations seek to remedy these deficiencies in a number of different manners, however only one of these that is of particular use for magnetically anisotropic systems is examined here.
Static correlation arises as a limitation of using a single Slater determinant to describe the electronic wavefunction. Cases with complete or near orbital degeneracy simply cannot be accounted for in the HF method and multiple determinants are required, an approach generally known as Multi-Configurational SCF (MCSCF). In reality, the electrons in a molecule can occupy any orbital (whilst obeying the Pauli exclusion principle); a situation referred to as Configuration Interaction (CI). The CI wavefunction is a set of linear combinations of all the possible determinants and directly accounts for the static correlation by allowing the electrons to freely occupy the degenerate or low-lying MOs. In fact, in the limit of full CI the dynamic correlation is also completely accounted for, and if coupled with an infinite basis set, the numerical solution approaches the true solution to the non-relativistic BO electronic many-body wavefunction. While performing this calculation would be the perfect scenario, the number of determinants rises extraordinarily quickly with the number of electrons and
MOs, as described by the Weyl formula (53),12,* becoming computationally prohibitive and therefore this limit is never achieved in practice. The calculation then becomes a balancing act; achieving the desired level of accuracy while encompassing the necessary physics is not always a simple task.
2 + 1 + 1 + 1 = (53) + 1 표 +표 + 1 푆 푁2 2푁 푑푒푡 푒 푒 푁 표 �푁 � �푁 � 푁 − 푆 푆
* The terms in brackets are binomial coefficients and is the number of orbitals.
푁표 43
Luckily, there are a number of strategies which can be employed to assist with such a task – the focus here is on the use of Complete Active Space SCF (CASSCF) calculations.
In the CASSCF method, a subset of the total MO space is chosen to be ‘active’ in which full CI is performed, while the remaining doubly occupied MOs are ‘inactive’ and those unoccupied are ‘external’, Figure 4. e benet of this approach is that only determ inants with signicant contributions to the wavefunction are included, greatly adding to the describable physics over HF theory while drastically reducing the number of determinants from full CI. In some cases it is a simple task to determine which orbitals should be active and in others it can be much more dicult, however the choice is highly dependent on the information required from the calculation.
Figure 4 | Schematic of the MO space used in the CASSCF approach.
Although the CASSCF method is very successful at accommodating static correlation in the many-body wavefunction, the dynamic correlation is not fully accounted for due to the use of a one-electron MF potential. With a good approximation to the zeroth order wavefunction, the eects of dynamic correlation can be accounted for with PT, where the eect of full CI can be approximated. Approaches such as Complete Active Space
Perturbation eory (CASPT2) 13 or N-Electron Valence State Perturbation eory
(NEVPT2)14 generally consider a large number of single and double electronic 44 excitations on top of the CASSCF wavefunction in order to approximate the full CI energies, thus recovering the dynamic correlation.
Relativistic effects become important for the electronic structure of molecules containing heavy elements or even just molecules with unquenched orbital angular momentum. Fully relativistic calculations with the Dirac equation are only currently tractable for molecules of a few atoms and therefore most approaches employ pseudo- relativistic approximations. On top of the non-relativistic approach, this usually consists of two steps resulting from the Douglas-Kroll-Hess transformation:15
1) use of a contracted basis set to account for the scalar relativistic effects; and
2) inclusion of the SOC.
While the use of an appropriate basis set is straightforward, the inclusion of SOC is somewhat more troublesome. Ideally the SOC would be employed during the orbital and CI optimization of the CASSCF wavefunction, however this is not currently a wide- spread approach in electronic structure packages. More commonly the SOC is applied after the CASSCF wavefunction in either a perturbative or non-perturbative manner.
Connection of the ab initio and parametric approaches
In the limit of a perfect ab initio calculation with no approximations, all interactions and effects would be implicitly accounted for and the magnetic properties could be directly obtained. However, such a calculation is a distant dream and the approximation techniques outlined previously must be used. Due to these approximations, results from even the most high-level ab initio calculations cannot match reality. By contrast, the parametric approach is simply designed to reflect experiment and therefore the errors are confined to the simplifications within the chosen model. In complicated situations however, the number of model parameters is much greater than can be reliably determined from the available experimental data and the problem is said to be over- parameterized. Simplifications are usually made so that some parameters can be
45 removed, but even so, it is not always possible to find a unique solution. In such cases, ab initio calculations can yield approximate values for the parameters such that it may be possible to find a solution for the parametric model. From another perspective, it could be viewed that due to the approximations made in the inherently inflexible ab initio approach, the parametric model provides an opportunity to correct the calculation by direct comparison to experiment. In reality, it is often a combination of the two perspectives that highlight the synergistic manner in which the parametric and ab initio approaches work together.
The restriction of the parametric basis to that of the angular momentum of the metal ion(s) implies an ionic form of the metal ligand interactions, i.e. the open shell orbitals remain purely metallic in character, which is in stark contrast to the ab initio MO approach. However, even in MO approaches, the magnetic orbitals usually maintain mostly metal character. Therefore in the simplest approximation, only the interactions with the d- or f-orbital populations are required, leading a relatively straightforward
CASSCF approach in which the nascent metal orbitals define the active space. This type of active space is described as minimal because it is likely that the electrons involved in the coordination bonds to the metal ion (i.e. the ligand lone pairs) have some metal character and therefore should really also enter the multi-configurational wavefunction.
Despite this, the minimal active space is usually quite successful in describing qualitative, and sometimes quantitative, features of the system. Indeed, this technique is very successful for 3d and 4f complexes, which form the vast majority of studied systems, because the orbitals do not mix too much with those of the ligand. Complexes of the 4d, 5d and 5f elements are not so straightforward, as these metals possess more radially extensive magnetic orbitals which may have significant ligand interactions. In these cases more care must be taken in the choice of active space to correctly describe the magnetic properties.
46
Experimental observables
The electronic structure, once determined by the methods outlined in the preceding sections, can be utilized to calculate many experimental observables and thus can be judged against available data. The magnetization of any electronic state is defined as the gradient of its energy as a function of magnetic field and thus the equilibrium molar magnetization is the thermally averaged sum over each state, (54).* The molar magnetic susceptibility is the derivative of the molar magnetization vector and therefore defines a second-rank tensor, (55).16,†
1 = 푑푖푚 (54) −퐸푖 푖 푘퐵푇 훼 휕퐸 푀 퐵 � − 훼 푒 푍휇 푖=1 휕퐵
푑푖푚 푑푖푚 −퐸푖 2 −퐸푖 ⎡ 휕퐸푖 휕퐸푖 푘퐵푇 휕 퐸푖 푘퐵푇 ⎤ = 퐵 , 푍 �� 훼 훽 푒 − 푘 푇 � 훼 훽 푒 � (55) 10 ⎢ 푖=1 휕퐵 휕퐵 푖=1 휕퐵 휕퐵 ⎥ 푁퐴 ⎢ ⎥ 휒훼 훽 2 푑푖푚 푑푖푚 푘퐵푇푍 ⎢ −퐸푖 −퐸푖 ⎥ ⎢ 휕퐸푖 푘퐵푇 휕퐸푖 푘퐵푇 ⎥ − �� 훼 푒 � �� 훽 푒 � ⎢ 푖=1 휕퐵 푖=1 휕퐵 ⎥ EPR spectra can also be calculated,⎣ considering the net absorption of ⎦microwave intensity by the sample, (56) and (57), where , , , is a lineshape function of the energy difference between and , the microwave푓�퐸푖 퐸푗 휂푖푗 frequency푣� and the linewidth
.‡ 퐸푖 퐸푗 푣
휂푖푗
, −퐸푖 −퐸푗 퐵 퐵 = 푖 푗∈푑푖푚 + 푘 푇 푘 푇 (56) 2 2 �푒 − 푒 � ′ ′ 퐼 � ���푗�푇��푥���⃑��푖�� ��푗�푇��푦���⃑��푖�� � 푖<푗 × , , , 푍 푓�퐸푖 퐸푗 휂푖푗 푣�
* Note that the magnetization is a vector and thus (54) expresses a single Cartesian component, . = −퐸푖 is the partition function, is the Boltzmann constant and is the temperature. 훼 † is the푑푖푚 Avogadro푘퐵푇 constant. 푍 ∑푖=1 푒 푘퐵 푇 ‡ The and unit vectors are mutually orthogonal to the static magnetic field in unpolarized 푁퐴 perpendicular′ mode′ EPR. 푥���⃑ 푦���⃑ 47
= 푁퐼 + (57) 휇퐵 푇��퐵�⃑� � �퐿�⃑푖 ∙ 퐼 푆⃑̂푖 ∙ 푔푖� ∙ 퐵�⃑ �퐵�⃑� 푖=1 Conclusion
This text has outlined some ways in which electronic, and hence magnetic, properties may be calculated. While not an exhaustive survey of all electronic structure methods, a flavour of some of the most common approaches that are particularly relevant to magnetism theory has been given. The subsequent chapters present some examples of how such techniques may be applied in the study of magnetic molecules, highlighting the pivotal role of electronic structure in the understanding of complex magnetic phenomena.
References
1 E. Schrödinger, Phys. Rev., 1926, 28, 1049.
2 P. A. M. Dirac, The Principles of Quantum Mechanics, Oxford University Press, 1958.
3 E. U. Condon and G. Shortley, The Theory of Atomic Spectra, Cambridge University
Press, 1951.
4 B. G. Wybourne, Spectroscopic properties of rare earths, Interscience Publishers, 1965.
5 T. Veszpremi,́ Quantum chemistry: fundamentals to applications, Springer, 1999.
6 P. A. M. Dirac, Proc. R. Soc. Lond. Ser. A, 1928, 117, 610.
7 P. Pyykko, Chem. Rev., 1988, 88, 563.
8 W. Kutzelnigg and J. D. Morgan III, Z. Phys., 1996, 36, 197.
9 A. Z. Capri, Nonrelativistic Quantum Mechanics, World Scientific, 2002.
10 J. J. Sakurai and S. F. Tuan, Modern quantum mechanics, Addison-Wesley Longman,
2010.
11 A. R. Edmonds, Angular momentum in quantum mechanics, Princeton University
Press, 1974.
12 J. Paldus, Phys. Rev. A, 1976, 14, 1620.
48
13 K. Andersson, P. A. Malmqvist, B. O. Roos, A. J. Sadlej and K. Wolinski, J. Phys.
Chem., 1990, 94, 5483.
14 C. Angeli, S. Borini, M. Cestari and R. Cimiraglia, J. Chem. Phys., 2004, 121, 4043.
15 M. Reiher, Theor. Chem. Acc., 2006, 116, 241.
16 R. Boča, Theoretical Foundations of Molecular Magnetism, Elsevier, 1999.
49
50
4. Paper one: “On the Possibility of Magneto-Structural Correlations:
Detailed Studies of Dinickel Carboxylate Complexes”
J. P. S. Walsh, S. Sproules, N. F. Chilton, A.-L. Barra, G. A. Timco, D. Collison, E. J. L. McInnes and R. E. P. Winpenny, Inorg. Chem., 2014, 53, 8464.
51
52
Article
pubs.acs.org/IC
On the Possibility of Magneto-Structural Correlations: Detailed Studies of Dinickel Carboxylate Complexes † † ‡ † § † James P. S. Walsh, Stephen Sproules, , Nicholas F. Chilton, Anne-Laure Barra, Grigore A. Timco, † † † David Collison,*, Eric J. L. McInnes, and Richard E. P. Winpenny † School of Chemistry and Photon Science Institute, The University of Manchester, Manchester M13 9PL, United Kingdom ‡ WestCHEM, School of Chemistry, University of Glasgow, Glasgow G12 8QQ, United Kingdom § Laboratoire National des Champs Magnetiqueś Intenses, 25, rue des Martyrs, B.P. 166, 38042 Grenoble Cedex, France
*S Supporting Information
μ ABSTRACT: A series of water-bridged dinickel complexes of the general formula [Ni2( 2- μ t t ′ t ′ t OH2)( 2-O2C Bu)2(O2C Bu)2(L)(L )] (L = HO2C Bu, L =HO2C Bu (1), pyridine (2), 3- methylpyridine (4);L=L′ = pyridine (3), 3-methylpyridine (5)) has been synthesized and structurally characterized by X-ray crystallography. The magnetic properties have been probed by magnetometry and EPR spectroscopy, and detailed measurements show that the axial zero-field splitting, D, of the nickel(II) ions is on the same order as the isotropic exchange interaction, J, between the nickel sites. The isotropic exchange interaction can be related to the angle between the nickel centers and the bridging water molecule, while the magnitude of D can be related to the coordination sphere at the nickel sites.
■ INTRODUCTION significantly, groups now routinely report field-dependent magnetization alongside variable-temperature susceptibility, With growing interest in the use of molecular nanomagnets in 1 and there is also now significant use of inelastic neutron applications such as quantum computing and magnetocaloric scattering (INS), largely driven by Güdel in the first instance.11 refrigeration,2 it is becoming increasingly important to be able The number of parameters used in the spin Hamiltonian has to relate the magnetic properties of molecular nanomagnets to also increased, primarily due to the realization that the their chemical structure. For spin-only systems, where the anisotropy of the single ions is frequently of significance in orbital angular momentum is quenched, there has been some 3−6 determining physical behavior. Recent work has also suggested success in relating the form of the exchange interaction that the form of the exchange interactions used is also vital: e.g., (either ferro- or antiferromagnetic) to structural parameters, 12 anisotropic exchange in {Cr7M} rings and antisymmetric and such studies often use the term magneto-structural 13 exchange within {Ru2M} triangles. correlations to describe these relationships. A major goal in the field of single-molecule magnets (SMMs) An important question has been which structural parameter is to increase the size of the barrier to magnetic relaxation. The to use in these correlations. Two of the most commonly height of this barrier depends upon both the total ground state encountered parameters are the bridging bond angle, as in spin of the molecule, S, and its axial anisotropy, D, according to fi ’ Hat eld s famous correlation for hydroxide-bridged copper(II) the equation dimers,7 and the bond distance between the metal and the 5 ̈ 2 bridge, as used by Gorun and Lippard and developed by Gudel ΔUDSeff = (1) and Weihe8 and later Christou and co-workers.9,10 These correlations have been of huge importance in developing the Given the apparent dominance of the spin term in this area of molecular magnetism, but they were all derived during a relationship, much of the early effort in the field was spent time when chemists generally only measured variable-temper- trying to maximize S. However, it was soon realized that the ature susceptibility as a magnetic observable, and when the only anisotropy could not be neglected and that even a huge ground- terms in the spin Hamiltonian were the Zeeman term and the state spin could result in a poor SMM if the anisotropy was 14 exchange interaction. This modeling approach makes the negligible. This has encouraged a shift toward the use of ions assumption that the exchange interaction, J, is dominant and with a large intrinsic anisotropy, since these are more likely to is often called the “strong-exchange” limit, or the “giant spin translate to a large D in clusters (the cluster D approximates to approximation” (GSA). Now, encouraged by friendly physicists, our physical Received: May 6, 2014 measurements are more comprehensive. Perhaps most Published: July 25, 2014
© 2014 American Chemical Society 8464 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
53
Inorganic Chemistry Article
Table 1. Crystallographic Data for Compounds 3−6
34 5 6 · formula C40H58N4Ni2O9 C42H72N2Ni2O13 C44H66N4Ni2O9 C2H3NC40H78Mg2−xNixO17 fw 856.3 930.4 953.5 879.64 cryst syst monoclinic monoclinic orthorhombic monoclinic
space group P21/cC2/c Pna21 P21/n a,Å 10.7520(4) 24.2188(8) 20.204(1) 12.0516(6) b,Å 20.1962(5) 19.6774(9) 10.7563(6) 19.9828(8) c,Å 42.546(1) 10.4136(5) 23.1760(1) 22.9732(12) β, deg 92.358(2) 98.088(4) 90 103.790(5) V,Å3 9231.1(5) 4913.4(4) 5036.6(5) 5373.1(4) T,K 100(2) 100(2) 150(2) 150(2) Z 84 4 4 ρ −3 calcd,gcm 1.232 1.258 1.257 1.087 λ,Å/μ,mm−1 0.71073/0.867 0.71073/0.825 0.71073/0.802 0.71073/0.104 fl θ no. of r ns collected/2 max, deg 36116/52.74 9491/52.74 12471/52.74 17587/52.74 no. of rflns unique/I >2σ(I) 18841/13977 5031/4074 6920/5113 10874/6792 no. of params/restraints 1064/23 282/0 584/2 568/0 R1/goodness of fit 0.0598/1.050 0.0489/1.050 0.0569/1.039 0.0643/1.029 wR2 (I >2σ(I)) 0.1496 0.1332 0.1263 0.1415 residual density, e Å−3 0.91/−0.69 1.28/−0.52 0.58/−0.84 0.43/−0.36 a tensor sum of the single-ion anisotropy terms). Lanthanides a literature procedure,19 whereas 2 and 3 were prepared by 20 have proven to be very promising in this regard, with many of modification of published methods. μ μ t t t 2 the recent energy barrier record holders utilizing their high [Ni2( 2-OH2)( 2-O2C Bu)2(O2C Bu)2(HO2C Bu)2(C5H5N)2]( ). A light 15 green solution of 1 (1.00 g, 1.10 mmol) in Et2O (10 mL) was treated intrinsic spin and anisotropy terms. with pyridine (0.18 mL, 2.32 mmol). After 1 h of stirring at ambient Similar success might be possible using transition-metal temperature, MeCN (4 mL) was added and the solution left to stand elements, with the added advantage of a more diverse chemistry overnight, after which time diffraction-quality crystals had formed. and a significantly greater natural abundance. Unfortunately, These were collected by filtration, washed with cold MeCN, and dried fl the most anisotropic of the d-block ions are, by definition, not under a ow of N2. Yield: 0.33 g (33%). Anal. Found: C, 53.11; H, 16 well described using spin-only models, and certainly not with 7.79; N, 3.36. Calcd for C40H68N2Ni2O13: C, 53.24; H, 7.60; N, 3.10. μ μ t t 3 models based on the GSA. There is therefore a real need for [Ni2( 2-OH2)( 2-O2C Bu)2(O2C Bu)2(C5H5N)4](). The same proce- dure was used as described for 2 using excess pyridine (0.50 mL, 6.45 chemists and physicists to devote more research toward mmol). Yield: 0.41 g (44%). Anal. Found: C, 55.98; H, 6.93; N, 6.43. understanding exactly which factors dictate the exchange Calcd for C40H58N4Ni2O9: C, 56.10; H, 6.83; N, 6.54. μ μ t t t 4 interactions between anisotropic ions. The question arises: [Ni2( 2-OH2)( 2-O2C Bu)2(O2C Bu)2(HO2C Bu)2(CH3C5H4N)2](). which, if any, magnetic parameter should be used within these The same procedure was used as described for 2 using 3- correlations? This is especially important, as it dictates the methylpyridine (0.20 mL, 2.26 mmol). Yield: 0.30 g (29%). Anal. Hamiltonian used to fit the data. Found: C, 54.20; H, 7.86; N, 2.91. Calcd for C42H72N2Ni2O13:C, 54.22; H, 7.80; N, 3.01. A well-cited magneto-structural study by Halcrow et al. μ μ t t 5 [Ni2( 2-OH2)( 2-O2C Bu)2(O2C Bu)2(CH3C5H4N)4](). The same reveals a linear relationship between isotropic exchange, J, and procedure was used as described for 2 using excess 3-methylpyridine − − 9 the Ni O Ni angle in oxo-bridged nickel(II) cubanes but (0.50 mL, 5.65 mmol). Yield: 0.30 g (29%). Anal. Found: C, 57.94; H, does so using only the temperature-dependent susceptibility, 7.51; N, 6.06. Calcd for C44H66N4Ni2O9: C, 57.92; H, 7.44; N, 6.14. μ μ t t t 6 under the assumption that the effect of the zero-field splitting [Mg2−xNix( 2-OH2)( 2-O2C Bu)2(O2C Bu)2(HO2C Bu)4](). · · · · (ZFS) is negligible at higher temperatures: i.e., it implicitly uses 4MgCO3 Mg(OH)2 4H2O (8.00 g, 17.1 mmol), 2NiCO3 3Ni(OH)2 4H2O (0.5285 g, 0.8994 mmol), and pivalic acid (40.0 g, 39.2 mmol) the GSA. More recent work by Hill and co-workers has fl ° revealed that the magnitude of the ZFS in nickel(II) is actually were stirred under re ux (160 C) for 24 h. The mixture was then cooled to room temperature and dissolved fully in an excess of diethyl non-negligible in these compounds.17 A similar correlation of J − − ether (200 mL). MeCN (30 mL) was added with thorough stirring, with Ni O Ni angle has also been proposed by Thompson and the solution was left to stand partially open to allow slow 18 and co-workers. evaporation. Large single crystals formed after 2 days. Yield: 6.32 g Here we report a magnetostructural study on a family of five (39.9%). Anal. Found: C, 54.47; H, 8.61; Mg, 5.65; Ni, 0.58. Calcd for structurally related nickel(II) dimetallics where the zero-field C40H78Mg1.95Ni0.05O17: C, 54.40; H, 8.90; Mg, 5.23; Ni, 0.66. splitting is on the same order of magnitude as the exchange X-ray Crystallography. The single-crystal structures of 1 and 2 have been reported previously.19,20 Single crystals of 3−6 were interaction between the ions. The aim is to examine whether ff fi mounted in the nitrogen cold stream of an Oxford Di raction we can still nd similar magneto-structural correlations when a XCalibur 2 diffractometer. Graphite-monochromated Mo Kα radiation more complex Hamiltonian is required, and where more data (λ = 0.71073 Å) was used throughout. Final cell constants were are available. obtained from least-squares fits of all measured reflections. The structures were solved by direct methods using SHELXS-97.21 Each Δ ■ EXPERIMENTAL SECTION structure was completed by iterative cycles of F syntheses and full- matrix least-squares refinement. All non-H atoms were refined Synthesis. All reagents, metal salts, and ligands were used as anisotropically. Difference Fourier syntheses were employed in μ μ obtained from Sigma-Aldrich. [Ni2 ( 2 -OH2 )( 2 - positioning idealized methyl hydrogen atoms, which were assigned t t t O2C Bu)2(O2C Bu)2(HO2C Bu)4](1) was synthesized by following isotropic thermal parameters (U(H) = 1.5Ueq(C)) and allowed to ride
8465 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
54
Inorganic Chemistry Article
Figure 1. Structures of the neutral complexes 1−6 in the crystal state. Data for 1 and 2 are from refs 26 and 20, respectively. All hydrogens are omitted, with the exception of those found crystallographically on the bridging water. Hydrogen bonds are indicated by dashed lines. on their parent C atoms (C−H 0.93 Å). Some pivalate groups bridging pivalates, each metal also bears a monodentate pivalate exhibited rotational disorder. This was modeled by allowing for two ligand that provides a stabilizing hydrogen bond at ∼2.5 Å to conformations of the tert-butyl group and refining their occupancy fi 2 21 the bridging water molecule. The coordination sphere of each factors. All re nements were against F and used SHELXL-97. metal ion is completed by two pivalic acid groups, and the Crystallographic data are collected in Table 1. CCDC reference numbers: 926884−926886 (3−5) and 999472 (6). complex has an overall neutral charge. Physical Measurements. Electronic absorption spectra were Substitution of the terminal pivalic acid groups in 1 occurs in collected on a PerkinElmer Lambda 1050 spectrophotometer. IR two stages. The pivalic acid cis to the bridging water, which is spectra of neat powders were recorded using a Thermo Scientific more labile than its trans counterpart, departs first. Compounds Nicolet iS5 FTIR spectrometer equipped with an iD5 ATR. Variable- 2 and 4 are generated by treatment with 2 equiv of pyridine and temperature (2−300 K) magnetic susceptibility measurements were 3-methylpyridine, respectively, in 30% yield (Figure 1). Further fi recorded in a 0.1 T magnetic eld on a SQUID magnetometer substitution utilizes an excess of pyridine and 3-methylpyridine (Quantum Design MPMS-XL). The experimental magnetic suscept- to afford 3 and 5, respectively, in similar yields (Figure 1). ibility data were corrected for underlying diamagnetism using ’ Infrared (IR) spectra of this series do not display any tabulated Pascal s constants, and the simulations of both magnetization ν and susceptibility were performed using PHI.22 Q-band EPR data were terminal (OH) stretches from either the bridging water or collected on a Bruker EMX spectrometer, and high-frequency, high- pivalic acid ligands. This is due to the formation of field EPR spectra were recorded at the LNCMI-CNRS at Grenoble on intramolecular hydrogen bonds, with the aforementioned a home-built spectrometer.23 EPR spectra were simulated using interaction between the bridging water and the available 24 EasySpin. Analytical data were obtained by the microanalytical oxygen atom of the monodentate pivalate ligands, and also service of The University of Manchester. between the pivalic acid protons and their neighboring Computational Details. All CASSCF calculations were performed 25 monodentate and bridging pivalate groups. This results in a with MOLCAS 7.8 using the RASSCF, RASSI, and SINGLE_- − ν ANISO modules. In all cases the ANO-RCC basis sets were used, weakening of the O H bond, shifting it to the (CH) region − −1 where the metal ion of interest was treated with TZVP quality, the first (2800 3000 cm ; Figure S1, Supporting Information). coordination sphere (and bridging water hydrogen atoms) was treated The change in electronic structure upon substituting pivalic with VDZP quality, and all other atoms were treated with VDZ quality. acid for stronger field pyridine ligands is evident in the The two electron integrals were Cholesky decomposed with the electronic spectra of this series (Figure 2). The low-energy default settings. region (<25000 cm−1) shows three ligand field (LF) transitions at ∼9000, ∼15000, and ∼25000 cm−1 (Table 2). This profile ■ RESULTS AND DISCUSSION 2+ 28 bears a striking resemblance to that of [Ni(OH2)6] , and the Synthesis and Characterization. The entry point for this spectra have been interpreted assuming approximate octahedral μ μ t t t 8 3 → 3 series is [Ni2( 2-OH2)( 2-O2C Bu)2(O2C Bu)2(HO2C Bu)4] symmetry at each d ion. The lowest energy A1g T2g (1), whose preparation involves heating nickel carbonate or excitation is a measure of the LF, following the trend 1 < 2, 4 < nickel hydroxide in pivalic acid.19 The compound contains two 3, 5. The additional methyl substituent in 4 and 5 has no effect nickel(II) ions bridged by one water and two pivalate ligands on the transition energies. − (Figure 1).26 Structures of this type are well-known for nickel The LF splitting of 1 at 8440 cm 1 matches that of the 27 28 and a host of other divalent metal ions. In addition to the hexaaquanickel(II) ion, which is not unexpected for a NiO6
8466 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
55
Inorganic Chemistry Article
comparison to NiO5N and NiO4N2), thus decreasing the ZFS effects. Crystal Structures. The structures of 3−6 have been determined by single-crystal X-ray diffractometry and con- trasted with the structures of 1 and 2. Salient metric parameters are collated in Table 3. Compounds 1−5 all contain the μ t μ {Ni2( 2-O2C Bu)2( 2-OH2)} core and vary only in the remaining two coordination sites at each nickel center, which are incrementally changed from two pivalic acid groups to two pyridine ligands. The oxygen atom of the water molecule (Ow) adopts a pseudotetrahedral geometry. The water protons are μ t aligned essentially parallel with each {Ni( 2-O2C Bu)Ni} plane, which in turn lie at an angle of ∼80° to each other. The − Figure 2. Electronic absorption spectra of 1 5 recorded in Et2O hydrogen bond between the water molecule and the terminal solutions at ambient temperature. pivalates is invariantly ∼2.5 Å across the series. The nickel − a centers display only slight distortion from regular octahedral Table 2. Assignment of LF Transitions in 1 5 geometry, with angles between adjacent donors less than 5° 12345 away from normal. ff 3A → 3T (F) 8440 8830 9380 8830 9360 The e ect of introducing pyridine ligands is assessed by 2g 2g μ 3 → 1 monitoring the structural parameters of the {Ni2( 2-OH2)} A2g Eg(D) 13350 13380 13370 13430 13310 fi 3 → 3 unit (Table 3). The rst substitution forming 2 and 4 is A2g T1g(F) 14580 15100 15800 15120 15810 − − ··· 3 → 3 accompanied by an elongation of the Ni Ow,Ni Ot, and Ni A2g T1g(P) 24830 25190 26160 25280 26230 ∼ ∼ ∼ a −1 Ni distances by 0.07, 0.06, and 0.1 Å, respectively. In Energy in cm . contrast, the Ni−O bonds with the bridging and terminal pivalate ligands are essentially unchanged, with the terminal − ff coordination sphere. A uniform increase of ∼400−500 cm 1 is pivalates cis to the substitution site slightly more a ected than observed when the π-donating pivalic acid is replaced by a σ- those in the trans position. The addition of pyridine decreases donating pyridine to generate NiNO and NiN O centers in 2 the overlap between the metal and bridging water, lengthening 5 2 4 − − and 4 and in 3 and 5, respectively. The two higher energy the bond and slightly reducing the Ni Ow Ni angle. It also transitions are similarly shifted. Each complex also exhibits a weakens the bond with the remaining pivalic acid ligand, fl 3 → 1 peak that we assign as the spin- ip A2g Eg excitation, whose promoting a second substitution. The pyridines lie parallel to 3 → intensity is enhanced by proximity to the spin-allowed A2g each other at distances typical for this ligand, and the additional 3 T1g transition. This is most clearly seen in the spectrum of 1, methyl substituents in 4 have no bearing on the overall − ··· − − where the peaks at 13350 and 14580 cm 1 appear to have the topology. The Ni Ni distance exceeds 3.5 Å, and the Ni Ow ∼ ° same intensity and are difficult to differentiate. However, in 3 Ni angle expands to 115 when all four pivalic acids are − − and 5, the stronger ligand field blue-shifts the spin-allowed replaced by pyridine. If the Ni Ow Ni angle were the excitation, leaving a weak shoulder to lower energy. Because dominant structural parameter, then we would expect this peak is essentially independent of the crystal field, it is compounds 1, 2, and 4 to be similar to each other and 3 → 1 fi ff assigned as the A2g Eg transition in all ve compounds. di erent from compounds 3 and 5. fi | | 3 Given that the zero- eld splitting, D , of the A2g term is, to a Compound 6 is the nickel-doped magnesium analogue of first approximation, inversely proportional to the magnitude of compound 1 and was prepared so that we could directly the ligand field, we would expect this zero-field splitting to be measure the single-ion parameters of nickel in a near-identical larger in 2 and 4 than in 3 and 5, and this is indeed what is environment, but in the absence of exchange coupling. The observed (vide infra). The smaller magnitude of |D| in 1, metric parameters are not expected to be identical between 1 despite having a ligand field strength smaller than that of 2 and and 6, given the significant difference in the metal radii between 4, might be explained by 1 having a coordination environment magnesium and nickel and the fact that the space group is ff (NiO6) that is more appropriately treated as octahedral (in slightly di erent (Pbca in 1, P21/n in 6), but they are actually
Table 3. Salient Average Bond Distances (Å) and Angles (deg) for 1−6a
123456 Ni···Ni 3.361(1) 3.465(2) 3.5092(6) 3.4760(7) 3.511(1) 3.462(1) Ni−O−Ni 111.2(1) 110.7(3) 115.8(1) 110.6(1) 114.7(2) 112.5(1) − Ni Ow1 2.037(2) 2.106(4) 2.075(3) 2.087(2) 2.085(5) 2.080(2) − Ni Ob2 2.018(3) 1.992(5) 2.045(3) 2.034(2) 2.061(5) 2.032(2) − Ni Ob3 2.032(3) 1.983(5) 2.027(3) 1.998(2) 2.023(5) 2.063(2) − Ni Ot3 2.053(3) 2.069(5) 2.069(3) 2.070(2) 2.058(5) 2.092(2) − Ni L1 2.070(3) (O) 2.142(5) (O) 2.096(4) (N) 2.118(2) (O) 2.086(7) (N) 2.073(2) − Ni L2 2.080(3) (O) 2.095(7) (N) 2.128(4) (N) 2.089(3) (N) 2.135(7) (N) 2.097(2) a ∑ Values were calculated using ( xi/n), where x is the bond metric and n is the number of values averaged. The subscript w denotes the oxygen of the bridging water group, b the bridging pivalate, and t the terminal pivalate. L represents the atoms at the labile terminal positions. 1−3 denote pairs trans to each other.
8467 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
56
Inorganic Chemistry Article
Figure 3. (A) Overlay of the temperature dependence of the magnetic susceptibility recorded on powders of 1−5 under a static field of 0.1 T. (B) Field-dependent magnetization of compounds 1−5 measured at 2 and 4 K. Solid lines in all graphs represent simulations using the parameters in Table 6. quite close and fall within the ranges set by 1−5. We can equivalent, even though this is only strictly true for the therefore be confident that the parameters we obtain will be of symmetry-related ions in 2 and 4, and the exchange is treated as relevance to the studies on the pure compounds. isotropic. Magnetometry. The temperature dependence of the χ χ ̂ ̂ ̂ ⃗ ̂ ̂ ̂ ̂ ̂ ̂ product MT (where M = molar magnetic susceptibility and H = gSβ()12+ S· H+ S 1122··DD S+ S·· S− 2 JSS 12· (2) T = temperature) for 1−5 are shown in Figure 3A. For all χ compounds, the room-temperature MT values of around 2.5 Good fits to the data can be obtained using this model but − − cm3 K 1 mol 1 correspond well to the expected spin-only value are ambiguous with regard to the sign of the axial ZFS − for two uncoupled S = 1 ions with giso = 2.2 2.3. The value is parameter, D, and are unable to offer a conclusive measure of constant until around 80 K for all compounds. At low the exchange interaction, which appears to be smaller than the temperature, the plot follows markedly different profiles for 1 ZFS in 2−4. In fact, the isotropic exchange parameter, J, can be and 4, which rise, suggesting ferromagnetic coupling, in fi varied over a range of values (keeping the tted D and giso comparison to those for 2, 3, and 5, which exhibit a sharp values fixed) before adversely affecting the“goodness” of the drop at low temperatures. It is tempting to attribute this drop simulations. These ranges for the value of J (in units of cm−1) to an antiferromagnetic exchange (indeed, this was done by are as follows: 1, 2.30−2.90; 2, 0.10−0.35; 3, −0.05 to −0.20; others29 for compound 2), but this neglects the effect of the − − − fi 4, 0.55 0.85; 5, 0.25 to 0.45. zero- eld splitting (ZFS) or at least assumes that it is much As a result, it is impossible to deduce the exchange (and, smaller than the exchange. To illustrate this point, we modeled importantly, to compare values across the series) from fitting the susceptibility alone for 2 and obtained a J value of −0.5 −1 − the magnetic data alone; further data are clearly required to cm (using the 2J convention). In reality, the ZFS can easily understand the magnetic behavior of these simple compounds be on the order of the exchange interaction in compounds fi unambigiously. Here, the additional data are EPR spectroscopy containing octahedral nickel(II) and may even be signi cantly at 331 GHz on pure and doped samples and CASSCF larger.30 calculations to yield estimates of the g values and anisotropy In compound 1, the upturn in the susceptibility below 50 K − − parameters for single-ion sites. culminating in a χ T value of 3.9 cm3 K 1 mol 1 at 2.5 K, M INS data for compound 1 were actually collected over a before a sharp downturn due to ZFS, is consistent with an S =2 decade ago;19 INS and magnetic susceptibility were used to ground state of two ferromagnetically coupled nickel(II) ions. model the exchange and ZFS parameters, but the poor quality We rule out the possibility of weak intermolecular interactions fi as the cause of this downturn by noting that in compounds 1−5 of the INS data meant that the rhombic term of the zero- eld the intermolecular Ni···Ni distances are never below ∼9Å. splitting tensor, E, was neglected and only isotropic exchange χ was employed, a scheme that we will show is inappropriate for Compound 4 also exhibits an upturn in MT upon cooling below ∼40 K that is likely due to a ferromagnetic interaction, these systems (vide infra). The INS data are reproduced in the albeit weaker than that in 1, with ZFS again resulting in the Supporting Information and are consistent with the model we onset of a sharp downturn (∼10 K) that prevents the arrive at from the present EPR studies. susceptibility reaching the value for a pure S = 2 state. Single-Ion Anisotropy: EPR Spectroscopy. To deter- The field-dependent magnetization curves are given in Figure mine the single-ion parameters (g, D, and E)in1 without the 3B. Compound 1 exhibits a sharp rise at low fields, with the 2 K added complexity of exchange coupling, we synthesized a data reaching saturation above 5 T. In contrast, compound 2 diamagnetic magnesium analogue doped with 5% nickel (6). At exhibits a very shallow rise that fails to reach saturation even at this doping level, the amount of pure Ni−Ni molecule is 7 T. Compounds 3−5 exhibit behavior that falls between these expected to be virtually undetectable in comparison to the two extremes. amount of Mg−Mg (diamagnetic) and Mg−Ni molecules, and For all five compounds, we begin by fitting the temperature- thus the spectrum should be that of the individual nickel ions dependent susceptibility alongside the field-dependent magnet- (i.e., the exchange coupling interaction is effectively turned off). ization using the spin Hamiltonian given in eq 2. In this model, The powder HFEPR spectrum (Figure 4 and Figure S3 the single-ion anisotropies of the two nickel ions are assumed (Supporting Information)) reveals sharp transitions localized at − − − to be axial (Dxx = Dyy = D/3 and Dzz = 2D/3) and also 5.0 5.5 T, with broader features over the range 8.5 13.5 T,
8468 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
57
Inorganic Chemistry Article
called “double shell effect”) proved only to affect the results in a minor fashion (Table 5), and therefore this was not investigated
Table 5. Calculated g, D, and E Parameters for the S =1 Ground State with 8 Electrons in 10 Orbitals, for 1
−1 −1 compd, site gx gy gz D (cm ) E (cm ) 1, Ni1 2.32 2.32 2.29 +4.0 −0.3 1, Ni2 2.31 2.31 2.29 +3.0 −0.2
for the remaining complexes. It is also possible to extract the orientations of the D tensor and the g matrix (see the Supporting Information). The calculated parameters for 1 and the parameters obtained Figure 4. Powder electron paramagnetic resonance spectra of 6 from simulations of 6 differ, chiefly in the magnitude of D (ab measured at 331.2 GHz at 5 and 15 K, with the simulation shown in initio, +4.4/+3.2 cm−1; experimental, +1.6/+1.1 cm−1). The red. gx1 = gy1 = 2.32, gz1 = 2.25, gx2 = gy2 = 2.25, gz2 = 2.19, D1 = +1.6 −1 −1 −1 −1 observed axiality of g is reproduced and is of the same sense (gz cm , E1 = +0.35 cm , D2 = +1.1 cm , E2 = +0.20 cm . Asterisks 32,33 indicate the appearance of a pair of features upon warming to 15 K < gx = gy), which is consistent with a positive D. What is that is recreated in the simulated spectrum. most noticeable about the calculated parameters is the extreme sensitivity of D to the nickel coordination environment. This parameter varies from +8.7 to 3.0 cm−1 with only minor and resembles the spectrum expected of an S = 1 ion with a changes in coordination sphere. This covers as wide an energy rhombic ZFS. range as the observed exchange interactions in nickel These spectra can be modeled as two nonequivalent nickel compounds.9,18 The experimental D value for the isolated sites with the following parameters: gx1 = gy1 = 2.32, gz1 = 2.25, nickel(II) ions in 6 suggests that the calculated values are −1 −1 gx2 = gy2 = 2.25, gz2 = 2.19, D1 = +1.6 cm , E1 = +0.35 cm , D2 somewhat overestimated. The calculated g values are also much −1 −1 = +1.1 cm , E2 = +0.20 cm . Reassuringly, the simulation less sensitive to coordination environment than is found recreates the appearance of a pair of features at around 11.5 T experimentally. (marked with an asterisk) upon warming from 5 to 15 K, which Dinickel Compounds: EPR Spectroscopy. Compounds can be attributed to transitions into the mS = 1 state from the 1−5 were first measured in the powder state at lower increasingly thermally populated mS = 0 state. frequencies (see the Supporting Information). These spectra Single-Ion Anisotropy: Computational Modeling. are extremely difficult to interpret, due to a very limited number Taking inspiration from previous success with cobalt analogues of transitions falling within the available magnetic field range. 31 of these complexes, we modeled compounds 1−5 using Such complexity often arises in compounds where a significant complete active space calculations to generate estimates for the ZFS (which is typical for nickel) is acting on weakly coupled, anisotropic g values and individual site ZFS parameters. non-Kramers ions. As an example we show the Q-band Compounds 2 and 4 possess nickel(II) sites that are related spectrum of compound 1 (Figure 5). Examples for the other by symmetry (2-fold rotation), while compounds 1, 3, and 5 compounds are given in the Supporting Information. have independent sites, with compound 3 having not two but four independent nickel sites (two molecules per asymmetric unit). In all cases, the nickel site that is not the focus of the calculation is replaced by a diamagnetic zinc(II) ion. The active space was chosen as the five 3d orbitals of the nickel(II) ion, where all 10 S = 1 and 15 S = 0 configuration state functions were calculated and mixed by spin−orbit coupling. From these calculations, the gx, gy, gz, D, and E values for the S = 1 ground multiplet can be extracted (Table 4). Expanding the active space to include the 4d orbitals of the nickel(II) ion (the so-
Table 4. Calculated g, D, and E Parameters for the S =1 Ground State with 8 Electrons in 5 Orbitals, for 1−5
−1 −1 compd, site gx gy gz D (cm ) E (cm ) Figure 5. Q-band EPR of 1 measured in the powder state. The red 1, Ni1 2.34 2.34 2.31 +4.4 −0.3 trace is a simulation using the parameters in Table 6. 1, Ni2 2.34 2.33 2.31 +3.2 −0.2 2, Ni1 2.35 2.33 2.28 +8.7 −1.3 3, Ni1 2.29 2.30 2.31 −2.4 +0.5 To allow for the large ZFS, we need to move to frequencies 3, Ni2 2.28 2.29 2.31 −3.0 +0.7 where hν is larger than the ZFS. We therefore collected powder 3, Ni3 2.31 2.30 2.29 −1.7 +0.5 spectra (Figure 6) at a much higher frequency (331.2 GHz) 3, Ni4 2.29 2.29 2.28 +2.0 −0.4 and over a much larger field range (0−16 T). Although the 4, Ni1 2.34 2.33 2.28 +6.4 −1.3 definition of the features is greatly improved, the spectra are 5, Ni1 2.31 2.30 2.29 +3.0 −0.9 still remarkably complicated for such simple compounds, and 5, Ni2 2.30 2.30 2.28 +2.5 −0.6 the variation between spectra is considerable.
8469 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
58
Inorganic Chemistry Article
to keep gi, Di, and J coincident in all compounds to reduce the number of parameters and neglected the fact that the metal centers in 3 and 5 were crystallographically nonequivalent (i.e., the parameters for all sites were taken as identical, assuming a pseudosymmetry). fi χ We started by simultaneously tting the MT(T)and M(H,T) data to eq 2 to obtain initial values of J and Di, before introducing these parameters into eq 3 to calculate fi HFEPR spectra, now introducing Ei and re ning J and Di.Asa next step, we introduced Euler angles relating the Di tensors (assuming they are related by a 2-fold axis as in 2 and 4): these angles were fixed from CASSCF calculations (see the Supporting Information). This only made significant improve- ments inand hence were only retained forcompounds 2 and 4, which we assume is because of their much larger Di values (and possibly because they have true crystallographic equivalence of the centers). The final parameters (Table 6) χ were then used to recalculate MT(T) and M(H,T) (Figure 3). For compound 1, where we had the doped materials available, the single-ion parameters were fixed from the doping study. Because we have more information for this complex (hence fewer free variables), we investigated the effect of a small anisotropic component to J (on the order of the dipolar interaction) and found this gave an improvement to the fi − Figure 6. Solid-state high- eld EPR (331.2 GHz) spectra of 1 5. calculated EPR spectra. For 2−5, where we do not have the Experimental traces are in black, with simulations using the parameters doped analogues, J was held as isotropic. in Table 6 plotted in red. The final parameter sets used to generate all of the simulations shown in this paper (including the magnetic Compounds 1, 3, and 5 exhibit their most intense features data) are given in Table 6. The fits to the observed HFEPR between 9−14 T, with effectively no transitions at lower fields. spectra are remarkably good for 1, and the main features of 2− Although compounds 2 and 4 display transitions in this same 5 are also simulated. However, in each case additional region, they also exhibit features at lower fields, between 1 and experimental features are observed that do not arise from this 4 T, that are of a much higher intensity. These low-field simple model. The predicted trend in the magnitude of D from transitions are a signature of significantly larger ZFS terms in ab initio calculations (2, 4 > 1, 3, 5) is supported, but the these examples. calculated values themselves do not give good simulations. A The combination of the doped study and computational key result of this study is that the low-field features observed for − work gives us a guide to the single-ion parameters. To interpret 2 and 4 can only be reproduced with |D| >5cm 1. the complex spectra of 1−5, we have used the spin Hamiltonian In 2−5, there is no obvious benefit to using anisotropic g − − − given in eq 3, where now Dxx = D/3 + E, Dyy = D/3 E, and values, and so isotropic g values are retained. The small Dzz =2D/3. anisotropy in the g values is predicted by ab initio methods, and the magnitude of the g values from both experimental and ab Ĥ = β()SSHSSSSJSŜ ·+gĝ · ⃗ + ̂ ·· D̂ + ̂ ·· D̂ − 2̂ · ̂ 1 1 2 2 112212 12 initio methods are within the expected range for nickel(II) ions, (3) albeit generally overestimated in the latter. This is almost identical with the Hamiltonian used to fit the Experimentally, the sign of D is positive for all compounds, magnetic data, except that we have introduced rhombic ZFS which is in general agreement with ab initio methods, which (E) terms and also the possibility for anisotropic exchange. In predict a positive D in all compounds except for 3, where in fact all cases, the reference frames for the gi matrix and Di tensors three of the nonequivalent centers are predicted to be negative are coincident for a given metal site; however, the reference and the remaining one positive. frames may differ between different nickel sites. Additionally, J Assessing the simulation parameters given in Table 6 is always fixed in the global reference frame. We initially chose alongside the structural parameters in Table 3 allows us to
Table 6. Electronic Parameters Used in the Global Simulations of 1−5
1 2 345
gxx, gyy 2.32/2.25 2.24 2.20 2.26 2.26
gzz 2.25/2.19 2.24 2.20 2.26 2.26 D (cm−1) +1.60/+1.10 +7.40 +2.10 +5.40 +1.80 E (cm−1) +0.35/+0.20 +2.45 +0.10 +1.40 +0.10 −1 − − Jxx, Jzz (cm ) +2.40 +0.35 0.1 +0.70 0.3 −1 − − Jyy (cm ) +2.55 +0.35 0.1 +0.70 0.3 R (deg)a 0, 0, 0 +109, −88.5, −71.0 0, 0, 0 +70.6, +102, −109 0, 0, 0 aEuler rotations of one nickel site in relation to the other in the ZY′Z″ convention.
8470 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
59
Inorganic Chemistry Article comment on a number of possible correlations. First, the ■ ASSOCIATED CONTENT magnitude of the axial parameter, D, is significantly larger in 2 * fi S Supporting Information and 4. This can be explained by the presence of a well-de ned Figures and tables giving infrared spectra for 1−5, full-range axis along the nitrogen atom in the NiNO5 coordination sphere HFEPR spectra for 6, low-temperature Q-band EPR spectra for for these compounds. The addition of a further nitrogen atom 2−5, Euler rotations of g/D from CASSCF calculations, and in 3 and 5, yielding NiN2O4 with oxygen atoms cis to each the INS spectrum of 1 using data from ref 19. This material is other, is accompanied by a reduction in D, as there is no longer available free of charge via the Internet at http://pubs.acs.org. a unique metal−donor atom direction. This magneto-structural ff correlation is very clear. The dramatic di erence in the ■ AUTHOR INFORMATION measured EPR spectra of 1 in comparison with those of 2 and 4 is then due to the much larger change in D between the Corresponding Author * complexes. Second, the exchange is ferromagnetic in nature for E-mail for D.C.: [email protected]. − − 1, 2, and 4, which have the smallest Ni Ow Ni angles of the Notes series. It is worth noting that, despite the ferromagnetic The authors declare no competing financial interest. χ exchange observed in 2, the product MT falls at low temperature, which is due to D being 20 times as large as J. ■ ACKNOWLEDGMENTS This research was funded by the EPSRC (UK); R.E.P.W. holds ■ CONCLUSION a Royal Society Wolfson Research Merit Award. J.P.S.W. thanks Detailed magnetic and EPR studies carried out on five closely the North West Nanoscience DTC for a Ph.D. studentship. ’ related dinickel compounds and on a doped diamagnetic N.F.C. thanks the University of Manchester for a President s analogue of the parent compound have shown that incredibly Doctoral Scholarship. We thank the EPSRC UK National rich and diverse data can be obtained from seemingly simple Electron Paramagnetic Resonance Service at the University of compounds. At first glance, there is no simple correlation Manchester. between structure and magnetic or spectroscopic behavior, e.g. − − ■ REFERENCES compounds 1 and 2 have very similar Ni Ow Ni angles, but in the former case variable-temperature susceptibility measure- (1) Leuenberger, M. N.; Loss, D. Nature 2001, 410, 789−793. χ − ments show an upturn in MT at low temperature and in the (2) Sharples, J. W.; Collison, D. Polyhedron 2013, 54,91 103. latter a downturn. (3) (a) Ohba, S.; Kato, M.; Tokii, T.; Muto, Y.; Steward, O. W. Mol. − However, detailed analysis shows that in these five simple Cryst. Liq. Cryst. 1993, 233, 335 344. (b) Ribas, J.; Escuer, A.; Monfort, M.; Vincente, R.; Cortes, R.; Lezama, L.; Rojo, T. Coord. compounds there are two correlations. The major correlation is − fi Chem. Rev. 1999, 193, 1027 1068. (c) Ruiz, E.; Cano, J.; Alvarez, S.; between the axial zero- eld splitting parameter, D, and the Alemany, P. J. Am. Chem. Soc. 1998, 120, 11122−11129. coordination geometry; where there is a single unique axis, due (d) Rodriguez-Fortea, A.; Alemany, P.; Alvarez, S.; Ruiz, E. Chem. to the presence of a N donor in a NiO5N donor set, the D value Eur. J. 2001, 7, 627−637. (e) Mialane, P.; Duboc, C.; Marrot, J.; is around 3 times larger than in a NiO6 donor set or in a cis Riviere, E.; Dolbecq, A.; Secheresse, F. Chem. Eur. J. 2006, 12, 1950− NiO4N2 donor set. The measured thermodynamic and 1959. (f) Thompson, L. K.; Mandal, S. K.; Tandon, S. S.; Brisdon, J. spectroscopic properties vary most due to this correlation. N.; Park, M. K. Inorg. Chem. 1996, 35, 3117−3125. When this is allowed for, we find that the correlation9,18 (4) (a) Escuer, A.; Aromi, G. Eur. J. Inorg. Chem. 2006, 4721−4736. (b) Gomez,́ V.; Corbella, M.; Roubeau, O.; Teat, S. J. Dalton Trans. between bridging angle and the sign of the magnetic exchange − interaction still appears to be present. This supports the 2011, 40, 11968 11975. (5) Gorun, S. M.; Lippard, S. J. Inorg. Chem. 1991, 30, 1626−1630. hypothesis that the dominant superexchange pathway in these 31 (6) (a) Gregoli, L.; Danieli, C.; Bara, A.-L.; Neugebauer, P.; compounds is via the bridging water molecule. Pellegrino, G.; Poneti, G.; Sessoli, R.; Cornia, A. Chem.Eur. J. 2009, Magneto-structural correlations continue to appear, but these 15, 6456−6467. (b) Maganas, D.; Krzystek, J.; Ferentinos, E.; Whyte, should be treated with considerable skepticism unless one A. M.; Robertson, N.; Psycharis, V.; Terzis, A.; Neese, F.; Kyritsis, P. Hamiltonian is used to simulate all the data. For example, we Inorg. Chem. 2012, 51, 7218−7231. (c) Maurice, R.; de Graaf, C.; could easily have simulated the magnetic susceptibity data of Guihery, N. J. Chem. Phys. 2010, 133, 084307. (d) Barra, A.-L.; 1−5 to a simple isotropic Zeeman plus exchange Hamiltonian Caneschi, A.; Cornia, A.; Gatteschi, D.; Gorini, L.; Heiniger, L. P.; − and described the trends in J with respect to structure. This Sessoli, R.; Sorace, L. J. Am. Chem. Soc. 2007, 129, 10754 10762. would have had no physical meaning, because modeling the (7) Crawford, V. H.; Richardson, H. W.; Wasson, J. R.; Hodgson, D. J.; Hatfield, W. E. Inorg. Chem. 1976, 15, 2107−2110. magnetization data (requiring local anisotropy terms that are (8) Weihe, H.; Güdel, H. U. J. Am. Chem. Soc. 1997, 119, 6539− larger than J) shows not only that such a Hamiltonian is ffi 6543. insu cient but also that the J values so determined are wrong (9) Halcrow, M. A.; Sun, J.-S.; Huffman, J. C.; Christou, G. Inorg. (even giving the wrong sign). This would not be helped by Chem. 1995, 34, 4167−4177. fitting the magnetization data to a second, different (10) Cañada-Vilalta, C.; O’Brien, T. A.; Brechin, E. K.; Pink, M.; Hamiltonian (often based on a giant spin approximation), to Davidson, E. R.; Christou, G. Inorg. Chem. 2004, 43, 5505−5521. determine global zero-field splitting parameters, as this is only (11) Basler, R.; Boskovic, C.; Chaboussant, G.; Güdel, H. U.; Murrie, ≫ M.; Ochsenbein, S. T.; Sieber, A. ChemPhysChem 2003, 4, 910−926 appropriate when J Di. Hence, care needs to be taken in attempting to correlate spin Hamiltonian parameters to and references therein. (12) (a) Piligkos, S.; Bill, E.; Collison, D.; McInnes, E. J. L.; Timco, structure when the anisotropy terms are larger than or 34 G. A.; Weihe, H.; Winpenny, R. E. P.; Neese, F. J. Am. Chem. Soc. comparable to the exchange, and this requires that the data − 35 2007, 129, 760 761. (b) Piligkos, S.; Weihe, H.; Bill, E.; Neese, F.; El are treated with a single Hamiltonian. Any meaningful model Mkami, H.; Smith, G. M.; Collison, D.; Rajamaran, G.; Timco, G. A.; should fit both variable-temperature susceptibility, variable-field Winpenny, R. E. P.; McInnes, E. J. L. Chem. Eur. J. 2009, 15, 3152− magnetization, and other spectroscopic data where available. 3167.
8471 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
60
Inorganic Chemistry Article
(13) Magee, S. A.; Sproules, S.; Barra, A.-L.; Timco, G. A.; Chilton, N. F.; Collison, D.; Winpenny, R. E. P.; McInnes, E. J. L. Angew. Chem., Int. Ed. 2014, 53, 5310−5313. (14) Ako, A. M.; Hewitt, I. J.; Mereacre, V.; Clerac,́ R.; Wernsdorfer, W.; Anson, C. E.; Powell, A. K. Angew. Chem., Int. Ed. 2006, 45, 4926− 4929. (15) See for example: Blagg, R. J.; Ungur, L.; Tuna, F.; Speak, J.; Comar, P.; Collison, D.; Wernsodrfer, W.; McInnes, E. J. L.; Chibotaru, L. F.; Winpenny, R. E. P. Nat. Chem. 2013, 5, 673−678. (16) Zadrozny, J. M.; Xiao, D. J.; Atanasov, M.; Long, G. J.; Grandjean, F.; Neese, F.; Long, J. R. Nat. Chem. 2013, 5, 577−581. (17) Wilson, A.; Lawrence, J.; Yang, E.-C.; Nakano, M.; Hendrickson, D. N.; Hill, S. Phys. Rev. B 2006, 74, 140403. (18) (a) Nanda, K. K.; Thompson, L. K.; Bridson, J. N.; Nag, K. J. Chem. Soc., Chem. Commun. 1992, 1337−1338. (b) Thompson, L. K.; Brisdon, J. N.; Nag, K. J. Chem. Soc., Chem. Commun. 1994, 1337− 1338. (19) Chaboussant, G.; Basler, R.; Güdel, H.-U.; Ochsenbein, S.; Parkin, A.; Parsons, S.; Rajaraman, G.; Sieber, A.; Smith, A. A.; Timco, G. A.; Winpenny, R. E. P. Dalton Trans. 2004, 17, 2758−2766. (20) Eremenko, I. L.; Nefedov, S. E.; Sidorov, A. A.; Moiseev, I. I. Russ. Chem. Bull. 1999, 48, 405−416. (21) Sheldrick, G. M. Acta Crystallogr., Sect. A 2008, A64, 112−122. (22) Chilton, N. F.; Anderson, R. P.; Turner, L. D.; Soncini, A.; Murray, K. S. J. Comput. Chem. 2013, 34, 1164−1175. (23) Barra, A. L.; Brunel, L. C.; Robert, J. B. Chem. Phys. Lett. 1990, 165, 107−109. (24) Stoll, S.; Schweiger, A. J. Magn. Reson. 2006, 178 (1), 42−55. (25) (a) Karlström, G.; Lindh, R.; Malmqvist, P.-Å.; Roos, B. O.; Ryde, U.; Veryazov, V.; Widmark, P.-O.; Cossi, M.; Schimmelpfennig, B.; Neogrady, P.; Seijo, L. Comput. Mater. Sci. 2003, 28, 222−239. (b) Veryazov, V.; Widmark, P.; Serrano-Andres,́ L.; Lindh, R.; Roos, B. O. Int. J. Quantum Chem. 2004, 100, 626−635. (c) Aquilante, F.; De Vico, L.; Ferre,́ N.; Ghigo, G.; Malmqvist, P.; Neogrady,́ P.; Pedersen, T. B.; Piton̆ak,́ M.; Reiher, M.; Roos, B. O.; Serrano-Andres,́ L.; Urban, M.; Veryazov, V.; Lindh, R. J. Comput. Chem. 2010, 31, 224−247. (26) Rajaraman, G.; Christensen, K. E.; Larsen, F. K.; Timco, G. A.; Winpenny, R. E. P. Chem. Commun. 2005, 3053−3055. (27) Aromí, G.; Batsanov, A. S.; Christian, P.; Helliwell, M.; Parkin, A.; Parsons, S.; Smith, A. A.; Timco, G. A.; Winpenny, R. E. P. Chem. Eur. J. 2003, 9, 5142−5161. (28) Cotton, F. A.; Wilkinson, G. Chemistry of the Transition Elements. In Advanced Inorganic Chemistry, 4th ed.; Wiley: New York, 1980; p 786ff. (29) Eremenko, I. L.; Golubnichaya, M. A.; Nefedov, S. E.; Sidorov, A. A.; Golovaneva, I. F.; Burkov, V. I.; Ellert, O. G.; Novotortsev, V. M.; Eremenko, L. T.; Sousa, A.; Bermejo, M. R. Russ. Chem. Bull. 1998, 47, 704−718. (30) Titis,̌ J.; Boca,̆ R. Inorg. Chem. 2010, 49, 3971−3973. (31) Boeer, A. B.; Barra, A.-L.; Chibotaru, L. F.; Collison, D.; McInnes, E. J. L.; Mole, R. A.; Simeoni, G. G.; Timco, G. A.; Ungur, L.; Unruh, T.; Winpenny, R. E. P. Angew. Chem., Int. Ed. 2011, 50, 4007−4011. (32) Mabbs, F. E.; Collison, D. Electron Paramagnetic Resonance of d Transition Metal Compounds; Elsevier: Amsterdam, 1992; p 482. (33) Atanasov, M.; Comba, P.; Helmle, S.; Müller, D.; Neese, F. Inorg. Chem. 2012, 51, 12324−12335. (34) Inglis, R.; Jones, L. F.; Milios, C. J.; Datta, S.; Collins, A.; Parsons, S.; Wernsdorfer, W.; Hill, S.; Perlepes, S. P.; Piligkos, S.; Brechin, E. K. Dalton Trans. 2009, 18, 3403−3412. (35) Herchel, R.; Boca,̆ R.; Krzystek, J.; Ozarowski, A.; Duran,́ M.; van Slageren, J. J. Am. Chem. Soc. 2007, 129 (34), 10306−10307.
8472 dx.doi.org/10.1021/ic501036h | Inorg. Chem. 2014, 53, 8464−8472
61
SUPPORTING INFORMATION
On the possibility of magneto-structural correlations: Detailed studies of di-nickel carboxylate complexes
James P. S. Walsh,† Stephen Sproules,†,‡ Nicholas F. Chilton,† Anne-Laure Barra,§ Grigore A. Timco,† David Collison,†* Eric J. L. McInnes† and Richard E. P. Winpenny†
†School of Chemistry and Photon Science Institute, University of Manchester, Manchester, M13 9PL, United Kingdom. ‡WestCHEM, School of Chemistry, University of Glasgow, Glasgow, G12 8QQ, United Kingdom. §Laboratoire National des Champs Magnétiques Intenses, 25, rue des Martyrs, B.P. 166, 38042 Grenoble Cedex, France. *[email protected]
−1 1 Table S1. Summary of the vibrational spectra (cm ) in the methyl ν(CH) region for 1–5. 2 1 2 3 4 5 3 i 2966.0 2980.9 2970.8 2972.7 2970.3 2961.6 2949.6 2959.2 2951.0 4 ii 2930.3 2926.0 2921.1 2924.0 2920.2 5 iii 2905.7 2904.3 2898.0 2902.8 2899.0 iv 2870.5 2870.5 2963.8 2868.6 2864.3
4000 3600 3200 2800 2400 2000 1600 1200 800 Table S2. Eigenstates (in cm−1) of the lowest energy manifold -1 Wavenumber (cm ) calculated for 1 using the parameters in Table 6. Figure S1. Overlay of the IR spectra of 1–5 S Energy (in cm−1) measured using crushed polycrystalline 0 15.99 samples. 1 11.79 10.73 10.47 2 2.280 331.2 GHz, 15 K 2.188 1.104 '' / dB 331.2 GHz, 5 K
d 0.2624 0
Table S3. Euler rotations of one nickel site in relation to the other in the ZY´Z ´´ convention. Extracted from CASSCF calculations as 0 2 4 6 8 10 12 14 16 detailed in the text. Field (T) Figure S2. Powder electron paramagnetic Compound R (°) resonance spectra of 6 measured at 331.2 1 +48.1,–139,–144 GHz at 5 and 15 K, with the simulation 2 +109,–88.5,–71.0 shown in red. gx1 = gy1 = 2.32, gz1 = 2.25, gx2 −1 3 +75.0,–34.0,–6.71 = gy2 = 2.25, gz2 = 2.19, D1 = +1.6 cm , E1 = −1 −1 −1 +123,–102,+2.81 +0.35 cm , D2 = +1.1 cm , E2 = +0.20 cm . 4 +70.6,+102,–109 5 +24.9,–18.8,–140
1
62
34.13 GHz, 10 K 2 3 34.09 GHz, 10 K '' / dB '' / dB d d
400 800 1200 1600 0 400 800 1200 1600 Field (mT) Field (mT)
34.23 GHz, 10 K 4 5 34.13 GHz, 5 K '' / dB '' / dB d d
0 400 800 1200 1600 0 400 800 1200 1600 Field (mT) Field (mT)
Figure S3. Q-band EPR spectra of 2 (top left), 3 (top right), 4 (bottom left), and 5 (bottom right).
−1 Energy transfer (cm ) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
0.06 5 K 0.05
0.04
0.03
0.02 Intensity (arb. units) 0.01
0.06 1.5 K 0.05
0.04
0.03
0.02 Intensity (arb. units) 0.01
0.00
0.0 0.5 1.0 1.5 2.0 Energy transfer (meV) Figure S4. INS spectrum of 1 (data reproduced from Ref. 19). Absorptions occur over the energy range of 7.5–14.0 cm−1, which is consistent with the lowest energy manifold calculated using the parameters for 1 in Table 6, and shown in Table S2. 2
63
64
5. Paper two: “Large Zero-Field Splittings of the Ground Spin State
Arising from Antisymmetric Exchange Effects in Heterometallic
Triangles”
S. A. Magee, S. Sproules, A.-L. Barra, G. A. Timco, N. F. Chilton, D. Collison, R. E. P. Winpenny and E. J. L. McInnes, Angew. Chem. Int. Ed., 2014, 53, 5310.
65
66
Angewandte. Communications
DOI: 10.1002/anie.201400655 Magnetic Anisotropy Large Zero-Field Splittings of the Ground Spin State Arising from Antisymmetric Exchange Effects in Heterometallic Triangles** Samantha A. Magee, Stephen Sproules, Anne-Laure Barra, Grigore A. Timco, Nicholas F. Chilton, David Collison, Richard E. P. Winpenny, and Eric J. L. McInnes*
5 [3, 4] Abstract: [Ru2Mn(O)(O2CtBu)6(py)3] has an S = /2 ground values. The unusual EPR signatures of some trimetallic state with a very large zero-field splitting (ZFS) of D = Cu[7] and FeS[8] enzymes can be explained by these effects. 1 1 2.9 cm , as characterized by EPR spectroscopy at 4– ASE effects on S > =2 states are much less studied, although it 330 GHz. This is far too large to be due to the MnII ion (D has been proposed as a mechanism for otherwise-forbidden 1 < 0.2 cm ), as shown from the {Fe2Mn} analogue, but can be magnetization quantum tunneling steps in Mn12 and other modeled by antisymmetric exchange effects. single molecule magnets[9] and also for the origin of high- order ZFS effects.[10] The magnetic anisotropy in transition-ion clusters is of Belinsky[11] and Tsukerblat et al.[12] have calculated the 3 fundamental importance in areas such as molecular magnet- effects of the ASE on the maximum S = /2 state (as ground or [1] II IV ism, for example giving rise to memory effects, through to excited state) in Cu 3 and V 3 triangles. They showed that [2] the characterization of metalloenzyme active sites. When components of the ASE within the M3 plane could break the the ground state can be described by a total electron spin S > degeneracy, that is, introduce a ZFS. However, in-plane 1 =2, arising from dominant isotropic exchange, the magnetic components are symmetry forbidden when the M3 triangle anisotropy tends to be dominated by the zero-field splitting lies on a mirror plane.[5] They further showed that symmetry- (ZFS) of the (2S + 1)-fold multiplet. This is generally assumed allowed components normal to the triangle do not split the to be dominated by the projection of the local ZFSs of the quartet unless there is a large isosceles distortion.[11] Here we [3] 1 1 metal ions. When the local spins are s = =2 (and have no show that this can explain the huge ZFSs in the S > =2 ground 6 ZFS) or when they are intrinsically isotropic, such as the S state of the heterometallic triangle [Ru2Mn(O)- II Mn ion, it is well understood that anisotropic components of (tBuCO2)6(py)3] ({Ru2Mn}, py = pyridine; Figure 1), which the exchange are the main contributions to the ZFS. In we have characterized by multifrequency EPR spectroscopy contrast, the general significance of the antisymmetric from 4 to 330 GHz. 1 component of the exchange interaction on S > =2 states is {Ru2Mn} and its {Fe2Mn} analogue both crystallize in the
not so clear: here we introduce an example where it provides P21 space group with one molecule per asymmetric unit (ESI): the source of very large ZFS effects. the Mn site is not crystallographically resolved. They have the Antisymmetric exchange (ASE; also known as Dzyalosh- classic structure of basic metal carboxylate triangles,[13] with inski–Moriya exchange) is the origin of spin canting (weak bridging pivalate and terminal pyridine groups (Figure 1). ferromagnetism) in extended lattices. In terms of molecular The acetate analogues of these complexes have been systems, ASE was first observed and discussed in trigonal reported,[14,15] and we find similar magnetic susceptibility (c) [4,5] 3 1 clusters of half-integer spins. This is because they provide behavior. {Ru2Mn} has cT= 4.42 cm Kmol (Figure 2), the the simplest discrete systems in which spin frustration can be studied,[6] since antiferromagnetic coupling in an equilateral 1 triangle leads to two degenerate S = =2 lowest energy states (2E term). ASE provides a mechanism for breaking the 2E degeneracy, and these effects can be quantified by EPR spectroscopy as they are manifested as unusual effective g-
[*] S. A. Magee, Dr. S. Sproules, Dr. G. A. Timco, N. F. Chilton, Prof. D. Collison, Prof. R. E. P. Winpenny, Prof. E. J. L. McInnes School of Chemistry and Photon Science Institute The University of Manchester Oxford Road, Manchester, M13 9PL (UK) E-mail: [email protected] Dr. A.-L. Barra Laboratoire National des Champs Magn tiques Intenses UPR 3228 CNRS, UJF-INSA-UPS, BP 166 38042 Grenoble Cedex 9 (France) [**] This work was supported by the EPSRC (UK), including the National [26] Figure 1. Molecular structure of {Ru2Mn}. Scheme: Ru/Mn (large EPR Facility and Service, and The University of Manchester. spheres), O (gray), N (white), C (black), H omitted for clarity. Average
Supporting information for this article is available on the WWW M···M and M-O distances: 3.36(4) and 1.94(4) in {Ru2Mn} and
under http://dx.doi.org/10.1002/anie.201400655. 3.32(4) and 1.91(5) in {Fe2Mn}, respectively.
5310 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 5310 –5313
67
Angewandte Chemie
Figure 2. cT(T) values for {Ru2Mn} (circles) and {Fe2Mn} (squares) measured in an applied magnetic field (H) of 0.1 T, with simulations (lines) as described in the text. Inset: M(H) at 2 and 4 K.
5 value expected for an isolated S = /2 state, hence the magnetic properties appear at first sight to be those of the isolated MnII ^ ion. Using the isotropic Hamiltonian Hiso [Eq. (1)]: X Figure 3. EPR spectra of {Ru2Mn} at 10 K, measured at: a) 3.87 GHz H^ ¼ g bsˆ H 2J sˆ sˆ 2J sˆ sˆ þ sˆ sˆ ð1Þ iso i i i 1 1 2 2 1 3 2 3 on a CH2Cl2/toluene solution; b) 220.8; and c) 331.2 GHz on a poly- ^ ^ crystalline sample. Simulations (red) based on Hamiltonian Hiso þ Hanti with the parameters in the text; the S-band simulation included an 1 III 5 5 II where s = s = = (Ru , low-spin d ) and s = / (Mn , high- isotropic 55Mn hyperfine coupling (Asˆ Iˆ ; I = 5/ )of 1 2 2 3 2 3 3 2 5 4 1 3 [20] spin d ), this corresponds to the j 0,s3i ground state (j S12,Si). A j A j =85 10 cm (typical for this coordination environment ). Gaussian linewidth parameters: a)15 (x,y) and 3 (z)G, with an A-strain simple Kamb treatment gives this ground state for J1/J2 > 3.5, of 2%; b,c) 1500G Gaussian linewdith with 5% strain in dz (see text with the first excited state (j 1,3/2i) at a relative energy of 2 for definition of the z direction). 7J2 2J1. Test calculations show that j J1 j must be greater than 3 1 ca. 10 cm and that J1/J2 must be greater than ca. 10 for there to be no evidence of excited-state population in cT(T) at high ions with O,N donor sets are j D j< 0.2 cm 1.[20] To test the temperature. Large j J1 j values are justified by the large radial model we have prepared and studied the equivalent 3d II extent of the 4d wavefunctions, and indeed direct overlap as complex {Fe2Mn} in which the Mn ion has the same ^ opposed to super-exchange is possible. Couplings of this coordination environment. Fitting cT(T) for {Fe2Mn} to Hiso III [16] 5 1 magnitude have been observed in homometallic Ru cages. with s1 = s2 = /2 gives J1 = 63.5 and J2 = 21.9 cm with g = Fitting the low-temperature drop in cT(T), and magnet- 2.0 (Figure 2), similar to its acetate analogue.[21] This gives the 5 ^ ization (M) data (Figure 2), for an isolated S = /2 with HZFS j 1,3/2i ground state, as confirmed by low-temperature [Eq. (2)], gives the axial ZFS parameter j D j= 3.0 cm 1 (E = magnetization and EPR measurements. Modeling the latter [17] ^ 3 1 0, with fixed g = 1.98). with HZFS for S = /2 gives D =+0.25, j E j= 0.04 cm (fixed g = 2.0; Figures 2 and 4). While the j 1,3/2i ZFS has contri- ^ ^2 SðS þ 1Þ ^2 ^2 butions from all three metal ions, assuming it arises entirely HZFS ¼ gibSˆ H þ D S þ E S S ð2Þ z 3 x y II 1 from Mn (an over-estimation) gives DMn =+0.13 cm . This is in the range known for O,N-donor six-coordinate MnII,[20]
This is much larger than estimated previously for the but negligible compared to the ground-state ZFS of {Ru2Mn}. [15] 1 acetate analogue by low-frequency EPR spectroscopy; Hence {Ru2Mn} has a contribution of about 3 cm to its ZFS therefore, we have measured the value of D directly by from another source. high-frequency EPR spectroscopy.[18] Spectra at low frequen- Given the predictions for ASE effects in the high-spin 1 1 [11] cies are those of an axial effective spin /2 with geff,? 6 and state of s = =2 triangles, we have investigated this model by 1 ^ ^ [4] geff,k 2, consistent with a ground-state M = /2 Kramers the Hamiltonian Hiso þ Hanti, where di are the ASE vectors, 55 ^ doublet, with resolved Mn hyperfine for frozen solution and Hanti is defined in Equation (3). samples (Figure 3a and see Figure S1 in the Supporting Information). Spectra recorded at 220 and 330 GHz H^ d s s d s s s s 3 5 anti ¼ 1 ˆ1 ˆ2 þ 2 ˆ2 ˆ3 þ ˆ3 ˆ1 ð Þ (Figure 3) unambiguously define the S = /2 multiplet, and [19] ^ 1 simulation with HZFS gives D =+2.9 cm (E = 0) with g = 1.98 (see Figure S2 in the Supporting Information). If we neglect the torsion angles of the terminal pyridine
In this simple model the ZFS of the j 0,5/2i state should ligands, then {Ru2Mn} is an isosceles triangle with C2v correspond to that of the isolated MnII site. However, this is symmetry. We define the normal to the trimetallic plane as absurd: the largest values reported for six-coordinate MnII the z axis to maintain consistency with the literature reports
Angew. Chem. Int. Ed. 2014, 53, 5310 –5313 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org 5311
68
Angewandte. Communications
There are no first-order corrections to the eigenvalues, but each ground-state M component mixes with all excited states with DM = 0. The three Kramers doublets are separated and, 1 3 3 5 to second order, the j M j= /2 to /2 gap is half the /2 to /2 gap. 5 This is the form of an isolated S = /2 state under Hamil- tonian (2), with separations of 2D and 4D, respectively. To second order, D is then given by Equation (4), 2dzJ þ dzJ 2 D ¼ 2 1 1 2 ð4Þ ðÞ2J1 7J2 ðÞ2J1 2J2 ðÞ2J1 þ 5J2
where the denominator is the product of the gaps to the three excited states. Excellent simulations of the EPR spectra (Figure 3 and see Figure S3 in the Supporting Information), including variable-temperature and hyperfine effects, are obtained by ^ ^ full diagonalization of Hiso þ Hanti (with gMn = 1.98 and gRu = 2.0; the simulations are insensitive to the latter), and fits to the low-temperature cT(T) and M(H) are indistinguishable from those in Figure 2. For these calculations we have taken 1 J1 = 1000 cm with J1/J2 = 10 (see above). Clearly, more than one set of dz,dz components can generate a given Figure 4. EPR spectra of polycrystalline {Fe2Mn} measured at 10 K and 1 2 z z 1 a) 9.76 and b) 33.97 GHz (5 K), with simulations (red) based on D value. If we take d1 ¼ 0, then we find d2 ¼66 cm ^ z Hamiltonian HZFS with the parameters in the text. Gaussian linewidth (Figure 3; calculations are insensitive to the sign of di ); of 300 G. z much larger values of d1 are required to generate equivalent D values [see Eq. (4)]. Adjustment of J1 and J2 will give z on M3 triangles. In this case, all three metal–metal vectors lie different di values, but the conclusion does not change: very 1 on a mirror plane and the only non-zero components of di are large ZFS effects (several cm ) have been introduced to the z 1 di . Such components do not lead to ZFS in the high-spin S > =2 ground state by ASE effects in second-order. (There [11,12] 5 states of equilateral triangles. In our case we have are also higher-order contributions: the calculated j M j= /2 to 3 3 1 a system that is isosceles not only in terms of the strength of /2 gap is not exactly double the /2 to /2 gap from ^ ^ the J coupling (often seen in homometallic triangles), but in diagonalization using Hiso þ Hanti.) Moreover, the simple
the identity of the spins themselves. description of the magnetic properties of {Ru2M} as being ^ Hamiltonian Hiso gives the 24 24 energy matrix compris- “those of the isolated M ion” is wrong.
ing the ground j 0,5/2i (with eigenvalue + 3J1/2) and excited The ASE parameters we have found here are entirely z 1 j 1,3/2i ( J1/2 +7J2), j 1,5/2i ( J1/2 +2J2), and j 1,7/2i ( J1/ reasonable: di values of greater than 100 cm have been ^ z II [5,7,22] 2 5J2) states. Applying Hanti, with only di 6¼ 0, as a perturba- found for antiferromagnetically coupled Cu triangles. It tion to this coupled basis gives the non-zero matrix elements has been argued that the latter are due to the favorable [7] (labeling states as j S12,S,Mi): alignment of ground- and excited-state d orbitals, because ASE arises from exchange between the electronic ground state of one ion and the excited state of another through spin–
orbit coupling (SOC). In {Ru2M}, the ASE will be favored both by the large SOC of RuIII, and the strong exchange interactions arising from the large radial extent of the 4d orbitals. The perturbative expressions above show that this can also be viewed as mixing of the ground-state MnII
functions (S12 = 0) with the “ferromagnetic” excited state
(S12 = 1) of the {Ru2} unit. This is equivalent to an alternative description of the electronic structure proposed for some oxo- bridged RuIII oligomers, where direct exchange between the 4d functions results in delocalized singlet and “low lying” (in electronic spectroscopy terms) triplet states.[23,24] In summary, we have shown the ASE interaction can lead to very large spin ground-state ZFSs in polymetallic com- plexes, even through a second-order perturbation on the isotropic exchange. The model reproduces spectroscopic observations across two orders of magnitude in field/fre- quency regime of EPR spectroscopy. The results show that these effects cannot be ignored, particularly when 2nd and 3rd
5312 www.angewandte.org 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 5310 –5313
69
Angewandte Chemie
row metal ions are involved, as is becoming popular as a route Jones, E. K. Brechin, V. Mosser, W. Wernsdorfer, Phys. Rev. B to introduce magnetic anisotropy into molecular magnets. 2008, 78, 132401. Furthermore, it is possible that ASE is much more widespread [10] N. Kirchner, J. van Slageren, B. Tsukerblat, O. Waldmann, M. Dressel, Phys. Rev. B 2008, 78, 094426. than generally imagined in coupled systems, and assigned to [11] M. I. Belinksy, Inorg. Chem. 2008, 47, 3521; M. I. Belinksy, Inorg. anisotropic exchange or local ZFS effects; for example, Chem. 2008, 47, 3532. 1 a ground state D = 0.25 cm for {Fe2Mn} can equally well be [12] B. Tsukerblat, A. Tarantul, A. M ller, Phys. Lett. A 2006, 353, 48. z 1 derived from d2 ¼0.67 cm , similar to values observed in [13] R. D. Cannon, R. P. White, Prog. Inorg. Chem. 1988, 36, 195. [5] {Fe3} triangles. This is important because different aniso- [14] A. Ohto, Y. Sasaki, T. Ito, Inorg. Chem. 1994, 33, 1245; H. tropy terms in the spin Hamiltonian result in different mixing Kobayashi, N. Ury , R. Miyamoto, Y. Ohba, M. Iwaizumi, Y. between ground and excited states, which have important Sasaki, A. Ohto, T. Ito, Bull. Chem. Soc. Jpn. 1995, 68, 2551. [15] H. Kobayashi, N. Ury , I. Mogi, R. Miyamoto, Y. Ohba, M. consequences for phenomena such as magnetic relaxation, Iwaizumi, Y. Sasaki, A. Ohto, M. Suwabe, T. Ito, Bull. Chem. including, as discussed by others, a possible origin of Soc. Jpn. 1996, 69, 3163. “forbidden” quantum tunneling steps and avoided crossings [16] A. Upadhyay, J. Rajprohit, M. K. Singh, R. Dubey, A. K. observed in several molecular nanomagnets.[9,25] Srivastava, A. Kumar, G. Rajaraman, M. Shanmugam, Chem. Eur. J. 2014, DOI: 10.1002/chem.201304826. Received: January 21, 2014 [17] Modeling of magnetic data used PHI software, see N. F. Chilton, Published online: April 15, 2014 R. P. Anderson, L. D. Turner, A. Soncini, K. S. Murray, J. Comput. Chem. 2013, 34, 1164. [18] On a home-built instrument: A. L. Barra, A. K. Hassan, A. Keywords: antisymmetric exchange · electronic structure · . Janoschka, C. L. Schmidt, V. Sch nemann, Appl. Magn. Reson. EPR spectroscopy · exchange coupling · zero-field splitting 2006, 30, 385. [19] EPR analysis and simulations used Weihe s SimEPR, with some calculations using routines as described by Piligkos et al., see [1] R. Sessoli, D. Gatteschi, J. Villain, Molecular Nanomagnets, C. J. H. Jacobsen, E. Pederson, J. Villadsen, H. Weihe, Inorg. Oxford University Press, Oxford, 2006. Chem. 1993, 32, 1216; S. Piligkos, E. Bill, D. Collison, E. J. L. [2] W. R. Hagen, Biomolecular EPR Spectroscopy, CRC Press, McInnes, G. A. Timco, H. Weihe, R. E. P. Winpenny, F. Neese, J. Taylor and Francis Group, 2009. Am. Chem. Soc. 2007, 129, 760; S. Piligkos, H. Weihe, E. Bill, F. [3] A. Bencini, D. Gatteschi, EPR of Exchange Coupled Systems, Neese, H. El Mkami, G. M. Smith, D. Collison, G. Rajaraman, Springer, Berlin, 1990. G. A. Timco, R. E. P. Winpenny, E. J. L. McInnes, Chem. Eur. J. [4] B. S. Tsukerblatt, M. I. Belinskii, A. V. Ablov, Dokl. Akad. Nauk 2009, 15, 3152. SSSR 1971, 201, 1410; M. I. Belinskii, B. S. Tsukerblatt, A. V. [20] C. Duboc, M.-N. Collomb, F. Neese, Appl. Magn. Reson. 2010, Ablov, Mol. Phys. 1974, 28, 283; B. S. Tsukerblat, B. Y. Kuyav- 37, 229. skaya, M. I. Belinskii, A. V. Ablov, V. M. Novotortsev, V. T. [21] A. B. Blake, A. Yavari, W. E. Hatfield, C. N. Sethulekshmi, J. Kalinnkov, Theor. Chim. Acta 1975, 38, 131; Y. V. Rakitin, Y. V. Chem. Soc. Dalton Trans. 1985, 2509. Yablokov, V. V. Zelentsov, J. Magn. Reson. 1981, 43, 288. [22] For recent examples, see A. Escuer, G. Vlahopoulou, F. Lloret, [5] For a recent review, see R. Bocˇa, R. Herchel, Coord. Chem. Rev. F. A. Mautner, Eur. J. Inorg. Chem. 2014, 83. 2010, 254, 2973. [23] T. R. Weaver, T. J. Meyer, S. A. Adeyemi, G. M. Brown, R. P. [6] O. Kahn, Chem. Phys. Lett. 1997, 265, 109. Eckberg, W. E. Hatfield, E. C. Johnson, R. W. Murray, D. [7] J. Yoon, E. I. Solomon, Inorg. Chem. 2005, 44, 8076; J. Yoon, Untereker, J. Am. Chem. Soc. 1975, 97, 3039. E. I. Solomon, Coord. Chem. Rev. 2007, 251, 379. [24] J. A. Baumann, D. J. Salmon, S. T. Wilson, T. J. Meyer, W. E. [8] Y. Sanakis, A. L. Macedo, I. Moura, J. J. G. Moura, V. Papaef- Hatfield, Inorg. Chem. 1978, 17, 3342; Y. Sasaki, A. Tokiwa, T. thymiou, E. M nck, J. Am. Chem. Soc. 2000, 122, 11855; F. Ito, J. Am. Chem. Soc. 1987, 109, 6341. Tiago de Oliveira, E. L. Bominaar, J. Hirst, J. A. Fee, E. M nck, [25] F. Cinti, M. Affronte, A. G. M. Jansen, Eur. Phys. J. B 2002, 30, J. Am. Chem. Soc. 2004, 126, 5338. 461. [9] M. I. Katsnelson, V. V. Dobrovitski, B. N. Harmon, Phys. Rev. B [26] CCDC 926362 and 926363 contain the supplementary crystallo- 1999, 59, 6919; I. Chiorescu, R. Giraud, A. G. M. Jansen, A. graphic data for this paper. These data can be obtained free of Caneschi, B. Barbarba, Phys. Rev. Lett. 2000, 85, 4807; C. M. charge from The Cambridge Crystallographic Data Centre via Ramsey, E. del Barco, S. Hill, S. J. Shah, C. C. Beedle, D. N. www.ccdc.cam.ac.uk/data_request/cif. Hendrickson, Nat. Phys. 2008, 4, 277; S. Bahr, C. J. Milios, L. F.
Angew. Chem. Int. Ed. 2014, 53, 5310 –5313 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org 5313
70
Supporting Information Wiley-VCH 2014
69451 Weinheim, Germany
Large Zero-Field Splittings of the Ground Spin State Arising from Antisymmetric Exchange Effects in Heterometallic Triangles** Samantha A. Magee, Stephen Sproules, Anne-Laure Barra, Grigore A. Timco, Nicholas F. Chilton, David Collison, Richard E. P. Winpenny, and Eric J. L. McInnes* anie_201400655_sm_miscellaneous_information.pdf
71
Supplementary Information
Synthesis. Solvents and reagents were from commercial sources and used without further purification, with the exception of manganese pivalate which was made by literature methods.[1]
{Ru2Mn}: Pivalic acid (22.1 g, 219 mmol) and RuCl3.xH2O (0.15 g, 0.57 mmol) in 4:1 EtOH/H2O (25 mL) were heated at 70 °C for 10 min. After cooling, manganese pivalate (0.52 g, 2.01 mmol) was added with stirring. After standing overnight, the purple solution was filtered, the solvent removed under vacuum, and the resultant oil filtered to remove excess manganese pivalate. The filtrate was diluted with MeCN (10 mL) and excess pivalic acid removed by successive cooling (-40 °C) and filtration. The solvent was removed, the residue stirred in distilled water (200 mL) for 30 min then filtered to give a dark purple solid which was washed with 4:1 H2O/MeCN (3 × 10 mL) and air dried (0.1 g). The solid was dissolved in MeCN (30 mL) and pyridine (10 mL), stirred overnight and filtered. The solvent was removed and the purple residue recrystallised from a minimum amount of pyridine (25 mg, 26%). Microanalysis: (C45H69MnN3O13Ru2) calcd: C, 48.38; H, 6.23; N, 3.76; Mn, 4.92%. Found: C, 48.38; H, 6.32; N, 3.79; Mn 4.47%. IR (KBr): 1405, 1369, 1356 sh cm–1. ESI-MS + 1 +ve (m/z): 1018 [M – (C5H9O2)] . H NMR (CDCl3): δ 2.04 s (36H, CH3), δ 3.13 s (18H, CH3) ppm.
{Fe2Mn}: Pivalic acid (50.0 g, 490 mmol) and Fe(NO3)3.9H2O (10.0 g, 24.8 mmol) were heated for 4 h with stirring at 160 °C, in a 2-neck round-bottom flask with condenser, to remove NO2. Manganese acetate (3.00 g, 12.2 mmol) was added under N2, and acetic acid removed by addition of toluene (2 x 100 mL) and distillation of the azeotrope. After cooling to room temperature, MeCN (50 mL) was added and the solution stirred for 15 min. The solid was filtered off and washed with MeCN (3 × 15 mL). The solid was dissolved in CH2Cl2 (0.2 g in 20 mL) and excess pyridine (200 mg) with stirring for 20 min. After removing the solvent under vacuum, the residue was recrystallised from diethyl ether/MeCN to give large brown crystals of {Fe2Mn} (0.102 g, 53%). Microanalysis:
(C45H69Fe2MnN3O13) calcd: C, 52.64; H, 6.77; N, 4.09; Fe, 10.88; Mn, 5.35%. Found: C, 52.60; H, 6.89; N, 4.14; Fe, 10.76; Mn, 5.20%. IR (KBr): 1408, 1371, 1357 sh cm–1. ESI-MS +ve (m/z): 846 [M + 1 – (C5H9O2) – (C5H5N)] . H NMR (CDCl3): δ 3.09 s (18H, CH3), δ 4.33 s (36H, CH3) ppm.
1
72
X-ray crystallography. X-ray diffraction data were collected on an OXFORD Diffraction XCaliber2 CCD diffractometer using Mo Kα radiation. The structures were solved with SHELXS, and SHELXL was used for the refinement.[2] Crystallographic information files were produced using OLEX2.[3]
Table S1. Crystallographic Data for 1 and 2
1 2
Empirical formula C45H69MnN3O13Ru2 C45H69Fe2MnN3O13
Formula weight 1117.11 1026.67
T, K 100(2) 100(2)
λ, Å 0.71073 0.71073
Crystal system monoclinic monoclinic
Space group P21 P21
a, Å 11.6515(5) 11.6211(5)
b, Å 19.772(1) 19.7837(6)
c, Å 11.8487(6) 11.9437(4)
β, ° 106.661(5) 106.955(4)
3 V, Å 2607.1(2) 2626.3(2)
Z 2 2
–3 Density (calculated), Mg m 1.423 1.298
3 Crystal size, mm 0.5 × 0.5 × 0.3 0.5 × 0.3 × 0.3
Theta range for data collection, ° 3.08 – 28.51 2.98 – 28.02
Reflections collected 10925 19757
Independent reflections 7574 [R(int) = 0.0363] 9845 [R(int) = 0.0476]
Completeness, % 99.7 99.8
Data / restraints / parameters 7574 / 2 / 598 9845 / 2 / 598
2 Goodness-of-fit on F 0.993 1.025
Final R indices [I > 2σ(I)] R1 = 0.0363, wR2 = 0.0843 R1 = 0.0524, wR2 = 0.1043
R indices (all data) R1 = 0.0446, wR2 = 0.0871 R1 = 0.0859, wR2 = 0.1353
–3 Largest diff. peak and hole, e Å 0.930 and -0.560 0.551 and -0.659
2
73
Physical measurements. Magnetic measurements were performed in the temperature range 1.8 – 300 K using a Quantum Design MPMS-XL SQUID magnetometer equipped with a 7 T magnet. Corrections for diamagnetism were made using Pascal’s constants and magnetic data were corrected for diamagnetic contributions from the sample holder. Fits were performed using the program PHI.[4] S-, X-, Q- and W-band EPR spectra were measured on Bruker instrumentation at the EPSRC National UK EPR Facility and Service at The University of Manchester. High-frequency, high-field EPR spectra were recorded at the LNCMI-CNRS at Grenoble on a home-built spectrometer.[5] Simulations and data analysis were performed using software written by Weihe,[6] Piligkos[7] and some test calculations performed with Stoll’s EasySpin.[8]
[1] M. A. Kiskin, I. G. Fomina, G. G. Aleksandrov, A. A. Sidorov, V. M. Novotortsev, Y. V. Rakitin, Z. V. Dobrokhotova, V. N. Ikorskii, Y. G. Shvedenkov, I. L. Eremenko, I. I. Moiseev, Inorg. Chem. Commun. 2005, 8, 89. [2] G. M. Sheldrick, Acta Cryst. 2009, A64, 112. [3] O. V. Dolomanov, L. J. Bourhis, R. J. Gildea, J. A. K. Howard, H. Puschmann, J. Appl. Cryst. 2009, 42, 339. [4] N. F. Chilton, R. P. Anderson, L. D. Turner, A. Soncini and K. S. Murray, J. Comput. Chem. 2013, 34, 1164. [5] A.L. Barra, A.K. Hassan, A. Janoschka, C.L. Schmidt, V. Schünemann, Appl. Magn. Reson. 2006, 30, 385. [6] C. J. H. Jacobsen, E. Pederson, J. Villadsen and H. Weihe, Inorg. Chem. 1993, 32, 1216. [7] S. Piligkos, E. Bill, D. Collison, E.J.L. McInnes, G.A. Timco, H. Weihe, R.E.P. Winpenny and F. Neese, J. Amer. Chem. Soc. 2007, 129, 760; S. Piligkos, H. Weihe, E. Bill, F. Neese, H. El Mkami, G. M. Smith, D. Collison, G. Rajaraman, G. A. Timco, R. E. P. Winpenny and E. J. L. McInnes, Chem. Eur. J. 2009, 15, 3152; [8] S. Stoll and A. Schweiger, J. Magn. Reson. 2006, 178, 42.
3
74
Fig. S1 Multi-frequency EPR spectra of {Ru2Mn} at 10 K recorded in CH2Cl2/toluene solution at S- (3.874 GHz), X- (9.388 GHz) and Q-band (33.95 GHz), and as a polycrystalline solid at W-band (93.99 GHz).
4
75
Fig. S2 331.2 GHz EPR spectrum (black) of polycrystalline {Ru2Mn} at 10 K, with simulation (red) -1 according to Hamiltonian �!"# (2) with S = 5/2, giso = 1.98 and D = +2.9 cm , E = 0 and a Gaussian linewidth of 1500 G with 5% D-strain.
5
76
Fig. S3 Variable temperature 331.2 GHz EPR spectra (black) of polycrystalline {Ru2Mn} at (top to bottom) 5, 10 and 15 K. Simulations (red) with Hamiltonian �!"# + �!"#$ and the parameters in the ! main text (Gaussian linewidth of 1500 G with 5% strain in �!).
6
77
Fig. S4 Variable temperature 331.2 GHz EPR spectra of polycrystalline {Fe2Mn}.
7
78
6. Paper three: “An electrostatic model for the determination of magnetic anisotropy in dysprosium complexes”
N. F. Chilton, D. Collison, E. J. L. McInnes, R. E. P. Winpenny and A. Soncini, Nat. Commun., 2013, 4, 2551.
79
80
ARTICLE
Received 15 Jul 2013 | Accepted 4 Sep 2013 | Published 7 Oct 2013 DOI: 10.1038/ncomms3551 An electrostatic model for the determination of magnetic anisotropy in dysprosium complexes
Nicholas F. Chilton1, David Collison1, Eric J.L. McInnes1, Richard E.P. Winpenny1 & Alessandro Soncini2
Understanding the anisotropic electronic structure of lanthanide complexes is important in areas as diverse as magnetic resonance imaging, luminescent cell labelling and quantum computing. Here we present an intuitive strategy based on a simple electrostatic method, capable of predicting the magnetic anisotropy of dysprosium(III) complexes, even in low symmetry. The strategy relies only on knowing the X-ray structure of the complex and the well-established observation that, in the absence of high symmetry, the ground state of dysprosium(III) is a doublet quantized along the anisotropy axis with an angular momentum 15 quantum number mJ ¼ ± /2. The magnetic anisotropy axis of 14 low-symmetry mono- metallic dysprosium(III) complexes computed via high-level ab initio calculations are very well reproduced by our electrostatic model. Furthermore, we show that the magnetic anisotropy is equally well predicted in a selection of low-symmetry polymetallic complexes.
1 School of Chemistry and Photon Science Institute, The University of Manchester, Oxford Road, Manchester M19 3PL, UK. 2 School of Chemistry, University of Melbourne, Parkville, Victoria 3010, Australia. Correspondence and requests for materials should be addressed to N.F.C. (email: [email protected]) or to A.S. (email: [email protected]).
NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications 1 & 2013 Macmillan Publishers Limited. All rights reserved.
81
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3551
he fascinating magnetic properties of the lanthanides have complexes, the ground Kramers doublet of DyIII is strongly continued to be a highly topical and strongly multi- axial with the principal values of the g-tensor approaching those ±15 6 Tdisciplinary research area for over 60 years. Such is the of the mJ ¼ /2 levels of the atomic multiplet H15/2 diversity of this field that their application reaches from magnetic (gx ¼ gy ¼ 0, gz ¼ 20). This empirical observation suggests that a resonance imaging and cell labelling1,2, to potential building simple, but appropriate, variational ansatz for the many-electron blocks of quantum computers3. The pursuit of such applications ground state wavefunction of these low-symmetry complexes relies on detailed knowledge of the magnetic anisotropy, which, consists of the atomic functions w±(a,b) corresponding to the 15 6 while being completely defined in cases of high symmetry, mJ ¼ ± /2 states of the multiplet H15/2. The variational is difficult to elucidate in low-symmetry complexes. Much parameters to be optimized in w±(a,b) consist of the two polar recent work, where 4f complexes have shown slow magnetic angles a and b, which specify the orientation of the quantization 4–7 relaxation and unprecedented non-collinear magnetic textures axis with respect to the low-symmetry crystal field VCF defined by at the single-molecule level8–10, also depends on understanding the ligands. To determine these angles, and hence the full g-tensor the orientation of the magnetic anisotropy. Of all the lanthanide of the ground Kramers doublet of the DyIII complex, we can use III ions, it is Dy that continues to prove the most interesting, the variational principle and minimize the energy E 15=2ða; bÞ¼ providing unexpected examples of new magnetic phenomena, by hiw ða; bÞjVCF j w ða; bÞ with respect to all possible virtue of its unique magnetic anisotropy11. However, because of orientations (a,b) of the quantization axis. the intricate electronic structure of lanthanide complexes, simple This proposed strategy is readily mapped onto a classical models that can predict magnetic anisotropy in molecular solids electrostatic energy minimization problem. Following the work of of low symmetry are still missing. Sievers21, the many-electron wavefunctions w±(a,b) can be ðÞa;b The single-ion properties of 4f metal ions, whether in mono- or described by an electron density distribution r 15=2ðÞy; f , where polymetallic complexes, are difficult to elucidate owing to the y and f are polar angles defined in the reference frame of VCF, shielded nature of the 4f orbitals giving rise to weak interactions expressing the angular dependence of the axially symmetric with the surrounding environment. Recent advances in post aspherical electron density. This aspherical f-electron density can Hartree-Fock multi-configurational ab initio methodology have be written as a linear combination of three spherical harmonics made accurate quantum chemical calculations on paramagnetic 4f Y2,0(y,f), Y4,0(y,f) and Y6,0(y,f), where the coefficients of each compounds possible12. The Complete Active Space Self Consistent are fully determined by angular momentum coupling and average Field (CASSCF) method can accurately predict the magnetic atomic radial multipole moments21. In the particular case of 13,14 III ðÞa;b properties of lanthanide complexes , and calculations of this Dy , r 15=2ðÞy; f can be approximated by an oblate spheroid type have become an indispensable tool for the explanation of distribution owing to the dominant contribution of the 6,15–17 22,21 increasingly interesting magnetic phenomena .These quadrupolar term Y2,0(y,f) to the expansion . calculations are especially useful in cases of low symmetry, where As the crystal field is a one-electron potential, the many- previous methods have provided intractable, over parameterized electron variational integral E±15/2(a,b) can be exactly recast into problems18,19. Although CASSCF ab initio calculations are a simple electrostatic energy integral, describing the interaction extremely versatile and implicitly include all effects required to between the electric potential generated by the crystal field ðÞa;b elucidate the magnetic properties, the results offer little in the way VCF(y,j) and the Sievers charge density r 15=2ðÞy; f associated of chemically intuitive explanations and to obtain reliable results with the f-electrons in the central DyIII ion (equation (1)). requires considerable intervention by expert theorists equipped with Zp Z2p access to powerful computational resources. ðÞa;b E 15ðÞa; b ¼ VCFðÞy; j r ðÞy; f sinðÞy dydj ð1Þ Recently, some of us have applied a simple electrostatic model 2 15=2 to rationalize the unexpected direction of the calculated magnetic y¼0 j¼0 anisotropy in two related sets of monometallic 4f complexes20. III This was based on the aspherical electron density distributions of Thus, we arrive at the hypothesis that in low-symmetry Dy the lanthanide ions, pioneered by Sievers21, and the design complexes, the many-electron ground state wavefunction and, principles for the exploitation of f-element anisotropy outlined by hence, the orientation of the magnetic anisotropy axis, can be Rinehart and Long22. Other groups have also been coming to determined simply by solving a classical electrostatic energy similar conclusions17. Although the use of crystal field methods to minimization problem. model anisotropic magnetic data is widespread23–25, models for the prediction of magnetic anisotropy in low symmetry are Constructing the crystal field potential of charged ligands.To few26,27. These methods are based on the diagonalization of a use this hypothesis, we must determine the explicit form of crystal field Hamiltonian, which, especially in cases of low VCF(y,j) appearing in equation (1), by using an appropriate symmetry, requires a large number of parameters that often can model for the charge distribution on the ligands. This may appear only be reliably determined by fitting experimental data. Such an a difficult problem; here, we use a simple model for charged approach can obscure the rationalization of magnetic anisotropy ligands that are common in many low-symmetry DyIII com- and its predictive power is uncertain. plexes. The charge on the ligands is expected to have a dominant Here we report a quantitative method based on a straightfor- role in the determination of the electrostatic potential experienced ward electrostatic energy minimization for the prediction of the by DyIII, and thus, we can calculate the electrostatic field pro- orientation of the ground state magnetic anisotropy axis of duced by charge on the ligands within a minimal valence bond dysprosium(III) ions, which does not rely on the fitting of (VB) model. Within this model, the charge is delocalized as a experimental data, requiring only the determination of an X-ray resonance hybrid that can be seen as a weighted sum of all crystal structure. ‘chemically stable’ Lewis structures (Fig. 1); this is a representa- tion of the leading contribution to the full VB wavefunction. By taking the sum of the partial charges accumulated by each atom Results in the VB resonance hybrid, qn, we arrive at a very simple frac- Many electron wavefunction and electrostatic minimization. tional charge distribution for the ligand, where, typically, very few An increasing number of ab initio CASSCF calcula- atoms will accommodate a charge and most will remain neutral. tions6,14,15,20,28 have shown that in most low-symmetry Our strategy is to construct the crystal field potential solely from
2 NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved.
82
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3551 ARTICLE
1 1 H 1 H 1 H 1 2 – / + / + /5 + /5 – / – /3 2 5 C N C 3 CH3 O HC C C C O +1 – paaH* +1 NO3 Na HC NH O O N C 1 1 O O 2 1 N O 1 1 – /3 – /3 2 – / – /2 1 + /5 H + /5 – /3 3 – /2 1 O – /2 N O +1 +1 Na – Na3DOTA Na O N 1 – /2 1 O H 1 – 1 2 – /2 1 1 – /3 2 acac : R ,R = CH3 – /2 R C R O N C C tfpb–: R1 = CF3, R2 = Ph – 1 2 O O tta : R = CF3, R = 2-thienyl 1 1 – 1 2 O – / – /3 hfac : R ,R = CF3 1 3 – /2 Br
1 H 1 Ph – /5 C – /5 –1 HCC Ph Si S Ph SiS– 1 – 3 – /5 MeCp HC CH Ph NH O 1 –1/ –1 – /5 5 –1 3– O teabmpH3 –1 N O H N HN Br teaH2– –1 –1 i – i PrO N O Pr O HO O –1 Br
Figure 1 | Partial charges assigned to the formally charged ligands in complexes 1–17. The zwitterionic N-(2-pyridyl)-acetylacetamide (paaH*) has two formal charges owing to the deprotonation of the a-carbon and protonation of the pyridyl nitrogen. Each pendant carboxylate arm of the macrocyclic 0 00 000 trisodium 1,4,7,10-tetraazacyclododecane N,N ,N ,N -tetraacetate ligand (Na3DOTA ) has a single negative charge that is delocalized evenly over the two oxygen atoms; three of the four acetate arms bind sodium cations, which each have a single positive charge. The aromatic anion of methylcyclopentadienyl (MeCp ) has a single negative charge that is delocalized evenly over the five cyclic carbon atoms. Not shown: Compounds 12 and 13 contain Zn2 þ ions, which have formal charges of positive two in our model. Compound 17 contains a central oxide (O2 ) ligand, which has a formal charge of negative two in our model. the fractional charges determined by the VB resonance hybrid, determination of magnetic anisotropy in cases of low excluding neutral atoms entirely (see Fig. 1 illustrating the frac- symmetry. In high symmetry, the orientation of the ground tional charges of the resonance hybrids for the ligands of interest Kramers doublet is pre-determined; note that in this case if ðÞa;b here). The partitioning of the charge without any need for VCF(y,j) does not stabilize the r = ðÞy; f electron density 15 2 15 computation illustrates the elegance of our model. along the symmetry axis, then mJ ¼ ± /2 will not be the Once the partitioning of the charge over the ligand is ground state. There is no such restriction in low-symmetry determined, the resulting partial charges are arranged around environments. the central DyIII ion using the known X-ray crystal structure of the complex, allowing the electrostatic potential to be easily Monometallic complexes. The energy spectra and g-tensors of calculated using crystal field theory29 (equation (2), where the ground 6H multiplets for compounds 1–14, as calculated (R ,y ,j ) are the spherical coordinates of the nth charged atom, 15/2 n n n ab initio, are given in Supplementary Tables S1–S18. Table 1 see Methods section). presents, for each complex, the principal value of the diagonal g- X Xk m tensor of the ground Kramers doublet (gz) and the electrostatic 4pð 1Þ k VCFðÞy; j ¼ hr iYk;mðÞy; j deviation angle, defined as the angle between the electrostatic ; ; 2k þ 1 anisotropy axis and the ab initio anisotropy axis along which g is k¼2 4 6 m¼ k ð2Þ z X defined. qnYk;mðyn; j Þ n We note that the anisotropy axis is accurately predicted by Rk þ 1 n n employing this minimal VB model, without taking into account different electron withdrawing or donating groups in the charged Minimization of the electrostatic energy in equation (1) in ligands, for example, hfac will have less electron density at the conjunction with this minimal VB model yields an orientation oxygen donor atoms compared with acac because of the of the anisotropy axis, which compares remarkably well electron withdrawing nature of the CF groups in the former with that obtained via rigorous ab initio calculations 3 ligand. (Table 1). To exemplify this correlation, we have calculated the magnetic properties of 14 low-symmetry monometallic dysprosium(III) complexes using the CASSCF ab initio Polymetallic complexes. Calculating the ab initio properties of methodology (see Methods) and compare them directly with the following polymetallic complexes is extremely computation- our electrostatic model. In this work, we focus on the ally expensive. Hence, to demonstrate the power of our simple
NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications 3 & 2013 Macmillan Publishers Limited. All rights reserved.
83
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3551
Table 1 | Comparison of ab initio and electrostatic calculations for DyIII complexes.
w Compound gz Electrostatic deviation (°) Reference 34 1 [Dy(acac)3(H2O)2] 19.62 10.9 43 2 [Dy(acac)3(phen)] 19.55 10.0 44 3 [Dy(acac)3(dpq)] 19.42 2.9 44 4 [Dy(acac)3(dppz)] 19.57 6.1 45 5 [Dy(tfpb)3(dppz)] 19.48 9.0 35 6 [Dy(tta)3(bipy)] 19.76 12.4 35 7 [Dy(tta)3(phen)] 19.66 8.0 46 8 [Dy(tta)3(pinene-bipy)] 19.81 6.9 47 9 [Dy(hfac)3(dme)] 19.65 11.0 3 þ 20 10 [Dy(paaH*)2(H2O)4] 19.78 2.4 1 þ 20 11 [Dy(paaH*)2(NO3)2(MeOH)] 19.68 7.4 1 þ 7 12 [DyZn2(teabmpH3)2(MeOH)] 19.98 6.6 1 þ 7 13 [DyZn2(teabmpH3)2] 19.90 8.0 2 þ z y 14,37 14 [Dy(DOTA)(H2O)Na3] 19.46 14.8
bipy, 2,20-bipyridine; dme, dimethoxyethane; dpq, dipyridoquinoxaline; dppz, dipyridophenazine; pinene-bipy, 4,5-pinene bipyridine; phen, 1,10-phenanthroline. wAngle between anisotropy axis calculated by ab initio and electrostatic methods. z Average gz value over four calculations, see Supplementary Tables S14–S18. yAngles between the experimentally determined and the ab initio and electrostatic anisotropy axes are 3.9° and 12.1°, respectively.
approach, we have performed a semi-quantitative comparison interact strongly with the two basal oxygen atoms. If, however, between the electrostatic calculations and published ab initio the quantization axis was parallel with the base and top of the results for three compounds: [Dy(MeCp)2(Ph3SiS)]2 15 (ref. 28) trapezium, there would be less interaction with the negative i [Dy6(teaH)6(NO3)6] 16 (refs 30,31) and [Dy5O( PrO)13] 17 charges thus stabilizing the orientation. This is the orientation of (refs 32,33). The methodology for the calculation of the the anisotropy axis (Fig. 2c and Supplementary Fig. S9) calculated electrostatic anisotropy axes in polymetallic complexes is by our electrostatic model and it provides a simple explanation identical to that of monometallic complexes and is performed for the ab initio results. Analogous arguments can be made for for each DyIII ion independently. The DyIII ions that are not the compound 11 where the b-diketonate oxygen donors are in a focus of the calculation are treated as part of the ligand and are similar trapezium-shaped arrangement, however, the coordina- given a þ 3 charge. The charged ligands in complexes 15–17 are tion environment now contains two chelating nitrate anions. The given in Fig. 1, which describes the charge partitioning based on oxygen atoms in NO3 have a larger negative partial charge than the minimal VB model. those in the b-diketonates, but this is offset by the positive charge on the nitrogen atom, which has an attractive effect on the electron density. Therefore, more-or-less the same anisotropy axis Discussion as in compound 10 is observed for compound 11, along the qn The form of the potential (equation (2)) contains terms of Rk þ 1 for diketonate-diketonate vector (Supplementary Fig. S10). each charged atom in the ligand, which implies that the closern to Compounds 1–10 have distorted square anti-prismatic geo- the DyIII ion and larger the magnitude of the charge, the greater metries and the calculated anisotropy axis of the ground state is its effect on the orientation of the anisotropy axis. Complexes 1–9 not found to be coincident with the pseudo-fourfold axis. This contain three b-diketonate ligands in a ‘paddle-wheel’-like observation, shown here to be due to simple electrostatic arrangement, with two b-diketonate ligands trans- to each other arguments, is contrary to many reports in the literature that and the third trans- to a neutral ligand (Fig. 2a). If the employ a fourfold axial interpretation to model the magnetic data ðÞa;b quantization axis of the r 15=2ðÞy; f electron density was along (refs 25,34–36). Clearly, in these cases, the electrostatics are more the ‘paddle-wheel’ axis, then the radial plane of the approximately important than pseudo-symmetry. ‘oblate’ density would be coincident with all three charged Compounds 12 and 13, chosen as a departure from b- ligands, representing a high-energy orientation. Therefore, the diketonate-based complexes in addition to their very interesting anisotropy axis is perpendicular to the ‘paddle-wheel’ axis, and we magnetic properties, are intimately related and can be inter- find that it passes through the two trans-b-diketonate ligands. converted reversibly by drying or soaking in methanol, via a single The radial plane of the oblate electron density is thus coincident crystal to single crystal transformation7. The difference between with only one charged ligand as opposed to two (Fig. 2b and the dysprosium(III) coordination environments is the removal of Supplementary Figs S1–S8). a terminal methanol molecule, changing the local symmetry from For compound 10, the negatively charged oxygen atoms of the a distorted pentagonal bipyramid to a distorted octahedron. two b-diketonate ligands are the closest to DyIII and therefore Considering the local DyIII coordination environment, a pair of have the greatest effect on the orientation of the anisotropy axis. trans- phenoxo oxygen atoms in both 12 and 13 have much These four atoms are roughly coplanar with the dysprosium(III) shorter Dy–O bond lengths than all others, at 2.21(2) Å compared ion and are arranged in a trapezium (Fig. 2c). The oblate electron with 2.39(2) Å for 12 and 2.186(4) Å compared with 2.3(1) Å for ðÞa;b density r 15=2ðÞy; f will be in a high-energy configuration when 13, thus defining both geometries as axially compressed. In both the quantization axis is normal to this plane of four negative cases, the three metal atoms are roughly coplanar with the charges, and therefore the minimum electrostatic energy and equatorial planes of the coordination polyhedra. The close oxygen anisotropy axis will lie in the plane of the b-diketonate oxygen atoms define an axially repulsive potential for dysprosium(III), atoms. The DyIII ion is much closer to the base of the trapezium which, coupled with the attractive nature of the Zn2 þ cations in and therefore if the quantization axis was to bisect the two the equatorial plane, explains the observed magnetic anisotropy parallel edges, then the radial plane of the electron density would axes (Supplementary Figs S11 and S12).
4 NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved.
84
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3551 ARTICLE
a H c R1 C R2 C C HC
HC
H O O C R1 R1 C H
C O O C C Dy NH HC CH 3 H N C O O C CH R2 R2
C
EE C 2.765 OO
H C H C 2.752 2.766 Dy b C C OO 4.308 CH H N
3
C
NH
H C
C H
HC
HC
Figure 2 | Magnetic anisotropy and electrostatic potential of b-diketonate species. (a) Idealized ‘paddle-wheel’ geometry of complexes 1–9 viewed down the ‘paddle-wheel axis’. (b) Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy axes for the ground Kramers doublet in 1, ðÞa;b 21 with the form of the oblate r 15=2ðÞy; f f-electron density using the representation of Sievers , magnified approximately 5 for clarity; Dy ¼ green, O ¼ red, C ¼ grey and H ¼ white. (c) Idealized electrostatic potential generated by a pair of paaH* ligands in the trapezium-shaped coordination environment present in the bis-b-diketonate-dysprosium(III) plane of compounds 10 and 11 and the resultant anisotropy axis (green line); red ¼ positive potential, blue ¼ negative potential.
600 a
400 a a
) b –1 200 360 c 270 180 0 b 135 Energy (cm 180 (°) b 90 –200 90 (°) c 45 0 0 –400 c
ðÞa;b Figure 3 | Orthogonal configurations for the magnetic anisotropy axis in compound 14. The form of the oblate r 15=2ðÞy; f f-electron density, visualized in the central structural diagrams, follows the representation of Sievers21, and has been magnified approximately 5 for clarity; Dy ¼ green, Na ¼ yellow, O ¼ red, N ¼ blue, C ¼ grey and H ¼ white. The rods in the central structural diagrams represent the magnetic anisotropy axes, identified by various methods: green ¼ electrostatic calculation, dark blue ¼ ab initio calculation (this work), light blue ¼ ab initio calculation14 and pink ¼ experimentally determined. The energies of each configuration have been calculated using our electrostatic approach. The electrostatic energy surface is calculated by considering all possible orientations (a,b) of the anisotropy axis in the potential generated by the charged ligands of compound 14. The electrostatic energy ðÞa;b surface shows double degeneracy of the minimum, maximum and saddle points due to the axial symmetry of the r 15=2ðÞy; f electron density.
2 þ In [Dy(DOTA)(H2O)Na3] 14, for which the magnetic configuration (Fig. 3a). Therefore, the anisotropy axis is anisotropy axis has been determined experimentally, the perpendicular to the pseudo-tetragonal axis of the molecule. dysprosium(III) ion is encapsulated by the macrocyclic DOTA4 The radial plane of the ‘oblate’ density is attracted by two Na þ ligand (ref. 37). The H atoms of the apical water molecule were ions more strongly than just a single Na þ ion (Fig. 3b), thus not found experimentally, so were placed in calculated positions determining the observed orientation of the anisotropy axis based on crystallographically characterized water molecules (Fig. 3c), which agrees well with the experimentally determined bound to LnIII ions (ref. 20). If the anisotropy axis was and ab initio axes (Table 1). We have rotated the apical water coincident with the pseudo-tetragonal axis, the radial plane of molecule in the ab initio calculations and found no dependence the oblate electron density would have a large interaction with the on the orientation of the ground state anisotropy axis to this four negatively charged acetate groups, yielding a high-energy perturbation, contrary to ref. 14.
NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications 5 & 2013 Macmillan Publishers Limited. All rights reserved.
85
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3551
Compound 15 contains two dysprosium(III) ions bridged by presented, we propose that this model can be used to aid in the two anionic Ph3SiS ligands and each capped by two classical rational design of molecular architectures displaying novel organometallic ligands, MeCp (ref. 28). The two ions are magnetic properties, exploiting and stabilizing the strong axiality related by inversion symmetry and therefore possess the same of the ground state of low-symmetry dysprosium(III), through single-ion electronic structure. The electrostatic potential at each the use of formally charged ligands. The simplicity of the proposal paramagnetic ion is dominated by the two MeCp ligands, is so profound that the model resonates strongly with the which are closer to the DyIII ion than the sulfur atoms at conclusions drawn in the 1950s and 1960s that the bonding of the 2.65(3) Å (average Dy–C distance) compared with 2.76(2) Å. This lanthanides is almost purely ionic (refs 38,39). leads to the anisotropy axis of the ground state lying By entirely neglecting the influence of neutral ligands in our perpendicular to the Dy-S-S-Dy plane (Supplementary model, we have shown the dominant nature of charged ligands in Fig. S13), in good agreement with ab initio calculations (ref. 28). the determination of the magnetic anisotropy of dysprosium(III) The hexametallic dysprosium(III) wheel (16) contains highly complexes. Compounds lacking any charged ligands are rare, but anisotropic paramagnetic centres, yet due to crystallographic S6 would likely show magnetic anisotropies that are much more symmetry, possesses a diamagnetic ground state with a toroidal sensitive to the type of ligands present, with contributions due to moment. Each DyIII ion is encapsulated by the teaH2 ligand dipoles and higher order multipoles as well as the spatial extent of and also has one chelating nitrate anion30. Applying our ligand electron density becoming important. The minimal VB electrostatic model to this compound yields the anisotropy axis model for the partitioning of charges on ligands works well for of each dysprosium(III) site in excellent agreement with the ab the formally charged ligands presented here. Other more general initio results (Supplementary Fig. S14)31. The anisotropy axis for schemes for the partitioning of atomic charges over the ligands each DyIII ion is canted around the ring in an alternating up/ are also being investigated, which may offer an improvement over down manner, which, due to the S6 symmetry of the molecule, the minimal VB model. causes the net cancellation of the out-of-plane magnetization in We are also extending the method to other lanthanide ions, the ground state31, leading to a toroidal moment8, similar to that examining whether this approach can work for other oblate ions 9,10 III observed in a Dy3 triangle . It is remarkable that such a simple (for example, Tb ) and whether the reverse electrostatic principles electrostatic approach can rationalize such complex physics. will apply to prolate ions (for example, ErIII). Although the Compound 17, with one of the highest energy barriers to the treatment presented here cannot be rigorously applied to non- reversal of the ground state magnetization, contains five Kramers ions in low-symmetry environments, TbIII complexes that III dysprosium(III) ions arranged in a pyramid, with each Dy possess a pseudo-doublet ground state with mJ ¼ ±6(gx ¼ gy ¼ 0, ion at the centre of an axially compressed octahedron. The gz ¼ 18) should follow similar electrostatic arguments to those equatorial plane of each ion is formed by four bridging discussed here for the determination of the magnetic anisotropy. isopropoxide (iPrO ) ligands and the axial positions are Conversely, preliminary results suggest that the ground state III occupied by the single m5-oxide bridge at the centre of the wavefunctions of Er ions in low-symmetry environments are not 32 molecule and a terminal isopropoxide ligand . The oxygen atom well defined and consist of strongly mixed mJ states, precluding the of the terminal iPrO ligand is substantially closer to the DyIII application of the treatment presented here. ion than all other donor atoms, at 2.04(1) Å compared with The work presented here is an advance not only for the chemistry 2.35(8) Å, and the central m5-oxide has a double negative charge. and physics communities involved with molecular magnetism, but These two features define a strongly repulsive axial potential for also for all areas concerned with the magnetic and spectroscopic ðÞa;b the r 15=2ðÞy; f electron density, where the energy is minimized properties of the lanthanides. Specifically, this work provides some when the quantization axis is coincident with this direction much-needed insight into the complex and continually intriguing (Fig. 4). The presence of the other charged ligands and the magnetic properties of dysprosium(III). High-level quantum trivalent dysprosium ions is a small perturbation in this case, chemical calculations such as CASSCF are computationally because of the strongly directional nature of the almost linear expensive and require intervention by experts to produce reliable iPrO -Dy-O2 axis. The results obtained using our electrostatic results. The complementary approach we outline here is available to model compare extremely well with those obtained using ab initio any chemist with minimal computational requirements. calculations33 (see Supplementary Table S19). Calculation of the ground state magnetic anisotropy axis of III Methods Dy in low-symmetry environments, employing an electrostatic Electrostatic calculations. The electrostatic calculations were performed on the minimization strategy, shows how simple chemical intuition can complete monometallic structures (excluding lattice solvent or non-coordinated aid in the understanding of a complex problem of electronic counter ions) using the reported X-ray geometry with no optimization. The charges structure. Given the success of the electrostatic model in the cases were assigned to the ligands as described in the text and all other atoms did not contribute to the potential. The angles describing the orientation of the ðÞa;b r 15=2ðÞy; f electron density with respect to the experimental geometry, (a,b), were then optimized to minimize the electrostatic energy. In applying this strategy, we have evaluated the uncertainty associated with using X-ray coordinates by moving the atomic positions by a random factor within the estimated standard deviations and found that the change in the orientation of the anisotropy axis over all monometallic compounds studied is on the order of 1°. We have elected to use the Freeman and Watson values for the average radial integrals (ref. 40), /rkS,and have investigated the effect of these values on the calculated anisotropy direction and also found deviations on the order of 1° when their values are altered non- systematically by up to 20%. The electrostatic calculations were implemented in a FORTRAN program, MAGELLAN, which is available from the authors upon request.
Figure 4 | Ground state magnetic anisotropy of compound 17. The blue Ab initio calculations. CASSCF calculations were performed with MOLCAS 7.6 rods represent the orientations of the anisotropy axes for each of the five (refs 12,41,42) on the same geometry as used for the electrostatic calculations. The III ANO-RCC-VTZP, VTZ and VDZ basis sets were used for the dysprosium ion, first Dy ions in complex 17 as calculated by our electrostatic model; coordination sphere atoms and all other atoms, respectively. The calculations Dy ¼ green, O ¼ red, C ¼ grey and H ¼ white. employed the second order Douglas–Kroll–Hess Hamiltonian, where scalar
6 NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved.
86
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3551 ARTICLE
relativistic contractions are taken into account in the basis set and the spin-orbit 26. Baldovı´,J.J.et al. Rational design of single-ion magnets and spin qubits based coupling is handled separately in the RASSI module. The active space for dys- on mononuclear lanthanoid complexes. Inorg. Chem. 51, 12565–12574 (2012). prosium was, in all cases, nine electrons in seven orbitals. It was found that the 27. Baldovı´, J. J., Borra´s-Almenar, J. J., Clemente Juan, J. M., Coronado Miralles, E. effect of the spin quartets and doublets was negligible on the orientation of the & Gaita-Arin˜o, A. Modeling the properties of lanthanoid single-ion magnets magnetic anisotropy and therefore only the 21 spin sextets were included in the using an effective point-charge approach. Dalton Trans. 41, 13705–13710 state averaged CASSCF and spin-orbit coupling calculations. The effect of CASPT2 (2012). corrections for dynamic electron correlation were investigated, but found to have a 28. Tuna, F. et al. A high anisotropy barrier in a sulfur-bridged organodysprosium minor role in the orientation of the magnetic anisotropy, where the extent of the single-molecule magnet. Angew. Chem. Int. Ed. 51, 6976–6980 (2012). basis set had much more of an impact, and therefore CASPT2 corrections were not 29. Hutchings, M. Point-charge calculations of energy levels of magnetic ions in performed. Cholesky decomposition of the two-electron integrals was performed to crystalline electric fields. Solid State Phys. 16, 227–273 (1964). save disk space and computational time. These three approximations combined led 30. Langley, S. K., Moubaraki, B., Forsyth, C. M., Gass, I. A. & Murray, K. S. to differences in the calculated ground state orientation for 1 of less than 1°, which Structure and magnetism of new lanthanide 6-wheel compounds utilizing is comparable with the uncertainty in the calculation of the electrostatic orienta- triethanolamine as a stabilizing ligand. Dalton Trans. 39, 1705–1708 (2010). tions, vide supra. 31. Ungur, L. et al. Net toroidal magnetic moment in the ground state of a {Dy6}- triethanolamine ring. J. Am. Chem. Soc. 134, 18554–18557 (2012). References 32. Blagg, R. J., Muryn, C. A., McInnes, E. J. L., Tuna, F. & Winpenny, R. E. P. 1. Bottrill, M., Kwok, L. & Long, N. J. Lanthanides in magnetic resonance Single pyramid magnets: Dy5 pyramids with slow magnetic relaxation to 40 K. imaging. Chem. Soc. Rev. 35, 557–571 (2006). Angew. Chem. Int. Ed. 50, 6530–6533 (2011). 2. Thompson, K. H. & Orvig, C. Editorial: Lanthanide compounds for therapeutic 33. Blagg, R. J. et al. Magnetic relaxation pathways in lanthanide single-molecule and diagnostic applications. Chem. Soc. Rev. 35, 499–499 (2006). magnets. Nat. Chem. 5, 673–678 (2013). 3. Aromı´, G., Aguila`, D., Gamez, P., Luis, F. & Roubeau, O. Design of magnetic 34. Jiang, S.-D., Wang, B.-W., Su, G., Wang, Z.-M. & Gao, S. A mononuclear coordination complexes for quantum computing. Chem.Soc.Rev.41, 537–546 (2012). dysprosium complex featuring single-molecule-magnet behavior. Angew. 4. Ishikawa, N., Sugita, M., Ishikawa, T., Koshihara, S. & Kaizu, Y. Lanthanide Chem. Int. Ed. 49, 7448–7451 (2010). double-decker complexes functioning as magnets at the single-molecular level. 35. Bi, Y. et al. Capping ligand perturbed slow magnetic relaxation in dysprosium J. Am. Chem. Soc. 125, 8694–8695 (2003). single-ion magnets. Chem. Eur. J. 17, 12476–12481 (2011). 14– 5. Habib, F. et al. The use of magnetic dilution to elucidate the slow magnetic 36. Ritchie, C. et al. Utilizing the adaptive polyoxometalate [As2W19O67(H2O)] relaxation effects of a Dy2 single-molecule magnet. J. Am. Chem. Soc. 133, to support a polynuclear lanthanoid-based single-molecule magnet. Inorg. 8830–8833 (2011). Chem. 50, 7004–7014 (2011). III III 6. Langley, S. K. et al. Heterometallic tetranuclear [Ln 2Co 2] complexes 37. Car, P.-E. et al. Giant field dependence of the low temperature relaxation of the including suppression of quantum tunneling of magnetization in the magnetization in a dysprosium(III)–DOTA complex. Chem. Commun. 47, III III [Dy 2Co 2] single molecule magnet. Inorg. Chem. 51, 11873–11881 (2012). 3751–3753 (2011). 7. Liu, J.-L. et al. Switching the anisotropy barrier of a single-ion magnet by 38. Bleaney, B. & Stevens, K. W. H. Paramagnetic resonance. Rep. Prog. Phys. 16, symmetry change from quasi-D5h to quasi-Oh. Chem. Sci. 4, 3310–3316 (2013). 108–159 (1953). 8. Soncini, A. & Chibotaru, L. Toroidal magnetic states in molecular wheels: 39. Wybourne, B. G. Spectroscopic Properties of Rare Earths (Interscience interplay between isotropic exchange interactions and local magnetic Publishers, 1965). anisotropy. Phys. Rev. B 77, 220406 (2008). 40. Freeman, A. J. & Watson, R. E. Theoretical investigation of some magnetic and 9. Chibotaru, L. F., Ungur, L. & Soncini, A. The origin of nonmagnetic Kramers spectroscopic properties of rare-earth ions. Phys. Rev. 127, 2058–2075 (1962). doublets in the ground state of dysprosium triangles: evidence for a toroidal 41. Karlstro¨m, G. et al. MOLCAS: a program package for computational chemistry. magnetic moment. Angew. Chem. Int. Ed. 47, 4126–4129 (2008). Comput. Mater. Sci. 28, 222–239 (2003). 10. Luzon, J. et al. Spin chirality in a molecular dysprosium triangle: the archetype 42. Veryazov, V., Widmark, P., Serrano-Andre´s, L., Lindh, R. & Roos, B. O. of the noncollinear Ising model. Phys. Rev. Lett. 100, 247205 (2008). MOLCAS as a development platform for quantum chemistry software. Int. J. 11. Gatteschi, D. Anisotropic dysprosium. Nat. Chem. 3, 830–830 (2011). Quant. Chem. 100, 626–635 (2004). 12. Aquilante, F. et al. MOLCAS 7: the next generation. J. Comput. Chem. 31, 43. Chen, G.-J. et al. Coordination-perturbed single-molecule magnet behaviour of 224–247 (2010). mononuclear dysprosium complexes. Dalton Trans. 40, 5579–5583 (2011). 13. Bernot, K. et al. Magnetic anisotropy of dysprosium(III) in a low-symmetry 44. Chen, G. et al. Enhancing anisotropy barriers of dysprosium(III) single-ion environment: a theoretical and experimental investigation. J. Am. Chem. Soc. magnets. Chem. Eur. J. 18, 2484–2487 (2012). 131, 5573–5579 (2009). 45. Wang, Z.-G. et al. Single-ion magnet behavior of a new mononuclear 14. Cucinotta, G. et al. Magnetic anisotropy in a dysprosium/DOTA dysprosium complex. Inorg. Chem. Commun. 27, 127–130 (2013). single-molecule magnet: beyond simple magneto-structural correlations. 46. Wang, Y., Li, X.-L., Wang, T.-W., Song, Y. & You, X.-Z. Slow relaxation Angew. Chem. Int. Ed. 51, 1606–1610 (2012). processes and single-ion magnetic behaviors in dysprosium-containing 15. Guo, Y.-N. et al. Strong axiality and Ising exchange interaction suppress zero- complexes. Inorg. Chem. 49, 969–976 (2010). field tunneling of magnetization of an asymmetric Dy2 single-molecule magnet. 47. Fatila, E. M., Hetherington, E. E., Jennings, M., Lough, A. J. & Preuss, K. E. J. Am. Chem. Soc. 133, 11948–11951 (2011). Syntheses and crystal structures of anhydrous Ln(hfac)3(monoglyme). Ln ¼ 16. Habib, F. et al. Supramolecular architectures for controlling slow magnetic La, Ce, Pr, Sm, Eu, Gd, Tb, Dy, Er, Tm. Dalton Trans. 41, 1352–1362 (2012). relaxation in field-induced single-molecule magnets. Chem. Sci. 3, 2158–2164 (2012). 17. Boulon, M. E. et al. Magnetic anisotropy and spin-parity effect along the series of Acknowledgements lanthanide complexes with DOTA. Angew. Chem. Int. Ed. 125, 368–372 (2013). This work was supported by the EPSRC (UK). N.F.C. thanks the University of Man- 18. Sorace, L., Benelli, C. & Gatteschi, D. Lanthanides in molecular magnetism: old chester for a President’s Doctoral Scholarship and Mr. J.P.S. Walsh for assistance with tools in a new field. Chem. Soc. Rev. 40, 3092–3104 (2011). figures. R.E.P.W. thanks The Royal Society for a Wolfson research merit award. 19. Sessoli, R. & Luzo´n, J. Lanthanides in molecular magnetism: so fascinating so challenging. Dalton Trans. 41, 13556–13567 (2012). Author contributions 20. Chilton, N. F. et al. Single molecule magnetism in a family of mononuclear N.F.C. and A.S. designed the research and performed the calculations. E.J.L.M., D.C. and b-diketonate lanthanide(III) complexes: rationalization of magnetic anisotropy R.E.P.W. provided detailed critique of the approach. N.F.C. and A.S. wrote the manu- in complexes of low symmetry. Chem. Sci. 4, 1719–1730 (2013). script with input from the other authors. 21. Sievers, J. Asphericity of 4f-shells in their Hund’s rule ground states. Z. Phys. B Con. Mat. 45, 289–296 (1982). 22. Rinehart, J. D. & Long, J. R. Exploiting single-ion anisotropy in the design of Additional information f-element single-molecule magnets. Chem. Sci. 2, 2078–2085 (2011). Supplementary Information accompanies this paper at http://www.nature.com/ 23. Jiang, S.-D. et al. Series of lanthanide organometallic single-ion magnets. Inorg. naturecommunications Chem. 51, 3079–3087 (2012). 24. Yamashita, K. et al. A luminescent single-molecule magnet: observation of Competing financial interests: The authors declare no competing financial interests. magnetic anisotropy using emission as a probe. Dalton Trans. 42, 1987–1990 Reprints and permission information is available online at http://npg.nature.com/ (2013). reprintsandpermissions/ 25. Pointillart, F. et al. A series of tetrathiafulvalene-based lanthanide complexes displaying either single molecule magnet or luminescence—direct magnetic and How to cite this article: Chilton, N.F. et al. An electrostatic model for the determination photo-physical correlations in the ytterbium analogue. Inorg. Chem. 52, of magnetic anisotropy in dysprosium complexes. Nat. Commun. 4:2551 doi: 10.1038/ 5978–5990 (2013). ncomms3551 (2013).
NATURE COMMUNICATIONS | 4:2551 | DOI: 10.1038/ncomms3551 | www.nature.com/naturecommunications 7 & 2013 Macmillan Publishers Limited. All rights reserved.
87
Supplementary Figure S1 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 2. Dy = green, O = red, N = blue, C = grey and H = white.
88
Supplementary Figure S2 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 3. Dy = green, O = red, N = blue, C = grey and H = white.
89
Supplementary Figure S3 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 4. Dy = green, O = red, N = blue, C = grey and H = white.
90
Supplementary Figure S4 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 5. Dy = green, F = yellow, O = red, N = blue, C = grey and H = white.
91
Supplementary Figure S5 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 6. Dy = green, S = purple, F = yellow, O = red, N = blue, C = grey and H = white.
92
Supplementary Figure 6 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 7. Dy = green, S = purple, F = yellow, O = red, N = blue, C = grey and H = white.
93
Supplementary Figure S7 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 8. Dy = green, S = purple, F = yellow, O = red, N = blue, C = grey and H = white.
94
Supplementary Figure S8 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 9. Dy = green, S = purple, F = yellow, O = red, N = blue, C = grey and H = white.
95
Supplementary Figure S9 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 10. Dy = green, O = red, N = blue, C = grey and H = white.
96
Supplementary Figure S10 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 11. Dy = green, O = red, N = blue, C = grey and H = white.
97
Supplementary Figure S11 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 12. Dy = green, Br = yellow, Zn = teal, O = red, N = blue, C = grey and H = white.
98
Supplementary Figure S12 – Comparison of the ab initio (blue rod) and electrostatic (green rod) anisotropy directions for the ground Kramers doublet in 13. Dy = green, Br = yellow, Zn = teal, O = red, N = blue, C = grey and H = white.
99
Supplementary Figure S13 – The ground state anisotropy directions (blue rods) for the DyIII ions in 15, calculated with the electrostatic model. Dy = green, S = purple and C = grey; H atoms not shown for clarity.
100
Supplementary Figure S14 – The ground state anisotropy directions (blue rods) for the DyIII ions in 16, calculated with the electrostatic model. Dy = green, O = red, N = blue and C = grey; H atoms not shown for clarity.
101
Supplementary Table S1 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 1. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.01 19.62 - 2 167.4 0.17 0.27 15.89 7.1 3 252.7 1.81 2.43 11.75 11.7 4 309.7 1.40 2.00 7.56 14.5 5 338.3 2.65 7.73 11.63 89.5 6 431.3 0.01 0.28 15.29 86.3 7 475.9 0.09 0.22 17.69 70.7 8 549.3 0.02 0.04 19.11 60.2
Supplementary Table S2 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 2. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.01 19.55 - 2 142.2 0.16 0.24 15.84 16.4 3 211.0 1.20 1.59 12.42 17.8 4 262.3 3.82 5.37 8.44 14.0 5 310.5 2.25 4.03 9.68 73.6 6 350.2 1.12 2.37 16.52 72.3 7 457.2 0.01 0.06 17.61 85.0 8 530.1 0.01 0.03 18.90 57.0
102
Supplementary Table S3 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 3. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.01 0.02 19.42 - 2 133.3 0.37 0.52 15.90 7.1 3 202.5 2.67 4.75 11.17 25.0 4 240.6 2.07 5.78 7.32 89.4 5 284.0 0.52 3.30 10.05 67.1 6 313.1 1.05 3.05 15.36 76.6 7 444.4 0.00 0.02 19.21 84.2 8 545.1 0.00 0.02 19.71 52.9
Supplementary Table S4 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 4. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.01 19.57 - 2 159.8 0.17 0.26 15.72 1.8 3 239.0 1.95 2.96 11.54 12.4 4 276.4 1.95 5.75 10.95 72.4 5 321.3 1.11 4.26 9.42 69.4 6 365.1 1.56 3.38 15.85 67.3 7 448.3 0.05 0.29 18.81 89.7 8 519.1 0.04 0.09 19.47 52.0
103
Supplementary Table S5 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 5. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.01 0.02 19.48 - 2 134.1 0.61 1.60 14.88 6.9 3 163.9 0.65 2.15 16.66 89.3 4 211.7 1.02 4.42 9.70 9.5 5 243.8 3.58 4.60 11.42 51.8 6 273.7 0.63 1.76 16.46 84.0 7 336.8 0.02 0.04 19.67 89.5 8 476.2 0.00 0.01 19.86 49.6
Supplementary Table S6 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 6. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.01 19.76 - 2 142.4 0.09 0.65 16.50 63.7 3 161.7 1.67 2.00 13.63 22.9 4 201.1 3.28 4.62 12.42 44.7 5 234.2 2.47 3.73 12.04 67.1 6 305.9 0.74 1.07 16.12 79.9 7 415.0 0.02 0.21 17.53 80.5 8 470.6 0.05 0.20 18.57 51.5
104
Supplementary Table S7 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 7. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.01 0.01 19.66 - 2 134.2 0.72 0.91 15.79 11.2 3 181.3 3.34 4.90 10.68 25.9 4 216.5 1.38 5.99 6.78 38.7 5 243.8 1.81 2.78 12.29 56.3 6 298.8 0.77 1.23 16.03 79.1 7 411.6 0.05 0.14 18.56 85.7 8 516.5 0.02 0.04 19.49 53.0
Supplementary Table S8 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 8. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.01 19.81 - 2 94.3 0.12 0.41 18.84 78.6 3 159.6 2.12 2.41 13.99 18.7 4 185.9 3.65 5.75 11.39 36.1 5 215.1 2.24 2.80 13.55 63.1 6 281.8 0.51 0.78 16.44 84.8 7 409.1 0.07 0.08 19.17 49.0 8 472.0 0.02 0.15 19.49 89.1
105
Supplementary Table S9 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 9. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.01 0.01 19.65 - 2 118.5 0.58 0.96 15.96 12.8 3 164.9 3.85 5.73 10.06 51.8 4 196.4 3.70 5.05 7.46 48.8 5 228.6 4.02 4.93 10.95 53.8 6 273.2 1.31 2.27 16.04 85.9 7 344.8 0.10 0.23 19.11 89.1 8 457.1 0.01 0.02 19.70 51.0
Supplementary Table S10 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 10. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.00 19.78 - 2 250.5 0.05 0.07 16.40 2.0 3 408.4 1.18 1.90 12.50 1.5 4 508.3 5.35 5.95 6.92 3.5 5 600.3 2.07 2.19 10.39 89.4 6 689.8 0.36 0.56 14.89 88.0 7 774.5 0.01 0.03 19.30 67.9 8 817.8 0.01 0.02 19.48 72.8
106
Supplementary Table S11 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 11. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.01 0.02 19.68 - 2 113.6 0.23 0.44 16.56 7.9 3 194.9 3.43 4.14 11.47 25.5 4 249.7 3.57 4.19 10.01 79.9 5 307.3 0.59 0.89 15.47 80.2 6 338.5 0.22 1.77 15.54 83.8 7 394.3 0.77 1.51 17.04 83.6 8 432.4 0.08 1.68 17.65 81.0
Supplementary Table S12 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 12. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.00 0.00 19.98 - 2 270.0 0.46 0.55 18.61 84.8 3 294.6 0.14 0.25 16.58 4.6 4 346.8 1.34 1.84 14.28 84.0 5 410.8 4.57 5.92 8.84 59.0 6 447.3 1.53 2.01 12.25 26.7 7 484.8 0.41 0.50 16.21 49.2 8 526.3 0.02 0.10 17.57 50.3
107
Supplementary Table S13 – Energies and g-tensors for the Kramers doublets of the ground multiplet of DyIII for compound 13. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.01 0.02 19.90 - 2 287.8 2.38 6.82 11.77 56.2 3 324.0 3.02 5.50 8.40 42.5 4 406.6 0.37 2.38 14.22 76.5 5 481.0 2.03 7.31 9.69 46.8 6 559.0 2.19 2.44 15.59 57.6 7 697.9 0.43 0.74 16.48 82.7 8 816.8 0.09 0.20 19.09 82.5
Supplementary Table S14 – Energies and g-tensors for the Kramers doublets of the III ground multiplet of Dy for compound 14, H2O rotation A. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.07 0.14 19.67 - 2 94.3 1.10 1.43 17.67 77.3 3 167.2 3.03 3.94 11.18 20.7 4 241.4 0.99 6.74 8.78 85.3 5 311.4 2.85 4.35 10.50 77.1 6 374.0 0.55 1.79 13.55 85.6 7 449.5 0.68 0.82 16.44 85.2 8 621.6 0.01 0.03 19.64 86.9
108
Supplementary Table S15 – Energies and g-tensors for the Kramers doublets of the III ground multiplet of Dy for compound 14, H2O rotation B. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.14 0.29 19.51 - 2 72.5 0.96 1.11 17.97 78.1 3 156.7 4.05 4.30 11.18 23.9 4 228.4 0.01 7.23 7.90 84.9 5 299.7 2.28 5.38 10.50 73.8 6 361.5 0.95 2.29 13.32 90.0 7 441.9 0.77 1.01 16.50 82.3 8 611.3 0.02 0.06 19.63 87.5
Supplementary Table S16 – Energies and g-tensors for the Kramers doublets of the III ground multiplet of Dy for compound 14, H2O rotation C. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.30 0.83 18.93 - 2 45.8 0.23 1.14 17.61 80.8 3 144.6 4.27 6.46 9.82 40.2 4 209.8 1.92 5.75 7.81 70.8 5 284.1 3.18 4.93 10.47 74.0 6 346.5 0.34 1.59 13.55 85.5 7 424.3 0.78 0.85 16.55 83.1 8 595.7 0.01 0.04 19.65 87.2
109
Supplementary Table S17 – Energies and g-tensors for the Kramers doublets of the III ground multiplet of Dy for compound 14, H2O rotation D. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.05 0.08 19.75 - 2 117.8 1.30 2.39 16.58 67.3 3 178.9 1.26 3.79 10.96 28.4 4 265.0 2.37 6.69 8.99 85.2 5 329.8 1.35 4.65 10.55 74.8 6 391.0 1.31 2.71 13.26 88.0 7 466.7 0.66 0.87 16.47 84.3 8 637.4 0.03 0.05 19.63 87.1
Supplementary Table S18 – Energies and g-tensors for the Kramers doublets of the III ground multiplet of Dy for compound 14, average of the four H2O rotations A – D. The angles between the principal axis of each doublet and the ground state are also given.
-1 Doublet Energy (cm ) gx gy gz Angle (°) 1 0.0 0.1(1) 0.3(3) 19.5(4) - 2 83(31) 0.9(5) 1.5(6) 17.5(6) 74.9 3 162(15) 3(1) 5(1) 10.8(7) 27.3 4 236(23) 1(1) 6.6(6) 8.4(6) 79.8 5 306(19) 2.4(8) 4.8(4) 10.50(4) 74.8 6 368(19) 0.8(4) 2.1(5) 13.4(2) 87.5 7 446(18) 0.72(6) 0.89(8) 16.49(4) 84.0 8 616(18) 0.02(1) 0.05(1) 19.637(7) 86.9
Supplementary Table S19 – Angles between anisotropy axes and terminal iPrO--Dy bond for dysprosium(III) sites in complex 17.
Method Dy1 (°) Dy2 (°) Dy3 (°) Dy4 (°) Dy5 (°) Ab initio 5.8 4.1 2.7 4.7 5.2 Electrostatic 4.6 1.9 5.0 4.0 4.0
110
7. Paper four: “The first near-linear bis(amide) f-block complex: a blueprint for a high temperature single molecule magnet”
N. F. Chilton, C. A. P. Goodwin, D. P. Mills and R. E. P. Winpenny, Chem. Commun., 2015, 51, 101.
111
112
ChemComm
View Article Online COMMUNICATION View Journal | View Issue
The first near-linear bis(amide) f-block complex: a blueprint for a high temperature single molecule Cite this: Chem. Commun., 2015, 51,101 magnet†
Received 21st October 2014, Accepted 5th November 2014 Nicholas F. Chilton, Conrad A. P. Goodwin, David P. Mills* and Richard E. P. Winpenny* DOI: 10.1039/c4cc08312a www.rsc.org/chemcomm
We report the first near-linear bis(amide) 4f-block compound and show (CASSCF) ab initio calculations12 that are often employed to 7,13 that this novel structure, if implemented with dysprosium(III), would examine 4f complexes, pioneered by Chibotaru. Electrostatic have unprecedented single molecule magnet (SMM) properties with an approachessuggestthattheoptimalligandenvironmentto 1 III energy barrier, Ueff, for reorientation of magnetization of 1800 cm . exploit the oblate spheroidal electron density of Dy is axial, where rigorously axial systems have the benefit of maintaining a Since their initial discovery,1 single molecule magnets (SMMs) single, unique quantization axis for the total angular momentum 14 have been lauded as candidates for high density data storage mJ states. A set of unadulterated mJ states implies that the devices.2 A major breakthrough in the field3 occurred in 2003 with probability of quantum tunnelling of the magnetization (QTM) 2 the observation of SMM behavior in a monometallic {TbPc2} is reduced, therefore increasing magnetic relaxation times. 1 4 complex with an energy barrier, Ueff = 230 cm . The ensuing The simplest axial ligand environment is a linear two-coordinate 5 decade saw rapid growth in lanthanide SMMs with the Ueff barrier complex with donor atoms exclusively on a single Cartesian axis; the 1 7 to magnetization reversal increased to 652 cm for another Ueff barrier is so large for the {Dy5}and{Dy4K2} alkoxide complexes 6 1 derivative of {TbPc2}, and 585 cm for a polymetallic Dy@{Y4K2} because of the strongly axially repulsive crystal field potentials along 7 III complex. The highest blocking temperature TB (i.e. the tempera- the local z-direction of each Dy .Othercompoundssuchas 16 ture at which hysteresis is observed) was also increased to 14 K, [(C8H8)2Ln] (ref. 15) or Cloke’s bis(arene) lanthanide complexes 3 3 8 via an N2 radical bridge in a {Tb2N2 }complex. are sometimes described as linear, but lack donor atoms directly on Although three of these milestones employ the TbIII ion, by far the axis. Linear 3d-metal compounds also show remarkable magnetic III 17 the most utilized lanthanide ion in SMMs is Dy by virtue of its behaviour with very high Ueff values. A one coordinate lanthanide 9 + unique electronic structure. Apart from a radical-bridged complex [DyO] has been considered theoretically with a very large Ueff 3 10 III 14 {Dy2N2 }complex, nearly all polymetallic Dy -based SMMs predicted, however such an entity is not chemically feasible. possess negligible interactions between magnetic spin centres, Very low coordination numbers for 4f-ions are difficult to and instead rely on the single ion anisotropy of DyIII (i.e. the local achieve as these are large, electropositive ions, which require a i crystal field environment) to provide the barrier to the reversal of sterically demanding ligand. Such a pro-ligand HN(Si Pr3)2 was i i magnetization. Intra- or intermolecular interactions are often designed, and synthesised from ClSi Pr3 and LiHN(Si Pr3), and this III i detrimental to the performance of Dy SMMs so that doping a was converted to the group 1 transfer agent [KN(Si Pr3)2]withKH. i small amount of the paramagnetic ion into a diamagnetic host Reacting two equivalents of [KN(Si Pr3)2] with samarium(II) diio- III 7 lattice (usually the Y analogue) often results in an increased Ueff. dide yields the mononuclear homoleptic bis(amide) complex, i i An electrostatic model for the design of ideal ligand environ- [( Pr3Si)2N–Sm–N(Si Pr3)2] 1 (Fig.1,seeESI† for details). ments to exploit the maximal anisotropy of DyIII has been Complex 1 is the first near-linear f-element complex, with an postulated,11,12 and shown to be in good agreement with multi- N–Sm–N angle of 175.52(18)1 in the solid state (Fig. 2, see ESI† configurational complete active space Self consistent field for details); this near-linearity contrasts with the bent C–Ln–C II 18–20 angles of [Ln {C(SiMe3)3}2] complexes (Ln = Sm, Yb, Eu). The School of Chemistry, The University of Manchester, Oxford Road, Manchester, M13 9PL, UK. E-mail: [email protected], [email protected] † Electronic supplementary information (ESI) available: Full synthetic details, crystallography, NMR spectroscopy, magnetism, and ab initio and magnetic relaxation methodologies. CCDC 1017031. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4cc08312a Fig. 1 Synthetic route to 1.
This journal is © The Royal Society of Chemistry 2015 Chem. Commun., 2015, 51,101--103| 101
113
View Article Online
Communication ChemComm
Fig. 2 Molecular structure of 1. Hydrogen atoms omitted for clarity. Selected bond lengths (Å) and angles (1): Sm1–N1 2.483(6), Sm1– N2 2.483(6), Sm1 C7 3.180(8), Sm1 C16 3.169(8), Sm1 C19 3.082(8), Sm1 C34 3.224(8), N1–Sm1–N2 175.52(18), Sm1–N1–Si1 109.9(3), Sm1– N1–Si2 111.6(3), Si1–N1–Si2 138.5(4), Sm1–N2–Si3 109.8(4), Sm1–N2–Si4 110.8(3), Si3–N2–Si4 138.8(4). Fig. 3 Electronic states and magnetic transition probabilities for the 6 ground H15/2 multiplet of 2 in zero field. The x-axis shows the magnetic bulky iPr groups are vital for the isolation of a homoleptic complex, moment of each state along the main magnetic axis of the molecule. Relaxation commences from the | 15/2 state and only includes pathways as [Sm{N(SiMe ) } (THF) ] exhibits additional O-donors.21 The i 3 2 2 2 which reverse the magnetization. Relaxation probabilities are calculated Sm–N distances in 1 [2.483(6) Å] are longer than those observed based on a magnetic perturbation and are normalized from each departing in [Sm{N(SiMe3)2}2(THF)2] [mean Sm–N 2.433(9) Å] but this is state (see ESI† for details). compensated by 1 exhibiting four short Sm Cmethine distances [Sm C 3.082(7)–3.224(7) Å] that are closer than the analogous III 6 Sm Cmethyl contacts observed in [Sm{N(SiMe3)2}2(THF)2][Sm C is excellent (Fig. S2 and S3, ESI†). Dy has a H15/2 ground multiplet, 21 3.32(1)–3.46(1) Å]. The approximately planar SmNSi2 fragments in which is split by the crystal field into eight Kramer’s doublets with
1 are staggered with respect to each other (twist angle of 44.421), total angular momentum projections mJ = 1/2, 3/2,... 15/2. The with the deviation from 901 attributed to agostic Sm Cmethine ab initio calculations show that the lowest six Kramers doublets are interactions. the almost pure mJ states of mJ = 15/2, 13/2, 11/2, 9/2, 7/2 Formally each nitrogen atom carries a single negative charge and 5/2, sharing a common quantization axis (Fig. 3 and Tables S1 and the SmII ion is divalent, with an [Xe]4f6 configuration. The f 6 and S2, ESI†). The two most energetic doublets are strongly mixed; a 7 configuration leads to a formally diamagnetic F0 ground state, characteristic of low symmetry complexes due to the lack of a 14 with close lying excited states that provide a non-zero magnetic rigorous molecular CN axis. Along the main magnetic axis moment at room temperature. Magnetic measurements on 1 these two states can be expressed as |cabi = 64%| 3/2i + give a room temperature magnetic moment of 3.62 mB that falls 26%|81/2i and |ccdi = 68%| 1/2i + 31%|83/2i and (Table S2, towards zero at low temperature (Fig. S2 and S3, ESI†). This is ESI†), giving the most energetic Kramers doublet a large gy value of clearly incompatible with interesting low temperature magnetic B17.5 perpendicular to the main magnetic axis. behaviour. However, the structure of 1 is close to the ideal linear Magnetic relaxation in lanthanides follows three possible arrangement to stabilize the large angular momentum states of routes: (1) QTM within the ground doublet (e.g. | 15/2i - DyIII and produce monstrous uniaxial magnetic anisotropy. |+15/2i in Fig. 3), (2) thermally assisted QTM (TA-QTM) via Such a DyIII compound is challenging to make; we believe a excited states (e.g. | 15/2i - | 13/2i - |+13/2i - |+15/2i), or i route via the heteroleptic [Dy{N(Si Pr3)2}2I] treated with the potas- (3) an Orbach process composed of direct and/or Raman sium salt of a large anion might work through precipitation of a mechanisms (e.g. | 15/2i - | 13/2i - |+15/2i). The most potassium iodide. Other routes canbeimagined,andherewe probable pathway depends on the composition of the states present predictions of the magnetic properties of such a complex, involved and their interactions with phonons. For example, the III intending to inspire synthetic work towards the linear Dy slow magnetic relaxation for {Dy4K2} was shown to occur via the complex, and, more ambitiously, the isoelectronic TbII analogue. first or second excited states (TA-QTM), depending on the i i + III The properties of [( Pr3Si)2N–Dy–N(Si Pr3)2] 2 are predicted by number and location of neighboring Dy ions providing a CASSCF/RASSI/SINGLE_ANISO22 ab initio calculations (see ESI† for source of transverse magnetic field.7 The states with opposing details) employing the structure of 1,whereSmII has been replaced magnetic projections are mixed proportionally to the product of by DyIII. The validity of the method was tested by calculating the the transverse field and the transverse g-factors and therefore variable temperature magnetic behavior of 1, where the agreement TA-QTM will occur via the excited state which has transverse
102 | Chem. Commun., 2015, 51,101--103 This journal is © The Royal Society of Chemistry 2015
114
View Article Online
ChemComm Communication g-factors above a certain threshold or where its main magnetic suggest that such hypothetical complexes are now chemically axis is non-collinear with that of the ground state. All non-QTM feasible. These metal–organic compounds are becoming of great transitions are induced by the vibrational modes of the lattice importance in molecular magnetism.8,10,27,28 (phonons) which create local oscillating magnetic fields This work was supported by the EPSRC (grant number through modulation of dipolar fields as well as an oscillating EP/K039547/1) (UK). N.F.C. thanks The University of Manchester crystal field potential.23 To a first approximation, we can for a President’s Doctoral Scholarship. R.E.P.W. thanks The associate the probability of a phonon induced transition with Royal Society for a Wolfson research merit award. We would the average magnetic13,14,24 and crystal field perturbation like to thank J. P. S. Walsh for assistance with graphics. matrix elements (see ESI† for details). Compared to all known DyIII complexes the calculated Notes and references properties for 2 areuniquewithverysmalltransverseg-factors and a common principal axis for the lowest six Kramers doublets. 1 R.Sessoli,D.Gatteschi,A.CaneschiandM.A.Novak,Nature, 1993, 365, 141. 2 D. Gatteschi, R. Sessoli and J. Villain, Molecular Nanomagnets, This suggests that both the probability of QTM within the ground Oxford University Press, 2006. doublet and TA-QTM is vanishingly small until the two most 3 L. Sorace, C. Benelli and D. Gatteschi, Chem. Soc. Rev., 2011, 40, 3092. energetic doublets. Orbach relaxation is also strongly disfavoured 4 N. Ishikawa, M. Sugita, T. Ishikawa, S. Koshihara and Y. Kaizu, J. Am. Chem. Soc., 2003, 125, 8694. in the low lying states as magnetic transition probabilities due to 5 D. N. Woodruff, R. E. P. Winpenny and R. A. Layfield, Chem. Rev., phonons are miniscule (Fig. 3). Efficient magnetic relaxation will 2013, 113, 5110. only occur via the highest energy doublets (Fig. 3, Fig. S4 and 6 C. R. Ganivet, B. Ballesteros, G. de la Torre, J. M. Clemente-Juan, E. Coronado and T. Torres, Chem. – Eur. J., 2013, 19, 1457. Tables S4 and S5, ESI†). Therefore the ab initio calculation 7 R. J. Blagg, L. Ungur, F. Tuna, J. Speak, P. Comar, D. Collison, E 1 predicts Ueff 1800 cm for 2 –fargreaterthananycomplex W. Wernsdorfer, E. J. L. McInnes, L. F. Chibotaru and R. E. P. to date. Whilst such calculations may over-estimate the energies Winpenny, Nat. Chem., 2013, 5, 673. 25,26 8 J. D. Rinehart, M. Fang, W. J. Evans and J. R. Long, J. Am. Chem. Soc., of the crystal field states, we can predict a TB in excess of 77 K as 2011, 133, 14236. such temperatures are often around 1/20th of the Ueff value if QTM 9 D. Gatteschi, Nat. Chem., 2011, 3, 830. 10 J. D. Rinehart, M. Fang, W. J. Evans and J. R. Long, Nat. Chem., 2011, within the ground doublet is disfavored, e.g. the TB/Ueff ratios for 3 3, 538–542. {Tb2N2 }, {Mn12}and{Mn6} are approximately 1/16, 1/15 and 11 J. D. Rinehart and J. R. Long, Chem. Sci., 2011, 2, 2078. 1/13 cm 1 K 1, respectively. Calculations for the TbII analogue 3, 12 N. F. Chilton, D. Collison, E. J. L. McInnes, R. E. P. Winpenny and which is also a 4f9 ion, predict analogous behavior to 2 (Table S6, A. Soncini, Nat. Commun., 2013, 4, 2551. III 13 L. Ungur, J. J. Le Roy, I. Korobkov, M. Murugesu and L. F. Chibotaru, ESI†). The high local symmetry at the Dy site implies that the Angew. Chem., Int. Ed., 2014, 53, 4413. nuclear quadrupole and hyperfine interactions will be axially sym- 14 L. Ungur and L. F. Chibotaru, Phys. Chem. Chem. Phys., 2011, 13, 20086. metric, preventing efficient QTM within the lower energy doublets. 15 (a) F. Mares, K. Hodgson and A. Streitwieser, Jr., J. Organomet. Chem., 1970, 24, C68; (b) K. R. Meihaus and J. R. Long, J. Am. Chem. To examine the stability of 2, we have performed ab initio Soc., 2013, 135, 17952. calculations for modified geometries where the N–Dy–N angle and 16 For example, J. G. Brennan, F. G. N. Cloke, A. A. Sameh and the Dy–N bond lengths have been altered by 0.51 and 0.01 Å, A. Zalkin, Chem. Commun., 1987, 1668. 17 J. M. Zadrozny, D. J. Xiao, M. Atanasov, G. J. Long, F. Grandjean, respectively (Fig. S5, ESI†). The results show that 2 is stabilized F. Neese and J. R. Long, Nat. Chem., 2013, 5, 577. when the Dy–N bond length is shortened and the N–Dy–N angle is 18 C. Eaborn, P. B. Hitchcock, K. Izod and J. D. Smith, J. Am. Chem. closer to 1801 compared to 1, yielding more favorable electronic Soc., 1994, 116, 12071. 19 C. Eaborn, P. B. Hitchcock, K. Izod, Z.-R. Lu and J. D. Smith, properties. These calculations do not take into account the inclu- Organometallics, 1996, 15, 4783. sion of a counter-ion in the structure, which may have conse- 20 G. Qi, Y. Nitto, A. Saiki, T. Tomohiro, Y. Nakayama and H. Yasuda, quences for crystal packing and the local structure of 2. Tetrahedron, 2003, 59, 10409. 21 W. J. Evans, D. K. Drummond, H. Zhang and J. L. Atwood, Inorg. Compound 1 is the first near-linear bis(amide) 4f-block Chem., 1988, 27, 575. complex. It allows us to propose a blueprint for the first 22 (a)G.Karlstro¨m, R. Lindh, P.-Å. Malmqvist, B. O. Roos, U. Ryde, generation of ‘high-temperature’ SMMs, with blocking tem- V. Veryazov, P.-O. Widmark, M. Cossi, B. Schimmelpfennig and P. Neogrady, et al., Comput.Mater.Sci., 2003, 28, 222; (b) V. Veryazov, peratures exceeding that of liquid N2 (77 K). The synthesis of P. Widmark, L. Serrano-Andre´s, R. Lindh and B. O. Roos, Int. J. Quantum the proposed archetypes, viz. the DyIII and TbII analogues of 1, Chem., 2004, 100, 626; (c)F.Aquilante,L.DeVico,N.Ferre´,G.Ghigo, is currently underway in our laboratory, however we believe this P. Malmqvist, P. Neogra´dy,T.B.Pedersen,M.Pitonˇ´ak, M. Reiher and B. O. Roos, et al., J. Comput. Chem., 2010, 31,224;(d)L.F.Chibotaruand is a target many other groups should be pursuing. Calculations L. Ungur, J. Chem. Phys., 2012, 137, 064112. on other fn ions suggest that f9 is ideal; even for the oblate f8 23 A. Abragam and B. Bleaney, Electron Paramagnetic Resonance of TbIII analogue, 4, we find that the pseudo-doublets show strong Transition Ions, Oxford University Press, 1970. 24 L. Ungur, M. Thewissen, J.-P. Costes, W. Wernsdorfer and L. F. Chibotaru, mixing between the | mJi and |+mJi projections, (Tables S7 and S8, Inorg. Chem., 2013, 52, 6328. ESI†), which would lead to strong zero-field QTM. 25R.Marx,F.Moro,M.Do¨rfel,L.Ungur,M.Waters,S.D.Jiang,M.Orlita, J. Taylor, W. Frey, L. F. Chibotaru and J. van Slageren, Chem. Sci., 2014, While 2 would have a huge Ueff, an even higher Ueff barrier 5, 3287. might be possible if dianionic monodentate ligands could be 26 E. Moreno Pineda, N. F. Chilton, R. Marx, M. Do¨rfel, D. O. Sells, i i incorporated, e.g. [( Pr3Si)2C–Dy–C(Si Pr3)2] , containing dianionic P. Neugebauer, S.-D. Jiang, D. Collison, J. van Slageren, E. J. L. McInnes and R. E. P. Winpenny, Nat. Commun., 2014, 5, 6243. methanediides. Our preliminary results suggest this could raise Ueff 27 F. Moro, D. P. Mills, S. T. Liddle and J. van Slageren, Angew. Chem., byafactorof1.2to1.3.Theincredibleadvancesmadeinlow Int. Ed., 2013, 52, 3430. coordination number metal–organic compounds in the last decade 28 R. A. Layfield, Organometallics, 2014, 33, 1084.
This journal is © The Royal Society of Chemistry 2015 Chem. Commun., 2015, 51,101--103| 103
115
Supplementary Information for: The First Linear f-block Complex and the
Route to a Liquid N2 Temperature Single Molecule Magnet Nicholas F. Chilton, Conrad A. P. Goodwin, David P. Mills* and Richard E. P. Winpenny*
Contents
1. General Synthetic Methods ...... 2
†† 2. Preparation of [K{NSiiPr3)2}] (KN ) ...... 2
3. Preparation of [(iPr3Si)2N-Sm-N(SiiPr3)2] (1) ...... 3
4. Crystallographic Details and Data...... 3
5. Nuclear Magnetic Resonance Spectroscopy ...... 4
6. Magnetic Measurements ...... 4
7. Ab initio Method ...... 6
8. Magnetic Relaxation...... 12
9. Supplementary References ...... 14
1
116
1. General Synthetic Methods All manipulations were carried out using standard Schlenk techniques, or an Innovative Technology PureLab HE glovebox, under an atmosphere of dry argon. Solvents were dried by refluxing over potassium and degassed before use. All solvents were stored over potassium mirrors except for THF which was stored over activated 4 Å sieves. d6-benzene was distilled from potassium, degassed by three freeze-pump-thaw cycles and stored under
i argon. ClSiPr 3 was dried over Mg turnings and KH was obtained as a suspension in mineral oil and was washed with hexane (3 x 50 mL) and dried in vacuo before use. All other reagents were used as received. 1H, 13C and 29Si NMR spectra were recorded on a Bruker 400 spectrometer operating at 400.2, 100.6 and 79.5 MHz respectively; chemical shifts are quoted in ppm and are relative to TMS (1H, 13C, 29Si). FTIR spectra were recorded as nujol mulls in KBr discs on a Perkin Elmer Spectrum RX1 spectrometer. Elemental microanalyses were carried out by Mr Stephen Boyer at the Microanalysis Service, London Metropolitan University, UK.
†† 2. Preparation of [K{N(SiiPr3)2}] (KN )
Liquid ammonia (28 mL, 1.2 mol) was added to a pre-cooled (–78 °C) solution of ClSiiPr3 (17.618 g, 91.38 mmol), allowed to warm to –60 °C and stirred for 1 hr. The colorless mixture was warmed to room temperature, giving a white precipitate. This was filtered, cooled to –78 °C and BunLi (2.5 M, 36.4 mL, 91.2 mmol) was added drop-wise and the reaction mixture allowed to warm to room temperature and refluxed for 1 hour. Volatiles were removed in vacuo and the yellow oil heated to 140 °C in vacuo for 3 hours to afford
[Li(NHSiiPr3)]n as a yellow oil in essentially quantitative yield. ClSiiPr3 (20.49 mL, 18.46 g,
95.76 mmol) was added to a solution of [Li(NHSiiPr3)]n (16.352 g, 91.2 mmol) in THF (30 mL) and refluxed in a sealed ampoule for 4 days. The pale yellow solution was filtered and volatiles removed in vacuo. Distillation (170 °C oil bath, 10-2 Torr) gave a mixture of
HN(iPr3Si)2 and iPr3SiCl. The mixture was then heated (<100 °C) in vacuo to give
HN(iPr3Si)2 in essentially quantitative yield. A portion of HN(iPr3Si)2 (10.428 g, 31.63 mmol) in toluene (20 mL) was added drop-wise to a pre-cooled (-78 °C) slurry of KH (1.52 g, 38 mmol) in toluene (10 mL) and refluxed for 3 hours. Filtration, followed by removal of volatiles in vacuo, afforded a beige solid. The solid was washed with hexanes (3 x 5 mL) and dried in vacuo to afford the product as an off-white powder, with multiple crops obtained from the washings. Yield: 7.041 g, 61 %. Anal. Calcd. for C18H42Si2NK: C, 58.78; H, 11.51; 1 N, 3.81. Found: C, 58.66 H, 11.46; N, 3.90. H NMR (d6-benzene, 298 K) δ: 1.03 (sept, JHH =
2
117
13 1 7.5 Hz, 6H, CH(CH3)2), 1.43 (d, JHH = 7.5 Hz, 36H, CH(CH3)2). C{ H} NMR (d6-benzene, 29 298 K) δ: 17.48 (s, CH(CH3)2), 20.85 (s, CH(CH3)2). Si NMR (d6-benzene, 298 K) δ: -16.31 (s). IR v/cm-1 (Nujol): 1238 (m), 1217 (m), 1175 (s), 1152 (s), 1099 (m), 1068 (m), 1002 (m), 974 (m), 876 (s), 649 (s), 611 (s).
3. Preparation of [(iPr3Si)2N-Sm-N(SiiPr3)2] (1) A solution of KN†† (2.942 g, 8 mmol) in toluene (15 mL) was added drop-wise to a pre- cooled slurry (–78 °C) of [SmI2(THF)2] (2.194 g, 4 mmol) in toluene (10 mL) and allowed to warm to room temperature with stirring. The mixture was stirred for 4 days, with the precipitation of a pale solid. The supernatant was filtered and the pale solid extracted with toluene (3 x 8 mL). Volatiles were removed in vacuo. The dark green solid was extracted with hexane (3 x 4 mL), concentrated to 2 mL and stored at –25 °C to give the product as dark green blocks (1.786 g, 55 %). Anal. Calcd. for C36H84Si4N2Sm: C, 53.56; H, 10.49; N,
3.47. Found: C, 53.39 H, 10.40; N, 3.38. Magnetic moment (Evans method, d6-benzene, 298 1 K): μeff = 3.73 µB. H NMR (d6-benzene, 298 K) δ: 0.20 (br, 72 H, CH(CH3)2), 6.81 (br, 12 H, 13 1 CH(CH3)2). C{ H} NMR (d6-benzene, 298 K) δ: 29.77 (s, CH(CH3)2), 35.28 (s, CH(CH3)2). 29 -1 Si NMR (d6-benzene, 298 K) δ: 6.06 (s). IR v/cm (Nujol): 1076 (s), 1057 (s), 944 (m), 882 (m), 697 (m), 660 (m).
4. Crystallographic Details and Data CrysAlisProS1 was used for control and integration and SHELXTLS2 and OLEX2S3 were employed for structure solution and refinement and for molecular graphics.
–1 Crystal data for 1: C36H84N2Si4Sm, Mr = 807.77 g mol , space group Pbca, a = 20.509(2), b 3 = 16.0788(19), c = 26.515(2), α = β = γ = 90, V = 8743.5(15) Å , Z = 8, Zʹ = 1 , ρcalcd = 1.227 –3 –1 g cm ; MoKα radiation, λ = 0.71073 Å, μ = 1.477 mm , T = 150 K. 18707 points (7676 unique, Rint = 0.130, 2θ < 50.0°). Data were collected on an Agilent Technologies Supernova diffractometer and were corrected for absorption (transmission 0.908 – 1.000). The structure was solved by direct methods and refined by full-matrix least-squares on all F2 values to give
2 2 2 2 2 1/2 wR2 = {[w(F0 – Fc ) ]/[w(F0 ) ]} = 0.1085, conventional R = Σ(|Fo| – |Fc|)/Σ|Fo| = 2 2 2 2 2 ½ 0.0728 for F values of 7676 with F0 > 2σ(F0 ), S = [Σw(Fo – Fc ) /(n + r – p)] = 0.908 for 412 parameters. Residual electron densities were 0.99 e Å–3 maximum and –1.21 e Å–3 minimum. CCDC 1017031 (1) contains the supplementary crystallographic data for this
3
118
paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
5. Nuclear Magnetic Resonance Spectroscopy Complex 1 exhibits simple 1H, 13C and 29Si NMR spectra, indicative of a symmetrical species in solution on the NMR timescale. The highly temperature-dependent chemical shift of the methine protons in a variable temperature study is attributed to the paramagnetism of the SmII center (μeff = 3.73 μB at 298 K, Evans method), which is comparable to 18 [Sm{N(SiMe3)2}2(THF)2] (μeff = 3.45 μB at 298 K).
Figure S1 - Variable temperature 1H NMR spectrum of 1.
6. Magnetic Measurements The magnetic properties of 1 were measured with a Quantum Design MPMS XL7 SQUID magnetometer, from 2 – 300 K in a field of 0.1 T. A fresh crystalline sample was ground and fixed with eicosane in an NMR tube under an inert atmosphere. The NMR tube was flame sealed under vacuum and mounted in a straw for attachment to the sample rod. The measurement was corrected for the diamagnetism of the straw, eicosane and the sample, the latter with Pascal’s constants, but not for the NMR tube. The room temperature χMT value of 3 -1 1.64 cm mol K (3.62 μB, very similar to that measured by the Evan’s method) reduces
4
119
3 -1 slowly with cooling until 100 K where it drops rapidly to 0.04 cm mol K (0.56 μB) at 2 K, and is in good agreement with that predicted by ab initio calculations for 1, Figure S2. The χ vs. T plot, Figure S3, shows the characteristic plateau at low temperatures of a temperature independent paramagnetism effect due to mixing of paramagnetic states into the formally
7 diamagnetic F0 ground term. The sharp rise at the lowest of temperatures is due to a small paramagnetic impurity of SmIII. The small differences between the calculated and experimental traces are due to the very subtle nature of the electronic structure of SmII, a feature owed to the strongly mixed close lying excited states.
Figure S2 – Experimental and calculated χMT vs. T for 1.
5
120
Figure S3 – Experimental and calculated χM vs. T for 1.
7. Ab initio Method Ab initio calculations were performed with MOLCAS 7.8 using the RASSCF, RASSI and SINGLE_ANISO modules. In all cases the 4f ions were treated with the ANO-RCC-VTZP basis set, the N donors and the Si atoms with the ANO-RCC-VDZP basis set, while all C and H atoms were treated with the ANO-RCC-VDZ basis set. For all calculations the 4fn configuration was modelled with a complete active space of n electrons in 7 orbitals. For DyIII and TbII calculations, 21 sextets, 224 quartets and 158 doublets were included in the orbital optimization and 21 sextets, 128 quartets and 130 doublets were mixed by spin-orbit coupling. For the TbIII calculation, 7 septets, 140 quintets and 195 triplets were included in the orbital optimization and 7 septets, 105 quintets and 91 triplets were mixed by spin-orbit coupling. For the SmII calculation, 7 septets and 140 quintets were included in the orbital optimization and mixed by spin-orbit coupling. The SINGLE_ANISO module was employed to calculate the crystal field decomposition for the spin-orbit eigenstates and to yield the crystal field parameters for the ground spin-orbit multiplet of DyIII (Table S3) and TbII. The crystal field parameters were used with PHIS4 to examine the composition of the
6
121
̂ 푞 wavefunctions using the Hamiltonian (expressions for the 푂푘 operators can be found in the PHI User manual at www.nfchilton.com/phi):
푘 ̂ 푞 ̂ 푞 퐻퐶퐹 = ∑ ∑ 퐵푘푂푘 푘 = 2,4,6푞 =‒ 푘 The small difference in the energies for the ab initio and the crystal field calculations (Tables S1 and S2) is due to the simple nature of the crystal field model, however in this case the departure is very small owing to an extremely strong crystal field potential. The reduction of the principal gz values from those expected for pure mJ states in the ab initio calculation is due to covalent effects14 which are excluded in the crystal field parameterization therefore recovering the expected values.
Table S1 - Ab initio calculated electronic states for 2.
-1 E (cm ) gx gy gz gz angle (°) 0 0.0000 0.0000 19.9044 - 526 0.0002 0.0002 17.0068 0.3 1026 0.0009 0.0011 14.1772 0.5 1426 0.0337 0.0335 11.5169 0.8 1682 0.9868 0.9224 8.9785 2.6 1803 1.6038 3.0759 6.4529 17.6 1836 9.9041 8.9619 1.9276 2.1 1861 2.3210 17.5423 0.5019 3.9
Table S2 - Crystal field calculated electronic states for 2.
-1 E (cm ) gx gy gz gz angle (°) Wavefunctions 0 0.0000 0.0000 20.0000 - 100%| ± 15 2⟩ 517 0.0003 0.0003 17.3327 0.3 100%| ± 13 2⟩ 1032 0.0011 0.0014 14.6629 0.4 100%| ± 11 2⟩ 1427 0.0324 0.0316 11.9909 0.8 100%| ± 9 2⟩ 1675 0.9743 0.9112 9.2952 2.4 96%| ± 7 2⟩ + 4%| ∓ 7 2⟩ 94%| ± 5 2⟩ + 2%| ± 1 2⟩ + 2%| ∓ 5 2⟩ 1798 1.1326 2.7141 6.5704 12.4 + 1%| ± 3 2⟩ + 1%| ∓ 3 2⟩ 64%| ± 3 2⟩ + 26%| ∓ 1 2⟩ + 3%| ± 1 2⟩ 1836 10.0160 8.6757 2.2354 2.1 + 3%| ∓ 3 2⟩ + 2%| ∓ 5 2⟩ + 1%| ± 5 2⟩ 68%| ± 1 2⟩ + 31%| ∓ 3 2⟩ + 1%| ± 5 2⟩ 1861 1.7752 17.6994 0.3270 4.6 + 1%| ∓ 1 2⟩
7
122
Table S3 - Ab initio calculated crystal field parameters for 2. Parameter Value (cm-1) ‒ 2 퐵 2 -1.1917562647038E-01 ‒ 1 퐵 2 -2.1119600305588E-02 0 퐵2 -1.1350199169445E+01 1 퐵2 9.7268699983100E-02 2 퐵2 1.4987459279149E-01 ‒ 4 퐵 4 -7.6752762196053E-04 ‒ 3 퐵 4 -5.4536142276867E-04 ‒ 2 퐵 4 1.0486686143292E-03 ‒ 1 퐵 4 -1.6575458595647E-03 0 퐵4 -7.2335435797888E-03 1 퐵4 -2.0658529932308E-04 2 퐵4 3.6533101842767E-03 3 퐵4 -4.2448546105741E-04 4 퐵4 -9.6344270575107E-04 ‒ 6 퐵 6 -4.2317898861512E-05 ‒ 5 퐵 6 -5.2516010707221E-05 ‒ 4 퐵 6 1.2655146551751E-06 ‒ 3 퐵 6 -1.0642106266008E-05 ‒ 2 퐵 6 -1.7313470374153E-05 ‒ 1 퐵 6 3.8030782269202E-06 0 퐵6 4.6813385093314E-05 1 퐵6 1.3122617258695E-05 2 퐵6 -6.5696276247257E-05 3 퐵6 1.5735760988224E-05 4 퐵6 -2.7776053728306E-07 5 퐵6 -4.7115889505093E-06 6 퐵6 -3.2828109496254E-05
8
123
Table S4 - Magnetic transition probabilities for 2.
| ‒ 15 2⟩ | ‒ 13 2⟩ | ‒ 11 2⟩ | ‒ 9 2⟩ | ‒ 7 2⟩ | ‒ 5 2⟩ | ‒ 푎푏⟩ | ‒ 푐푑⟩ | ‒ 13 2 ⟩ 100