close

Вход

Забыли?

вход по аккаунту

?

Organic and Organometallic Molecular Magnetic MaterialsЧDesigner Magnets.

код для вставкиСкачать
REVIEWS
Organic and Organometallic Molecular Magnetic MaterialsDesigner Magnets
Joel S. Miller" and Arthur J. Epstein*
Magnets composed of molecular species
o r polymers and prepared by relatively low-temperature organic synthetic
methodologies are a focus of contemporary materials science research. The anticipated properties of such molecularspecies-based magnetic materials, particularly in combination with other
properties associated with molecules
and polymers. may enable their use in
future generations of electronic, magnetic, and/or photonic/photronic devices
ranging from information storage and
magnetic imaging to static and low-frequency magnetic shielding. A tutorial of
typical magnetic behavior of molecular
materials is presented. The three distinct
models (intramolecular spin coupling
through orthogonal orbitals in the same
spatial region within a molecule/ion, intermolecular spin coupling through
pairwise "configuration interaction" be-
tween spin-containing moieties, and
dipole-dipole, through-space interactions) which enable the design of new
molecular-based magnetic materials are
discussed. To achieve the required spin
couplings for bulk ferro- or ferrimagnetic behavior it is crucial to prepare materials with the necessary primary, secondary, and tertiary structures akin to
proteins. Selected results from the
worldwide effort aimed at preparing
molecular-based magnetic materials by
these mechanisms are described. Some
organometallic solids comprised of linear chains of alternating metallocenium
donors (D) and cyanocarbon acceptors
(A) that is. . . . D'+A'-D'+A'- . ' , exhibit cooperative magnetic phenomena.
Bulk ferromagnetic behavior was
first observed below the critical (Curie)
temperature T, of 4.8 K for [Fe"'(C,Me,),]'+ [TCNE]'- (Me = methyl;
1. Introduction
Magnets have captivated mankind from time immemorial
and have played a prominent role in the development of both
science and technology as well as in the evolution of society."]
In perhaps the earliest example of technology transfer involving
condensed matter science, Chinese workers exploited the funda-
[*I
['I
Dr. J. S. Miller"]
Central Research & Development
Experimental Station E328
Du Pont. P.O. Box 80328, Wilmington. DE 19880-0328 (USA)
Dr. A. J. Epstein
Department of Physics and Department of Chemistry
The Ohio State University
Columbus, OH 43210-1106 (USA)
Telefax: Int. code + (614)292-3706
New Address:
Department of Chemistry
University of Utah
Salt Lake City. U T 841 12 (USA)
Telefax: Int. code + (801)581-8433
Angew.
C'/iiw?.
Int. Ed. Engl. 1994, 33. 385-415
0 YCH
TCNE = tetracyanoethylene). Replacement of Fe"' with Mn"' leads to a ferromagnet with a
of 8.8 K in agreement
with mean-field models developed for
this class of materials. Replacement
with Cr"', however, leads to a ferromagnet with a T, lowered to 3.65 K which is
at variance with this model. Extension
to the reaction of a vanadium(0) complex with TCNE leads to the isolation of
a magnet with a
z 400 K, which exceeds the thermal decomposition temperature of the material. This demonstrates that a magnetic material with a T,
substantially above room temperature is
achievable in a molecule/organic/polymeric material. Finally, a new class of
one-dimensional ferrimagnetic materials based on metalloporphins is discussed.
mental observation of magnetic behavior that had been discovered by ancient Grecians by invention of the "south-pointing''
compass.['b1 The development of present day
and theoretical condensed matter physics as well as quantum
mechanics and molecular chemistry are also deeply indebted to
magnetism. Magnets have become ubiquitously woven into the
fabric of everyday life and are indispensable in our present
world. This is reflected by the annual sales of magnets in the
western-world which reached $2 x lo9['] in 1990. Applications
of magnets span from magneto-mechanical machines (frictionless bearings, medical implants, magnetic separators, etc.)
to magnetic shielding and electromagnetic radiation absorption
(deflection of static and low-frequency magnetic fields), to
acoustic devices (loudspeakers, microphones, etc.), to telecommunication/information technology (switches, sensors, magnetic resonance imaging, as well as magnetic and optical disks,
etc.), and to motors and generator^.'^] Futuristic "smart" materials and systems will rely upon "smart" switches, sensors, and
transducers that undoubtedly will be comprised in part by mag-
Verlu~sgesellschuffrmhH. 0-69451 Wcinheim, 1994
0570-0833/94/0404-03853 10.00+ .25/0
385
REVIEWS
netic materials. Magnets consequently are an important focus of
modern materials research programs, with emphasis on improvement of their technologically useful properties.
Current magnetic materials share the generic attributes of
a ) being atom-bascd. b) having d or f orbital transition- or lanthanide-metal-based spin sitcs. and c) possessing extended network bonding in at least two dimensions. Furthermore. thcy d )
are preparcd by high-temperature metallurgical methodologics.
Metamorphosis from such atom-based inorganic materials to
molecule-based organic;polymcric materials should enablc the
a) specific alteration of the magnetic properties by cstablished
organic chemistry methodologies, b) combination of magnetic
properties with other mechanical. electrical, and/or optical
propertics, and c) simplicity of fabrication often cnjoyed by organic materials, that is. low-tcmperature synthesis as opposed to
the high-temperature metallurgical methods currently utilized.
These qualities may herald the use of organic materials in futurc
generations of electronic. magnetic, photonic and/or photronic
devices. The discovery of molecule-based magnets parallels thc
discovery of cooperative electrical phenomena in organic;
molecular superconductors with the advent of cooperative magnetic phenomena in organic/molecular materials.
The issue of what constitutes a molecule-based magnet frequently arises. As noted previously classical magnets are com-
J. S. Miller. A. J. Epstein
prised solely of a high density of d or f orbital mctal spin sitcs
and are prepared by metallurgy. An organic magnetic material,
in contrast. intist posscss at least one type of spin site t h a t is
based on s and p orbitals and bc prepared by organic chemistry
methodology; that is. an organic component of thc material
plays an active role in the resultant magnetic bchavior. Intcrmcdiatc between thcsc extremes lie solely metal-spin-site-b~iscd
magnetic ma teria Is prcpa red and modified by orca n i c chemist I-y
methodologies ( i t . , an organic component of the material pla) s
a passive role in the resultant niagnctic behavior). I n principle.
organic-based materials prepared by metallurgy (i.c., high-temperature pyrolysis of organic compounds"') lic between thcsc
cxtremes. Although various specific criteria for ii molecule,
organic magnetic matcrial have heen applied. we define organic'
polymeric magnetic materials as those materials prepared
through the cstablished low-temperature synthetic procedure of
organic. organometallic. and coordination metal synthetic
chemistry. We iisc the term "molecular species" to rcfcr to any
of the above be it a molecule, ion, or polymer. This fcaiure is
common to all thc materials studied by the majority of the
scicntists that have participated at confcrcnces on [his topic.'"
It should be noted that the ferromagnetic Ziiitl phase
Ca,,MnBi, coinposed of isolated ions.'51 tho ferroniagnet
RbzCrC1, coniposcd of a covalently bonded two-dimensional
Arthur J. Epstein 1vus borri in Brooklj.17, N P I York
~
and received his B. S. in Ph!.sics fiom
the Polj.tcd~nicInstituft. ?/'Brooklyn in 1966 and his Ph. D. in Phj~.ric.r,
fj.oni tlie Uniivrsity
of Penn.ryliiania in 1971. After thirteen years l i t the Xerox WcJhsterRcsearch C e ~ i t~rlierc,
~,
IIP was Principal Scicnti,c.t,he joined The Ohio State L'niversit?, in 1985 as Professor of'
and Prof&.sor ofChmii.rtrj.. H P hns heen The Director of'thc Ohio Siotc Cnii
f i ~ rMatcv-ids Rr.~rurcIi.since 1989. He also has h a w adjuncr Profi~s.sorof P
at the Universitj. of Florida and IL'US a Visiting Prqfkxsor at Uniiwxit+ Puris-Surl and The
Technion-hue1 Institute of Technology. He is a Regional Editor.fiir Sytirlirtic Metals. His
current research interests include experimental and theoreticul study of'magnetic, electronic/optical, and transport phenomena of sj,ntlietic magnets (molecular, organic and po1j.mc.ric magnets) and .sjxthetic metals (conducting poljwiers, orgunic and nd~c,iilrir
niatcrinls) ,
including the stud), ojexcitations and their dynamics. He also is active in [lie e.xplorntion of
potential applications of' unconventional magnetic, electronic, and optical muterials,for ii,hicli he consults ivitli , s o w r i l companies.
In addition to having originated tiwntl; patents and editing three c,onfhrence proceedings, he hus puhlishetl over 350 paper.s in
these and other areas.
1
REVIEWS
Designer Magnets
network structure.lhl as well as several examples of magnetic
materials prepared by the pyrolysis of organic moleculesr71are
typically prepared at high temperatures and as such are not
within the direct focus of this review.
Molecular-species-based ferromagnetic compounds although
postulated in the 1960s have been realized only within the past
decade.", x . 9 1 I n this review we will briefly summarize the
essence of the magnetic behavior important for understanding
the physical properties of this class of materials". l o ] and review
some 01' the research results achieved worldwide in this rapidly
developing area of contemporary inquiry.
2. Magnetic Behavior-A
Tutorial
Magnetic behavior is typically detected by a material's attraction (or repulsion) to a magnet. It arises from the intrinsic spin
of an electron and how the electron spins on adjacent atoms (or
molecules) couple with each other. Every electron has a minute
magnetic moment associated with the quantum-mechanical
"spin". The electrons reside in orbitals and each orbital may
have two electrons --one "spin-up". that is, spin aligned with an
applied magnetic field (7). and one "spin-down", that is, spin
aligned opposing an applied magnetic field (l)~-~-such
that the
spins cancel. A radical is an atom. ion. o r molecule with an odd
number of electrons (or an even number of electrons 17 in more
than 1 7 2 orbitals. Thus, a radical must have at least one orbital
with an unpaired electron and hence have a net spin.
The unpaired electron has a spatial distribution, termed spin
density. which can be distinct from the charge and electron
densities of a molecular species. For a radical anion such as the
tetracyanoethenide radical, [TCNE]' -, the additional electron is
added to the lowest unoccupied molecular orbital (LUMO) of
the closed shell TCNE. The original L U M O of T C N E is now a
partially occupied molecular orbital (POMO), if we follow a
restricted open-shell Hartree-Fock ( R O H F ) formalism in
which all of the orbitals are doubly occupied except for the
POMO. (Note that P O M O refers to orbital occupancies that
may lead to states that can be different.) For [TCNE]'- the
POMO is a TE*orbital ofh,, symmetry,["] and there are several
nodes in this orbital which are dominated by a large node between the central carbon atoms participating in the TCNE
double bond (Fig. I top). In the R O H F formalism, the spin
density of the POMO represents the excess spin density in the
molecular species. and this spin density must be in one direction
(positive) and is termed a spin. Furthermore, for [TCNE]'- this
excess spin density can only reside in the p orbitals, and it should
not have EPR hyperfine splittings because there is no mechanism for the spin to interact with the s orbitals which have the
proper symmetry to interact with the nuclear spins. The simplest
way to couple all of the orbitals is to use the unrestricted
Hartree Fock ( U H F ) model in which the restriction of double
occupancy in an orbital is relaxed. In the U H F model each
orbital is singly occupied and one thus solves for positive ( a )or
negative (P) spin d e n ~ i t i e s . ~ '(By
~ " ]convention for a radical, the
positive spin density (aspin) is greater than the negative spin
density (b spin)). Because the a-and P-spin densities are calculated individually, coupling of all of the orbitals is possible so
that regions exist with excess a and p spins. The amount of
~
unpaired spin density on an atom within the molecular species
need not, however, correspond to the atom contributions in the
POMO (Fig. 1 middle). Furthermore, the regions of excess spin
are now no longer determined by the R O H F POMO; thus mixing with s orbitals leads to the observation of EPR hyperfine
splittings.
Fig. 1. Electron structure information for
[TCNE]'-. All calculations were done on
[TCNE]'- with the local density functional
method using the optimum calculated geometry [I I ] and the unrestricted spin formalism
with the program DGauss and visualizations
with UniChem [14a]. Highest occupied ?-spin
orbital (top): spin density (middle): electrostatic potential (ESP) energy surface at
k90 kcalmol-' (3.9 ev) (bottom). ESP surface was obtained by the interaction of a particle
of 1 + charge with the unrelaxed molecular
electron density. Positive regions (violet) correspond to repulsive interactions and negative
regions (green) correspond to attractive regions (from ref. [14b]). (Figures kindly provided by D. A. Dixon and S. C. Walker).
Spin polarization can be described by either valence bond
(Fig. 2) or molecular orbital approaches. Spin polarization arises because there are different exchange interactions between the
unpaired a-spin electrons in the POMO and the other cc-spin
electrons as compared to no exchange with the (I-spin electrons.
For example, for the Li atom the 2s electron polarizes the I s
electrons because there are different effective potentials for the
a-and (3-spin I s electrons. Relaxing the double occupancy for
the Is orbital will lead to a lower energy. Another way to consider such spin polarizations, which has been used to explain the
spin density in the CH; radical['3b1(positive (a)spin density on
C and negative ((I) spin density on H), is the use of two simple
rules: 1) an intraatomic Hund's rule in which electrons on an
atom tend to have parallel spins and 2) electrons that form a
chemical bond tend to have antiparallel spins. Because the aand (3-spin densities can interact, spin polarization can occur, by
which an unpaired electron can polarize the paired spins in an
adjacent o or TC bond in such a way that one of the paired
electrons in the bond is closer to one atom than another (Fig. 2).
A way to view the charge distribution in a molecular species,
in addition to the use of Mulliken
is to calculate the
electrostatic surface potential (ESP). The interaction energy (attractive or repulsive) of a test charge (in this case I + ) with the
unrelaxed electron density is thus calculated as illustrated for
[TCNE]' - in Figure 1 bottom. The ESP surface shows that there
is significant negative charge on the nitrogen atoms even though
one formally would place most of the negative charge on the two
carbon atoms of the double bond.
387
REVIEWS
J. S. Miller, A. J. Epstein
thermal energy. Such spins do not couple (that is, align cooperatively), instead they form a weak magnet (paramagnet Fig. 3 a).
When the radicals come sufficiently close to each other, the
coupling energies increase, ultimately influencing the alignment
of the neighboring spins. If this alignment occurs in such a
manner that the spins oppose each other, this leads to antiferromagnetic behavior (Fig. 3 b). Bulk ferromagnetic behavior, albeit rare, occurs when the spins in a solid align in the same
direction (Fig. 3c). resulting in a net magnetic moment. Thus,
ferromagnetism requires that the individual unpaired spins interact collectively with each other aligning themselves. Ferrimagnets can occur when there are different moments (essentially
a different number of spins) on neighboring sites which align
antiferromagnetically but d o not cancel; this leads to a net moment for the solid (Fig. 3 d). It is vital to note that this commer-
a) paramagnet
disorded spins (2D)
Fig. 2. Illustration of the orbital interactions for spin polarization in a o bond (top).
between rt orbitals through two cr bonds (center left), between a orbitals through cr
bonds and x orbitals (center right). and between a orbitals rotated by 90 through
o bonds (bottom).
The net S = 1/2 spin density (positive or negative) on each
atomic site of a molecular species is determined by adding up
separately the contribution to each atomic site of every “upspin” and “down-spin’’ filled electron orbital, that is, the sum of
all a-spin electron densities minus the sum of all f3-spin electron
densities. The POMO electron and spin densities differ from the
molecular/ion charge densities. The total charge density is a sum
of the charge density associated with each occupied energy level
(both x and p spin) at each atomic site less the positive nuclear
charge, and is illustrated by the electrostatic surface generated
when a point charge is brought toward a molecular species
(Fig. 1 bottom). The POMO electron density, the unpaired spin
density, and the total charge density of a molecule will therefore
differ from each other as shown in Figure 1 for [TCNE]’-.
The wavefunction and therefore the POMO valence-electron
density is determined computationally, however, electron and
spin densities can be determined experimentally by diffraction
methods. The valence-electron density can be determined from
the low-angle scattering in a high-resolution single-crystal structural determination[12e1or preferably from the difference between results obtained by using single-crystal X-ray and neutron
diffraction studies at the same temperature. The absolute spin
density can be directly determined from single-crystal polarizedneutron diffraction studies at low temperature. Absolute spin
densities may also be obtained experimentally from analysis of
either the N M R contact shifts[’2b1or detailed analysis of the
EPR hyperfine couplings constants ;[I5] however, the relaxation
time of the species must be amenable for these resonance measurements to be applicable.
Radicals are usually far enough apart that their magnetic
coupling energy is small compared to the coupling- breaking
388
i t l t l t
c) ferromagnet
ordered (aligned)spins (2D)
‘ l ‘ t ’ t
111111 1/1/11
e) ordered (opposed) spins (2D)
Fig. 3. Two-dimensional spin alignment for a) paramagnet, b) antiferromagnet,
c) ferromagnet. d ) ferrimagnet. and e ) canted ferromagnet behavior.
cially useful highly magnetic behavior is not a property of a
molecular species; it is a cooperative solid-state (bulk) property.
Three-dimensional (3D) interactions are typically necessary to
achieve long-range magnetic order and bulk ferromagnetic behavior; however, two-dimensional (2D) interactions in some
circumstances suffice, as is probably the situation for
Rb,CrCI,[6a1 in which interlayer spacings can be increased by
replacing R b + with bulky alkylammonium counterions (vide
infra), but the is not significantly altered. For strictly one-dimensional(1 D) nearest neighbor interactions, long-range order
can only occur at 0 K. However, for the special case when a
chain has anisotropic interactions with randomness in the anisotropy, hysteretic effects mimicking three-dimensional order
may occur (vide infra). For ferrigmagnets (e.g., chain systems
comprised of alternating quantum ( S = 1/2) and classical
( S > 1/2) spins) the net magnetization may reverse direction as
a function of increasing temperature (ix., compensation may
occur).
As noted previously, magnets are characterized by their response to a nearby magnet, or more specifically to an applied
magnetic field, H . For ideal, noninteracting spins a net magnetic
moment or magnetization, M , is induced in the solid when exA n p t . Chem. Inr. Ed. En,$ 1994. 33, 385-415
REVIEWS
Designer Magnets
posed to an applied magnetic field, H ; where M is proportional
to H [Eq. (l)]. The proportionality constant is termed the molar
M=xH
temperature dependence of M is described by the Brillouin function [Eq. (8)]
(1)
magnetic susceptibility x, and it has a temperature dependence
that is characterized by the Curie expression [Eq. (2)l. The Curie
where
B = - 2 s + 1 coth(2S2 s + 1 x)
2s
-
coth
(&)
and
constant C in emuKmol-'. is defined according to [Eq. (3)], in
('=
NgzpiS(S
3k ,
+1)
(3)
which S is the spin quantum number (one-half of the number of
unpaired electrons per radical), N is Avogadro's number, g is the
Lande factor, p , is the Bohr Magneton, and k , is the Boltzmann
constant. Sometimes the spins experience an effective parallel
(or antiparallel) exchange field due to cooperative interactions
with the neighboring spins which increases (or decreases) the
measured susceptibility from that predicted for independent
spins by the Curie law. The high-temperature susceptibility data
often can be fit to the Curie-Weiss law [Eq. (4)l; where for
-'1
C
The Brillouin function no longer applies when there are significant ferromagnetic or antiferromagnetic interactions among
the spins. However, for T $ 0 an effective temperature given by
T,,, = T - B may be used in the argument of the Brillouin function to obtain a "fitting" of the data.r381Table 1 summarizes
typical expected values for xT, perf,and M , as a function of S.
Table 1. Representative values of %T,perr.and M , as a function of S and the Lande
,? value.
(4)
1:2
2x1 2
parallel (ferromagnetic) and antiparallel (antiferromagnetic) interactions, 0 is greater or less than zero, respectively. The value
of 0 can be determined from the x- = 0 intercept obtained in
the linear extrapolation of the plot of 1 - l versus T at high
temperature: note the value of x is temperature-dependent.
Chemists frequently report the effective moment perfor sometimes simply x T . The effective moment is defined in Equation (5).'lhaI
'
2
2
1
2
3x1/2
2
1/2 + l
2
312
2
4x112
2
2x 1
2
122 3 / 2 2
2
1;2 + 2
+
0.375
0.750
1.000
1.125
1.375
1.875
1.500
2.000
2.250
3.375
1.73
2.45
2.83
3.00
3.32
3.87
3.46
4.00
4 24
5.20
5 5x5
11 170
11 170
16755
16755
16755
22 340
22 340
22 340
27925
3
6
8
9
11
15
12
16
18
27
1
2
2
3
3
3
4
4
4
5
',
For a simple system comprising one mole of noninteracting
spins (i.e.. 0 = 0), and where S is a valid spin quantum number,
per,is temperature-independent [Eq. (6)].
The magnetization M also has a characteristic applied magnetic field dependency that enables the rapid qualitative determination of the magnetic behavior. At low temperatures and
high magnetic fields (conditions under which the magnetic energy (gSp,H)r'6b1is comparable in magnitude to the thermal energy ( k , T ) ) .the magnetization no longer obeys Equation (l), but
approaches the limiting value or saturation magnetization M ,
[Eq. (7)]. M , has the units of emuGmol-', though it sometimes
is reported in units of pB molecule-
'. F o r independent spins the
The temperature dependencies of 1 , ~ -and pefrare illustrated
for independent spins (Curie), and spins with ferro-, and antiferromagnetic interactions in Figure 4. A schematic illustration of
a number of field-dependent magnetization behaviors is shown
in Figure 5.
At sufficiently low temperature the spins may order. This
temperature is the critical temperature T,. If the spins align
parallel to each other (ferromagnet, Fig. 3 c), a macroscopic
spontaneous magnetization in zero applied field (i.e.,
M(H,,,,,,, = 0) > 0) is present, and the critical temperature in
this case is sometimes referred to as the Curie temperature T,.
If neighboring spins are aligned antiparallel (antiferromagnet,
Fig. 3 b), there is no net macroscopic moment and the susceptibility is anisotropic. Here the critical temperature is sometimes
referred to as the Nee1 temperature TN. As noted earlier, ferrimagnetism can occur when the antiferromagnetically aligned
spins have differing local moments, resulting in the incomplete
cancellation of the parallel and antiparallel spin sublattices
which leads to a reduced, but nonzero, moment (Fig. 3d). In this
case the critical temperature is also sometimes referred to as the
Nee1 temperature TN.
389
J. S. Miller. A. .I.
Enstein
REVIEWS
0
50
T/K+
100
Fig. 4. The susceptibility y (aj. the reciprocal susceptibility z - ' extrapolated from
the high-temperature region (b). and the effective moment {I.,, (c) as a function ol
temperature for independent g = 2 , S = 1/2 spins as well as ferromagnetically coiipled (Q = 10 K ) and antiferromagnetically coupled (0 = 10 K) spins. The CurieWeiss law. (Taken from ref. [9]).
~
For ferro- and ferrimagnets well below the critical temperature, the atomic or molecular magnetic moments are essentially
all aligned on a microscopic scale. Actual samples, however, are
composed of small regions termed domains, within which the
ferromagnet
spontaneous magnetization
M (H=O)> 0
M,= gJ
v -
coupling
ferrimagnet
T
M
emu G rnol
coupling
antiferromagnetic
coupling
-no "
HIG-t
diamagnetic
Pig. 5. Schcmatic illustration of the magnctlzntion .M as a function of appl~cd
magnetic field H for sc~crdltlpes of commonly ohservcd mci_pneticbehavior (from
ref. [Y]).
390
local magnetization is essentially saturated. The relative direction of the magnetic moment in adjacent domains is not parallel.
This phenomenon has its origin in the lowering of the total
energy of the magnet by transformation from a configuration of
all parallel moments (resulting in a high energy stored in the
magnetic field generated by these moments) to an array of domains that point in various directions (resulting in little energy
stored in magnetic fields but at an energy cost of forming "domain walls" between the domains). The coercive field H , is the
reverse magnetic field necessary to reduce the magnetization of
a sample to zero starting in a saturation condition magnetization. "Hard" magnets have large values of H , ( > 100 Oe),
whereas "soft" magnets have low, sometimes unobservable
values of H , ( < 10 Oe). Large values (hundreds of Gauss) of 11,
are necessary for magnetic storage of data. while low values
(milliGauss) are necessary for ac motors. An example of a
molecular-species-based soft ferromagnet is the complex
[Fe"'(S2CNEt2)2]CI,which has been characterized by magnetic
susceptibility, heat capacity, and 57FeMossbauer spectroscopy.
It exhibits a T, of 2.46 K, but does not exhibit hysteretic effects.["] Magnetic materials that are subdomain in size exhibit
superparamagnetic (high-spin) behavior. for example. naturally
occurring ferritin.[18"]The magnetization and coercivity are i n portant parameters i n ascertaining the commercial utility of a
magnet. Table 2 summarizes values for some of the more conimon magnetic materials and new molecular-species-based m a g nets.
It is interesting to ponder the limiting size of a molecule beyond which its spins will no longer remain aligned within a
single magnetic domain. Calculations of the size of a single
magnetic domain are very complex due to dependencies on the
geometry, magnetic anisotropy etc.,L18h.d1
and to the best of our
knowledge this question has not been experimentally or theoretically addressed for a molecular-species-based system. As a preliminary model one may consider typical spin magnitudes in a
known single-domain magnetic material. The single domain of
magnetite, Fe,O,. which is present in magnetotactic bacteria,""' has a volume of about 1 x 10' A3.[191and thus contains
1.2 x 10' Fe,O, units with four net spins each or about 5 x 10"
total spins (ix., 0.05 spin per A3). For iron, if one assumes an
effective critical domain radius of 75 ,& for a spherical singlc
domain.["'] each single-domain particle has 1.5 x 10' iron
atoms and about 3.3 x 10" spins (0.2 spin per A3).Likewise. for
Nd2Fe B. if one assumes a conservative single-domain radius
It
of 150 A.['*'] 2 x 10" spins (0.1 spin per A,) are present per
domain. Assuming that the single domain contains at least 106
spins. it is feasible that a molecular-species-based material, containing a single spin per repeat unit with an equivalent weight of
100 daltons, and with about 10" repeat units (thus a molecular
weight of lo8 daltons), might also be a single domain. The
spatial dimensions of such a domain would be quite large and its
spin density would be relatively Ion. Note that the domain size
also depends strongly on the shape of the particle; for example.
altering the shape of an iron particle from a sphere to a prolate
ellipsoid with an aspect ratio of ten increases the domain size
Thus. materials with a
from 1.5 x lo5 to 9.6 x 10' Fe
high aspect ratio such as linear and minimally branched polymers may have a substantially greater number of spins and
larger dimensions and still remain single magnetic domains.
REVIEWS
Designer Magnets
orientation at neighboring sites in the ordered state may be
canted with respect to each other.
Metamagnetism is a magnetic-field-dependent transformation from an antiferromagnetic state to a high-moment ferromagnetic state; that is. the spin alignment depicted in Figure 3 b
is transformed to that depicted in Figure 3 c by application of an
applied magnetic field. Decamethylferrocenium 7.7.8,g-tetracyano-p-quinodimethanide,[Fe(C,Me,),] [TCNQ]. is a molecular-species-based metamagnet (see Section 5 ) . A spin glass occurs when there are local spatial correlations in the directions of
neighboring spins, but no long-range order. For a spin glass the
spin alignment is as described for a paramagnet (Fig. 3 a). However, unlike a paramagnet where the directions of the spins vary
If the relatively large single-domain molecular species remain
far apart from each other so that the interaction among these
molecules is very weak. the temperature and magnetic field dependence of the measured magnetization would reflect the behavior of sinple-domain particles. In this case, the magnetization of the single-domain molecule follows the direction of the
applied magnetic field by rotation of the magnetic moment. If
an uniaxial anisotropy is present, there will be a barrier to the
rotation of the magnetic moment in response to the changes in
the magnetic field. The "supermagnetic" behavior of single-domain magnets differs in time dynamics from that of the reversal
of magnetization within a multimagnetic domain particle due to
domain wall motion."0.
Table 2. Selected magnetac properties of representative magnetic materials.
Materid
fl IK1
1101
1408
650
673
92
3.6
23
19
30
3
35
22.2
11.6
10.5
22.6
27.8
3.7
61
1
0.7
> 350
Mag.M
netism [a]
Imol55.8
5X.Y
58 7
232
445
1081
84
365
409
388
567
465
417
269
385
403
654
668
370
617
454
531
502
527
527
530
453
552
896
980
278
483
350
FO
FO
FO
FI
FO
FO
FO
FO
FO
FO
CW
FI
FI
FO
FI
FI
FI
tI
CW
FO
FO
MM
FO
FO
FO
FO
FO
MM
MM
FO
FO
MM
FI
~
T
[KI
Density
'1
Saturation Magnetization [emuCr]
I ~ c r n - ' ] [mol-'1
7.86
8.89
8.91
5.1
8.59
7.6
4.89
2.91
3.26
1 .5
1.63
1.8
1.8
1.65
2
2.02
1.61
1 61
1.57
1.49
1.31
1.26
1.31
1.21
1.25
1.26
I .31
1.39
1043
1394
627
858
993
585
387
52.4
55
2.46
8.3
14
4.6
13
30
4.6
8.1
7.6
I5
14
4.x
2.55
3.65
3.1
6.2
x.x
8.5
2.8
18
0.6
0.38
> 350
~
1.33
1.41
1.12
~
~
[M-']
[g-']
[cm-"]
12175
9 500
3 200
21 300
66050
1x1 100
8 800
22700
22 500
14 500
12 000
22 300
22300
6400
22 300
22 300
22 000
22 000
3 900
27 590
16 300
18 000
17 700
12175
9 500
3 200
7100
I t 000
11 300
8 800
22 700
22500
14500
12000
11 150
11 150
6400
11 150
1 1 150
22 000
22 000
3 900
13795
16300
18 000
17700
218
161
54.5
91.8
148
168
55.4
33.6
32.9
10.5
44.7
35.9
33.9
35.2
1715
1434
486
468
1215
1273
512
181
179
56.1
34.4
86.3
96.3
39.3
116
112
54.1
53
16.5
66.6
47
42.7
46.1
22 500
13 500
20000
24 200
1 2 500
30000
2 800
11 125
6 000
22 500
13 500
20000
24 200
6250
30000
2 800
11125
6 000
42.7
25.5
44.2
43.8
53.4
32.1
57.8
60.9
30.6
10.1
23
17.1
40.7
14.2
25.8
105
62.2
55
37.4
21.1
48
53.5
Z3.X
58
CoerX-ray
civity [h]
H , [GI (T[K]I lbl
1
SC
sc
-7
213
56000
25 000
650
SC
SC
P
none
sc
sc
SC
SCn
Pn
60 (4.7)
I0 ( 2 )
80 (4.2)
60 ( 4 . 2
50(1.31
SC
320 (1.21
sc
< 5 (1.2)
SC
6200 (4.2)
160 (5)
1000 (2)
H,, 1.6 kG
none
none
none
3600
1200 (2)
H,. 800 G
375 (5)
1: (0.44)
H,, 100 G
60 (300)
SC
SC
UC
UC
SC
UC
P
SC
SC
P
D
~
ferrimagnet, MM = metamagnet. C;W = canted!weak ferromagnet. [b] D = disordered, P = powder diffraction. Pn = neutron powder diffraction. S(' = single crystal. SCn = neutron single crystal. UC = unit cell. H,, = metamagnet critical field. [c] Abbreviations H,Pc = phthalocyanine.
[a] F O = ferrom;ignetic. FI
=
Many other magnetic phenomena are possible. Canted ferromagnetism. metamagnetism, and spin-glass behaviors are examples. A canted ferromagnet can result from the relative tilting
of ferromagnetically coupled chains such that, though the material IS ferromagnetic, there is a reduction of the moment (see
Fig. 3e). Sometimes a canted ferromagnet is referred to as a
weak ferromagnet. A molecular-species-based example is
phthalocyaninatomanganese(I1) (MnPc). This complex has a T,
of 8.3 K . but a saturation magnetization corresponding to only
72 % of the fully aligned spins due to the canting arising from
the herringbone structural
In other cases, the spin
with time, the spins remain fixed in their orientations for a spin
glass. A correlated spin glass has spins pointing in similar directions for short distances, but no long-range order. [V(TCNE),] . y(MeCN) at low temperature is an example of a correlated spin glass (see Section 9).
Like the familiar gas/liquid critical behavior, critical behavior
occurs at the onset of cooperative magnetic ordering near the
critical temperature T,. The critical exponents can be compared
to theoretical expectations to elucidate the dimensionality and
anisotropy of the spin interactions which dominate the formation of the ordered spin state.
391
J. S. Miller, A. J. Epstein
REVIEWS
3. Origin of Cooperative Magnetic Behavior in
Molecular-Based Magnetic Materials
Creating a magnetic molecular material demands the alignment of the spins in all three directions of the solid material.[221
The “exchange interaction” J describes the type (ferro- or antiferromagnetic) and degree of coupling between pairs of spins
and is to first order a two-body interaction. It i s thus essential
to have magnetic exchange interactions in all three directions
and any de novo design of new materials must include an exchange interaction between neighboring spins in all directions.[221In order to accomplish this, it is, of course, necessary
to examine the sign and magnitude of the exchange between
pairs of molecular-species-based spins.
The magnetic exchange interaction is a direct consequence of
quantum mechanics. The total electronic wavefunction can be
factored into spatial and spin terms. Because electrons are
fermions, the total electronic wavefunction must be antisymmetric. The spin (magnetic) exchange has its origin in the requirement that the total electronic wavefunction must change
sign under the interchange of two electrons. For example, consider bringing two H atoms together to form H,. If the two
electron spins are antiparallel (71: S , S , = 0), the real-space
wavefunction is symmetric or bonding. On the other hand if the
two electron spins are parallel (TT: S, S , = l), the real-space
wavefunction is antisymmetric or antibonding. The difference in
energy of the symmetric and antisymmetric wavefunctions is
related to the exchange energy. The lower energy of the realspace bonding wavefunction for H, results in an antiferromagnetic exchange. since the pair of electron spins are antiparallel.
The close approach of two spin-containing molecular species
( A and B) leads to a similar interaction. The new total wavefunction is often calculated in a configuration interaction (CI) (i.e.,
perturbation theory) approach. That is, the total wavefunction
for the combined pair of molecules is approximated by admixing the ground state with an excited state arising from the promotion of an electron from the partly filled (spin-containing)
orbital of A to a virtual (empty) orbital of B as well as the
opposite case for partially occupied orbitals of B and virtual
orbitals of A . This leads to a new set of energies for each spinsymmetry state. The spin state of the lowest energy configuration determines whether the pair of molecular species are coupled antiferromagnetically (antiparallel spins on the adjacent
molecules) or ferromagnetically (parallel spins on the adjacent
molecules). The difference in energy between this lowest energy
configuration and the lowest energy configuration of the opposite spin state is related to the exchange energy.[23.241 The relative energies of the differing configurations depend strongly
upon the charge transfer integrals between the selected orbitals
of A and B (i.e., i and j . respectively) ([j&), and also upon the
difference in energy between the ground state of isolated A and
B a n d the various excited state configurations (AEYB).The magnitude of each contribution to the total pairwise-exchange interaction is proportional to (PYB)2/AEYB;
the sign is determined by
the total spin of the excited state.[25.261
(The formation of this
“incipient” chemical bond is sometimes termed “kinetic” exchange.) If the charge transfer integral between A and 5 is zero
(i.e., orthogonal orbitals), the antisymmetrized real-space total
wavefunction and symmetrized spin-space wavefunction (high-
+
+
392
spin state) is lower in energy than the symmetrized real-space
total wavefunction and antisymmetrized spin-space total (lowspin state), that is, Hund’s rule applies. (Of course, if A and 5
approach close enough to form a covalent bond. the perturbation approach of CI must be replaced by calculating the full
wavefunction of the new molecule derived from A and B.)
It is not possible to calculate accurately the total energies and
transfer integrals for pairs of molecular species each containing
hundreds of electrons. Therefore a number of simplifying special cases of the configuration interaction concept have been
emphasized by several research groups as a practical guide to
enable the synthetic design of molecular-species-based magnets.
These simplifying approximations differ in complexity and in
their usefulness to the synthetic chemist. The simplest intramolecular approach is to consider molecular species with
spins in orthogonal orbitals. The next simplest would involve
spin polarization through bridging groups with filled orbitals
connecting two spin-contdining species. The next level of interaction involves the partial transfer of electrons between different molecular species. The simplest intermolecular approach
utilizes the CI of the spin-containing molecular species, each
considered as a whole unit, and where only interactions between
the POMO of one molecular species and the POMO of its neighbor are examined. Other still more complex cases include the CI
between the next highest HOMO (NHOMO) and/or the next
lowest LUMO (NLUMO) of a pair of inolecular species. An
even more complex variant is to consider the excitations of
occupied A orbitals to virtual orbitals on A as well as to virtual
orbitals of B. Again, it is important to emphasize that the CI
models give two-body (pairwise coupling) terms. This is insufficient to account for three-dimensional magnetic behavior, for
which configuration interactions in all three directions are crucial.
Increasing the complexity of CI calculations more accurately
describes the system, however, it makes the model less useful to
the practicing synthetic chemist. As a practical guide to the
synthetic chemist we have classified the three distinct mechanisms for spin coupling (Scheme 1).
1. Ferromagnetic exchange resulting from orthogonal orbitals (no
C1)-Hund’s rule (intramolecular spin coupling)
2. Configuration interaction (CI) to determine dominant ferro- or antiferromagnetic exchange
2.1. Intramolecular species excitations (spin polarization)
2.2. Intermolecular species excitations
2.2.1. Involving only the POMOs on adjacent molecular species
2.2.1 . I , Without spatial constraints
2.2.1.2. With spatial constraints
2.2.1. Involving NHOMOs and/or NLUMOs on adjacent molccular species
2.2.2.1. Intermolecular species excitations
2.2.2.2. Intramolecular species excitations
2.2.3. Excitations between localized and delocalized (metallic)
states (intramolecular spin coupling)
3 . Dipole-dipole (through-space) exchange (intermolecular spin coupling)
Scheme 1 Classification of three mechanisms of spin coupling.
Often in discussions of the exchange mechanisms emphasis is
given to the chemical stabilization of a spin state capable of
exchange. This leads to identification of exchange mechanisms
by their chemical or physical nomenclature. An example of this
REVIEWS
Designer Magnets
is ferromagnetic exchange between polarons created in conjugated segments of block copolymers. In this case, polarons are
the means of creating spin-containing units. The exchange between these polarons comes from CI through the intervening
moieties (vide infra). The above is a classification of operative
spin-exchange mechanisms, not the means to achieve spins.
Frequently it is unclear which mechanism is the primary exchange mechanism in a particular chemical system. Particularly
for molecular-species-based systems with complex compositions
and structures, it is probable that more than one mechanism
plays a significant role (e.g., ferromagnetic coupling of spins
within a one-dimensional chain due to orthogonal orbitals
(mechanism 1) and ferromagnetic o r antiferromagnetic coupling between chains due to configuration interaction (mechanism 2) or possibly dipole-dipole interactions between chains
(mechanism 3)). Furthermore, since there are many levels of
complexity of the CI model (mechanism 2), there are many ways
to describe this exchange mechanism and properly truncated
exhaustive calculations are required to provide a detailed understanding.
Examples of these models elaborated in terms of their chemical embodiments are described below. Note that the suggested
mechanisms may not stand the test of time, but evolve with
increased understanding of the structures and properties. Each
of the mechanisms given in Scheme 1 is discussed below.
These materials are not normally attributed to molecular-species-based chemistry and thus might be considered by some to
be inappropriate to be discussed in this review. Nonetheless,
ferromagnetic materials can be also prepared by substitution of
the alkali metal cation with a quaternary ammonium ion,
[RNH,]' (R = Me, Et. PhCH, etc.). These fascinating ferromagnets have relatively high values ( > 50 K) which are essentially independent of interlayer separation and are reasonably
transparent.[61 These layered compounds, in principle, lend
themselves to form Langmuir-Blodgett-like monolayer films
which would be suitable to test concepts of two-dimensional
magnetic behavior.
The general idea of designing a material with spins residing in
orthogonal orbitals on adjacent sites has also been pursued by
0. Kahn et al. as well as others; however, only intriguing highspin small molecules have been reported.[271Recently extendednetwork solids of [Bu,N][MCr(ox),] (ox = oxalato; M = Fe,
Co, Ni, Mn, Cu) have been reported to exhibit ferromagnetic
behavior.['']
It is important to reemphasize that highly magnetic behavior
is not a property of a molecular species; it is a cooperative
solid-state (bulk) property. Thus, to achieve bulk ferromagnetic
behavior in the context of the aforementioned high-spin molecular species, intermolecular interactions, perhaps as described
below. must be operative or the molecular species need to be
sufficiently large to be considered a magnetic domain (vide
supra).
3.1. Intramolecular Ferromagnetic Exchange Resulting
from Orthogonal Orbitals (No CIFHund's Rule
The essence of achieving high-spin behavior by using this
approach is the presence of unpaired electrons in orthogonal
orbitals (zero quantum mechanical overlap), which, however,
reside in the same spatial region. The greater the latter condition
is fulfilled, the stronger the ferromagnetic exchange. In this case
the antibonding real-space total electron wavefunction is lower
in energy than the bonding real-space total electron wavefunction so that the high-spin state is stabilized. This has been
demonstrated for orthogonal orbitals on the same site, for example. carbene, : CH, ( S = 1) and Mn" ( S = 5 / 2 ) , and multiple
sites in the same spatial region, for example, trimethylenemethane. :C(CH,), ( S = I ) and 0, ( S = l ) . This essence of
Hund's rule is schematically illustrated in Figure 6.
A
A
orbital overlap
spin density
Fig. 6. Schematic illustration of the relative energy, orbital overlap, and spin density
for orthogonal POMOs on A .
The concept of having a large number of unpaired electrons
residing in orthogonal orbitals that are ferromagnetically coupled by CI through-bond coupling or superexchange (vide infra)
through a spinless bridging halide has been exploited by P. Day
et al. in their studies of Rb,CrCI, and related materials which
have an extended-network structure in two dimensions.""]
3.2. Configuration Interaction to Determine Dominant
Ferro- or Antiferromagnetic Exchange
The orthogonal-orbital mechanism (see Section 3.1) provides
a means to achieve a high-spin ferromagnetically coupled
molecular species in which the spins reside in the same spatial
region. In order to magnetically couple spins residing in distant
spatial regions within a molecular species, the total wavefunction of the system must be considered, and it is typically evaluated by configurational interactions. Additionally, the CI can be
applied between molecular species within a solid to achieve the
ferromagnetic coupling between these species needed to induce
bulk ferromagnetic behavior. These are complex, time-consuming computations that are often simplified by limiting the number of CIS considered. The details of the CI, of course, depend
on how the problem is formulated, that is, what is given for the
ground state and excited state wavefunctions. The simplest approximation employs only the POMOs on adjacent molecular
species.
3.2. I . Intramolecular Excitations (Spin Polarization)
As discussed above, an unpaired electron spin in a POMO can
polarize the paired electron spins in an orthogonal bond connecting it to another moiety with a POMO. This polarization is
illustrated by the coupling between a carbon n electron and a
proton nucleus through the hyperfine interactions. Although a
valence bond-like picture is usually utilized (see Fig. 2 top), a
full molecular orbital calculations reveals that the proper description of the spin polarization requires the admixture (config393
.I.
S. Miller, A. J. Epstein
REVIEWS
uration interaction) of a small amount of an antibonding excited
state into the ground state to obtain the spin polarization.l'zal
Hence the spin polarization mechanism is an intramolecular
configuration interaction mechanism.
Spin polarization is an intramolecular effect that is used
extensively to interpret EPR hyperfine coupling constants.[I2.3 0 , 3 ' 1 Hyperfine splittings are observed for unpaired
electrons in proximity to ring hydrogens in aromatic compounds, and although unpaired spin density is not delocalized
directly onto the hydrogen atom, unpaired spin density is induced at the hydrogen nucleus and observed (through the hyperfine splitting) by spin polarization (i.e.. an unpaired electron
in a C-based p orbital polarizes the paired spins in an orthogonal C - H cr bond so that one of the paired electrons is more in
the vicinity of the C atom than another, even though there is no
unpaired spin density delocalized onto it; Fig. 2 top).[L2c.dl
With the knowledge of the spin-polarization mechanism and
that the ground states of dioxygen and some carbenes were
high-spin triplets due to spins residing in orthogonal orbitals, K .
Itoh and N . Mataga in 1968 suggested[2y1that large planar
alternate hydrocarbons comprised of m,fu-substituted triplet
diphenylcarbene moieties would have a high-spin (ferromagnetically coupled) ground state in accord with Hund's rule and spin
polarization. Proposed examples of high-spin multiplicity polymers[29bIinclude 1.
CI approach (tight-binding calculation of energy bands) that
includes the effects of on-site Coulomb repulsions (Hubbard
model) to describe the antiferromagnetic coupling between adjacent n electrons within the C, ring and the bridging carbene
moieties, resulting in a net ferromagnetic coupling. Nasu accounts for the ferromagnetic coupling between the nonbonding
and n electrons of the carbene by using a periodic Kondo model
(i.e.. Hund's rule for orthogonal
K. Tanaka et al.
similarly calculated spin polarization by which long-range magnetic coupling is possible in poly(pheny1nitrenes) using a tightbinding calculation of energy bands (CI).134b1
The efficacy of the
polyene bridges in pairs of spins has been recently demonstrated
by using bipyridylpolyene-bridged bimetallic
2a
*P
3
1
H. Iwamura et al.[321have prepared several high-spin polycarbene radicals including 2 a (S = 4). The largest compounds
characterized. albeit in situ. are the hexacarbene 3 (S = 6)[32h1
and a nonacarbene (S = 9).[32c1The magnetization of 3 confirms the S = 6 ground state, however. the temperature dependence of the susceptibility demonstrates only antiferromagnetic
coupling between the S = 6 radicals.[32b1The high-spin ( S = 4)
ground state in 2 has been attributed to four S = 1 carbene sites
(arising from spins in orthogonal orbitals in the same spatial
region, Fig. 6 right) (vide supra) coupled by spin polarization as
illustrated in 2a with arrows denoting the direction of the polarized spins.[331(Note that for ortho- or puru-substituted analogues spin polarization results in antiferromagnetic coupling
and diamagnetic behavior.) The spin polarization in an idealized
poly(tn-diphenylcarbene) was evaluated by K. Nasu by using a
394
q.
A. A. O v ~ h i n n i k o v [ has
~ ~ ] developed a simple approach to
account for the CI between spins on arrays of adjacent carbon
atoms each with a single p, electron. He noted that the net S for
planar alternate hydrocarbons (e.g.. 2-6) can be calculated by
Equation (9). where 12 is the number of nonstarred atoms and n*
is the number of starred atoms (adjacent atoms are alternately
starred (*) and nonstarred and identically denoted atoms are
not adjacent to each other). For example, trimethylenemethane
(4) has an S = 1 ground state. and larger molecules could have
large values of S. Polymers such as 5 are predicted to have a
,
s is the number of repeat units
large value for S (S = . ~ / 2where
within the polymer).[35.361 Chemical reactivity arising from the
strong tendency of unpaired electrons to form covalent bonds
suggests that stable high-spin materials are difficult to prepare
following this methodology. Bond distortions as well as inter-
REVIEWS
Designer Magnets
r
1
L
4 S = l
5 s=x/2
molecular coupling which pair-up electrons must also be avoided.
High-spin radicals based upon several triarylmethyl radical
sites are under investigation by several research groups worldwide. An S = 5 radical, 6, has been prepared.[37iPerchloropoly-
7
8
9
netically coupled polar on^.[^^"^ Conducting oligomers based
upon poly(hms-acetylene), poly(pyrro1e) or similar polymers
were identified as candidates to test this approach. Upon, for
example, p-"doping", a cation radical (positive polaron) segment is formed, which may be schematically represented as
shown in Scheme 2. Ferromagnetic coupling of these radicals is
%tBu
p - "doping"
tBuw
6
tBu
triarylmethyl radicals have also been studied ; the largest example characterized is 7 (S = 3/2), which has been isolated in
two stereoisomeric forms.[38' Polymers based on triarylmethyl
radicals have been reported to exhibit saturation magnetizations ofthe order of 1 % of bulk.''. 391 These materials are poorly characterized.
Another example of a high-spin moiety sought as a building
block for an s:p orbital-based magnet is 1,3-cyclobutanediyl (S),
which has spins residing in different regions within the radical
and. like trimethylenemethane (4)has a strong triplet preference.[""]
Equation (9) is not sacrosanct and may fail for the prediction
of S = 0 compounds; thus it may not reliably guide the synthetic
chemist toward high-spin molecules. This is clear as tetramethyleneethane (9) is an unsolved puzzle. Equation (9) as
well as more sophisticated molecular orbital calculations lead to
the expectation that 9 has a singlet ground state; however, a
triplet ground state has been reported for 9 in its twisted as well
as its planar analogues.[411Thus, Equation (9) appears valid for
n*
17, but may not properly describe a system for n = n*. if
they are non-Kekule o r c r o s s - c o n j ~ g a t e d . [ ~ ~ ~ ~
H. Fukutome et al. suggested that block copolymers comprised of conjugated conducting oligomers and ferromagnetic
spin-coupling segments (for configuration interaction. e.g.,
rnetci-substituted benzene unit) will upon doping have ferromag-
+
Scheme 2
p r ~ p o s e d [ ~ " . to
~ ' ]be stabilized by a wzrtd-substituted diphenylcarbene moiety such as 10 (vide supra).
Synthetic chemists have just
started to prepare suitable examples of these block copoly10
mers to test this model. D. A.
Dougherty et al. reported[43di
that partial oxidation of 11 with either I, or AsF, leads to a
material for which the magnetization is best described by the
presence of spins with an S > 2 behavior, not an independent
spin (S = 1/2) behavior; however, antiferromagnetic coupling
among these spins and the low density of spins need to be circurn~ented.["~"~
.
3.2.2. Intermolecular Species Excitations
Configuration interaction can be applied between molecular
species within a solid to provide a means for magnetic coupling
among these species that is needed to achieve bulk magnetic
behavior. In the simplest case this can be done by considering
only the POMOs. It can also be done by considering the next
highest occupied (NHOMO) and/or next lowest unoccupied
molecular orbitals W L U M O ) or between a localized state and
a delocalized (metallic) state.
395
J. S. Miller, A. J. Epstein
REVIEWS
3.2.2.f. Involving the POMOs on A@icent Molecular Species
Dominant ferromagnetic or antiferromagnetic exchange determined by configuration interaction (CI) between POMOs on
adjacent molecular species can be enhanced by the spatial relationship of the atoms on adjacent molecules and their pairwise
contribution to the total exchange (1.e.. Heisenberg exchange).
model requires that at least A or Bpossesses a degenerate (accidental or intrinsic) orbital that is not half-filled, empty, or filled.
The magnetic properties of several electron-transfer salts based
on metallocenes and cyanohydrocarbons have been discussed
within this mode1.[8'~y~44bl
The configurational interaction
model for the degenerate (d, t ...) and nondegenerate (s) cases
(see Table 3 for abbreviations) can be represented by using
'I
forms for the nearest neighbor Hubbard
3.2.2.1.I. Without Spatial Constraints
The application of configuration mixing of a virtual triplet
charge-transfer excited state of a whole molecular species with
the ground state. in order to stabilize ferromagnetic exchange
.
coupling (kinetic exchange) for . . [A]'+ [B]'+ [A]' [B]'( A = donor. B = acceptor) chains. was introduced by H. M.
M ~ C o n n e l l . [ ~The
~ " ' large separations and energy differences
between ions on adjacent sites lead to small overlaps of the
frontier orbitals ("incipient" bonding) and prevent the formation of metallic energy bands for mixed-stack electron-transfer
salts.12h". 4 5 1 In contrast, the spins of the [A]'+[4'- repeat units
with half-occupied nondegenerate POMOs can couple antiferromagnetically (Fig. 7a). Admixture of the higher energy
charge-transfer states (Fig. 7 a: right) with the ground state
(Fig. 7a, left) lowers the total electronic energy and stabilizes
antiferromagnetic coupling ( T i ) which, if the number of spins on
the two moieties are different. leads to ferrimagnetic behavior.
Table 3. Representative magnetic couplings for homo- and heterospin systems with
sineh. doublv. and trinlv deeenerate POMO? Ial.
-x
B t A
.4 t B
12
AF
AF
FO
AF
FO
AF
FO
AF
FO
AF
FO
FO
FI
FI
t'
FO
d2
FO
FI
FO
FI
AF
FO
AF
AF
FO
AF
FO
AF
FO
AF
AF
FO
FI
FO
FI
FI
.4
I
a
(or
B)
s =1:2
s'
d'
d'
d'
t'
I'
d3
t'
S=l
S =1:2:l
S=123:2
t'
d'
d'
t'
d'
I'
1.3
t3
B
(or
A)
FI
Example
[C,,,]
'
~
'?
jMn"'Cp;]
[Cr"'Cp;]
' [TCNE]''+
[TCNE]'-, [V(TCNE),]
FI
FO
++ + - + + -I+
A
A
A
B
B
[a] A F = antiferromagnetic coupling; FI = ferrimagnetic coupling; FO = ferromagnetic coupling. Intriiisic or accidental orbital degeneracies: s = singly (a or b),
d = doubly ( e ) ,t = triply (t).
B
A
B
Fig. 7 . Schematic illustration of the stabilization of ferro- or antiferromagnetic
coupling by configurational interaction (CI) only involving POMOs on adjacent
sites ( A and B). If both A and B have a half-filled nondegenerate POMO (s'), the
B t A (or A t B) virtual charge-transfer excited state stabilizes antirerromagnetic
coupling (a). If B (or A ) has a nonhalf-filled degenerate POMO (e.g., d', assumed
here to be A ) , the B - A charge-transfer excited state formed (b) will stabilize
ferromagnetic coupling (or A t B charge-transfer excited states formed through
excitation of a "spin-up" B electron will stabilize antiferromagnetic coupling).
This energy reduction and spin delocalization does not occur
when the two electron spins are parallel (TT; ferromagnetically
aligned) in accord with the Pauli exclusion principle. Thus, antiferromagnetic coupling is stabilized. Antiferromagnetic coupling (0 = - 18 K) has been reported for the salt [TTF]'+
[Ni{S,C,(CF,),),]'(TTF = t e t r a t h i a f ~ v a l e n e ) ~which
~ ~ " ~ has
the . . . [A]'+[B]'-[A]*+[B]'- . . . structure in all three direct i o n ~ . '(Note,
~ ~ ~ ]however, ferromagnetic coupling (0 = 16 K )
has been reported (vide infra) for the isoelectronic
[TTF]" [Pt{S,C,(CF,),),]'which has a similar
structure, but is not i s o m o r p h o ~ s . [ ~ ~ ~ ~ )
For other electron configurations involving partially occupied degenerate orbitals pairwise ( A B ) ferromagnetic coupling
may be achieved (see Fig. 7 b). Magnetic couplings predicted by
this model for representative electronic configurations are summarized in Table 3. To achieve ferromagnetic coupling with this
+
396
Configurational mixing of a virtual singlet excited state with
the ground state for [A]'[B]- systems. which stabilizes antiferromagnetic coupling when the spin on the donor and that on the
acceptor are unequal and produces ferrimagnetic order,r44b]has
been applied['j, 47 to the high-temperature molecular-species-based magnet [V(TCNE),] . ,r (solvent) ( t 3 / s ' ) (vide infra).
Application of the model to [Cr"'Cp;]:'+ [TCNQ]'- or
[Cr"'Cp;]"+ [TCNE]'- (t3/s') leads to the expectation of antiferromagnetic coupling leading to ferrimagnetic behavior; however, ferromagnetic order is observed (vide infra). Other subsets
of this model have been the specific research focus of several
groups.[521
Key to achieving pairwise bulk magnetic ordering is the extension of the model beyond pairwise ( A B ) interactions in onedimension. Figure 8 illustrates how the energy of the ferromagnetically coupled system depicted in Figure 7 b is lowered upon
the admixture of the ground state of energy E,, with one or two
intrachain (a and a, b, respectively) virtual charge-transfer excited states of energy E,, and additionally with a second interchain
excited state (a, b, c).lSg 91
3.2.2.1.2. With Spatial Constraints
In 1963 McConnell suggested a mechanism for ferromagnetic
exchange that involved a spatial arrangement of neighboring
A i ? p w . Chcm. Inr. Ed. Engl. 1994, 33. 385-415
REVIEWS
Designer Magnets
I,
\
-
Fig 8. Schematic state diagram depicting the relatike energy of the ground Es. and
excited states Ec- before and after admixing (virtual [A]" [B]'- charge transfer) with
a single excited \tale a), a n intrachain pair of excited states a) and b). and wlth a
third interchain excited state aristng from admlxture with an out-of-registry acceptor to stahilkre ferromugnetic coupling (from ref. [X g]).
netic exchange in three dimensions is necessary for bulk ferromagnetic behavior.
Cleverly designed stable radicals were sought to test this model. The spin multiplicity of dicarbenes incorporated into rigid
[2.2]cyclophanes have been shown to be controlled by the overlapping of the spin-containing orbitals in the aromatic C ,
rings.[s41McConnell's m0de1"~1 suggests that the pseudo-meta
isomer 1 2 b should possess a singlet ground
state. This is in contrast
to the behavior of the
12aS=2
pseudo-ortho 12 a and
pseudo-para 12c iso0
mers of the bis(pheny1me-
thylenyl)[2.2]cyclophanes,
radicals, which possessed both positive and negative spin densities.[s31He claimed that radicals with "... large positive and
negative atomic x-spin densities ... [that] pancake ... so that
atoms of positive spin density are exchanged coupled ... to atoms
of negative spin density in neighboring molecules ... gives a ferromagnetic exchange interaction." This is illustrated in Figure 9
for a pair of identical ally1 radicals, A and A' which have positive
spin densities on the terminal carbon atoms and negative spin
density on the central carbon atom.['2b1F o r large separations
between A and A' spin interactions d o not occur (Fig. 9a). As
A and A' approach each other in such a way that regions of
positive and negative atomic x-spin densities can "pancake", net
ferromagnetic exchange resulting from the incomplete cancellation of antiferromagnetic coupled spin components arising from
pairwise (CI-like) interactions of atomic components of the radicals increasingly occurs (Figs. 9 b, c),To achieve ferromagnetic
exchange by this mechanism. routes to spin pairing, for example
bond formation, arising from the close approach of the A and
A', (see Fig. 9 d ) must be avoided. (Note that the molecule in
Fig. 9 d may be a triplet.) Like the mechanism 2.2.1.1 in
Scheme 1. this is a mechanism for ferromagnetic exchange between adjacent radicals, not bulk ferromagnetism. Ferromag-
A
A
A'
A
A'
A
A'
which both have a quin12b S = O
tet ground state. ESR
data on this system are
consistent with these results, suggesting that ferromagnetic exchange can
be achieved by this
McConnell
model.["J
Note that the ground
state spin multiplicity of
these isomers has been
described by CI by K . Yamaguchi et al.[ss"lTanaka et al. used
CI incorporating the spatial distribution of the spin to calculate
the favorable orientations for ferromagnetic exchange between
diphenylcarbenes as a function of relative rotation and tilt
angles of the molecules.[ssb1Achievement of bulk ferromagnetic
behavior in these materials has yet to be reported.
Similar to this McConnell approach using molecules with
registered spin distributions. the antiferromagnetic coupling of
alternating spin sites with different metal atoms with a different
number of spins per site leading to ferrimagnetic behavior has
been promoted by 0. Kahn.[sb]Likewise. D. Gatteschi and P.
ReyI8"l have focused on the same paradigm by utilizing extended network structures comprised of metal ions and nitroxide
radicals. Due to configuration interaction, adjacent spin sites
couple antiferromagnetically ( t i ) , whereas next-adjacent spin
sites couple ferromagnetically (TIT). Thus. through alternating
sites with a larger number of spins (e.g., 5 for ( S = 512) Mn")
with sites with fewer spins (e.g.. 1 for (S = 1/2) Cu" or nitroxide
ligands) (i.e., Tl), the Ss cannot cancel, thereby leading to a
ferrimagnetic system (see Fig. 3d). Kahn et al. have reported
several examples of highly magnetic systems developed by this
strategy using Mn"/Cu"; for example, [Cu"Mn" (obbz)] . H20[56a1 (13) (obbz = N,N'-bis(2-carboxyphenyl)oxamidato) has a T, of 14 K. This group also reported that the
A'
Fig. 9. Schematic illustration using the spin density of a pair of ally1 radlcak ( A and
A ' ) for the stahilirntion of ferromagnetic coupling by configurational interaction
(CI)involvine spin-exchange Nhich results from regions ofpositive spin density (not
of the entire POMOs) oE A interacting with regions of negative spin density o n A'
as these exemplary S = I ; ? radicals as they are brought closer together. Minimal
interaction5 occur at large separations (a). which increase as the separation decreases (b) and (c). until bonding occurs at short separations (d).
391
J. S. Miller, A. J. Epstein
REVIEWS
structurally characterized
[Mn"Cu"(pbaOH)] . 3 H,O (14)
(pbaOH = 2-hydroxy-l,3-propanediylbis(oxamato))has a of
4.6 K,15hh1and upon dehydration and concomitant structural
alteration to [Mn"Cu"(pbaOH)] . 2 H,O the
is increased to
30 K,[56'1
r,
The Gatteschi and Rey collaboration has focused on Mn",'nitroxide systems, such as the ferrimagnetic complex 15 with a T,
of 8.1 K.I5?"] They have also
F3C.,
reported on a nonstructurally
5 = + 0 5 F 3
characterized [ (Mn(O,CC,F,)Z~2 NITEt] (NITEt = ethyl nitronyl
F3C
0
CF3
nitroxide) compound with a 015
value of about 2 5 K and a
T, of 20.5 K.1s7h1The rare-earthN
containing analogue [{Dy"'(hfac),NITEtJ]
(hfac = hexafluoroacetylacetonate), in contrast, exhibits antiferromagnetic
ordering at 4.4 K ( H = - 6 . 1 K).[57C1
For 13-15 the interchain interactions may well involve the
spin-exchange mechanism illustrated in Scheme 3 ; here each
horizontal line depicts a chain of alternating S = 512
and antiferromagnetically
coupled S = 1 ;'3 (1) spins.
and the adjacent out-ofregistry chains are arranged so that the S = 512
spins
are close to
the antiferromagnetically
coupled S = 112
spins.
which leads to ferromagnetic coupling of antiferromagnetically coupled
spins.
Scheme 3
Note that these ferrimagnetic systems possess
a few common features; namely. they d o not obey the CurieWeiss expression, as x - ' ( T ) is not linear, but a fit to the higher
temperature susceptibility has 0 < 0. Also plots of I T and moment vs. temperature have a minimum value. A mathematical
model describing isolated linear chains of alternating S = 11'2
quantum spins and S > 1/2 classical spins has been derived by J.
Seiden and has been used for some of the aforementioned ferrimagnetic
y$-m
illustrated in Figure 10. As pointed out in the discussion of
mechanism 2.2.1.1 (Section 3.2.2.1 .I). the simplest application
of the CI model between partially occupied molecular orbitals
on adjacent molecules does not produce the correct sign of
the exchange interaction for [CrCpT] [TCNE][58",h1 (t3,'s')
(Cp* = C,Me,) as well as for [CrCpT][TCNQ] (t3/~1).[s"1
[Cr(C,Me,H,_,),]"[TCNE]'(s= 3.6) (0 > 10 K) ( S ~ / S ' ) . [ ~ ~ ~ ]
and [TTF]"[Pt{S2C,(CF,),),]'- ((I= 18 K)(s1/s').[46c~58u1
For
these materials the CI between POMOs predicts antiferromagnetic exchange (resulting in ferrimagnetism for the former pair
of examples). yet ferromagnetic exchange and, for the former
pair of compounds, ferromagnetic order was observed. Instead
, ! 1 ~ ~
ltl''
\
a)
NLUMO-
-+
-+
-*
3-
POMO
NHOMO+
(I)
u
NLUMO
B +A
NLUMO-
A
POMO to virtual NLUMO excitation
NLUMO-
B
(r)
(I)
B
A
B+A
A
B
NHOMO to POMO excitation
Pig. 10. Scheinatic illustration of the stabilization of ferromagnetic coupling by
configurational interaction (CI),involving only virtual cxcitation from a POMO o n
A to a virtual NLUMO on B (a) o r N H O M O on A to a POMO on B (b).
Configuration interactions can be more complex by inclusion
of either the NHOMO and/or NLUMO in addition to the POM O on one or more of the molecular species. In this case the CI
can be considered within each molecular species (e.g.. excitations from occupied A orbitals to virtual orbitals on A ) as well
as between species (e.g.. excitations from occupied orbitals on A
to virtual orbitals of B).
of considering an excitation between POMOs, an excitation
from the POMO of [CrCpT]'" to the NLUMO of [TCNE]'+ (or
alternatively an excitation from the POMO of [TCNE]'- to the
NLUMO of [CrCpT]'") results in the expectation of ferromagnetic exchange. (Similarly, an excitation from the NHOMO of
[CrCpT]'" to the POMO of [TCNE]'+ as well as the reverse
from the NHOMO of [TCNE]'- to the POMO of [CrCpT]'"
results in the expectation of ferromagnetic exchange.) The importance of "subjacent" orbitals has also been invoked to explain the observation of reactions forbidden by the WoodwardHoffmann rules.["] Thus, the extended configurational admixing model can account for the ferromagnetic coupling observed
for a pair of [CrCp;]"+ salts. Because of the large number of
electrons involved. an ab initio calculation to determine the
primary exchange mechanism is beyond current computational
capabilities. The necessity to invoke NHOMO -+ POMO excitation to stabilize ferromagnetic coupling was originally used to
explain the ferromagnetic coupling (0 > 10 K ) observed for
[Cr(C,Me,H,_ ,),]"[TCNE]'+ (s= 3.6).['8'1
3.2.2.2.1. Intermolecular Species Excitations
3.2.2.2.2. Intramolecular Species Excitations
Configuration interactions involving intermolecular excitations from a POMO to N L U M O or NHOMO to POMO are
As discussed previously one may consider CI by intramolecular excitations as well as intermolecular excitations. This special
3.2.2.2. Excitations Involving N H O M O und/or N L U M O
398
A i i , ~ yCheni.
~.
Inr. Ed. En,ql. 1994. 33. 385 415
Designer Magnets
REVINVS
case is sometimes described as spin polarization (or spin delocalization) or superexchange.['22.' 2 3 1 It depends upon which
energy levels are admixed with the chosen ground state wavefunction and it is closely related to the previous discussion. For
example. Figure 11 a illustrates an intramolecular species virtual
excitation from the POMO to the N L U M O on A followed by a
virtual excitation from the POMO on B to the N L U M O on A
and is referred to as spin delocalization, while Figure 1 1 b illustrates intramolecular species virtual excitation from the
N H O M O to the POMO on A followed by a virtual excitation
from the POMO on B to the N L U M O on A and is referred to
as superexchange. I t is important to note that these mechanisms
can give rise to either ferromagnetic or antiferromagnetic coupling.
+-
a)
+-* +
A
A
R
B
U
4-+
-
A
A
B
B
A
B
A
B
Fig. I 1 Scliemauc illustration of the stabilization of ferromagnetic coupling by
configuriitional interaction (CI). involving the virtual excitation of an electron from
a POMO to ti NLUMO on A followed by a virtual excitation from the POMO on
B to the N L U M O on A (spin delocalirationj (a), o r from a N H O M O to the POMO
o n A followed b> a virtual excitation of a n clectron from the POMO on 8 to the
N L U M O 011 A (superexchange) ( b j
Intramolecular species spin polarization was discussed under
mechanism 7.1 (Section 3.2.1) and can also be considered as an
intramolecular species effect. For [FeCpT]" [TCNE]'- A. L.
Buchachenko suggested that intramolecular spin polarization of
the Cp* tive-membered rings by the spin(s) of the transition
metal ions (e.g.. Fe"') should induce a small negative spin on
Cp* ring carbon atoms and result in net ferromagnetic coupling
to the adjacent [TCNE]'-.[331This was later quantified by Kahn
et a l , [ f > l a . hl They suggest a polarization (by CI arising from an
intra-ion excitation of electrons from the Cp* ligand to the
metal-ion) of the bonds between the ring C atoms of the Cp*
ligand. leading to a negative spin density on the carbon atoms
of the C'p* five-membered ring. the magnitude of which increases with the spin on the metal (S = 112, 1, and 312 for [FeCpT]'+,
[MnCpT]'+, and [CrCpT]'", respectively). In contrast. for
[NiCpr]' ' , a direct delocalization of spin density from the Ni"'
to the Cp* ring, resulting in positive spin density on the ring
carbon atoms is suggested. The ferromagnetic (for [FeCpT]".
[MnCpT]:',
and [CrCpT]" ') or antiferromagnetic (for
[NiCpT]") exchange with [TCNE]'- or [TCNQ]'- then results
from interaction of the spin on the Cp* ring carbon atoms with
that of the anion radical.[6'".b1 Recent N M R studies of
[M(C,EtMe,),]+ support the presence of negative spin densities
on the carbon atoms of the five-membered ring for M = Fe,
Mn, and Cr and positive spin density on the carbon atoms of thc
five-membered ring for M = Ni, consistent with the hypothesis.[621Polarized single-crystal neutron diffraction studies are in
progress to directly determine the spin density of the [FeCp:]'+
cation. Configuration interaction involving excitation from the
POMO of the nitroxyl radical to the higher occupied Cu"-based
orbitals has been suggested as the origin of the ferromagnetic
exchange for bischelated Cu" complexes with axially coordinated nitroxyl hgdnds. In contrast the CI between the POMO of the
nitroxyl ligand and the Cu" POMO gives rise to antiferromagnetic exchange for similar complexes with equatorially coordinated nitroxyl l i g a n d ~ . [ " ~ ]
Soos and McWillianis,["'dl as well as more recently Kollmar
and Kahn[61'1pointed out that many additional competing excitations need to be considered by CI than those described under
mechanism 2.2.1 . I . Furthermore, due to the almost-orthogonal
nature of the POMOs on the donor and acceptor for
[FeCp:] [TCNE]. the simplest POMO/POMO-based CI is less
likely than the many competing excitations; thus, the Cp*
ligand can provide a pathway for superexchange or spin delocalization. Yamaguchi et al.[641describes an intramolecular spin
polarization of the Cp* ring by the spin(s) of the transition metal
ions as partial charge transfer from the Cp* of an electron (with
spin parallel to that of the metal) to the metal atom followed by
the CI between the Cp* and the acceptor ([TCNE]'-). The net
effect is usually ferromagnetic: M+(T)Cp*(J)A(T). The sign of
the net exchange then depends upon the sign of the exchange
between Cp* and [TCNE]'-.
It is noted, however, that the T, values increase as
[MnCpT]" > [FeCpT]" > [CrCpT]'", and d o not increase
monotonically with S as expected (see Section 6). This suggests
that for the [CrCpT]:" salts, which are predicted to have antiferromagnetic coupling by the simplest POMO/POMO-based
CI model, the POMOjPOMO CI does not dominate, and ferromagnetic coupling by the more complex POMOiN HOMO/
NLUMO-based CI model or an intramolecular CI spin polarization model may be more appr0priate.1~". '". 6 ' c ] Thus.
further studies are necessary to firmly establish which specific
exchange mechanisms dominate in this series of electron-transfer salts.
The combination of CI using POMO/NHOMO/NLUMOs as
well as spin polarization have been used to explain the ferroniagnetic coupling reported for galvinoxyl (16a). Galvinoxyl has
received substantial attention because it exhibits relatively
strong ferromagnetic coupling (0 = 19 K) in the solid state. The
0 value is too large for dipole interactions (vide infra) alone to
be operative. Below 85 K it undergoes a first-order phase transition to a weakly magnetic
The details of this structural
transformation are unknown. A solid solution of 16a and hydroxygalvinoxyl (16 b) has been reported to exhibit ferromagnetic coupling (10 < 0 < 19 K), since the high-temperature
399
J. S. Miller, A. J. Epstein
REVIRNS
structure is maintained.[hsb1Hysteresis curves characteristic of a
bulk ferromagnet were not observed for either of these materi-
3.3. Dipole-Dipole (Through-Space) Exchange
als.l~sal
Dipole-dipole, through-space spin-spin interactions not involving the overlap of orbitals have not been formally discussed
as a mechanism to achieve bulk ferromagnetic behavior; however, the results of several studies show that through-space spinspin interactions can lead to cooperative magnetic behavior below about 1 K."'] This weak magnetic interaction is due to
magnetic fields generated by the magnetic moment associated
with each spin. It does not involve overlap of orbitals or spin
density as illustrated in Figure 12.
t Bu
m;
t
B
p
q
B
U
OH
0
t Bu
t Bu
t Bu
16a
t Bu
16b
More recently, the ferromagnetic ordering reported for /I-pO,NC,H,NIT (NIT = nitronyl nitroxide) (17) has been ascribed to a CI between the POMO on one site and a NHOMO
on another site.i661It, however, is plausible that nonbonding,
'4
-4
17
orbital overlap
through-space dipole-dipole interactions may be important,
and this system is described in greater detail under mechanism
3 (see Section 3.3).
3.2.2.3. Excitations between Localized and Delocalized
(Metallic) States (Intramolecular Spin Coupling)
A model of antiferromagnetic exchange (by configuration
interaction) between localized d electron spins of the organometallic radicals and a proposed band of detocalized organic
moiety based n electrons has been proposed by Tchougreeff and
Misurkin and applied to [FeCpT] [TCNE].[671This picture resembles those which arise in the Kondo problem[681and in the
Ruderman, Kittel, Kasuda, and Yoshida (RKKY) model for the
interaction of magnetic impurities in
They proposed
that the POMO b,, [TCNE]'- n* orbital hybridizes with the
delocalized n orbitals of the Cp* ligands and the overlapping 4s
and 4p orbitals of Fe"' to form a partially filled band of states.
They showed that antiferromagnetic exchange between the localized d electrons of [FeCpz]'+ and the delocalized n electrons
of the organic system could lead to ferromagnetic order in the
system of localized Fe d-orbital-based spins. This does not,
however, produce the correct saturation magnetization for
[FeCp;] [TCNE], because the delocalized n electrons would
have a reduced net spin, part of which would order antiferromagnetically with respect to the Fe d-orbital-based spins. Furthermore, the insulating nature of [FeCpz] [TCNE] indicates
that a delocalized n electron band model is inappropriate. However, a similar approach of interaction of localized metal-based
spins with delocalized n electrons was used to account for phenomena in selected phthalocyanine-based salts.'691
Recent work on the ferromagnetic Zintl phase A,,MnBi,,
(A = Ca and Sr) with critical temperatures of 55 and 33 K,
respectively, attributes the magnetic properties to the ferromagnetic coupling of spins between the localized Mn"' centers by the
conduction electrons (the R K K Y model).r51
400
spin density interactions
Fig. 12. Schematic illustration of the orbitals (a) and spin densities (b) for nonbonding dipole-dipole. through-space interactions involving a POMO on A .
Tanol suberate (18), which has a sheetlike crystal structure,
and exhibits Curie- Weiss behavior at higher temperature
(0 = 0.7 K) and metamagnetic behavior below a T, of 0.38 K, is
a candidate for this mechanism. Below the critical field ( H J of
100 G 18 is an antiferromagnet, while above 100 G 18 has a
high-moment state."'] Polarized-neutron studies were utilized
to determine that the spin density is solely located on the NO
moiety, equally between the 0 and N.[72a1Neutron scattering
has also determined that the N O groups couple ferromagnetically within layers separated by 5.9, 6.4,and 6.9 A. and couple
antiferromagnetically between the layers separated by 10.8 A.
Very recently 1,3,5,7-tetramethyl-2,6-diazaadamantane-2,6dioxyl(l9) was reported to exhibit ferromagnetic interactions in
the solid as evidenced by a 0 of 10 K and a fit of the M ( H ) data
to S = 6.[73"1
The relatively large value of 0 suggests that
through-space dipole-dipole interaction is not the only mechanism operative. Compound 19 recently has been reported to
order at 1.48 K.i73b1
19
Crystals of 17 fit the Curie-Weiss expression (0 zz 1 K) at
higher temperatures and have been reported to have a ferromagAngekv. Chrm. I n t . Ed EngI. 1994, 33. 385-415
REVIEWS
Designer Magnets
netic state for a
at 0.60 K.I6('] In contrast 7-p-O,NC,H,NIT
is an antiferromagnet.[66b]The origin of the exchange has been
ascribed to a C1 between the POMO on one site and a N H O M O
on another site (vide
However, due to the nonbonding interactions. dipole-dipole interactions may be important.
Under hydrostatic pressure the ferromagnetic coupling, as evidenced by an increase of 0, is increased by 40 ?LO at 9 kbar pressure.[""] The large pressure-dependence suggests that throughspace dipole-dipole interaction is insufficient to account for the
three-dimensional ordering. Zero-field muon spin resonance,
which detects the spontaneous static local magnetic fields in
ordered states. clearly shows a precession signal comparable to
that expected from a three-dimensional Heisenberg model and
a T, of about 0.67 K.[741Antiferromagnetic coupling has been
observed for several other nitronyl n i t r ~ x i d e s . ~ ' ~ ]
4. Progress in the Discovery of Magnetic Materials
Based on Molecular Species
Worldwide interest in molecular-species-based ferromagnets
is growing at the present time.[41Experimental evidence for bulk
ferromagnetic behavior in molecular-species-based compounds
has been limited to the electron-transfer salts comprised of
decamethylnietallocenes and tetracyanoethylene (TCNE)
and/or 7.7,8,8-tetracyano-p-quinodimethane(TCNQ), and [j-pN02C(,H4NIT (17). Ferromagnetic coupling has been reported
for [ (Me,N)2CC(NMe,),](C,o), though no hysteresis curves
were reported below a transition temperature of x 16 K , and the
saturation moment at 5 K corresponds to about 10% of the
expected value for only one spin per CbO."('] Recent studies
ascribe the high-temperature antiferromagnetic behavior
(0 = - 58 K ) to a superparamagnetic state above x 16 K.[76c1
Preliminary zero-field muon spin resonance, however, does not
show a change in the spectra as the temperature is reduced to
below T.[741
There is a rapid suppression of the T, with pressure
by about 1 0 K per kbar.[76d1
The unusually strong pressure-dependence combined with the relatively high conductivity
( 2lo-' S c m - ' ) of [(Me,N),CC(NMe,),](C,,)
led to the suggestion'^"dl that the transition at 16.1 K may be due to an itinerant ferromagnetic state. In contrast, ESR studies suggested[76b.c] that the radical spins in the magnetically ordered state
localize on the C,, moiety and are ferromagnetically correlated.
Further study is necessary to establish the low-temperature
magnetic state of this system and determine the origin of the
exchange interaction. Cooperative magnetic behavior is not observed for [(Me,N)zCC(NMe,),](C,,).[76b~'I
Evidence for ferrimagnetic ordering has been demonstrated
for several Mn"Cu" and Mn" (nitroxide) extended chain structures discussed previously as well as the room-temperature magnet [V(TCNE),] . ?(CH,CI,) and the recently reported mang a n e s e ( ~prophyrin-TCNE
~)
electron-transfer salt (vide infra).
Numerous attempts to prepare ferromagnetic polymers by
the topochemical polymerization of a dinitroxyldiacetylene, oxidation of 1.3.5-triaminobenzene or an indigo polymer, thermolysis of numerous hydrocarbons and polymers, hydrogen
abstraction from polymers based upon triphenylmethyl fragments. and electrodeposited polyaniline have been reported.
The stated magnetizations correspond to only a few percent of
the maximum theoretical magnetization. The poor chemical
characterization, low values of the saturation magnetization.
and marked variation between samples suggest that the ferromagnetic behavior in these infusible, intractable polymers may
not be intrinsic. This area has been recently reviewed."] In the
remainder of this review we focus on research primarily, but not
exclusively, carried out in our laboratories which was initiated
with the discovery of cooperative magnetic behavior in a metallocene-based electron-transfer salt.
5. Structure and Magnetic Properties of
Electron-Transfer Donor/Acceptor Salts
In 1979 [Fe"'CpT]''[TCNQ]'was shown to exhibit a highmoment ferromagnetic state above 1600 Oe (i.e.. cooperative
metamagnetic proper tie^)."^"] The linear chain structure was
established for the kinetic phase of [Fe"'Cp:]'+[TCNQ]'- [77h1
(Fig. 13). The moment p significantly deviates to positive values
Fig. 13. Alternating donor:acceptor
D" A ' - D ' A ' - . . . linear chain sti-uctui-e
of [Fe"lCp:l" [A]'. [A = TCNQ. TCNE. DDQC12. CJCN),]. [Fc"Cp,l[TCNEj.
and [Fe"'":]"
[C,(CN),]-. The structure shows v i w s of adjacent in-registry and
out-of-registry chains for A = TCNE.
' ' '
at low temperature from the temperature-independent Curie
behavior characteristic of ferrocenium salts with spinless anions.[', 77a1 The field-dependent magnetization A4(H) reveals
metamagnetic behavior, that is, for an applied field of less than
1600 Oe the magnetization is characteristic of an antiferromagnet; whereas above 1600 Oe a sharp rise and approach to magnetization saturation characteristic of a ferromagnet is observed
(cf. Fig. 5)."7a1 While several metaniagnets, for example, FeCI2.
had been c h a r a c t e r i ~ e d , this
~ ~ ~was
] the initial example in which
neither a one-, two- nor a three-dimensional covalently bonded
network structure is present. Subsequently, we sought to elucidate the structure-function relationship by the systematic modification of the acceptor A, the substituents on the C,Me, ring,
and the metal ion to identify the steric/electronic features necessary to stabilize ferromagnetism and ultimately design a molecular-species-based ferromagnet.
40 1
J. S. Miller. A. .I. Epstein
REVIEWS
On the supposition that a smaller radical anion would
have a greater spin density which could lead to increased spinspin interactions, we sought to identify stable radical
anions smaller than [TCNQ]'+. and thus selected [TCNE]'-.
[Fc"'CpT]'+[TCNE]'- was prepared and found to posscss the
same . . D'+A'-D'+A'- . . motif (Fig. 13).17y1With application of only the Earth's magnetic field a spontaneous magnetizaThe saturation magnetization for single
tion is
crystals aligned parallel to the . . D"A'-D'+A'. . . stacking
axis is 36 Yogreater than iron on a g-atom basis, and agrees with
the calculated saturation moment for ferromagnetic alignment
of the donor and the acceptor.[8g j1 The critical (Curie) temperature T, was determined to be 4.8 K . The magnetization versus
applied field data for [Fe"'CpT]'+[TCNE].- exhibits hystcrcsis
loops (Fig. 14) with a large coercive field of 1000 Oe a t 2 K.["]
2.0
1.5
1 .o
i-
0.5
!
1'
I 1
1 0 4 ~ ~0
emu G mol
I
-0.5
I
I
-1.0
I
!
!
J
-1.5
-1500 -1000 -500
0
500
1000
1500
HIG+
Pis. 14. The magnetization A4 as a function of applied field
[Fe"'Cp:]'-[TCNE]'shows hksteresis loops at 2 K .
I/
I'oI
It is sufficiently magnetic at low temperatures to be attracted to
a Co,Sm magnet (Fig. 15).
Fitting the single-crystal magnetic susceptibility data taken
for the applied magnetic field parallel to the C , axis of the donor
402
to different physical models aids in the understanding of the
microscopic spin interactions. Above 16 K interactions among
the nearest neighbor spins within individual one-dimensional
chains are sufficient to understand the magnetic behavior. The
Curie Weiss Law is sufficient to model the higher temperature
( T > 130 K j data with 0 = 30 K,[". 7 y 1 A one-dimensional
Heisenberg model with a ferromagnetic exchange coupling ( J j
of 19 cm- (27 Kj. however, better accounts for the data down
to 16 K.IZ5l Thesc models are insufficient to account for. the
magnetic behavior below 16 K a s long-range spin correlations
and three-dimensional spin interactions become increasingly
important. This continues until permanent long-range (three-dimensional) ferromagnetic order occurs at 4.8 K . Preliminary
studies ofthe pressure dependence of T, reveals that T, increases
with applied pressure by 0.21 K kbar-' and reaches 7.X K at
14 kbar applied
(This is in marked contrast to
decreasing T, with increasing pressure of the organic superconductors.[8"bI)Variation of the low-field magnetic susceptibilitl
with temperature above T . magnetizatioii urith temperature below T,,, and the magnetization with magnetic field al T. enabled
a measurement of three critical exponents (Table 4) for the magnetic field parallel to the chain axis. The obscrvcd valucs arc in
accord with a transition to a three-dimensional ordered magnetic state.
The configuration admixing of a virtual-triplct cxcitcd state
with the ground state model (mechanism 2.2.1.1 i n Scheme 1 )
provides a mechanism to stabili7e ferromagnetic interactions
within donoriacceptor (D:A) complexes and a simple. prelimnary framework to view this class of materials and mike some
predictions a s to the magnetic coupling of new materials. The
paradigm of this model is realized by this series of clcctrontransfer salts possessing a linear chain (one-dimensional) structure comprised of alternating cation donors (D) and anion acThis
ceptors (A) that is, . . D'+A'-D'+A'- . . (Fig. 13j.[8F.y1
model is described in further detail under the mechanism 2.2.1 . I
(Schemc 1 : Section 3.2.2.1.1). More complex models have been
invoked to describe this system and are described under the
mechanism 2.2.2 (Scheme 1 : Section 3.2.2.2).
To identify the structure-function relationship with the goal
of preparing a molecular-species-based ferroniagnet with a
higher critical temperature. thc properties of . . . DADA . . .
structured compounds based on [M(C,R,)J+ were studied.
Three modifiable entities are: replacemcnt of the Me groups
with H and Et, LISC of altcrnatc open and closed-shell anions,
and replacement of Fe"' with other metal ions.
REVIEWS
Designer Magnets
[F.L C.P ;I ' /
[MnCpf]: '
[C'i-cpq:'
[FeCp;]. '
[hln"p;y
[C'rC'pTl:' '
[TCN El'[TCNE]'
[TCK';E]'11-CN Q]'
[TCNQ]'[TC"Q]'~
~
12
I
3.2
1.2
1
3'2
1.2
1,2
1,2
1,.2
1,:
I:>
4.8
8.8
+ 16.9 [el
3.65 ti]
1.55 [a]
6.3 [b]
3.3 [c]
f32.3
f12.3
+ 22 6
f
10.5 [i]
+ 11.6 [fl
1000(2)
1200(4 2)
[Zl
0.09
:05
1.21
4.42
[25. 701
[XI1
0.5
1.1
[5XI
[77a]
3600(3)
[kl
[5')1
[a] M e ~ ; i i n . i ~ i i c tnit11
i c :I 1600 G critical field. [b] 50 G. 7<
from the maximum slope ofd.lil.dTis reported a s 6.2 K . [c] 15 G. T, from themaxirnum clope ol'd.l.l.d71\ reported
:I\ ;1 K Id] I10r pol!ci-)stalline samples. [el For cr)stals aligned parallel a n d perpendicular to the applied magnetic field. t i , = 30 K and 0 . = 10 K 1251. [I] 12.8 K (R. S.
1LlcLe;in. I. S. Millei- iinp~ihlisliedresults). [g] Nor observed [I>] Ci-iticolconstants [I] 9.0 K ( R . S McLean. 1. S. Miller unpublished resulrsj. [j] 0.15 C,. [k] Kol observed:
10.15 o c
Substituted C 5 rings may maintain the fivefold symmetry necessary to form a cation with degenerate partially occupied
molecular orbitals and a Kraniers doublet ( ' E ) ground state as
reported lo r decamethylferrocenium. Three other such ferrocenes have been studied: ferrocene. 1,2.3,4.5-pentamethylferrocene. a n d decaethylferrocene. Ferrocene is significantly more
difficult t o oxidize than decamethylferrocene. and is not oxidized by TCNE.[H31Nonetheles, [Fe"Cp,] [TCNE] forms and
belongs to the same structure type[s41(Fig. 23) and is diamagnetic.LX4hl
[FeCpCp*] (Cp = cyclopentadienide. C,H,) is a sufficiently strong donor to reduce TCNE, and the simple ( 1 : I )
one-dinicnsional salt iis well as three other phases were prepared.["] This 1 : 1 phase exhibits weak ferromagnetic coupling
a s evidenced from the fit of its susceptibility to the Curie-Weiss
expression uith 0 = 3.2 K. but cooperative three-dimensional
magnctic ordering is not observed to the lowest temperature
studied ( - 2 K).r851A linear chain structure is proposed for
[Fe"'(C''Eti)2]'+ [TCNE]'-. which also exhibits ferromagnetic
coupling as evidenced from the fit of its susceptibility to the
Curie- Weiss expression with 0 = 7.5 K . It also did not exhibit
cooperative three-dimensional magnetic ordering.18h1
To test the necessity of a ' E ground state, the TCNE electrontransfer salt with the lower symmetry [Fe(C5Me,H),] donor was
prepured.18-1 The magnetic susceptibility can be fit by the
Curie-Weiss expression with 11 =z 0 K. The absence of either
three-dimensional ferro- or antiferromagnetic ordering above
2.2 K in [Fe(C,Me,H),]'+[TCNE]'- contrasts with the behavior of [FeCpt]'+[TCNE]'-. The "Fe Mossbauer data are in
accord nith the absence of significant magnetic coupling
among the radical ions, and show only nuclear quadrupole
splitting for the [Fe(C,Me,H),]"
salts and not the zerofield Zeeman splitting observed for [FeCp:]'+[TCNE]'-,[7y1
[FeCp~]'-[TCNQ]--,''7b1 [FeCp:]'+[C,(CN),]'-.[R8'and
[FeCpT]' [DDQCl,]'- (DDQCI, = 2.3-dichloro-5.6-dicyanobenLoqiiinone).[8y1The lack of magnetic ordering apparently
arises 1.1-om poorer intra- and intermolecular overlap within and
between the chains leading to substantially weaker magnetic
coupling for [Fe(C,Me,H),]'+[TCNE]'-. This suppresses the
spin-ordering temperature. Alternatively, due to the overall C,,
symmetry the [Fe(C,Me,H),]'+
charge-transfer excited state
may be a singlet. not a triplet as expected for [FeCp:]",
and the
admixture of a singlet, not a triplet charge-transfer excited state
should lead to antiferromagnetic. not ferromagnetic. coupIing.18s. Since significant antiferromagnetic coupliiig was also absent for [Fe(C,Me,H),]'+[TCNE]'-, perhaps a reduced
overlap with neighboring radicals is the more important effect
of modification of the cation.
Substitution of TCNE with either TCNQ,["] C,(CN),.[881
TCNQI, ,["I (TCNQI = 2,Sdiiodo-TCNQ). DDQCl,,'*yl or
[M{S,C,(CF3),),,]'{tz = 2,
M = Ni,["'"] Pt[y2h1; 17 = 3,
strucM = M O [ " ~ ~acceptors
}
leads to
tured complexes with dominant ferromagnetic coupling (see
Tables 5 and 6).
Replacement of [TCNE]'h diamagnetic [C,(CN),]- also
leads to the formation of a
D ' + A - D ' + A - . . structured
phase; it exhibits only Curie susceptibility (11 = - 1 K)."91 The
[Co"'Cpz]+ [TCNE]'- complex with a diamagnetic donor has
been prepared and exhibits essentially the Curie susceptibility
anticipated for [TCNE]'- .'I
This demonstrates the necessity of
having both the D and A radicals for stabilizing ferromagnetic
coupling in the linear chain alternating . . . D'+A'- D' 'A*- . .
motif.
Attempts to prepare [M'"CpT]+ (M = Ru, 0 s ) salts of [TCNE]'- have yet to lead to compounds suitable for comparison
with the highly magnetic Fe"' phase.[y31Replacement of Fe"' in
[Fe"'Cp5]'' [A]'- (A = TCNE, TCNQ. C,(CN),) with doublet
Ni"'. triplet Mn"', and quartet Cr"' leads to compounds exhibiting antiferro-. ferro-, and ferromagnetic coupling. respectively
(see Tables4 and 5). This series enables the direct testing of
several models. The simplest POMO/POMO-based CI model
predicts antiferro-, ferro-. and antiferromagnetic coupling, respectively; whereas invoking the more complex NHOMOi
NLUMO excitation CT model predicts the observed antiferro-.
ferro-, and ferromagnetic coupling, respectively. More importantly, the anomalous behavior of [Cr"'Cp:]"' [TCNE]'- is
striking since its T, is less than that observed for
[Fe"'Cp:]'+ [TCNE]'- and [Mn"'CpT]:+ [TCNE]'-. and fails to
follow
x S(S + l), which suggests the competition of more
than one mechanism (vide infra).
Magnetic ordering is not observed for bis(dithio1ato)metallate salts of decamethylferrocenium; however, ferromagnetic coupling as evidenced by Curie- Weiss 0 constants
which range from 0 to 27 K is observed (Table 6 ) . This provides
an insight into a structure-function relationship. Of the compounds studied only [MCpT][M'{S,C,(CF,),),] (M' = Ni, Pt;
M = Fe.["'] Mn["]) and [FeCp:] [Mo{S,C,(CF,),J 3][y2c1
have
a one-dimensional chain structure. The M = Fe coinplexes have
the largest H values, the largest effective moments, and the most
pronounced magnetic field dependencies of the susceptibility.
The Pt analogue (0 = 27 K) possesses one-dimensional
403
J. S. Miller, A. J. Epstein
REVIEWS
Table 5 . Summary of thc Curie-Weiss H values and critical temperatures (7;)for
. . D'A
D * A . . . structures.
. . . D'+A"D"A'chains. whereas the Ni analogue ( 0 =
I 5 K) possesses zigzag one-dimensional . D'+A'-D'+A'- . .
chains and longer M . M separations (1 1.19 A vs. 10.94 A for
'
' '
the Pt analogue). Thus. the enhanced magnetic coupling the Pt
compound compared with that of the Ni compound presumably
arises from the stronger intrachain coupling. In contrast, for
8.8
[MnCpT] [M'{S,C,(CF,),)
(M' = Ni. Pd, Pt) weak ferromag37 7
3.65
netic coupling is observed (0 = 2.8 f 0.8 K), and metamagnetic
20
19
behavior with a 7; of 2.5 i- 0.3 K was reported.["] This is
19
comparable to the T, reported for metamagnetic
19
[FeCp~]'+[TCNQ]'-.[77"1
18
4.8 [h]
16.8 [gl
It is curious to note that, whereas replacement of [FeCpT]'+
44
(
S
= 1!2) in [FeCpT]'+[TCNE]'- with [MnCpS]:+ ( S = 1 ) in3.8
creases T, nominally in accord with theory (i.e.. 7; x S(S + 1).
2.75
0.75 [c]
vide infra),[8'1 and replacement of [FeCpT]'+ in
15.1
[FeCpT]'+[TCNQ]'- with [MnCpT]:+ destabilizes the meta15
[FeCp;]" [Ni[S,C,CCF3),: 4 15
magnetic behavior and stabilizes ferromagnetic ordering, how[MnCpf] ' [Me,DCNQI]'[CrCp;] .+[C,(CN),J13.8
ever, replacement of [FeCpT]" in [FeCp,*][M'{S,C,(CF,),) l]
13 5
[FeCpf]" [chloranil]'with [MnCp;]:' stabilizes metamagnetic behavior.[97]The ab12.8
3.5 [f]
[Ci-Cp:] ' ' [TCNQ]'12.3
[FcCpf]" [TCNQ]'
2.55 [d]
sence of simple trends emphasizes the important role of inter6.5 [el
[MnCp:] ' [TCNQ]'10.5
chain as well as intrachain interactions (vide infra).
12
In the opposite extreme [FeCp,*][Ni(S,C,(CN),),] possesses
11.8
11.o
[Cr(C,,Me,HJ,].' [TCNE]'.
isolated D'+A;-D'* dimers and has a zero 8, the lowest ,ueffr
[FeC'p:]'' [Me,DCNQI]'10.8
and has no field dependence of the
The ob[TTF]" [Pt(S,CZ(CFJ2j4 10.8 [J]
[FeCpr]" [DDQCl,]'served high moment is consistent with one spin possessing a
10.3
9.5
highly anisotropic g value per repeat unit. Intermediate between
8.4
the one-dimensional chain and dimerized chains structures are
7.5
6.1
[Fe(C,EtJ,]' ' [TCNQ]'the x- and [$phases of [FeCpT] [Pt{S,C,(CN),),] which have
3.3
[FeCpCp*]'+ [TCNE]'one-dimensional . . . D'+A'-D'+A'. . strands in one direc2.8 [d]
3.7
tion and . . . DAAD . . dimer units in another direction. Their
2.6
2.4 [d]
1.9
2.3 [d]
0 and peffare intermediate in value (Table 6). This is consistent
0.8
[Fe(C,Me,H),]' ' [TCNQ]'with one-third of the anions having a singlet ground state. Thus
-0.3
[Fe(C5Me4H)2].+[TCNE]'.
a correlation exists between the presence of one-dimensional
-0.6
[FeCp;]" [cyanil]- [I]
D ' I A ' - D" A'-1
[CoCpf]' [TCNE]'
chains and the presence and magnitude
-1
[FeCpr]" [C,(CN),CI]of
ferromagnetic
coupling
as evidenced by a positive 0 value.
-1.2
[FeCp;]" [C,(CN),]-6
[NiCp:]" [C,(CN),]'Insight into the effect of increasing the intrachain separation
-11.5
[NiCp:]" [TCNE]'was obtained through the study of a tris(dithio1ato)metallate
34 [i]
[TTF]'+ [Nil S,C,(CF,),) ,]'salt of [FeCp,*]'+, namely [FeCpT]" [Mo{S,C,(CF,),) 4 - .[92c1
[a] T , determined from a linear extrapolation of the steepest slope of the M(7') data
The salt possesses only parallel out-of-registry one-dimensional
t o the temperature lit which M = 0. [b] Unpublished data. [c] ac (100 H r ) measure. . . D' +A' - D' A' - . . . chains with intrachain M o . . Mo sepament. [d] Metamagnetic. [el 50 G . T , from the maximum slope of dM:dTis reportrations of 14.24 A. The 0 value is only 8.4 K, which probably
ed as 6.2 K . [t] 15 G, T , from the maximum slope of dM/dTis reported as 3.1 K.
[g] For crystals aligned parallel and perpendicular to the applied magnetic field:
reflects the enhanced shielding of the spin and reducing of the
0 = 30 K and O 1 = 10 K [XI. [h] 7.X K at 14 kbar [go]. [i] -18 K reported [46a].
spin-spin interactions due to the bulky CF, groups. [FeCpT]"
[j] 16 K reported [46c]. [k] Me2DCNQI = 2,5-dimeth~l-N.N'-dicyanoquinonediiinine. J. S. Miller. C. Vazquez. R. S. McLean. W. M. Reiff. S. Hunig, Mi,.Motrv.
[Mo{S,C,(CF,),},]'and
[TTF]"[Mn{S,C,(CF,),l,l'1993. 5 , 448.
(M = Ni, Pt) are the sole structures reported to date that only
possess out-of-registry parallel chains and d o not have parallel
chains in-registry. Thus. the antiferromagnetic s'/s' A'-/A'interchain interactions are eliminated; however, for the latter
Table 6. Summary of the Curie-Weiss 0 values and pZitvalues for [FeCp;] [M(S2C,(CN),),]. pair of electron-transfer salts antiferromagnetic s ' j s ' D'+/A'interchain interactions are present.
Anion
Structural arrrangement
Spin repeat
Susceptibility
These results support the utility of one-dimensional
unit
0 [Kl ~ . i r[PJ
. . . D'+A'-D'+A'- . . . chain structures for achieving signifi[ N I ( S , C , ( C N ) ~ ) ~ ] ' - D'' [A];-D'+ dimer [a]
D'
0
2.83
cant ferromagnetic coupling and ultimately bulk ferromagnetic
D"+l:3A'66
3.05
r - [ P t [ S , C , ( C N ) , ~ , ] ' ~ . . D'+A"D'' . . . sheets
I D - - D ' + [ A ] I - -.chains[a]
behavior as observed for [FeCp:]'+[TCNE]'With the wide
/I-[PtlS,C,(CN),),]'- 1 D - D ' + A ' - . . chains
D''+1,3A'9.8
3.10
variation of bis- and tris-dithiolates available other salts may
D' [A]: D' ' dimer [a]
exhibit stronger ferromagnetic coupling and possibly bulk ferro[ N I ( S ~ C ~ ( C E ~ ) ~ ]l ~
D]- '. -D ' + A ' - ' . . [b]
D" + A ' 15
3.73
[Pt:S2C2(CF,),J2].I D - - D ' + A ' - . chains
D" + A'.
27
3.16
magnetic behavior. Furthermore, many polymorphs with differ[Mo(S,C,(CF,),I,]'1D-.D'+A'. chains
D" + A'8.4 3.85
ing physical properties can easily, albeit sometimes irreproducibly, be prepared and the specific design of structure types,
[a] [A]: =isolated S = 0 [MIS,C-,(CN),),]:- dimer, [b] Zigzag chains.
[Fec'pf]' [C,(CN),,]'[FeCpf]' ' [PtlSLCL(CF3)ijl].[MnCp:] ' [DDQCl,]'[MnCpf] ' [TCNE]'[CrCpr] [TCNE]'[MnCp:] ' [DDQBr2]'[MnCpf] ' [DDQI,]'.
IFeCp;]" [broinanil]'[l-eCpf]'+[DDQBr,]'[MnCpf] * [C,(CN),]'[FeCp;]" [TCNE]'+
35
27
27
22.6
' +
~
~
-
+
+
404
A n g e ~Cheni. I n f . Ed Engl. 1994, 33,
385-415
REVIEWS
Designer Magnets
for example, one-dimensional . . . DADA . . chains, continues
to be an art.
an external field exists in the mean-field model whenever the
Equations (12) have nonzero solutions. The transition tempera(12a)
6. Calculation of the Critical (Curie) Temperature T,
Heisenberg along with D i r a ~ [solved
~ ~ ] expressions that relate
T, to the exchange integral ( J ) between sites i and j for the
general spin case assuming that only the number of equivalent
nearest neighbor sites ( z ) is important. This simplest mean-field
model leads to Equation
T, =
2JrS(S + 1)
3 k,
Equation (10) is constrained to have only one spin site S
and a unique J . These are poor assumptions for the
[MCpT]' [TCNE]'- system since the distinct spin sites may have
different values of S.and the crystal structure implies four distinctly different nearest neighbor exchange integrals, that is, J,,
JAA,
JDD,
and $,"(Fig. 16). A more appropriate mean-field expression for T, was therefore developed.r241In the simplest form,
since the values for the four exchange integrals are unknown, a
single effective 1,Jeff,is utilized. In this simplified description,
which may be inadequate for highly anisotropic materials, there
are two distinct spin averages (S, and SAcorresponding to the
D + and A'- sites, respectively). The Hamiltonian per molecule
(.X ) takes the form of Equation (1 1) : The directions of the vec-
tors (S,? and ( S , ) are along their respective effective magnetic
fields. A state of spontaneous spin polarization in the absence of
-
+-Jgi - - - - - - -
7J.
7
ture T,is the highest temperature for which this is valid. Near
the T, the self-consistent spin averages are small and a Taylor
series expansion of the equations reduces them to the coupled
linear Equations (12a and 12b).
Equations (12) have a solution only if the determinant of the
coefficients vanishes; a condition which leads to a quadratic
equation for 1/T = I/Tc,and only one of the roots gives a positive q .This formalism breaks down as J in one direction becomes large compared to the other interactions because a onedimensional system is formed in this limit and T, must approach
zero Kelvin. The prediction of a finite transition temperature for
a one-dimensional system is a well-known fundamental deficiency of mean-field models, and a more sophisticated method
is probably required to fully treat the strongly anisotropic case.
Nonetheless, the general solution for a single effective J , Jeff,for
Equation (12) is Equation (13).
Assuming that the intra- and interchain interactions remain
unchanged as the donor is varied, Jeffwill remain constant and
consequently the relative T, can be calculated for different values of SAand S, (Table 7). For [Fe"'Cp~]'+[TCNE]'-,T, is 4Jef,
within this mean-field model. Since the observed T, is 4.8 K, Jeff
is 1.2 K (0.83 cm-', 0.1 meV). This result points out that
though the value of Jeff
is indeed small, it is sufficient to enable
a T,at experimentally accessible temperatures due to the large
number of nearest neighbors.
The major advantage of the above equation is that T,can be
scaled for different S, values (Table 7). Thus, T, for a donor
(S, = 1) (with an S, = 1/2 acceptor) is 6.8 J or 1.7 times greater
than that for a S, = 1/2 donor (with an SA= l / 2 acceptor) system. Hence, since T, is 4.8 K for [Fe"'Cpr]'+ [TCNE]'-, and all
else being equal, T, would be expected to be 8.2 K for the
isostructural [Mn"'CpT]'+[TCNE]'-. This is in good agreement
with the reported value of 8.8 K for [Mn'l'CpT]'+[TCNE]'-.rs'l
(More complex equations utilizing four different Jvalues lead to
JDD-.
Jn*
-anion
----
Chring
Table 7. Calculated mean-field T , as a function of S, for S, = 1;2 systems.
.*----Fe
out-of-registry
in-registry
Fig. 36. Schematic illustration of the structure of orthorhombic
[Fe"'Cp;]'+[TCNE]'- showing intrachain, pairwise out-of-registry and in-registry
interactions.
Aiigiw. Clwii. Inr. Ed. Ei?gl. 1994, 33. 385-415
T.
SD
45
112
1
6.8 J
10 J
312
2
[a] Scaled: T,
13.7 J
=
scaled
[a]
1.oo
1.70
2.50
3 43
TJTC(S, = 112).
405
J. S. Miller, A. J. Epstein
REVIEWS
an improved scaling of 7;. as a function of S D . [ 2 4 , R Thus.
' 1 ) the
observed 7;. is in excellent agreement with the experimental values and suggests that if this mean-field model is correct. similar
exchange interactions operate for both [Fe"'CpT]''[TCNE]'and [Mn"'CpT].+[TCNE]'-.['""l
The S , = 3/2;S, = 1/2 case (i.e., [Cr"'CpT]''+ [TCNE]'-) was
also theoretically treated and a predicted scaled value for the T,
of 12 K was obtained. Experimental study of the magnetic susceptibility of [Cr"'CpT]''-[TCNE]'- shows a ferromagnetic
transition at 3.65 K substantially lower than that of either
[Fe"'CpT]'+[TCNE]'- or [Mn"'CpT]'+[TCNE]'- .["I This trend
is also observed for [Mn"'CpT]:+[TCNQ]'- (7; = 6.2 K)[821and
[Cr"'C"]'"[TCNQ]'+
(T, = 3.1 K)[591and may arise from the
Cr"' salts having significantly lower values of J than the Mn"'
and Fe"' salts.^'"'] The anomalously low value of
is further
compounded by the observed ferromagnetic exchange, in contrast to the prediction of antiferromagnetic exchange and a ferrimagnetic ground state for the configuration mixing model
described previously. (This model successfully yields the sign of
the magnetic exchange of all other linear chain metallocene
cases studied so far.) Experimental study of [CrCp;] [TCNE] is
complicated by its extreme sensitivity to the presence of oxygen
which modifies the observed exchange interactions.[s81
7. Dimensionality of the Electron-Transfer
Donor/Acceptor Salt System
The mean-field models described previously emphasize a
three-dimensional approach to magnetic ordering. The low value of the three-dimensional ordering temperature (T, = 4.8 K)
compared to the one-dimensional exchange interaction (as indicated by J = 19 cm-' (27 K)) for [FeCpT]" [TCNE]'-, suggests
the applicability of quasi-one-dimensional models for the onset
of three-dimensional cooperative ordering. For example, for a
tetragonal lattice with interacting nearest neighbor chains
[(Eq. (l4)].['"']
Using a value of J = 27 K for the intrdchain exchange and the
experimental value of Tc,a ratio of LJintra/LJnler =77 is obtained.
Thus, [FeCpT]'+[TCNE]'- and most of the other metallocene
electron-transfer salts are in the one-dimensional limit.
To test the critical importance of the one-dimensionality,
spinless S = 0 [CoCp;]' cations were randomly substituted for
the cation in the [FeCpT]" [TCNE]'- structure. This resulted in
the formation of random finite magnetic chain segments embedded into the linear chains.["51The dramatic, precipitous reduction of T, with increasing [CoCp;]' content is in excellent agreement with theoretical concepts['o31(Fig. 17, Table 5) for highly
anisotropic systems. With a 14.5 YOreplacement of Co"'for Fe"',
T, dropped from 4.8 to 0.75 K. The extreme sensitivity of the
three-dimensional ordering temperature T, stands as a cautionary note in attempts to observe a high T, for solids comprised of
high-spin oligomers. For such systems the ratio of the intra- to
the interoligomer exchange may be very large. Given the finite
length of oligomers. the results of the aforementioned doping
406
5.0
0
7
3.0
-
TcfK
2.5
-
4.5
4.0
3.5
1.5
-
1.0
-
0.5
-
2.0
0.0
0
0
0
0
'
experiments suggest a significant suppression of three-dimensional ordering.
8. Unusual Magnetic Properties of
[MnCp:]:+ IDDQCIJWith the goal of preparing additional molecular-speciesbased
ferromagnets,
the
ferroinagnetically
coupled
[FeCpi]'+[DDQCI,]'- electron-transfer salt was characterized
( 0 = 10 K)."" As T, is proportional to the spin magnitude
(Eqs. (10) and (1 3 ) ) , [MnCp;]'+[DDQ]'- was prepared anticipating that 7;. might occur at temperatures accessible in our
laboratories.[94] The high-temperature magnetic susceptibility
of [MnCpT]:+[DDQCI,]'- can be fit by the Curie-Weiss expression with a 0 value of 26.8 K, which suggests that the
strongest exchange interactions (the interactions within individual chains) are ferromagnetic. Hysteretic magnetic field dependent behavior. albeit complex, was observed below 2 7 K. The
150 to 2000 G magnetic field dependencies of the magnetization
for a sample previously aligned in zero field or in high field
(1 9.5 kG) is presented in Figure 18 for increasing and decreasing
magnetic fields. Above 2 3.8 K the magnetization exceeds the
expected value calculated from the Brillouin fiinction [Eq. (S)]
for fully aligned S = 1 and S = 1/2 spins, whereas dramatically
different behavior is observed below this temperature. Thus, the
data imply a complex magnetic phase diagram at low temperature (Fig. 19). If one assumes complete alignment of the crystals
with magnetic field parallel to the C , molecular axis. a samplehistory dependent saturation magnetization (M,) is observed
with values up to 24200 emuGmol-'. The data are consistent
with strong ferromagnetic coupling between adjacent radicals
within each chain and a net weak antiferromagnetic coupling
between the chains.['041This leads to metamagnetic behavior.
Thus, when an applied magnetic field is large enough. it becomes energetically favorable for the spins in all the chains to
align ferromagnetically. Below z 4 K there is an anomalous
behavior with large hysteresis and remnant magnetization.['041
For example, at 2 4 K the magnetization abruptly drops by
REVIRNS
Designer Magnets
Fig. 18. Stereovie% of the m o l x in;ignetization .I-I
as a function oftemperature for Iero-lield (blue
and cyan) and high-field (red and _ereen) coolcd
(previously ;iligned i n 19.5 kG) polycrystalline
sample of [MnCp:] [DDQCl,]'- The cyan a n d
green data were taken from decreiiwig applicd
fields. while the red and blue data \+eretaken from
increasing applied fields. (The actual field application aequence WIS 150. 300, 500. 1000. 2000. 1500. 750. 400, and 200 G prior to annealing at 25 K in zero field for 30 min and then applying 200. 400. 750. 1500. 2000. 1001).
acquired.
501). 300. ;rnd 1.10 Ci fields.)The pair of simihr cur\
t 2000 G ari5es from the two independent times the data
c~
+
more than an order of magnitude depending o n the applied field
to ii value lower than that calculated from the Brillouin function. At high temperature there is a field-dependent crossover
from a low-magnetization to a high-magnetization state. This is
suggestive of the presence of perhaps both metamagnetic and
possible lattice distortion (spin-Peierls-like) transitions.
However. since spin-Peierls transitions occur only in antiferromagnetic states, complex magnetic behaviors must be
operative in the material. Simple metamagnetic behavior
and
has been reported for [FeCpT]''[TCNQ]'- ["I'
(
M
=
Ni,
Pd,
Pt).[971
[MnCpTI'+[M(S,C,(CF,),) 4'
1500
111
ll
I000
HIG
500
netic-paramagnetic phase H I . A nonequilibrium phase diagram
wits derived by cooling the sample in a zero applied field above
8 K, followed by increasing the applied magnetic fields at constant temperatures (Fig. 19 b). This diagram shows the first-order boundary separating the different phases: the antiferromagnetic phase I, the intermediate phase 11b, and the ferromagnetic-paramagnetic phase 111. A nonequilibrium phase
diagram can be generated for the metastable ferromagnetic
state, which appears after the system is cooled from 10 K (in an
applied field exceeding 1500 Oe) to the desired temperature and
the field is decreased to the selected value and the temperature
is subsequently reduced. This phase diagram is also achieved if
a zero-field cooled sample is exposed to a sufficiently large external field (Fig. 19c). The field is then decreased t o the desired
value and the temperatureis subsequently raised. The ferromagnetic state is stabilized at low fields ( < 650 G) below 4 K by
using this
0
5000i
1
40001
1'
9. Room Temperature Polymeric Magnet
IV(TCNE),J * Y(CH,CI,)
b
Ill
Following the observation of bulk ferromagnetic behavior for
[MnCp:]:+[TCNE]'- we targeted the preparation of the
electron-transfer salt of [V(C,,H,)J
and TCNE.[47[V1(C6H6)J+,like [Mn"'CpT]'+. is an S = 1 cation possessing a
3E2p
ground state [a:&, ( zd:2d:ydi2 y.)] (Fig. 20).110s1
The addition of [V(C,H,),]' to an excess of TCNE in dichloromethane
results in an immediate black precipitate. The empirical composition of this material is [V(TCNE),] y(CH,C12) (.Y 2 2: J' z
1 / 2 ) . [ ' 0 6 ]However, due to the extreme insolubility ofthe precipitate. reactivity of the solvent, and extreme air and water sensi-
HIG
1,
----L,-
1000
0
,
,
,
, I,
,
T/K
,
;-;,,
,
8
4
3
1500
I
4
Fig. 19. Magnetic phase diagram (see
text) for [MnCpf] [DDQCI2]'- (from
ref. [104]).
+
On the basis of the experimental results three phase diagrams
were derived to describe the various transitions. The equilibrium
phase diagram for decreasing and increasing temperatures in
which the value of applied magnetic field is changed above a
three-dimensional antiferromagnetic ordering temperature of
8.5 K and is kept constant during the temperature scan is presented in Figure 19a. There are three distinct regions: antiferromagnetic phase I, mixed phase TI, and the coexisting ferromag-
[Mn"CpT]'
S = 112
[Vu( C,H,),J.
S = I I MniiiCp;I-+
or
[V1(C6H6),1:+
Fig. 20. Schematic illustration of the electronic configuration5 of [Mn"Cpf]'.
[Mn"'Cp:].', [V"(C,,H,),I'. and [V'(C,H,),I '
407
REVIEWS
J. S. Miller. A. J. Epstein
tivities. variations in composition as a function of preparation
conditions have been observed. The first step in the reaction is
electron transfer from [V(C,H,),]' to TCNE followed, unexpectedly, by loss of the benzene ligands. as is evident from the
IR spectra which lack vC-H absorption bands in the 3060 to
31 00 cm r e g i o ~ i . [ ~
4 y',~
The mass spectrum of the magnetic T
M
material has peaks assignable to CH,CI, and TCNE; however,
a peak at nz/z 78 assignable to C,H: was not observed. All of
the materials exhibit strong, broad absorption bands at 2099
and 21 88 cm which can be assigned to vCENor perhaps vN=c=c,
The breadth of the vCENabsorption bands and their relatively
low energy are consistent with the presence of reduced TCNE.
with its nitrogen atoms coordinated to vanadium. The oxidation
state of the vanadium or TCNE (Table 8) has yet to be determjned,[47.4Y.501
'
Tablc 8. Charge and spin state\ foi- V and TCNE a s a function of oxidation states.
Species
Oxidation state
V"
V'
V"
vlll
V'\
Vb
[TCN t]"
[TCNE]'
[TCNE]'~
Charge z
Spin State S
0
I+
Z f
3+
4 f
5+
0
I-
12
1
32
1
1 '2
0
0
1 2
0
?-
-
0
0 .
0
fi
O-
0
0.
0
7
-2000
-1000
0
1000
2000
H/Oe+
Fig 22. Hysteresis ( , M ( H ) )of [V(TCNE),] . i.(CH,CI,) iit room temperature. (The
data was taken on a vibrating sample magnetometer: T = 300 K , H , = 60 G (From
ref. [47]).
ganic-based material with a critical temperature above room
temperature. The critical temperature exceeds 350 K , the thermal decomposition temperature of the sample. A linear extrapolation of the magnetization to a temperature at which it would
vanish leads to an estimate of a
z 400 K. An independent
estimate to T, was obtained[481by using the empirical correlation between the spin-wave coefficient and T, for amorphous
From spin-wave theory and data taken below
60 K a spin-wave dispersion coefficient has been estimated for
and T, is estimated to be ~ 4 0 K
0
the V-based magnet.[*', 48.
in agreement with the linear extrapolation of M ( T ) .
[V(TCNE),] y(CH,Cl,) has a field-dependent magnetization
( M )between 1.4 and 350 K (Fig. 21). The nearly linear increase
0
50
100
150
200
250
300
TIK+
Fig. 21 Typical magnetization ( M ) [emuG mol- '1 as a function of temperature ( T )
at 15.X (4). 2.0 (*). 1.0 (A). 0.5 (v), and 0.15 (F) kG for [V(TCNE),] .j(CH,CI,),
1.0 kG ( 0 ) for [V(TCNE),] -r(MeCN). and 1.0 kG ( 0 ) for [V(TCNE),]. y(THF).
Fig. 23. Photograph of a powdered sample of the [V(TCNE),] j(CH,CI,) magnet
being attracted to a Co,Sm magnet.
'
of M with decreasing temperature is unusual and may reflect the
contribution of the two spin sublattices (V and TCNE) and/or
the effects of disorder. Hysteresis with a coercive field of 60 G
is observed at room temperature (Fig. 22). The strong magnetic
behavior of this material is readily observed by its attraction to
a permanent magnet at room temperature (Fig. 23). Thus, this
system is the first and only example to date of a molecular/or408
[V(TCNE),] . y(so1vent) has been prepared with a variety of
S = 0 solvents including THF, acetonitrile (MeCN), diethyl
ether, benzene, 1,2-dichloroethane, and hexane; CH,Cl,, THF,
and MeCN have been most extensively studied. The magnetization varies dramatically with the choice of solvent, and
sysAngew. C/io?z.I n [ . Ed EiiEllgl. 1994, 33. 385-415
Designer Magnets
REVIEWS
tematically decreases with the ability of the solvent to coordinate with the vanadium center (vide infra).
Due to the structural disorder and variable composition of
the magnetic material, the detailed structure has yet to be elucidated (vide infra). TCNE may bond to metals in many ways.
For early transition metals linear o r bent V-N CJ bonds are
anticipated. Presently we formulate each vanadium center as
being surrounded by up to six ligands which are primarily nitrogen atoms from different TCNE moieties. Chlorine atoms from
the weak CH,CI, ligand or from oxidative addition of CH,CI,
may also coordinate (Fig. 24). Any trace oxygen that is present
will strongly bond to the vanadium. The TCNE ligands may
bind up to four different vanadium centers through CJ N bonds.
This fragment may be planar or twisted (Fig. 24). Nonetheless,
its ability to bind to more than one vanadium atom enables the
construction of a three-dimensional network structure that supports the strong three-dimensional spin-spin coupling necessary
for a
of 400 K. By using T H F or MeCN as alternative solvents. materials with a T, < 350 K can be isolated and characterized. Given the geometrical constraints imposed by the coordination of TCNE to a metal, an open structure with unfilled
vanadium coordination sites is unexpected, and thus would be
consistent with the observed high surface area of ~ 1 0 m0 2 g - ' ,
which suggests internal porosity on the molecular level.
\
i
planar
twisted
system. Based upon the IR data and elemental analysis, the
composition of the precipitate is presently best formulated as
[V"(TCNE),] .1/2(CH,CI2) with S = 3/2 V" and two S = lj2
[TCNE]' ligands. Ferromagnetic coupling gives a total spin
S,,,,, of 5/2, and if one assumes that g is 2, the saturation magnetization ( M , ) is expected to be 28 x lo3 emuGmol-'. For antiferromagnetic coupling between V" and the two [TCNE]' ligands, and hence ferrimagnetic behavior, the S,,,,, is 112 and
M , is expected to be 5.6 x lo3 emuGmol-' ; the latter is in good
agreement with the value of 6 x lo3 emuGmol-' observed at
2 K and 19.5 kG.[471
The origin of the antiferromagnetic exchange between the
spins is important for the design of future high-temperature
molecular-based magnets. A variation of the CI mixing of virtual triplet excited states with the ground state to produce a ferromagnetic coupling model (mechanism 2.2.1.1 (Scheme 1 : Section 3.2.2.1 .I), proposed to account for the ferromagnetic
exchange in [FeCp:] [TCNE],[44h1
can account for the antiferromagnetic exchange of [V(TCNE),] .y(solvent) and the resulting
ferrimagnetic state with a high T,. The configurational mixing
suggested is that of a virtual singlet excited state with the ground
state for the V[TCNE], system, in which the spin on the V"
center (S= 3/2) and that on the acceptor (S= 1/2) itre unequal
and produce ferrimagnetic
The proposed bonding of the TCNE nitrogen atoms to the vanadium center leads
to a sizable orbital overlap; thus, this mechanism should produce a sizable exchange.
The degree of randomness in [V(TCNE),] y(solvent) appears
to increase for the CH,CI, < T H F < MeCN prepared materials, respectively.[11o1This is probably due to the increasing ability of the solvents to coordinate with the vanadium center which
in turn displaces some of the TCNE-based radicals and limits
the connectivity between vanadium sites. The random anisotropy then leads to the formation of a wandering axis ferrimagnet or even a correlated spin-glass (vide infra)." ' I These
types of local order are shown schematically in Figure 25 and
may be contrasted with the more rigid ordering indicated in
Fig. 24. Proposed local bonding about each TCNE and V center in
[V(TCNE),] y(CH,CI,).
Attempts to prepare magnetic materials from the reaction of
[Vo(C,H,)(C,H,)]['091 or [V"(C,H,),] and TCNE or with
~ o ( C , H , ) , ] and other acceptors such as perfluoro-TCNQ
(TCNQF,), C,(CN), , C,[C(CN),],, 2,3,5,6-tetrachlorobenzoquinone (chloranil), DDQCI,, and 2,3,5,6-tetracyanobenzoquinone (cyanil)['*] lead to the isolation of insoluble precipitates of unknown composition. These materials do not exhibit
field-dependent magnetic susceptibilities and their high-temperature susceptibilities can be fit to the Curie-Weiss expression
with 0 2 -1 K characteristic of weak antiferromagnetic coupling between spins. Likewise, in an attempt to prepare the Nb
analogue of the V-based magnet, TCNE was allowed to react
with [Nb(C,H,Me,),]; the rapidly formed precipitate exhibited
weak antiferromagnetic coupling.
Although the structure is as yet unknown, it is interesting to
speculate as to the type of magnetic coupling present in this
Angeu
Cliein. Inr. Ed. EnRl 1994. 33. 385-415
Fig. 25. Schematic illustration of
wandering axis ferrimagnet (top)
and correlated (ferrimagnetic)
spin-glass (bottom).
409
.I.S. Miller, A. J. Epstein
REVIEWS
Figure 3 d for a classical ferrimagnet. For the case where the
randomness is small compared to the uniaxial anisotropy, a
wandering axis ferrimagnet is obtained. This differs from the
usual ferrimagnet as the spins d o not completely align even at
relatively large applied fields. However. when the random anisotropy is larger than the uniaxial anisotropy the average direction
the spins point in varies over a spatial correlation length.
Extensive physical studies have now been carried out on the
[V(TCNE),] . J(CH~CI,)system as well as on similar materials
prepared from T H F and MeCN solvents. The results of temperature- and magnetic-field-dependent magnetization.1108, 21 ilC
susceptibility.[' 1 3 ] electron paramagnetic resonance."
I1 4 ] d c
and audio frequency (10' through 10' Hz) conductivity and
dielectric constant.[' "I and microwave frequency conductivity.
dielectric constant, and permeability"". ' I 6 ] lead to a self-consistent picture of the [V(TCNE),] )'(solvent) systems a s being
dominated by the presence of weak random anisotropy." ""]
That is. the magnetic phenomena are dominated by the effects
of variation of magnetic interactions from site to site due to the
solvent and disorder in the system.
Extensive M ( H , T ) studies of [V(TCNE),] y ( T H F ) and
[V(TCNE),] . .r(MeCN), in addition to studies of [V(TCNE),J
. y(CH,Cl,) provide strong support for the key role of random
anisotropy. The [V(TCNE),] . y(CH,Cl,) system has a in excess of room temperature. The T, for the materials prepared in
T H F and MeCN are 200 and :
140 K , respectively (Fig. 21),
although this transition temperature is sample-dependent and
varies with solvent content and structural order. The ac susceptibility studies suggest that the most disordered of these materials [V(TCNE),J . ?,(MeCN)is a reentrant spin-glass with a spinof = 10 K (again the transition
glass freezing temperature
temperature varies with solvent content and structural order).
The materials prepared in T H F and MeCN have only weak or
absent coercive fields in accord with the greater randomness.
The results of an equation-of-state analysis for the materials
prepared in MeCN in the vicinity of the three-dimensional ordering temperature T, :
138 K are shown in Figure 26. Here the
magnetization data for all H and Tstudied in the vicinity of 7;
collapses to universal curvcs for T > and T < T,. The values
of the effective critical exponents are in accord R;ith the random
anisotropy model." ''I
The EPR spectra of [V(TCNE),] ).(solvent) have been
recorded for CHZC12.T H E and MeCN solvents.[""] Thc rich
set of temperature-dependent spectra obtained provide an insight into the magnetism. A typical derivative spectrum at room
temperature for [V(TCNE),] . I(CH2CIZ)is shown in Figure 17.
1.5
1.0
c
0.5
,
T
0.0
-0.5
-1.0
-1.5
3100
3200
3300
3400
3500
3600
3700
3800
HIG+
(c)
-5011
++/
-
5
5
10
15
20
25
30
Fig. 26. Scaling plot of M ( 7 H ) for [V(TCNE),] . y(MeCN) The ordinate is the
7c)~'7,/]raised
iiaturiil logofMdi\.ided bythereduced temperature/[= ( I t 1 = l ( 7
to the /I power: the .lbscissa 1s H divided by Pi.uhere /j and fi are critical exponents
Note that the data coll.tpce onto the uni\eraal curves. the upper one for T < T, and
the lower one o f 7 > 7, ( i - d . [112a]).
~
410
-400
-200
0
200
400
Fig. 27. Top: Derib'itive EPR sifniil ks. f/ f o i - [V(TCNE),J i.(CH,CI,) iil 295 K .
/ = inlensit) (arbitrai-) units). B o t t o m Near zero-lield E P R vipnal of
[V(TCNE),] . J (CHICII)iit 200 K (from re(. [ 1141). Yote the greatly expanded vertical scale. The ahcissii i v i n units of Ciau\i. the ordinate I S the derivative of EPR
intensity in iirhitrary u n i t \
Four distinct resonance signals are observed: 1 ) the main resonance with ,p = 1.92, 2) an approximately half-field resonance,
3) a broad near-zero-field resonance. and 4) a narrow rero-field
antiresonance." "I These resonance signals vary systematically
with the composition ofthe sample (including differing solvents)
and the temperature. The latter two signals at 200 K are more
clearly shown in Figure 27b. The integrated intensity of the
main resonance for each of the compositions studied is proportional to the measured temperature-dependent dc magnetization, hence it is associated with the ferrimagnetic resonance.
Contrary to the usual magnetic systems, the linewidth reaches a
minimum at T,. with a critical behavior mimicking the critical
behavior of M ( T ) . as expected for a material with "sloppy"
waves (spin waves with wavevector q > where is the correla-
<.
<
REVIEWS
Designcr Magnets
tion length).[l"' The shift of the g value of the main resonance correlates with the magnetization and agrees in
magnitude with the expected values for demagnetization effects.
The broader low-field signal is largest in the materials prepared i n CH2CI,, weak in the systems prepared in T H E and
virtually nonexistent in the materials prepared in MeCN. The
peak-to-peak width and its sample variation suggest that this
low-field signal is similar to that obscrved for a domain wall likc
resonance. The sharp zero-field signal is unusual. It is present in
each of (hc compositions below T,, though for the samples prepared in MeCN it evolves into ;I resonance signal from an antiresonance below 2 t 5 K. H temperature associated with
freezing o u t of the spins into ii spin-glass phase.["31 Based on
this tempcrature dependence. this antiresonance may be caused
by a Ion-field magnetoresistance which makes it more difficult
for electrons to hop among charge carrier sites (presumably
TCNE), thereby reducing the sample absorbance of microwaves
a t higher magnetic fields.
The [V(TCNE),] . ?(solvent) materials are semiconductors
which have moderate conductivities of the order lo-' to
Sen-' a t room temperature depending upon the sol[ I 1 5 , I I61 In
general. the dc conductivity varies as
exp[ - ( T , ;T)',"], which is reminiscent of Mott variable-range
hopping." "I However. the strongly temperature-dependent frequenc!. dependence suggests that the behavior is more complex.
This effect was suggested to be a ramification of the correlated
spin-glx+ behavior of the [V(TCNE),] ,?(solvent) materials." 1 5 . 1 I (?I The charge transport probably involves charge hopping among anionic [TCNE]'- sites.
The I-ccults o f the physical studies reinforce the importance of
three-dimensional coupling in the [V(TCNE),] . >(solvent) system. Because of the three-dimensional network structure
present. Equation (10) (T'ible 7) may be used to provide an estimate o f the effective exchange interaction J,,, operative in the
room temperature magnet system (solvent = CH,CI,). If one
assumes ii T of 400 K. an average of five nearest neighbors, and
for S ( S + 1 ) the root-mean square of S,, = l;2. and S, = 313
(i.e.. [ ( 1 ' 2 ) ( I '2 1 ) ( 3 2 ) ( 3 ? + I)]'"). a value of J,,, of 70 K
( 5 0 cm I ) is obtained. This is less than a factor of three larger
than the intrachain exchange (Section 5 ) in [FeCpT][TCNE] for
which T = 4.8 K : yet Tcis nearly one hundred times larger. The
differences in transition temperatures do not correlate with the
difference in Jellalone, but also with the change in dimensionality from a quasi one-dimensional (for which Equation (14) is
operative) to essentially a three-dimensional system (for which
Equation ( 1 3) is operative). Hence. we conclude that to achieve
a high T in a molecular-species-based system it is a significant
advantage to increase the dimensionality and connectivity of the
spin network as well as J.
Thc solvents are suggested to effect the magnetic state
through several mechanisms. These mechanisms include
I ) reducing the number of nearest neighbor spins surrounding
the vanadium ion. 2) reducing the connectiveness among the
vanadium ions by the multicool-diiiating molecular unit (thereby effectively reducing the dimensionality of the magnetic state),
and 3) effecting the degree of crystalline order around each of
the vanadium ion sites (thus introducing random anisotropy
and also random exchange).
10. Metalloporphyrin-Based Magnets
We recently reported"
that [MnTPP]::+[TCNE]'- .
2 PhMe
(TPP = meso-tetraphenylporphinato) represents
a
new
structural
class
of
molecular
magnets.
[MnTPP]::'[TCNE]'2PhMe crystallizes a s parallel onedimensional ' . . [D]::' [A]'- [D]::+ [A]'- . . ' chains in which
the [TCNE]'+ binds identically to two [MnTPP]::+ (20) moieties
in a r,n,is-~c2-N-o-bound fashion
(Fig. 2821). The solid-state motif is
13).[A]'sinceelectron-transfer
than
[A]'- does
that not
for
the
distinctly
salts[MCpS]
(Fig. different
+
coordinate to the M in the latter system. Thus. the bonding of reduced
TCNE to Mn is a model for the
bonding of reduced TCNE to V in
the [V(TCNE),] . ?.(solvent) magnet.
g\
Nm~sMnoa~N
.
t
N
-
/
/
\
/
20 MnTPP
+
~
.Iil,~l'Il.
< /W/Ii.
/ill.
IX
€/1$/.
1994. 33. 385-415
Fig. 28. Scgmcnt of a one-dimensional . . . D ' A' D ' A ' . . chain 4iouing
[TCNE]' f ~ ~ m - ~ i , - N - ~ - b o n dtoi l(a)
l g [MnTPP]::+ (the toluene mnlcculc\ ol.solv,ition are not shown for clarity) and (b) [MnOEP]::'. Distances i n .&.
The susceptibility for [MnTPP]::' [TCNE]'+ . 2 PhMe can be
fit by the Curie-Weiss expression between 115 and 250 K
(0 = 61 K) and above 280 K ( 0 = -15 K). A minimum in the
value of Z T characteristic of one-dimensional ferrimagnetic behavior (vide supra)i27". 201 ISobserved at ~ 3 1 K0 , and field-dependent susceptibility is observed below 50 K. The saturation
magnetization M, of 30000emuGmol-' is in good agreement with the expected saturation magnetization of
41 1
J. S. Miller, A. J. Epstein
REVIEWS
'
29900 emuG mol- if one assumes ferromagnetic ordering of
the 5' = l / 2 [TCNE]'- spin and S = 2 [MnTPP]"' spins. A hysteresis curve with a coercive field of 375 G was obtained at 5 K
(Fig. 29), which demonstrates that [MnTPP] [TCNE] is a magnet at low temperature. Extrapolation of the steepest slope of
12000,
gooo~
terial does not exhibit any evidence for cooperative magnetic
ordering down to 2 K. The differences in the magnetic properties lie in the structural differences. Although both
[MnTPP]::+[TCNE]'- and [MnOEP]::+[TCNE]'- form parallel one-dimensional . . . [D]::+ [A]'+ [D]::+ [A]'- . . chains as
noted in Figure 28. they are dimeric instead of uniform for
[MnOEP]::'[TCNE]'-.
Also the rr*[TCNE]'- orbitals interact
differently with the Mn"' centers. Thus, uniform chains appear
to be important to achieve long-range magnetic order. This
necessity has also been identified to achieve high, metallike dc
electrical conductivity for one-dimensional molecule-based
metals such as those based on TCNQ.[451
11. Epilogue
-av
i
-3000
-9000
Since the initial reports of three-dimensional cooperative
metamagnetism in [FeCpT] [TCNQ]'77"1and ferromagnetism in
[FeCp;] [TCNE],['241there has been very rapid progress in the
development of materials. models for the origin and control of
the spin-exchange interaction. theories for the origin, and details
of the observed magnetic behavior such as T, and saturation
magnetization ( M J The chronology of increases in T, is compared to that of the organic and ceramic superconductors in
Figure 30. From these first examples of materials exhibiting
-12000
-3000-2000-1000 0
1000 2 0 0 0 3000
Fig. 29. Hystcresis data at 5 K for polycrystalline [Mn"'TPP]+[TCNE]'~. ZMePh.
(Recorded with a SQUID magnetometer: SQUID = superconducting quantum intcrference device).
the 3 G M ( T ) data to zero yields an estimate of a magnetic
ordering temperature of 18 K . A much weaker second transition is observed at 40 K which may reflect ordering to a
canted spin state or other complex magnetic behavior. Thus,
[MnTPP] [TCNE] with the tvans-p2-N-o-bonded [TCNE]'- is
an example of a new structure type of molecular-species-based
magnetic material. The presence of magnetic hysteresis and a
relatively high T, value for such a structurally (and probably
magnetically) one-dimensional material suggests the possible
role of random anisotropy o r other disorder in the production
of these "ordering" effects." * l a ]
As a consequence of the alternating S = 2 and S = 11'2 chain
structure, the [MnTPP]::+[TCNE]'+ . 2 PhMe one-dimensional
chain is an excellent model system for studying a number of
unusual magnetic phenomena, for example, the magnetic behavior of mixed quantum/classical spin systems.[' 'I Thus, for
example, [MnTPP]::+[TCNE]'+ 2 PhMe may reveal a compensation phenomenon upon further
To extend this system the
analogous TCNE electron-transfer
complex was prepared with
the easier to oxidize MnOEP
Et
Et
(OEP = octaethylporphine), 21.
Nw~~M~~--MN
[MnOEP]::+ [TCNE]'
exhibits
Et
weak
ferromagnetic
coupling
as ev\ N /
idenced by the fit of the susceptibilEt
El
ity to the Curie- Weiss expression
with a H of 7 K . However. this ma21 MnOEP
-g,
41 2
Fig. 30. Time evolution of the discovery of increasing critical temperatures, 7,.for
molecular-based magnetic materials as well as organic and oxide superconductors.
ET = bis(ethylenedithio)tetrathiafulvalene.
cooperative magnetic ordering, the field of molecular-speciesbased magnets has evolved. It now includes a wide range of
phenomena such as magnetism in molecular-species-based materials at temperatures significantly in excess of room temperature. As this research area has evolved. based upon the models
of spins in orthogonal orbitals in the same spatial region within
a molecular species in order to achieve intramolecular spin coupling and high-spin behavior. the pairwise configuration interAnjiru. Cliein. lnt. E d EngI. 1994, 33, 385-415
REVIRNS
Designer Magnets
action between molecular species and/or dipole -dipole,
through-space interactions to achieve intermolecular spin coupling. several chemical features have emerged as important considerations in designing new magnetic materials. Clearly, to prepare a molecular-species-based magnet, both the donor and
acceptor, if present. must be radicals. Such radicals need only
have one spin per site; however, a greater number of spins per
site is expected to lead to materials with a higher T,. For donor/
acceptor-based systems the competition between ferromagnetic
and antiferromagnetic interactions arises from D . . . D. D . ' A,
as well as A . . . A interactions. Subtle changes in the orbital
overlaps presumably lead to substantial changes in magnetic
coupling. Thus. as is the case with proteins. the primary, secondary, and tertiary structures are crucial for achieving the
desired cooperative magnetic properties. Currently the discovery of new molecular-species-based magnets is limited by the
rational design of solid-state structures which remains an art.
This is due to the formation of numerous polymorphs, complex
and solvated compositions, as well as undesired structure types.
The growth of crystals enabling the study of the single-crystal
structure and magnetic properties is also an important limitation. New radicals, donors, and acceptors as well as new structure types are necessary for this area of research to develop
further. Finally. one needs to understand the frontier orbital
overlaps as they contribute to the interchain and intrachain
coupling. Given the present rapid growth in this field it is clear
that major advances will be occurring over the next decade in
this new multidisciplinary branch of solid-state science.
We IIUIV hcvic.fited greatlj~.fromthe stimulating interactions and
ii'itli our colkagues, postdoctoral felcontinctctl coll~rhorcitioi?,~
l o w , mid . s t i r d m t s ,fbr they have made iniportant contributions
enuhling tlie r(ipid progress q f [he work reported herein. Special
thunks ure e . u t e i ~ l e dto L. Ahrarns, A . Bohni, M. J. Burk, J. C.
Crcluhrcw, F Chen (Houston), C. W. Cliu (Houston).K. R. Cronzuck, D. A . Dison. G. Du, A . Grand (Grmoble), R. Hqffinann
ICornc~ll).Z . J. Huung (Houston), G. A . Jones, J. Joo, N . J.
Jones, M . Laridjani (Orsay),L. Lardear, S. Long, W. Marsliall,
R. S. McLcan, B. G. Morin, K. S. Nurgyan, Z . Oblakowski,J. P.
Pouget (Orsuj-), Y 7: Ren (Houston),E. Ressouche [Grenohle),
J. Scliwizer iGrenohle), A . Suna, A . L . Tchougreejy (Cornell),
c'. Var_yurz,Y Y Xue (Houston),S. Zcnie (Miami), A . Zheluclev
iGrenoh/cJ, P. Zhou, and E: Zuo (Miami) ,for allowing us to
rcport i r w k in progress. The autliors,furtliernzore grat<fullj~acknowledge flic support that the Deppartment of Encrgj%,Division of
Materitrl.5 Scimicrs (Grcmt Nos. DE-FG02-86BR45271 and
DE-FG03-93ER45504) has given ,for these s1udie.s.
Received: December 2. 1992 [A 929 IE]
German version: Angel1 Chcin. 1994, 106, 399
[I] a ) A. Helleinans. B. Bunch, Time/uhlc>.sof Sc/eiice. Simon and Schuster. New
York. 1988: b) [ l a]. p. 19, c) [I a]. p. 61.
[?] .h.lugiier\ iii Your Future 1991. 5 (6). 22; W. G. Hart. Inti~rterh(hiiferencr on
Poli~i,icr-Botirlr~rl
Mugiicts '92. Rosemont. IL. USA, unpublished. and W. G.
Hart. personal communication.
[3] R. Wood. L'ndwsmiiding Mugnetrn?, Tab Books. Blue Ridge Summit. PA.
USA, 1988. Chapter 2; D. C. Mattis. The Theory of Mugnetism, Springer.
N e u York. 1981. Chapter 1 : A. L. Robinson. Scimce 1984223,920; K . H. J.
Buschou. . M U I P ~Sci.
. R e p 1986. 1. 1; M. Cohen, A d ~ u n r i n gMu/errul.\ Re.rcurc.li (Eds.. P. A. Psaras. H. D. Langford). National Academy Press. Washington. DC. USA. 1987, p. 91: J. 1. Croat, J. F. Herbst, MRSBuN. 1988, 1 3 ( 6 ) ,
37. R. M. White. S c i m c ~1985.
~
229, 4x07: W. E. Wallace. J. Lix-C'onimon
, V c , / . 1984. 101).H5: P. Chaudhari. M. Flemings. Mutcriu1,s Scwnce unrl G i g i ~ ~ " w I , f ~o r tli<, 1990.s, National Research Council. National Academy Press,
W:ishington. DC. USA, 1989. p. 94.
Angcn. C ' I I ~ I I I/lit.
. bd Engl.
1994. 33. 3x5-41 5
[4] a) Proi. Con/:Ferroniugnetic und High Spin Moleculur Bused ?Motcriul,\ (Eds.:
J. S. Miller, D. A. Dougherty) ( M o l . Cry.rt. Lfq. c'rysr. 1989. 176); b) Pro(
Conf. Molecular Mugneric Muteria1.s: & A 1 0 A R W Molecrrlur Mugiiel!( Mritcwiuls (Eds.: 0. Kahn, D. Gatteschi. J. S. Miller, F. Palacio). Kluwer, London, 1991. E I W : c) Proc. I n / . Symp. on Chemistry und Plly.cic,\ o/ Molci~ulur
Birsed Mugneric Mutrriuls (Eds.: H. Iwamura. J. S. Miller) (Idol. C r n t . Lry.
Crvsr. 1993, 232, 233).
[5] T. Y.Kuromoto, S. M. Kaurlarich, D. J. Webb, Clitwi. Muri,r. 1992, 3. 435.
[6] a) P. Day. Ace. Chein. Rrs. 1979, 14, 236; C. Bellito. P. Day. J. !Mafrr. Clion
1992.2.265. b ) Some A,CrX, materials have been prepared by low-temperature solution chemistry.
[7] J. S. Miller. A h . Murer. 1992. 4. 29X. 435.
[8] Reviews: a) A. L. Buchachenko. Rirc\. C h i Rrv. 1990. 59. 307: L.\p. KIiini.
1990. j9. 529: b) 0. Kahu, SirucI. Bonding Berlin 1987, 6R. 89: 0.Kahn. Y.
Journaux. in lnorgunic Muteriuls (Eds.. D. W. Bruce. D. M. O'Hare). Wile).
New York. 1993, p. 59; c) A. Caneschi. D. Gatteschi. R . Sessoli. P. Re). ACT.
Cheni. R r i . 1989.22.392;d ) L. Dulog. ,Yoc/ir. Cheiii. f i ~ hLoh.
.
1990.3K.44X:
e) H. Ishida, Encjl. Poljiii. Sci. Eng. 1985- 1989, p. 446; f ) T. Sugawara. Y d i
Goser Kuguk~rKlokurshi 1989. 47. 306; g) J. S. Miller. A. J. Epstein. W. M.
Reiff. Arc. Clirm. Res. 1988.21. 114: h) J. S. Miller, A. J. Epstein. W M. Reiff.
Science 1988. 240. 40; i) J. S. Miller. A. J. Epstein in h ' w A\pri.l.\ o/
Orgunir C/wii/i.s~r.v(Eda.: Z.Yoshida. T. Shiha, Y. Ohsiro). VC'H. New York.
1989. p. 237; j ) A . J. Epstein. J. S. Miller. Mol. Cri,,\/.L i q C'rv,s/. 1993.
228. 99.
[9] J. S. Miller, A. J. Epstein. W M. Reiff, Chtw. Rev. 1988. 8 N . 201
[lo] F. Palacio, F. J. Lazaro. A J. Van Duyneveldt. Mol. CnJr. Liy. ( ' r w . 1989
176, 289: F. Palacio in [4b] p. 1 : F. Palacio, J. Ramos. C. Castro. .Mu/. C r W
Liq. c'rysr. 1993. 232. 179.
[Ill D. A. Dixon. J. S. Miller. J. Ain. C/ieni. Soc. 1987, /fJY. 3656
[I21 a) A. Carrington. A. D. McLachlan. Introducirori io Mugiwtic Ke.soiiancc~.
Harper & Row. New York, 1967: b) [12a]. p. 223ff: c) [IZa]. p. X I X3:
d ) R. S. Drago. Physicul Mefhods of Cheini~rrr.Saunders. Philadelphia. PA,
USA. 1977. p. 329ff; e) P. Coppins. AIWIU.
Rrv. P/iy,s. Clwn. 1993, 43. 663.
1131 a ) Note that by allowing more freedom of the orbitals. a I o w r energy can he
reached with more relaxation o f t h e positions of the electrons Although (5'')
is a good quantum number for the R O H F wavefunction. i t IS not for the U H F
formalism: so the latter formalism must he used with caution. h) A. Szaho.
N. S. Oatlund. M o d e m Qiiaiitiim Chcwiistrj~:I i i t r o i l i ~ c r i o i 10
i Adruiicrrl € / P I rronic Sti-iri rirre Theory. MacMillan. New York. 1982; c ) W. J. Hehre. L. R;Idom, P. von R. Schleyer, J. A Pople. A h Iiiitio * M o k iulur Orhirul Tlwiir?,,
Wiley. New York, 1986.
[I41 a ) J. W. Andzelm in Densrh. F W I ~ / K JMrthods
~U/
m Clicwr.s/ri~(Eds.: J. Labanowski, J Andrelm), Springer, New York, 1991. p. 101 :J W Andrelm. E J.
Wimmer, Clicm. Plitn. 1992. 96. 1280: P/?yc.rcuB 1991. /72. 307. DGauss is ;I
local density functioiial program available from the Ci-a? Unichem Pi-ojcct.
b) A. Zheludev. A. Grand. E. Reswuche, J. Schweitzer. B. G Morin. A. J.
Epstein. D . A. Dixon, J. S. Miller. Angew. Chem. 1994, 106. A n p i i . Clwiii. I n / .
€d. Engl. 1994, 33. in press.
1151 M . S. Davis. K Morokuma, R. W. Kreilick. J. Am. Chem. .So<,.1972. 93. 5588.
1161 a ) ~ i s i n u n ~ t s o f e m u m o l ~ ' aTni ds i n K . b ) F o r a s y s t e m w i t h S = 1 : 2 . g = 2
the magnetic energy at 10" G ( 1 T) is 0.7 K 0.5 c m - ' . 62 KeV.
[17] H. H. Wickman. A. M. Trozrolo. H. J. Williams, G W. Hull, t R. Merritt.
P h p . Res. 1967, 155, 563: H. H. Wickman. J. Cliiwi. P/?ys. 1972. 56, 91h:
G. C. DeFotis. F. Palacio, C . J. O'Connor. S. N. Bhaatia. R. L. Carlin, J A m
Chon. Soc 1977. 99. 8314: N. Arai. M. Sorai. H. Suga. S. Seki. J. PI7ys. C%IW.
Solidy 1977. 36, 1231
a) S. M . Heald. E. A. Stern, B. Bunker, E M . Holt. S. L. Holt. J. Anr. C'hcm.
Soc. 1979. 101. 67; b) S. Chikazumi, Phvsics of iMugnc~ti.sni. Part 3. Wiley.
New York. 1964; c) A. H. Morrish, The Pliysicul Princ/phr of .Wugtwri.~m,
Wiley, New York, 1965; d) [~XC],Chapter 7: e) [IXc], p. 341 : f ) J S Cook.
P. L. Rossiter. Crit. Rev. Solid Siclie Murer. Sci. 1989. 15. 509
R. B. Frankel. R. B. Blakemore. J. Mugn. Mugii. Murt,r. 1980. l J - I X , 1562;
R. B. Frankel. G . C . Papaefthymiou. R B. Blakemore. W. O'Brian. Riodinu.
BJophy.5. Aclu 1983. 763%
147.
D. B. Awschalom. D. P. DiVincenzo. J. F. Smyth. S c i f i i i . ~1992. 25X. 414.
S. Mitra. A. K. Gregson. W. E. Hatfield, R . R. Weller. Inorg. C'heni. 1983, 22.
1729.
Generally rhrec-dimensional interactions are necessary: however. in bomc
circnmstanceS two-dimensional and even one-dimensional interactions for
alternating quantum (S = 1/2) and classical (5' > 1 2 ) spins. hysteretic effects
can occur. See for example. R. A. Serota, P. A . Lee, Phyr. Rrv. R 1986. 34.
1x06; R. Dickman. E. M . Chudnovsky. ihid. 1991.34. 4397.
a ) J. H . Van Vleck, The Theory of' E l w t r i c und Mugtietic Sr~si~i~~~tihilirirs,
Oxford University Press, London. 1932; b) J. H. Van Vleck. Rrr. Mod. PI?),.\.
1945, / 7 , 7. c) J. H. Van Vleck. ihid. 1953. 2.7. 220: d ) J. B. Goodenough.
Mugngneri.sm uiid the Chemrcul Bond. Wile?. New York. 1963: e) R. L. Carlin.
Mugneruchemistry, Springer. Berlin. 1986: f) C. Kittel, Inlroilirction l o S o l d
Sfate PIiisics. 5th. edition. Wiley. New York. 1976.
D. A. Dixon. A. Suna. J. S. Miller. A. J. Epstein in [4b] p. 171
S Chittapeddi, K. R. Cromack. J. S. Miller. A. J. Epstein, P h i ~ R. e v . L e t / .
1987.58, 2695.
41 3
REVINVS
[26] Z G. Sons i n P/ij,,sic\uwrl C/ioiiiiwi, o/ L o i ~ ~ - ~ i ~ ~ i e i i ..Solid\
~ r o i i(Ed.
u/
L. Alcaccr). Reidel. Dordrecht. 1980, p. 143: G . Beni. P. Pinem. D. Hone. f / i j , , \ ,
R n , . 5 1973. 8. 3389: C. Lyon-Caen. M. Cyrot. J P/ij..i. C. 1975. 8 , 2091
[?7] a) 0 .Kahn. Srriwr. Boiiiliiig 1987, 68, 89: h) 0. Kahn. An,arii~C/iciii. 1985. 97.
837: A i r g m C/?[,in./I?/. Ed. Ei?,q/.1985. 24. 834.
[?XI H . Tamaki. Z. J. Zhong. N . Matsurnoto. S. Kida. M. Koika\bt;i, N. Achiwii. H
Okiiua. J. Aiii. C%eiri Soc. 1992. 114. 6974.
[29] ti) S . Morinioto. K . Itoh. F. Tanaka. N . Mataga. Pwlwinrs of r / r c Sj,ilipowiiri
011 2Mo/wii/ur.Strut.rur(2,Tokyo. 1968. p 6: h) N. Mataga. T/Iw. ('/11111. Acre
1968, 10. 372.
I301 R. L. Carlin, ~~~[i,qri[,rii[,/r[,iiii,s/ri, Springer. Ne\\ York. 1986. p. 70-71.
[31] W. E. Hatfield in Tiicoi-I arid A/ip/i~~rio17.\
of M o l ~ ~ ~ ~PIirliililigii[,ri.siri
iiIu~:
(Eds.:
E. A. Boudrcaux. L . N . Mulay). Wiley. Neii York, 1976. Chapter 7.
1321 a ) H . Iwiimurti. P i w App/. C/imi. 1986. 58. 187: T. Sup;iii;irii. 5. Muratii. K
Kimura. H. 1w;unur:i. J. A m C ' / i w r . So<. 1985, 107. 5293. T Suga\\ar;i. S.
Bandow. K. Kiiniira, H . Iwainura. i h i d . 1984. 106. 6449: T. Suga\iar;i. S.
Bandow. K . Kiinura. H . Iwamui-a. K. Itoh. ihid. 1986. /OK. 368: b) N. Nakaniura. K . Inoue. H Iwamui-a. T. FtIJiOka. Y. Sawnki. [ h i d . 1992. 114. 14x4:
c ) H Iwamura. N. Nakamura. N . Kogii. S. Sasaki, Md. C'i,v,\r, Liq. C'rj ti.
1992. 718, 207: t I Iwamura. Pirrc. .4pp/. ( ' h i i . 1993. h.i> 57. N . Nakaniura.
ti. Inoue. A. 1\4amui-a. .Aii,yeii. Chnii 1993. 1/15. 91111. A r i , ~ m C. h i i . l i i t . Ed.
h g l . 1993. 32. X71
1331 A . L. Buchachcnko. Mol C'irinr. Liq C'ri \r. 1989. 176. 307.
[34] n) K Nasu, Plij5. Rcr. H 1986, 3.3, 330: b) K . Ynshizaua. A . Chano. A Ito.
K . TAnaka. T. Yamahe. .Svrrh. ,b'r,r. 1992.i2. 377: c) J A . Thomas, C. J. Jones.
J. A . McCleverty. D. Collison, F. E. Mahhs. <' J. Hiirding. M . G. Hutching.
J. C/u~i11.
Sor . C/icwl. Coiiii1iiiii. 1992. 1796.
1351 A A. O ~ h i n n i k o \ ,I%~,or.C/iii?i A ~ i o1978. 47. 297.
1361 D. J. Klein. C . J. Nclin. S. Alexander. F. A. Matsen. .I C ' / i m r . Phi:s. 1982. 77.
3101.
[37] A R a p . S. Uratnapanya. S. Thayumanavan. .I A m . C ' / i w r . Soc. 1992. 114.
18x4: S. Utainapanya, A. Rai,jca. i h d . 1991. 113. 9294.
[3X] a ) .I. Veciana. C. Rovira. N.Ventosa. M . 1. Crespo. F. Palacio. J. Aiii. C%wr.
S o c . 1993. 115. 57: hJ J. Veciana. C. Rovira. M . I . Crespo, 0. Armet. V. M .
Domingo. F. Palucio. J A i n . Cheiii Soc. 1991. 113, 2552.
[39] M . Ota. S Otani. C ' h c w . Lrrr. 1989. 1 1 79: M . Ota. S. Otani. M . Igartashi, hid.
1989. 1183: M Ota. S. Otani. K . Kobayashi. M . Igara\hi, Mol. C r w . Liq.
Crj,.\r. 1989. 176. 99, M . Ota, S. Otani. Marr~:.Rcs. So(..Sivip 1990. 173. 71.
[40] D. A . Dougherty. A < ( , .C'/wiii. Re\ 1991. 24. 88.
[41] P. Nnchli@dl~.K . D Jordan, J. A m . C/XWI.
Soic 1992. 114. 4743; ihid. 1993.
1li. 270.
[42] a) H. Fukutome. A. Takahashi. M . Oriiki. C h ~ n iP / i m Lcrr 1987. 133. 34:
b) K . Yamaguchi. Y. Toyoda. T. Fueno. Sriirh. M P I .1987. 19. 81.
[43] a ) D . A . Knisaki. W. Chang. D. A . Dougherty, ./. h i . U t i v i i . SIW.1991. 113.
2764: h) D. A. Douglierty. California Institute of Technology (Pasadena.
<'A), personal communication.
[44] a ) H . M . McConiiell. Pro(,. Rohwi A M.i./c/i F ~ ~ i i C'oit/.
d.
C'hciw. Rri. 1967.
11. 143: h ) J. S Miller. A J. Epstcin. J ,4itr. Chriii So( 1987. I O Y . 3x50.
[45] J. S. Miller. A i m li. K Auiil. Sci. 1978. 313. 25.
[46] a ) I . S. Jacobs. L. V. Interrante. H R . Hart. Jr., A I P Cotif Proc. 1975. 24.
355. I. S. Jacob.;. H. R. Hart. Jr.. L. V. Interi-ante, J. W Br'iy. J. S. Kasper.
G. D . Wiltkin\. D. E. Prober. W. P. Wolf. J. C. Bonner. /'/rjwici B ( A m r e r duiri) 1977. 86-SX 6 5 5 , h) J. S. Miller. J. C. Calahrese. unpublished I-ewlts.
c) J. C. Bonner, T. S. Wei, Cr D Watktna, H . W, J. Bliite. .I App/. P/ii..s. 1978.
49. 1321 : d ) J S Kasper. L. V. Interrdnte. Acrii C ' r i x a / / o , q r . Swr. B 1976, 32.
2914
[47] J. M. Manriquei.. G.T. Yce. R . S. McLeiin. A. J. Epstein. J. S Miller. Suniw
1991.2i27,1415.
14x1 A. J. Eprtein. J. S. Miller in C(iii/ir~qi/td
Pii1j.iiiw.s mid /Wiilei/ ,Mur<,riu/\: T l i ~ ~
/iircwiotiiw r i o i i ril C/zmiiid (iwd El(,( iroiii( S r ~ : i iruw
(
(Edr.: W. R Salmeck. I.
Luiidstl-om. B. Ranby) ( P ~ : o c,Vohc/
.
S ~ : i i i p No. NS-81). Oxford Univei-sit).
Pres\. Oxford. 1993. p. 475.
1491 J. S. Miller. G. T Ycc. J. M . Miini-iquez. A. J. Epstein. i n ref. [4X]. p. 461
01 J S Miller, A. 1. Epstein. ,!do/. C ' g x r . Liq. Ci-j\r. 1993. 233. 133.
I] A . 1. Tchougree l'l. R . Hoffmann. ./. P/zj.\. CYICJII. 1993. 97. 350.
[52] a) I-. P. Radliakrishnan. Z. Soos. H . Endres. L. J. A ~ e v x l o ..
I
Chiw Phi,.\.
1986. Hi. 1126: b) E. Doriiiann. M. J. Nouak. K . A Williiims. R . 0 . Angus.
Jr.. I-.. Wudl. J. ,4111. C i i ~ i i iSoc. 1987. 109. 2.594.
[53] H. M . McC'onnell. ./. C'/icwi. phi^ 1963. .W,1910.
1541 A . I i o k i i . S M u i a t a , T. Sugauara. H. 1namur;i. J. Atii. C.limi .So( 1985. 107,
17x6: ihid 1987. 109. 2631.
[551 a) K. Y a m a p e h i . 7. tueno. K . Nakasuji. I.Murata. Cheiii. Lrri. 1986. 62Y.
h ) K . Tanaka, T Takeuchi, K . Yoshi~awa.M. Toriumi. T. Yamahe. Swrh.
.MrJr. 1991, 44. I
1.561 a) F Nakatani. J. Y Carl-iat. Y. Journaux. 0. Kahn. F. Llorcl. J. 1'. Renard. Y.
Pel. .J. Slettrii. M . k c n l a ~ u c i -J. .In, C/i<wt.SOC. 1989. I / / .5739. b) 0 Kahn,
\i Pel. M. Verdagucr. J. P. Rennrd. J. Slecten. ihid. 1988. 110. 782: c ) F.
Nakatani. P. Bcrgcrat. F.. C'odjovi. C . Mathoniere. Y.Pei. 0 . Kahii. / i i w x .
< ' / i c i i i . 1991. 3 1 . 3977.
[57] a ) A. Caneschi. D . G:ittcschi, J P. Renard. P. Rey. K. Sessoli. I i i o r x . C7iciii.
1989. 3K. 3314: h) A . Caneschi. D. Galtcrchi. J. P. Renard. P. Re). R . Scraoli.
414
J. S. Miller, A. J. Epstein
J. ,4111. C/WI?I. S o ( . 1989. 111. 785; c ) C. Bellini. A. Caneschi. D. Gatteschi. R .
Sessoli. Arlv. M u r w 1992. 4.504: d) J. Scidcn. J P/ij,.rirlrrrLerr. 1983.44. L947.
[58] a ) F. Zuo. A J Epstein. C . V u q u e i . R. S. McLean. J. S. Miller. ./ .Malei..
C/wri. 1993.3. 215: h) F. Zuo. S. Zane. P. Zhou. A. J. Epstein. R. S. McLean.
J S. Miller. J. Ap/i/. P h i \ . 1993. 73. 5476. P. Zhou. M. Makivic. F. Zuo. S.
Zaiie. J. S. Miller. A. _I. Epstein. f / i j \. Rcl. 8.1994. 49: c ) J S. Miller, D. M.
O'Hare. A Chackrabortg. A. J. Epstein. J Am. ( % r n l . So( 1989. 111. 7851.
[jY] W. E. Broderick. B. M. Hoffman. J Air!. C'hein. SOC.1991, 11.1. 6334.
[hO] .I. A. Berson. L. Salem. J. Air[. C h e i i i . So(.. 1972. 94, 8917. H . E. Zimnierman.
. 4 ( i . (7/ic,iii
Rtd.\
1971. 4. 172.
[61] a ) C. Ko1lm;ir. M . Couty. 0. Kahn. .I h i .C/IPIII.
S o l . . 1991. 11s. 7994: h) C .
Kollniar. 0. Kahn. J. C/wiii. I-'/iq..s. 1992. 96, 2988, c ) C. Kollmar. 0. Kahn. J
. A i ~ r . C h i i . So( 1991. 113, 7987:(1) Z C; Sons. P C. M McWilliams, M o l .
C~:inr.LJI/.(ri \r. 1989. 176, 369.
[(12] .I. Bliiinel. N . Hehendani. P H u d e c x k . F. H . Klihlct-. LV Straiiss, J, h i .
C ' / l [ , I l l .S(J(' 1992. 114. '$223.
[6?] R. N. Musin. P V. Schaztncv. S. A. Maltnovskaya. / i i o r , a . ('heiii. 1992. 31.
411x
[64] K . Ylimaguchi. M. Okumura, T. Kauiiniura. T. Noro. K Nakasuji. Moi.
Cri.\r. Liq. C r i . \ t . 1992. 218. 229.
[65] '11 M . Kinoshita. M i i / . C'I:I,ST. Liq. C ' i y t . 1989. /76. 163: K . Asaga. T. Sugann.
M. Kinoshita. C % ~ v i r .P/ii:s. Lrrr. 1987. 141. 530: h) K.-M. Chi, 1 C C'alahreh e . J. S. Miller. S. I.Khan, M o / C i j t r Liq. C r ~ \ i 1989,
.
176. 185.
[66] a ) M Tamura. Y Nak;irau;i. D Shioiiii. K. No~a\b:i, Y. Hosokoshi. M .
Ishikaua. M . T;ikohashi. M. Kinoshita. C/i(wi. P/ij,.\. Lrrr. 1991. 186. 401 :
b) M . Kinoshita. Tokyo. Japan. personal communication: c ) K . A u a p . Y,
Miiruyaina. C/ietii. : ! 4 ( i r w . 1990. 3, 535. d ) K . Awagn. Y Maruyama. J, C ' h u
P/lj..\ 1990. 173. 33
[67] A. L. Tchougrerl'f. I.A. Misurkin. C/wiii. P$., 1991. 153, 371 : A. L. Tchougrccff. 1. A . Mi5urkin. Pha. Rn.. B 1992. 46. 5357.
[68] 1. Kondo. Soiid Store Phi.\ 1969. 773. 183.
(691 M . Y Ogana, B. 41. Hoffman. S. Lee. M. Yudkowsky. W. P. Halpcrin. Plijs.
Rcr Lcrr 1986. 1 7 . 1177: M Y.Ogawa. J. Martinsen. S. M Palmer. J. 1.
Stanton. J. T m a k a , R. L. Grccne. B. M. Hoffman. J. A. Ibcrs, J Atn. Chcm.
Soc. 1987. 109, 1 1 I S .
[70] J. S Miller. D. T. Glat/hofer. R L;i\ercanne. S. Chittipeddi. P Vacn. T. B.
Brill. M. D. Timken. C J. O'Connor. J. H. Z h m g . J C. ('alabrese. A. J. Epaleiii. C./i(v?i. ; b / u t ~1990.
~ 2. 60.
[71] G. Choutcau. C Vcyrct-Jeandcy. J. f / r i 5 , (f1irr.s)1981. 42. 1341
[72] a ) P J Bro\4n. A. ('apioniont. B. Gillon. J. Schweirer. J. Moxil. lMa,q~i.,Zliirr~:.
1979. 14. 289: h ) A. Benoit. J. Flouquet. B. Gillon, J. Schaeizcr. ;bid 1983.
31 33. 115s.
[73] a) R. Chiai-elli. A . Rassat, P. Rcy, J. C/rcni.Soc. Chorr. C o i i r i i i i r r i . 1992. 1081:
b) R. Chiarelli. M A. No\ak. A. Riissxt. J. L. Tholcncc. .Yrtiirrc 1993, 363.
147.
[74] Y. J. Ueniur,i. L P Le. G M Luke. S i i r r h . Mrr. 1993. 56. 2845.
[75] T. Sugann. T.Goto. M . Kinoshita, Solid Srritc C~iiriiiiiin.1991. X/1. 1021 : K
Aw'aga. T. Inahe. Y. M x u y a n i a . C'hoi7. Phj.i. Lctr. 1992, 190. 349: K . Awaga.
T. Inahe. U. Nagashiina. T. Nak;imur;i. M . Matsumoto. Y. Ka\iabata. Y.
Maruyama. i h i d 1992. 1711.
[76] a ) P. Allemand. K . Khemani. A. Koch. F Wudl. K. Holcrei-, S. Donovan. G.
GI-Liner.J. D . Thompson, Siioirr 1991, 3.7.3. 301: P. W. Stcplicn, D . COX.J. b'.
Laiihei-. L. Miltxi->. 1. B. Wilcy. P. Allemnnd. A Hirsch. K. Holczer, Q . LI.
.I D. T h o m p o i i . F. Wudl. Xrirurc 1992. <.Ti. 331 ; h) K. Tandk,!, A. A. Zakhidov. K . Ynshizii\\a. K . Okahara. T. Yiunahc. K . Yakushi. K. Kikuchi. S .
Suruki. 1. Ikcmoto. Y. Achiba. Phi,\ Lei/. A 1992. 164. 221. c) K . Tanaka.
A A . Ziikliidov. K . Yoshirawa. K . Okahara. T. Yamabe. K. Yakushi, K .
Kikuchi. S. S u ~ u k i I. . Ikeinoto. Y Achiha. submitted, d ) J. I). Thompson. C;.
Spar-n. E Diedei-icli. G. Griincr. K . Holcier. R . B. Kaner. R . L. Whetten. P.
Allemiind. Q Clien. F. Wudl. !llorrv. Rc.5. .So<. Si.irip S r i . f 1 . 1 ~ . 1992. 347.
315.
G. A. Cnndcl,i. I. Sv,art/endruher. J. S. Miller. M . J. Rice, J. A m C h i i
M . Reiff', L. D. Preston.
A. H . Reis. Ji- . E. Gel-bert. M. Exlinc. J. Troup. M . I) Wird. .I f / i n C h m i r .
1987. 91. 4344.
[78] E Stryje\cski. N Ciiordano, Adv. P/i!\. 1977. 26. 487.
1791 J. S. Millci-. J. C Cal;threre, H. Romnielmann. S. Chittapeddl. .I H . Zhang.
W. M . Reiff. A .I.Epatcin. J A m C/wiii. SOC.1987, 109. 769.
[XO] a ) Z. J. Huang. F.Uien. Y. T. Rcii. Y Y Xue. C. W. Chu. J. S. Miller. J. App/.
f / i I : s . 1993. 73. 6 W . h ) J. M. Williams, A. J. Scbultz, U . Grciscr. K. D . Carlson, A . M . Kini. H . H . Wang. W.-K Kwok. M.-H. Whangbo, J. E Schirher.
Siiriici, 1991. 253. 1501: CT.Spiirn. J D. Thompson. R. L. Whetten. S:M.
Huiing. R . B. Kaner. F. Diederich, G. Gruncr. K . Holcrer. Phi,.i. Rri,. Lrrt
1992. hK. 1128. J E. Schirher, D. L. O\erincyer. H H. Wang. 1. M . Williams.
K . I>. Ciirlson. A . M. Kini. U . Wclp. W.-K. Kwok PIzi,.\ii(i C ( . & i r s t i ~ r h i ? i J
1991. 17X. 137: G. Sparn. .I.I). Thompson. S:M.
Hunng. R. B. Kaner. F.
Diederich. G . Griincr. K . H o l c x r . .Scie~iw. 1991. 253. 1x2').
[Xl] G. T. Ycc. J. M . Mnnriquer. D . A. Dixon. R. S. McLeiin. D. M . Groski. R. B.
Flippen. K S . Naraynn. A. J. Epstein. J. S. Miller. A d i . M a / e r . 1991, 3. 309.
[82] \?! E. Broderick. J. A. 'I'hompson. E. P Day, B M. Hoflhlan. Sciwicc 1990,
3 4 ~410.
.
[77]
21)
So'.. 1979. 1/11, 2755: b) J. S. Miller. J. H . Zhang. W
Designer Magnets
[X3] J. L. Robbins. N. Edelstein. B. Spencer, J. C. Smart. J A m . Chem. Suc. 1982,
104. 1882.
[84] a) 0 W. Webster, W. Mahler, R. E. Benson. J A m . Ci7em. Soc. 1962.84,3678;
M. Rosenblum. R. W. Fish, C. Bennett. ibiii. 1964, 86. 5166; R. L. Brandon,
J. H. Osipcki. A. Ottenberg. J. Org. Chem. 1966. 31. 1214; E. Adman. M.
Rosenblum. S. Sullivan. T. N. Margulis. J. Am. Chem. Soc. 1967,89, 4540; B.
Foxman. Brdndeis University. Watham. MA, USA. personal communication; B. W. Sullivan, B. Foxman. OrgunomerullicA 1983. 2. 187; b) J. S. Miller.
J. C. Calabrese, W M. Reiff. D.T. Glatzhofer. unpublished results.
[85] J. S. Miller. J. C . Calabrese. D. T. Glatzhofer, unpublished results.
[86] K . M Chi. J. C. Calabrese. W. M. Reiff. 1. S. Miller. 0rgmometdiir.s 1991, 10.
688.
[87] J. S. Miller. D. T. Glatzhofer, D. M. O'Hare. W. M. Reiff. A. Chackraborty,
A. J Epstein. fnorg. Chcm. 1989, 27. 2930.
(881 J. S. Miller. J. H. Zhang. W M. Re;ff, J A m Ch1.m. Sor.. 1987. 109, 4584.
[89] E. Gerbert, A. H. Reis. J. S. Miller. H. Rommelmann, A. J. Epstein, J Am.
Chon Sol,.1982, 104. 4403, J. S. Miller, P. J. Krusic. D . A. Dixon. W. M.
Reiff. J. H Zhang, E. C. Anderson. A. J Epstein. ibirl. 1986. 108. 4459.
[90] A. H . Reia, Jr.. L. D. Preston. J. M. Williams. S. W. Peterson, G. A. Candela,
L.J Swartzendruber. J. S. Miller, J A m Chem. Soc. 1979. 101. 2756.
[Yl] J. S. Miller, D. A. Dixon, J. C. Calabrese. R. L. Harlow. J. H. Zhang, W. M.
Reiff. S Chittappeddi. M . A. Selover. A. J. Epstein. J. A m . Chem. So<. 1990.
1 I.?.5496.
[92] a ) J S Miller. J. C. Calabrese. A. J. Epstein. Inorg. Chcm. 1989, 27, 4230;
b) J. S. Miller. J. C. Calabrese, A . J. Epstein, unpublished results: c) W. B.
Heuer. D. O'Hare, M. L. H. Green, J. S. Miller. C/7em. Muter. 1990, 2, 764.
[93] D. O'Hare. J. S. Miller. J. C. Green, J. C. Chadwick, Orgunometullics 1988. 7 .
1135.
[Y4] .I S. Miller. R. S. McLean, C. Vazquez, G. T. Yee, K. S. Narayan, A. J. Epstein. J ,Miiicv. Ciiem. 1991. 1. 479.
[95] K. S. Nardyan. K . M. Kai. A. J. Epstein, J. S. Miller, J. A p p / . Phys. 1991, 69.
5953. K . S. Naraydn, B. G. Morin, J. S. Miller. A. J. Epstein. Phys. Rev B
1992. 46. 6195.
[96] M. D. Burk. J. C. Calabrese. J. S. Miller, unpublished results.
[97] b'.E Broderick, J. A. Thompson. B. M. Hoffman. Inorg. Chem. 1991, 30,
2960.
19x1 C. LBzqueL, D. A. Dixon. J. C. Calabrese. J. S. Miller, J Org. Chrm. 1993, 58.
65.
2nd edition, Oxford
[99] a ) P. A. M. Dirdc, The Principle.\ ufQuuntuni Uechunic.~,
Unibersity Press, Oxford. 1935; b) P. A. M. Dirac, Proc. R . Sac. London A
1926. 112. 661
[loo] Thou$h the observed
is in excellent agreement with this mean-field model
prediction. and suggests similar exchange interactions are operating for both
the [MCpr]' [ T C N E r - M = Fe. Mn) systems, the Curie-Weiss 0 value of
22.6 K ohseined for the [MnCpt] + [TCNE]'- in the high-temperature regime
doe\ not Ycale with the increased magnitude of the spin ( S )on the metallocene.
Instead i t is slightly less than the value or0 recorded for [FeCp:]' ' [TCNE]'-:
hence these mean-field scaling models must be used with caution.
[ l o l l Since a s described earlier J x I I . A E , ' , and since instead of the POMOs
being i i n o l x d in the charge-transfer excitation, the excitation involves the
Angiw. (%cm. 117l.Ed. Engl. 1994. 33. 385-415
REVIEWS
NHOMO and/or N L U M O ; thus, the A E will be greater and J will be
smaller.
[lo21 S. H . Liu, J. Mugn. Mugn. Maler. 1988, 82. B417.
[lo31 R. B. Stinchcombe in Phase Transitions and Criticul Phenumenu (Eds.: C.
Domb, J. L. Lebowitz), Academic Press, London. 1983, 7. 152.
[I041 K. S. Nardyan, 0. Heres, A. J. Epstein, J. S. Miller, J. Mugn. Mugn. Muter.
1992, 110. L6.
[I051 a ) J. L. Robbins. N. Edelstein. B. Spencer. J. C. Smart, J Am. C%cm.Siir.
1982,104.1882;b) F. G. N. Cloak, A. N. Din. J. C. Green. R. N.Perutz. E. A.
Seddon. Orgunomerallics, 1983. 2. 1150.
[lo61 For [V(TCNE),] .,Y(CH,CI,)a range of I and J is observed for preparations
in different solvents and wlth different mole ratios of the starting materials:
benzene in small amounts is present in some cases. For the material discussed
herein s = 2; v 5 1.'2. Trace elements were determined by ICP- Fe 41 ppm).
Co [none detected <25 ppm)], Ni [none detected < 50 ppm)]. Cr 34 ppm).
[lo71 F. E. Luborsky, in Fcwromugnrric Muti,riulc Vol. I (Ed.: E P. Woblfarth).
North Holland, Amsterdam, 1980, p. 452.
[lo81 A. J. Epstein, J. S. Miller, M o / . C r w . Liq. Crril. 1993.233. 171; P. Zhou. J. S.
Miller. A. J. Epstein. unpublished results.
[lo91 G. Engebretson, R. E. Rundle, J A m Chem. Soc. 1963. 85. 481
[llO] Z. Oblakowski, A. J. Epstein, M. LaridJanl, JLP. Pouget, J. S. Miller. unpublished results.
[ l l l ] a) E. Cbudnovsky. J Mugn. Mugn. Maler. 1988. 64, 5770; E. Chudnovsky.
ihrd. 1989. 79, 127 and references therein: b) P. Gebring, M. Salamon, A.
Moral, J. Arnaudas, P h u . R C F B
. Condiw. Maller 1990, 41. 9134, and references therein
[312] a) P. Zhou, B. G . M o m , J. S. Miller, A. J. Epstein, P/7j,.s. R F I ~B. 1993. 4K.
1325; b) P. Zhou, J. S. Miller, A. J. Epstein. Phjs. Lett. A 1993, 1x1. 71.
[113] B. G . Morin, P. Zhou, C. Hahm. A. J. Epstein. J. S. Miller, J. .4pp/. Ph.v.5 1993,
73. 5648.
[114] S. Long. J. S. Miller. A. J. Epstein, unpublished results.
[115] G . Du. J. Joo. A. J. Epstein, J. S. Miller, J Appl. P / 7 ~ s1993,
.
73. 6566.
[116] J. Joo. J. S. Miller, A. J. Epstein, unpublished results.
[117] D. L. Huber, P11,w. Lett. A 1971. 37, 285.
[118] N. F. Mott, E. Davis. Electronic ProcrsJes in Noncrwlailint~Sohd.~,Clarendon, Oxford, 1979.
[139] J. S. Miller. J. C. Calabrese. R. S. McLean. A J. Epstein. 4 d r . Mutrr. 1992, 4.
498.
[120] See for example: M . Verdauger. M. Julve, A Michdiowicz. 0 . Kahn. Inorg.
Chem. 1983, 22. 2624; M. Drillon. J. C . Giandurzo, R. Georges, Phjs. M t .
A 1983, 96.413.
[123] a) R. A. Serola, P. A. Lee, Pliy.7. Rev. B Coni1en.s h4uiti.r 1986, 34, 1806; R.
Dickman. E. Chudiiovsky, ibid. 1991. 44, 4397; b) J. Curely. R. Georges. M.
Drillon. i/id 1986, 33, 6243.
[122] P. Zhou, B. G. Morin. A . J. Epstein, R. S. McLean, J. S. Miller. J. Appl. P / i w
1993, 73. 6569.
[123] J. Curely, R. Georges, M. Drillion, Phys. Rrv. 8 Condimr. ,Mutrrr 1986. 33,
6243.
[124] J. S. Miller, A. J. Epstein, W. M. Reiff, Mol. C'ryTl. Liq. ('ryxt. 1986. 120.
27.
41 5
Документ
Категория
Без категории
Просмотров
1
Размер файла
3 386 Кб
Теги
organometallic, molecular, organiz, magnetic, materialsчdesigner, magnet
1/--страниц
Пожаловаться на содержимое документа