C1CS15047B

This article was published as part of the

Molecule-based magnets themed issue

Guest editors Joel S. Miller and Dante Gatteschi

Please take a look at the issue 6 2011 table of contents to

access other reviews in this themed issue

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online / Journal Homepage / Table of Contents for this issue

3336 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

Cite this: Chem. Soc. Rev., 2011, 40, 3336–3355

Molecular spintronicsw

Stefano Sanvito*

Received 23rd February 2011

DOI: 10.1039/c1cs15047b

The electron spin made its debut in the device world only two decades ago but today our ability

of detecting the spin state of a moving electron underpins the entire magnetic data storage industry.

This technological revolution has been driven by a constant improvement in our understanding on

how spins can be injected, manipulated and detected in the solid state, a field which is collectively

named Spintronics. Recently a number of pioneering experiments and theoretical works suggest that

organic materials can offer similar and perhaps superior performances in making spin-devices than

the more conventional inorganic metals and semiconductors. Furthermore they can pave the way

for radically new device concepts. This is Molecular Spintronics, a blossoming research area aimed

at exploring how the unique properties of the organic world can marry the requirements of

spin-devices. Importantly, after a first phase, where most of the research was focussed on exporting

the concepts of inorganic spintronics to organic materials, the field has moved to a more mature

age, where the exploitation of the unique properties of molecules has begun to emerge. Molecular

spintronics now collects a diverse and interdisciplinary community ranging from device physicists

to synthetic chemists to surface scientists. In this critical review, I will survey this fascinating,

rapidly evolving, field with a particular eye on new directions and opportunities. The main

differences and challenges with respect to standard spintronics will be discussed and

so will be the potential cross-fertilization with other fields (177 references).

1. Introduction

Most electronic devices, either for logic or sensing applications,

operate on the principle of detecting the variation of an

electrical current with an external stimulus. This can be the

gate potential in a transistor or the presence of the object to

detect in a sensor. Regardless of its nature the stimulus acts to

produce a change in the device electrostatic profile. When the

stimulus has a magnetic origin, then the detection is more

complicated, simply because a magnetic field is less efficient

then an electric one at driving the electron motion. For

example it takes a magnetic field of 103 T to produce a Lorentz

force equal to that of an electric field of 0.1 V nm�1 (assuming

that the electron moves at 105 m s�1). This means that for

magnetic detection a property different from the electrical

charge must be used. This is the spin.

The importance of the spin degree of freedom for the electron

motion in transition metal magnets was acknowledged a long

time ago by Mott,1 who first expressed the ‘‘two fluids’’

concept. This establishes that the current in a magnet is carried

by two fluids (interacting), characterized by electrons having

opposite spin directions. Such a concept was exploited only

50 years later with the discovery of the giant magneto-resistance

(GMR) effect in magnetic multilayers,2,3 which effectively

demonstrated that the electrical resistance of a magnetic device

can be modified by changing its magnetic texture. This signified

the beginning of the field of magneto-electronics or, as more

School of Physics and CRANN, Trinity College, Dublin 2, Ireland.E-mail: [email protected]; Fax: +353-1-6711759;Tel: +353-1-8963065w Part of the molecule-based magnets themed issue.

Stefano Sanvito

Stefano Sanvito completed hisundergraduate studies in Milan(Italy), before moving to theUniversity of Lancaster (UK),where he obtained a PhD intheoretical Physics. In 2002 hejoined the School of Physics atTrinity College Dublin as aLecturer, after having spenttwo successful years as apostdoctoral fellow at theUniversity of California SantaBarbara (USA). In 2006 hebecame associated Professorin Physics and in 2009 DeputyDirector of the Center for

Research on Adaptive Nanostructures and Nanodevices (CRANN).Stefano Sanvito leads the Computational Spintronics Group, alarge and dynamical theoretical/computational research groupthat investigates elementary properties of materials and of nano-devices using computer simulations. In particular the groupdevelops and maintains the code Smeagol, a state of the artpackage for materials specific electron transport calculations.

Chem Soc Rev Dynamic Article Links

www.rsc.org/csr CRITICAL REVIEW

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3337

commonly known, of Spintronics.4 At present, spintronics

encompasses a rather large and diverse range of phenomena

and experimental techniques. However, broadly speaking,

they all address the problem of injecting, manipulating and

detecting spins in the solid state environment.

Historically, spin-related phenomena have been first studied in

transition metals and subsequently in inorganic semiconductors.5,6

More recently considerable attention has been dedicated to less

conventional materials sets. These for instance include ferroelectric

insulators, presenting a spontaneous electrical polarization,7

and magnetic semiconductors, offering spin-selectivity to a tunnel

current.8 Among the new materials, organic molecules have

recently emerged as a completely revolutionary platform for

spintronics. This field, inspired by pioneering spin-transport

experiments9 and early theoretical predictions10 has already taken

the evocative name of Molecular Spintronics.

The rationale for spintronics to go from metals to inorganic

semiconductors is that of creating a single materials platform

where to integrate information-processing, traditionally

implemented in non-magnetic semiconductors, with data-storage,

commonly the domain of magnetic metals.11 Semiconductors,

in general, are more versatile than metals, since their

electronic properties can be modified by little changing their

electronic structure (the carrier density for instance). Further-

more the spin-relaxation time is usually much longer in

semiconductors than in metals12 and in a semiconductor one

can hope of controlling the spin–orbit interaction and perform

spin-manipulation13 (this last possibility has turned out to be

rather challenging in practice). The addition of other materials

to the spintronics portfolio is also driven by specific expectations.

For instance the idea of using ferroelectric insulators underpins

the prospect of making multi-functional devices, where the

magnetic information can be read and/or stored electrically.

But, what is the rationale for going organic?

In general, organic molecules can be prepared in a practically

infinite range of types and combinations, their properties can

be finely tuned and their degree of purity may have no equals

in the inorganic world. Furthermore molecules are synthesized

at low temperature and in the same conditions they can also be

processed. However, the most important feature is that their

electronic properties and their functionalities span an extremely

large range. For instance the conductivity of organic polymers

can be engineered over fifteen orders of magnitude14 and they

can display both magnetism15 and ferroelectricity16 at room

temperature. Furthermore, since most of the organic materials

are made of elements populating the upper rows of the

periodic table, both spin–orbit and hyperfine interaction, the

two principal ways for spins to interact with the environment,

are weak. As such organic materials appear as an incredibly

versatile and unique playground for exploring new spintronics

concepts and/or for implementing existing ones.

In this review I will discuss the most recent developments in

the field of molecular spintronics. In particular I will highlight

the differences with conventional spintronics in inorganic

metals and semiconductors, so that the reader will form an

opinion on what organic materials have different to offer to

spintronics. I will start by discussing some general concepts

common to both inorganic and organic spintronics and review

the main mechanisms that spins have available to interact with

the external world. Then, I will first overview the most recent

advances in large-scale devices and finally move to single

molecule junctions. Throughout the discussion I will also

make connections to other research fields which share with

molecular spintronics experimental techniques and concepts.

Little discussion will be dedicated to carbon-only inorganic

macromolecules such as graphite, graphene and carbon nano-

tubes. For these we refer the reader to the abundant recent

literature.17

In closing this introduction I wish to mention that a number

of comprehensive reviews on different aspects around molecular

spintronics have appeared recently. These include spin-transport

in organic semiconductors18,19 and in single molecules,20,21

quantum computing with molecules,22–24 molecular materials,25

organic radicals on surfaces,26 and spin-polarized scanning

tunnel microscopy (SP-STM).27

2. From spintronics to organic spintronics

There are a number of concepts, related to how spins move in

a solid state environment, which are common to any spin-devices

and can be applied to any materials classes. These define the

entire field of spintronics and will be the subject of the first half

of this section. Then I will discuss how specific materials

properties affect the dynamics of spins in materials, a discussion

which will reveal how peculiar organic molecules are with

respect to any other media. Finally I will overview the most

recent experiments and theories on large-area organic devices.

As such, this section describes how the ideas of conventional

spintronics have been exported to devices made of organic

semiconductors, i.e. it describes the evolution from spintronics

to Organic Spintronics.

2.1 Basic concepts

The general idea behind spintronics is that of detecting the

response of spins to an external stimulus and, by using such a

platform, implementing logic (either classical or quantum),

memory and sensing capabilities. Therefore, one has first to

understand how spins evolve in time in a solid state environment,

in particular when these are associated to moving electrons.

An ideal spintronics experiment consists of three main actions

(see Fig. 1). First one has to generate a (non-equilibrium) spin

population within a given material (injection), i.e. spins must

be prepared in the desired initial configuration. Then an

external stimulus should be applied to alter the initial spin

population in a controllable way (manipulation) and finally

the result of the manipulation should be detected (detection).

Of course the manipulation step can be eliminated and one

may simply look at how spins evolve, i.e. how they reach

equilibrium again. Importantly for each one of these steps a

number of experimental strategies are possible, depending on

the specific nature of the materials set used. Optical pump–

probe techniques based on Kerr/Faraday rotation are the

standard for inorganic semiconductors,28 while for metals

both electrical29 and optical methods30 can be used. In organic

molecules, with a very few exceptions, only electrical injection/

detection is possible, so that only this aspect will be discussed

in some details here.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3338 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

The prototypical electric spin-device is the spin-valve,31

which consists of two magnetic materials, usually metals,

sandwiching a non-magnetic one (middle panel of Fig. 1).

An electron current drives a non-equilibrium spin population

from the injector to the non-magnetic material. If this reaches

the detector the resistance of the entire device will depend on

the mutual orientation between the magnetization vectors

of the injector and the detector. This occurs essentially because

the individual resistances of the two spin-channels, in Mott’s

spirit,1 are different, so that a different equivalent spin-resistance

circuit is associated to the different magnetic orientations of the

electrodes magnetizations.32

In general the relative conductance of the two spin-channels

of a magnetic material defines its spin-polarization, P. This is a

property related to the electronic structure of the material

itself and to the experimental technique used to measure such a

polarization. In general one can write33

Pn ¼N"v

n" �N#v

n#

N"vn" þN#vn#; ð1Þ

where Ns is the material’s density of state (DOS) at the Fermi

level (EF) for spin s (s = m for majority electrons and s = k

for minority) and vs is the spin-dependent Fermi velocity. The

Fermi velocity is weighted differently depending on the particular

experiment, with n = 0 for photoemission measurements and

for tunneling across amorphous barriers, n = 1 for ballistic

transport and n = 2 for diffusive conduction. Note that, as a

consequence of the definition of Pn, the same material may

have a substantially different polarization depending on the

specific experiment carried out or on the specific length-scale

examined.34

Typically the place where spins are injected is different from

the one where they are detected so that electrons spins must be

driven across the non-magnetic material. In such a transfer

process they will continuously attempt to reach their equilibrium

state by interacting with the environment. At equilibrium in a

non-magnetic material there is no spin imbalance so that

during their motion electrons lose their initial spin-polarization.

Two quantities characterize the interaction of spins with their

surrounding: the spin-relaxation time, tS, and the spin-relaxation

length, lS. The spin-relaxation time is the average time that an

electron spin takes before changing its original direction. Note

that, in analogy with nuclear magnetic resonance,35 also in

spintronics one can define the longitudinal, T1, and the

transverse, T2, spin-relaxation times. However in a spin-valve

experiment there is no direct access to these two quantities

separately, since the injected electrons do not have a defined

phase relation (they are not coherent). Hence the more general

definition of tS must be used. Instead the average distance

travelled by a spin defines the spin-relaxation length. Clearly

tS and lS are simply related by lS = mtS, where m is the average

electron velocity.

The magnitude of lS relatively to the spin-valve non-

magnetic spacer thickness, L, determines the ability of the

spin-valve to work. In fact if lS o L, the spin-polarization of

the injected current will be completely lost by the time the

electrons reach the detector. As such the resistance of the

entire device will not depend on the direction of the magneti-

zation vectors of the electrodes. In contrast if lS 4 L some

spin-polarization will survive the motion in the non-magnetic

spacer and the total resistance will depend on the spin-valve

magnetic state. This simple concept however suffers of another

obstacle, present even if lS = N. This is known as the

resistance mismatch problem.36

The equivalent resistance of a spin-valve is obtained by

adding in series the resistances of the electrodes (spin dependent)

and that of the spacer (spin independent). As such, if the

resistance of the spacer is much larger than that of the

electrodes, the total resistance of the device will be only weakly

spin dependent. Unfortunately this is the case of inorganic

semiconductor spacers contacted by magnetic metals, so that

the direct spin-injection from metals to semiconductors has

been traditionally problematic and one usually needs to

include large spin dependent barriers in the device stack.37

Since the typical resistivity of an organic semiconductor is

significantly larger than those of its inorganic counterparts an

even more severe problem is expected in organic spin-valves.

This is however not the case, as we will see later on.

A final mention goes to tunneling junctions. In this case the

spacer is a wide-gap material and the electrons are not injected

but simply tunnel between the two electrodes. The MR is

determined by the electronic structure of the entire junctions.

When the tunnel barrier is amorphous the MR is controlled by

the spin-polarization of the electrodes,38 while for crystalline

barriers the precise symmetry of the wave-function of the tunneling

Fig. 1 Different experimental strategies for studying spin-dynamics

in semiconductors. In inorganic semiconductors both the creation of a

spin-polarized wave-packet and its detection at a different position can

be performed optically (upper panel), by using respectively circularly

polarized light (injection) and Kerr (or Faraday) rotation (detection).

In the case of spin valves (middle panel) both detection and injection

are performed by magnetic metals. Finally hybrid strategies do exist,

for instance where the injection is electrical and the detection is optical

(lower panel).

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3339

electrons dominate the transport.39–41 In tunnel junctions

made with organic materials no fully crystalline structure

can be made and most of the properties are dominated by

the interface between the electrodes and the relevant molecules

(see next sections).

2.2 Materials and devices considerations

Electron transport in organic materials is substantially different

from that in inorganic semiconductors and spin transport is no

exception. The conventional band picture, where electrons or

holes move as quasi-free particles (the effects of the band-

structure can be incorporated into the effective mass describing

the bands curvature) is replaced by one where electron hopping

between localized molecular orbitals provides the channel for

the electrons to move across the material. Thus, the typical

mobility of 450 cm2 V�1 s�1 for p-doped Si must be compared

with that of rubrene, one of the most conducting organic

semiconductors, which is only around 10 cm2 V�1 s�1.

The electrical response of a typical vertical device (a spin-valve

for example) made with organic materials depends on how

extended the transport channel is with respect to the coherence

length of the hopping process. If the device dimensions are

comparable with the coherence length, then the main transport

mechanism will be tunneling. In this case the conductance is

only weakly temperature dependent, it is proportional to

the device area and it decays exponentially with the barrier

thickness.42 Otherwise, when the coherence length is shorter

than the device size, the transport mechanism will be incoherent

electron diffusion. Also in this case the conductance is proportional

to the device area, but now it scales as a power low with both

the temperature and the layer thickness.43 Hence, establishing

the transport mechanism should always be the first step

in the characterization of an organic spin-valve,44 since there

is no spin-injection in the tunneling limit in contrast to the

diffusive one.

Let us now turn our attention to the spin-transport properties

of organic materials. These are summarized in Fig. 2, where we

present the lS–tS diagram for a representative sample of

different materials. The overwhelming message emerging from

the picture is that organic materials occupy the top-left corner,

i.e. they are characterized by long spin-diffusion times but

rather short spin-diffusion lengths. The relationship between

lS and tS can be easily rationalized by fact that the mobility in

organic materials is usually rather poor (note that for many

data in Fig. 2, tS is not directly measured but simply inferred

from the values of lS and m). However, why is the spin-

relaxation time typically so long?

Long tS’s in organic materials are well known in the electron

paramagnetic resonance (EPR) community,55 and they are

understood in terms of the absence of any efficient mechanism

for spin-relaxation. Spin–orbit interaction, critical in inorganic

semiconductors, is rather weak in organic media, mostly

because it scales as Z4 with the atomic number, Z, and because

most organic molecules are made of elements in the upper

rows of the periodic table. Just as an example, one should

recall that spin–orbit interaction produces a splitting of the

valence band of GaAs of 340 meV, while in C-diamond this

is only 13 meV.19 Notably spin–orbit interaction plays a

ubiquitous role in spintronics. On the one hand it is usually

the largest source of spin-dephasing, at least in inorganic

semiconductors, and on the other hand it is the main interaction

allowing one spin-manipulation and optical spin-detection.12

Thus, such a lack of significant spin–orbit interaction makes

polarized-light pump–probe optical techniques ineffective for

organic materials and most of the standard optical charac-

terization tools designed for the inorganic world cannot be

adopted for molecules.56

Also hyperfine interaction is rather inefficient, since there

are only a few nuclei of light elements possessing a nuclear spin

(except for H) and the typical molecular orbitals responsible

for electron transport are p-type delocalized mostly over

C atoms. Still hyperfine interaction appears to be as a key

ingredient for explaining the almost universal presence of the

low-field room-temperature MR in organic materials57 known

as the organic magneto-resistance (OMAR) effect.58

A further source of spin-scattering is provided by para-

magnetic impurities. The efficiency of such a spin-scattering

channel depends on the overlap between the electronic wave-

functions of the hopping electrons and that of the impurity.

This is expected to vary from material to material and general

rules are difficult to draw. Also in this case H is the most

notable scattering-center as well as spin-radicals, which might

be abundant in molecular materials.55 Note that scattering

to impurities overall contributes to the spin-relaxation of

conducting electrons proportionally to both the impurity

concentration and the electron density.

A natural question arising now is on how the physical–

chemical properties of molecular materials affect their spin-

transport characteristics. In general the electron transport

Fig. 2 Spin-relaxation time, tS, against spin-diffusion length, lS, for

different classes of materials. Note that non-magnetic organic materials

occupy the upper left corner, i.e. they are characterized by long tS andshort lS. The picture is adapted from ref. 44. Data in the figure

correspond to the following references: a,45 b,12 c,46 d,47 e,48 f,49 g,50

h,51,52 i,53 m,9 n.54 G. Szulczewski, S. Sanvito and J. M. D. Coey,

Nature Materials, 2009, 8, 693–695. Copyright (2009) by the Nature

Publishing Group.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3340 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

in organic media is highly sensitive to the molecules packing,

to their mutual interaction, to doping and to the system

dimensionality. Relatively high mobility is usually guaranteed

by p-conjugation and increases as the material’s crystallinity

improves. Currently the highest mobility ever recorded for an

organic field effect transistor is for rubrene single-crystals60

and it is of the order of 40 cm2 V�1 s�1. Much less however is

known about the influence of the physical–chemical properties

of molecular materials on their spin-transport. Apart from the

already mentioned relationship between lS and tS, which

essentially establishes that the spin-diffusion length in organic

materials is mobility limited, little is known about the role of

crystallinity and chemical composition over tS. In the case of

conducting polymers there is a significant body of information

obtained with EPR,55 which enables one the study of both spin

and spinless nonlinear excitations. The ratio of these quasiparticles

depends sensitively on various properties of the polymer and

the dopants introduced into it. As such both lS and tS becomes

sensitive to the polymer physical–chemical properties. Importantly

conducting polymers are quasi-1D objects and most of their

characteristics are determined by their dimensionality, so that

the known EPR results do not directly transfer to 3D molecular

crystals.

In summary organic materials are characterized by their

ability of sustaining long-living spin-states, since all the possible

mechanisms for spin-scattering are weak. This means that in

general there is not a single dominant interaction limiting tS,but a combination of all the possible scattering events

determines the dynamics. For instance spin–orbit interaction

has been recognized as the main source of spin-dephasing in

ultra-pure carbon nanotubes,59 but strong hyperfine coupling

can also be present in 13C-enriched ones.61 Similar experiments

have been carried out in either protonated or deuterated

polymers with the similar conclusion that hyperfine coupling

can play a crucial role in spin-dephasing.62 At the same time

however there is also a significant body of evidence in favor of

spin–orbit interaction.

One can then look ahead and ask what organic molecules

can bring to spintronics, or in other words, what range of

devices organic spintronics can embrace. Clearly whatever is

the application, this should benefit from the long tS but

should not be limited by the short lS. As such one should

envision applications where spins are manipulated or detected

not too far from the point of injection. This essentially means

making nano-scale devices where the typical channel length is

smaller than 20 nm. An intriguing prospect is that of coherent

spin-manipulation by using electron spin resonance (ESR)

either electrical or optical detected,63 which may lead to new

logic devices. At present only a few proposals for spin-logic

have been brought forward,64 but the situation might rapidly

change. Intriguingly the same ESR techniques can be used

as diagnostic tools for closely related technologies such as

organic light emitting diodes and photovoltaic cells just to

name two of them. A second option is that of looking at

single molecule devices. Here the issue of the injection and

spin-diffusion length becomes immaterial but one has to

understand how spin-manipulation can be carried out when

the electrons spend only a tiny amount of time on the active

region of the device (the molecule). The first class of devices is

discussed in the next section while single molecule junctions

are reviewed in section 3.

2.3 Early experimental progress discussed by concepts

The first demonstration of MR in an organic spin-valve was

provided by a lateral device and dates to almost a decade ago.9

The electrodes material of choice was La0.7Sr0.3MnO3,

while the organic layer was composed of sexithiophene (T6).

The typical mobility of T6 is of the order of 10 cm2 V�1 s�1

(p-type conduction), so that the appearance of a MR at room

temperature for channels with lengths ranging between 100 nm

and 200 nm was a big surprise. In fact in this situation of a

poorly conductive channel the resistance mismatch argument36

predicts essentially no MR. Notably a spin-valve can still

support MR even in the presence of large resistance mismatch

if the electrode is a perfect half-metal.36,65 This is, in principle,

the case for La0.7Sr0.3MnO3. However, one has to realize that

the half-metallicity of the current (100% spin-polarization) is

strictly valid only at low temperature even for band half-

metals. Furthermore it generally decreases as the temperature

gets larger. Recent experiments66,67 correlate the detection of

the MR to the persistence of the surface magnetism of

La0.7Sr0.3MnO3. In one case66 the MR is completely lost

at about 220 K, while in another it survives up to room

temperature,67 with the difference between the two experiments

being rooted in the different surface preparation. In any case

both experiments showMR at temperatures where the injected

current has a polarization smaller than 100%, i.e. they are in

defiance of the resistance mismatch argument. Such an early

experiment then posed the question of whether or not injection

was really occurring.

A second problem originating from this early experiment is

that there was no direct correlation between the MR and the

magnetic switching of the electrodes’ magnetization. This

is what defines the spin-valve operation. In particular the

resistance vs. magnetic field curve of a spin-valve should

display a characteristic ‘‘butterfly’’ shape, with abrupt changes

in resistance as the magnetic field sweeps across the coercive

field of the electrodes [see Fig. 3 top panel]. A clear demon-

stration of the spin-valve effect was provided two years later

for a vertical device using La0.7Sr0.3MnO3 and Co as the two

electrodes’ materials and tris(8-hydroxyquinoline)aluminium(III)

(Alq3) as organic spacer. In this case a significant MR was

found at low temperature for an organic layer thickness of

about 100 nm and persisting, although severely reduced, up to

thicknesses of about 200 nm. Considering that the typical

mobility of Alq3 (10�5 cm2 V�1 s�1, n-type) is significantly

lower than that of T6, the possibility of spin-injections appeared

even more puzzling.

An additional puzzling aspect of this second early experi-

ment was the fact that the MR had a negative sign. If one uses

the standard definition of MR, MR ¼ Rð0Þ�RðHÞRð0Þ , where R(H) is

the device resistance in a magnetic field H, then negative MR

means that an increase in magnetic field from H = 0 produces

a decrease in the device resistance. Since the two abrupt

changes in resistance in the R–H plot are attributed to the

switching of the magnetization of one of the two electrodes

towards the external field, one then concludes that the resistance

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3341

is lower when the magnetization vectors of the two electrodes

are anti-parallel to each other. This is in contrast to several

experimental observations in inorganic spin-valves,68,69

although it can be accounted for by a particular band alignment

between the organic layer and the electrodes.70

In fact by assuming tunneling transport one can express the

MR in terms of the electrode spin-polarizations, P1 and P2,

MR ¼ Rð0Þ � RðHÞRð0Þ ¼ 2P1P2

1þ P1P2; ð2Þ

where it is understood that the H = 0 state corresponds to a

perfect antiparallel alignment of the magnetization vectors of

the two electrodes, while R(H) is taken for a perfect parallel

configuration.38 If the spin-polarizations of the two electrodes

have opposite signs (one electrode is a spin-majority conductor

while the other is a spin-minority one), then a negative MR

will be expected. Such an interpretation however is in contrast

to the experimental evidence for majority spin injection at the

Co/Alq3 and Co/Al2O3/Alq3 interfaces42 and to the widely

accepted fact that La0.7Sr0.3MnO3 is a majority conductor.

Since negative MR has been found in many La0.7Sr0.3MnO3/

Alq3/INS/Co junctions (INS =Al2O3, LiF)51,67,71–73 a common

explanation must be found. Two elements may contribute

to solving the puzzle. Firstly one has to note that the spin-

polarization entering the Julliere’s formula [eqn (2)] is not a

well defined quantity. In particular it does not depend solely

on the electronic structure of the electrodes, but also on that of

the organic channel and on how the two materials interact

with each other. Secondly, eqn (2) is strictly valid only for

tunneling transport and so it may be inappropriate to use it for

a spin-injection problem. Both these hypotheses have been

recently verified. In support of the first idea a recent well-

controlled tunneling experiment on La0.7Sr0.3MnO3/Alq3/Co

junctions with nanoscale cross-section has shown large positive

MR at low temperature,74 interpreted in terms of the particular

bonding structure between Alq3 and the electrodes. At the

same time a model requiring majority spin-injection at both

the Co and La0.7Sr0.3MnO3 electrodes, but including the

electronic structure details of the Alq3 layer and its hopping

conductance, allows one to explain the negative MR.67

Although I will return to the two issues of the sign of the

MR and of the bonding between the organic and the magnetic

electrodes in the next sections, here I wish to stress that such a

controversy unveils our incomplete understanding around the

microscopic origins of the MR in organic materials. One

crucial aspect is that constructing a level-energy diagram for

an organic/inorganic interface is usually quite challenging. Let

us see in some detail what are the problems that might arise.

The first step in producing a level diagram is that of aligning

the Fermi level of the electrode with the valence band of

the semiconductor, or for extremely localized states, with the

molecule highest occupied molecular orbital (HOMO). This is

in itself a difficult task since chemical reactions might occur at

the interface and the level pinning might be determined by

hybrid interface states.75 Similarly interfacial electrical dipoles

may significantly alter the vacuum level alignment76 and

so may electron charging.77 Intriguingly such difficulties in

determining the alignment of the occupied states across the

junction also highlight the abundance of possibilities that the

organic world has to offer to spintronics. One can for example

envision situations where interfacial dipole engineering is

performed in order to improve the spin-injection efficiency

or to produce currents with a spin-polarization superior to

that of the magnetic metals forming the device. Likewise,

dipoles can be used to engineer the magnetic properties of a

metallic surface, as already obtained by using intense electric

fields.78 Notably it was recently demonstrated that functional

group engineering can be used to manipulate also the coercive

field of a permalloy,82 an effect that can be exploited for

the construction of future devices with tailored magnetic

properties.

The final step in drawing the level diagram consists in

aligning the organic semiconductor conduction band (or the

lowest unoccupied molecular orbital—LUMO) with respect to

the electrodes’ EF. The problem here is that one cannot simply

add the optical-gap to the HOMO, since the exciton binding

energy is crucial for photon absorption but bears little relevance

for electron transport. Importantly, misplacing the LUMO

may result in an erroneous assignment of the character

(electron or hole) of the injected electrons. Sometimes, but

this analysis is rarely carried out, a combination of direct and

inverse photo-emission can help in deducing the transport

HOMO–LUMO gap.83 Advanced electronic structure theory

based on first principles methods can also be a powerful tool

in hand. However, its use is complicated by the fact that a

correct description of the quasi-particle states is guaranteed

only when many-body methods (such as the GW scheme)

Fig. 3 Magneto-resistance of the La0.7Sr0.3MnO3/Alq3/Co spin-valve

reported in ref. 51. The top panel present the R–H plot defining the

magneto-resistance, the bottom left panel is the device scheme while

the bottom right one shows the dependence of the MR with the

layer thickness. Z. H. Xiong, D. Wu, Z. Valy Vardeny and J. Shi,

Nature (London), 2004, 427, 821–824. Copyright (2004) by the Nature

Publishing Group.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3342 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

are combined with an energy theory (such as density functional

theory—DFT).84 In this case however the computational over-

heads are significant and only systems comprising a limited

number of atoms can be described.

Before closing this section I wish to make some final

remarks on the role played by the interface between the

organic material and the magnetic electrode in determining

the spin-transport properties of spin-valves. In general it is

commonly accepted that maintaining good vacuum conditions

improves the device performance and increases the MR. This

fact is usually associated with a better quality of the electrode

itself. It is also true that in many vertical spin-valves incorporating

organic molecules the deposition of an inorganic insulator

(typically Al2O3) in between the bottom electrode and the

organic layer greatly enhances the MR.42 The reason for this is

not clear. Certainly organic materials in direct contact with a

metallic electrode may form a dipole layer at the interface.

This is widely documented for tunnel junctions79 and organic

light emitting diodes.80 One of the consequences of the dipole

layer is the formation of gap states, which may participate in

the electron conduction. Importantly gap and surface states

can be drastically modified by surface treatment. For instance

it has been recently demonstrated that surface treatment

in a rubrene-based field effect transistor may reduce the

electron trapping time at the interface by more than one order

of magnitude.81 This suggests that the quality of the interface

may significantly affect the transport properties of these

devices.

However, it is indeed much less clear how the same effects

influence the spin-transport properties, for instance the spin-

polarization. For inorganic semiconductors it has been demon-

strated theoretically that tunnel contacts can significantly

improve the spin injection.37 As such the manipulation of

the interface barrier certainly plays a role. To date however

our understanding is still incomplete and in particular it is

not known how the specific surface chemistry affects the spin-

polarization of the current.

2.4 Recent experimental advances and new directions

The early experiments described in the previous section kicked

off the field of organic spintronics, but at the same time left a

number of fundamental issues without a definite solution. The

search for these solutions is the main focus of the most recent

literature. In particular three conceptual questions remain:

1. Can tunneling be ruled out and can spin-injection be

uncontroversially demonstrated in organic junctions?

2. What is the main source of spin-relaxation?

3. What does determine the MR sign?

2.4.1 Tunneling or spin injection?. Demonstrating with

certainty whether the transport is dominated by tunneling or

spin injection is a rather subtle issue. The results of a recent

experiment by Barraud et al.74 help in putting the discussion

on a quantitative ground. The experiment consists in fabricating

Alq3-based tunnel junctions by nano-indenting a large-area

device with an atomic force microscope tip. In this particular

case the electrodes are respectively La0.7Sr0.3MnO3 and Co, so

that the junctions are composed of the same materials used in

earlier works.51,67 The only difference is that the nano-indentation

process allows one to define vertical devices with a rather

confined cross section (a few nm in diameter), and most

importantly to monitor the device resistance as a function of

the Alq3 thickness so that tunneling can be established with

certainty. Then, a huge MR is found at low temperature

and surprisingly this has a positive sign. I will come back to

the issue of the sign later, for the moment I will focus the

discussion on the resistances involved in the experiment.

The typical layer resistance of a large-area organic spin-valve

is about 25 kO mm�2, while the resistance of a single tunnel

junction with an Alq3 thickness of 2 nm is of the order of 108 O(see ref. 74). With these numbers in hand we can estimate that

the presence of tunnel junctions of this type, separated from

each other by about 20 mm, in an infinite resistive layer will

produce the same layer resistance of a typical large-area

organic spin-valve. The detection of current hot spots of a

few nanometres in size at such a low density is not an easy

task. It is then legitimate to ask whether the large-area

experiments measure spin-injection or simply spin-tunneling

through hot spots.

This is not simple to establish in organic materials. In

contrast to inorganic semiconductors in fact standard optical

pump–probe methods cannot be applied since they all rely on

spin–orbit interaction. Also the typically large resistivity in

organic materials seems to preclude the detection of either the

spin Hall or the Hanle effect, and more in general the use of

four probe non-local measurements. Thus one has little access

to a direct spatially resolved measure of the spin propagation

within the organic medium and has to depend on a less

direct characterization. Let me describe what are the options

available.

Firstly one has to conduct a thorough electrical charac-

terization and establish the dominant transport mechanism. In

particular the study of the dependence of the I–V curve on the

temperature and on the organic layer thickness is emerging as

a standard element of metrology for these junctions, although

it is not always conducted. When the analysis is carried out the

results appear broadly consistent with each other and point to

the fact that the MR is drastically reduced as one moves from

the tunneling to the diffusive transport regime. This has been

demonstrated true for a number of materials combinations:

(i) in CoFeB/Al2O3(1.5 nm)/Alq3/Co junctions85 the transition

from direct to two-step tunneling occurs for an Alq3 thickness

of about 2 nm, where the room temperature MR drops from

35% to B5% (see Fig. 4); (ii) in Fe/Rubrene/LaAlO3/

La0.7Sr0.3MnO3 junctions86 thicknesses in excess of 20 nm give

diffusive transport and small low-temperature MR (45%),

which disappears at about 100 K; (iii) similar results to those

obtained in Fe/Rubrene/LaAlO3/La0.7Sr0.3MnO386,87 have

been reported for Co/Al2O3/Rubrene/Fe;88 (iv) the persistence

of a low-bias room temperature 12% MR at the onset of the

diffusive transport has been demonstrated in CoFeB/MgO/

Alq3/Co.89 At the same time also some more negative results

have been reported with the complete absence of MR in the

diffusive transport regime for Fe/Alq3/Co junctions.90

Has the question been finally answered, i.e. has spin-injection

been demonstrated? It is important to realize that, although an

electrical characterization is a necessary condition to claim

spin injection, it is not a sufficient one. In fact one can still

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3343

have hot tunneling spots dominating the spin-transport over a

background of non-spin-polarized diffusive conduction. Thus

additional characterization must be performed. At present two

rather different methods have been used. These are respectively

two-photon photo-emission91 and muon spin resonance.92 The

key aspect is that both schemes possess some level of spatial

resolution in the spin-detection within the organic layer, so

that detailed evidence of the persistence of the spin-injection

can be collected. So far these two schemes have provided

ground for the spin injection in copper phthalocyanine91 and

Alq3,92,93 with typical spin-diffusion lengths in the few nano-

metre range. Hence, the next question is about what suppresses

the spin-polarization of the current in organic materials.

2.4.2 What is the main source of spin-relaxation?. First of

all let me recall how the spin diffusion length, lS, is commonly

extracted from a spin-valve measurement. The general idea is

to assume that there is little spin scattering at the inter-

face between the organic material and the electrodes and that

the spin polarization of the injected electrons is attenuated

exponentially as e�(d)/lS, where d is the layer thickness. This

leads to a modified version of the Julliere’s formula, which

now reads

MRðdÞ ¼ Rð0Þ � RðHÞRð0Þ ¼ 2P1P2e

�ðdÞ=lS

1þ P1P2e�ðdÞ=lS: ð3Þ

Eqn (3) is then used to fit the MR as a function of length. This

scheme however presents conceptual and practical problems,

which make the estimates of lS rather uncertain.94 Firstly the

Julliere’s formula is valid for tunneling but not necessarily for

diffusive transport. This means that the exponential decay of

the spin-polarization might not be appropriate for organic

materials. Secondly, there is now a substantial body of work

suggesting that the interfaces play a fundamental role in the

spin injection by acting either on the structural, the electronic

or both structural and electronic properties of the magnetic

electrodes. As such the spin polarization Pi becomes an ill

defined quantity and little can be said about lS.

Still, even with these uncertainties, the modified Julliere’s

formula can be used to obtain a semi-quantitative under-

standing of the spin-diffusion length, which typically in organic

materials is around a few nanometres. The question is then,

what is the limiting factor determining such a particular length

scale. Also in this case the answer is not trivial. As mentioned

before both the spin–orbit and the hyperfine interactions

are expected to be small in organic media. On the one hand

this means that the spins will interact weakly with their

environment and on the other hand that both the interactions

may contribute equally. Let us have a quick overview on the

current experimental situation.

There is one main argument supporting the spin–orbit

interaction as the main source of spin relaxation in organic

materials.95 This is based on a number of experiments92 reporting

a fast decrease of lS with temperature. Since the energy scale of

the nuclear spins dynamics is tiny, one should not expect a

severe temperature dependence of the hyperfine contribution

to lS. As a consequence the hyperfine coupling is ruled out and

spin–orbit interaction should be considered as the main cause

for the spin relaxation. The specific mechanism for the spin

de-phasing is then identified as the Elliott–Yafet96 one.52 The

identification is based on an observed enhancement of lS with

system confinement, which is consistent with the inverse

dependence of lS on the mobility predicted by the Elliott–Yafet

mechanism. A second piece of supporting evidence is provided

by the dependence of lS over the electric field.95 Although

this analysis is compelling, there are also several experi-

ments where lS varies only slowly with the temperature,66,67

which then brings the hyperfine interaction back into the

race.

In organic materials the primary channel for hyperfine

interaction is opened by the hydrogen protons via super-

exchange between the conducting p orbitals of carbon and

the s electrons of H.97 A key experiment demonstrates the

relevance of hyperfine interaction for the spin-diffusion in

organic materials.62 This consists in measuring both magnetic

resonance (optically detected) and the MR of a polymer in

either the hydrogenated or the deuterated form. Since the

deuterium nuclear magnetic moment is about 1/4 of that of

hydrogen, deuteration is expected to narrow the linewidth of

the photoluminescence peak as a function of an external

magnetic field. Furthermore deuterated polymers are expected

to give a significantly larger MR than the hydrogenated ones

in spin-valves of similar thicknesses. This indeed was observed

in the recent work of Nguyen et al.62 and in those of a few

other groups.98,99 As such, the importance of hyperfine inter-

action for the electron spin dynamics in organic materials

appears to be established with some certainty. However, also

in this case there is negative experimental evidence. In fact a

recent measurement on the response of an Alq3-based organic

light-emitting diode to a magnetic field suggests no variation

of the signal with deuteration, indicating little hyperfine

coupling with protons.100

In conclusion, although there is now significant experimental

evidence for both a hyperfine and a spin–orbit related spin

Fig. 4 Room-temperature resistance (circles, right axis) and MR

(squares, left axis) for variable Alq3 thicknesses. The junction is

CoFeB/Al2O3(1.5 nm)/Alq3/Co. The figure is from ref. 85. J. J. H. M.

Schoonus, P. G. E. Lumens, W. Wagemans, J. T. Kohlhepp,

P. A. Bobbert, H. J. M. Swagten and B. Koopmans, Phys. Rev. Lett.

2009, 103, 146601. Copyright (2009) by The American Physical Society.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3344 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

diffusion mechanism, there is still no consensus on what

is the primary interaction responsible for the spin diffusion.

This may simply point to the fact that the dominant mechanism

is materials specific, i.e. it may vary from material to material.

An interesting observation along this direction is the fact that

to date there is no single device displaying at the same time

both spin-valve behaviour and OMAR. More investigation is

certainly still needed on this particular aspect.

2.4.3 What does determine the MR sign?. Let us now return

to the discussion about the sign of the MR. This is an issue

that has puzzled the field of organic spintronics since its birth.

The puzzle stems from the fact that the same materials

combination, namely La0.7Sr0.3MnO3/Alq3/Co, can produce

either negative51,67 or positive74 MR. This fact is incompatible

with the Julliere’s formula, unless one assumes that the sign of

the spin-polarization of the injected electrons at one of the

two interfaces may change from device to device. Chemical

bonding can help in achieving this and the argument proceeds

as follows.

The spin-polarization entering the Julliere’s formula for

tunneling effectively corresponds to the electrodes DOS spin-

polarization (organic barriers are likely to be amorphous so

that the Julliere’s formula in the pure tunneling limit holds33).

However hybrid surface states may form as the result of the

interaction between the molecular orbitals of the organic

media and the electrodes’ magnetic surfaces. If these appear

at the Fermi level, they may alter the spin-polarization of the

tunnel electrons to a point that Pmay coincide with that of the

hybrid states themselves. The cartoon of Fig. 5 helps in

understanding this point.

In panel (a) the DOS of the electrodes and that of the

molecule in isolation are both presented, where for the sake of

simplicity I assume that only the HOMO is responsible for

the conduction. With this particular level alignment EF of

the electrodes is not resonant with any molecule states, the

transport is tunnel-like and the spin-polarization of the tunnel

carriers, P, coincides with that of the electrodes. When the two

materials are brought together interaction takes place and

two new main features appear. Firstly, the molecule DOS

broadens, reflecting the fact that the lifetime of the molecular

orbitals becomes finite as electrons can be transferred to and

from the electrodes. Such a broadening is spin-selective as it

depends on how strongly a particular molecular orbital interacts

with the extended electronic states of the metal. Since in

magnetic metals the Fermi surfaces of majority and minority

spins are different also the interaction with the molecular

orbitals may be different. In the cartoon for instance [panel

(b)] it is the minority spin-band that interacts more strongly

and consequently the broadening is larger for that spin direction.

Note that such a spin-dependent level broadening brings

only the minority component of the HOMO at the electrodes’

EF, so that P will change sign with respect to the situation of

panel (a).

The second effect is a shift of the molecule DOS with respect

to the electrodes’ EF. Since such a shift can also be spin-

dependent one may end up with a new spin-polarized molecular

orbital at EF. This case is presented in panel (c). Now the

largest DOS at the Fermi level for the hybrid state has

majority character and the sign of P changes again to return

to that of the situation described in panel (a). Importantly

if P is determined by an hybrid state the strength of the

interaction between the molecule and the electrodes also sets

the energy scale of the MR. Thus it is not surprising that in the

experiments of Barraud et al.74 the MR halves at only 25 mV

and disappears completely at 180 K.

The exciting prospect arising from the fact that the MR of

an organic device can be modified by molecular bonding is

that such a modification may be engineered. One can then

envisage specific custom made chemistry aimed at improving

the performances of organic spin-valves and of spin-devices in

general. This is an extremely tantalizing opportunity which has

given birth to the suggestive new field of Spinterface science.101

A final comment should be made on the case where the

transport is diffusive. In this situation spin-polarized tunneling

is only the first step of the transfer of a spin from one electrode

to the other. This is followed by diffusion and by a second

tunneling process to the drain. However, apart from possible

spin relaxation, no other event can change the spin-polarization

of the injected electrons. Furthermore, even if inelastic

processes occur, the vast majority of the electrons will exit

the organic media approximately at the same energy they

enter, since little electric field builds up in the junction. As

such one expects that the arguments provided above will still

hold true for the spin-injection limit.

2.4.4 Hybrid devices. In the introduction I have pointed

out that one of the benefits of using molecules for spin

Fig. 5 Schematic of the formation of an hybrid state between a

magnetic metal and a molecule: (a) When the magnetic metal (left) and

the molecule (right) are well separated, the overall DOS is simply the

superposition of the individual DOS of the two spin components

(blue represents the spin-up DOS and red the spin-down DOS)—that

is, a broad spin-polarized DOS for the metal and a series of discrete

energy levels for the molecule (here only the HOMO is represented). In

this case, the DOS of the metal alone determines the spin-polarization

of the tunnelling current. (b,c) When the molecule is brought into

contact with the metal the DOS gets modified in two ways: the energy

levels broaden (b) (the broadening is exaggerated in the figure) and

their position shifts in energy (c). In both cases new peaks in the DOS

might appear at the EF of the electrodes, arising from new hybrid

interfacial states. It is this new DOS that determines the spin-

polarization of the injected current, which can be dramatically different,

and even reversed, compared with the polarization of the electrodes

(as in b). Stefano Sanvito, Nature Physics, 2010, 8, 562–564. Copyright

(2010) by the Nature Publishing Group.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3345

applications is that molecules come with an immense range of

properties. However in all the discussion up to now the

organic media have simply replaced their organic counterparts

but no additional functionalities have been exploited. Some

very recent experiments however have explored this second

possibility and in particular have looked at magnetic and

configurational bistable molecules.

Robust magnetism well above room temperature occurs

rarely in organic materials, mainly because the typical density

of the magnetic ions is small and the magnetic interaction

is relatively short-ranged. A notable exception is that of

vanadium(TCNE: tetracyanoethylene)x, V(TCNE)x (x B 2),15

which orders ferromagnetically up to around 400 K. The

magnetic order originates from the interaction of the unpaired

p* orbitals of the (TCNE)� anion with the spins of V2+,

resulting in a magnetic moment of 1 mB per formula unit.

V(TCNE)x is also believed to be a half-metal so that it appears

as an ideal spin-injector for organic spin-valves. The idea is

also particularly attractive since the resistivity of V(TCNE)x is

considerably larger than that of a typical 3d magnetic metal,

so that the resistance mismatch obstacle36 should play only a

minor role.

Two experiments have successfully attempted to use V(TCNE)xas an injector of spin-polarized carriers. In the first102 the

spin-polarized electrons injected from V(TCNE)x in a GaAs/

AlGaAs light emitting diode are recombined to produce

polarized light. From the analysis of the light polarization as

a function of an external magnetic field it is found that

majority electrons are injected but the injection efficiency

remains quite low. Importantly the magnitude of the spin-

polarization of the current appears to be rather insensitive to

the temperature and to the bias.

In the second experiment103 a V(TCNE)x/rubrene/LaAlO3/

La0.7Sr0.3MnO3 spin-valve is made and the MR is recorded

(it is the order of 2%). The R–H plot shows the typical spin-

valve butterfly shape with the switching fields corresponding

to the electrodes coercive fields, i.e. the spin-valve MR effect is

established with certainty. Intriguingly the MRmagnitude first

increases with temperature up toB100 K and then it decreases

until vanishing at a temperature between 170 K and 220 K,

depending on the precise device stack. Such a non-monotonic

behaviour of the MR with temperature is ascribed respectively

to the temperature dependence of surface magnetization

of La0.7Sr0.3MnO3 for T 4 100 K and to the V(TCNE)xresistance increase as T decreases (for T o 100 K). Again the

efficiency of the injection seems to be rather temperature

independent in the range investigated.

These two results are extremely interesting since they

provide the first demonstration of spin-injection from an

organic magnet. As such they pave the way for all organic

spin-devices. These can be fabricated with inexpensive chemical

methods and potentially they can bring additional function-

alities to their inorganic counterparts. Further development

however is hampered by the limited choice of organic magnets

with Curie temperatures well above room temperature,

although the progress in the field in the last decade has been

significant.

A second fruitful research line is that aimed at producing

multi-functional junctions, i.e. junctions which react to stimuli

of different origin (say electrical and magnetic). Many examples

exists in the near field of molecular electronics and concern

devices made with molecules that can switch between alternative

geometrical configurations presenting different electronic

properties.104 This means that the I–V curve of the device

changes abruptly when the molecule is driven into the various

states.105 Generally the molecules are photo-switchable so that

most of the research is driven by the prospect of making smart

molecular opto-electronic devices. However a similar switching

activity can be triggered by an intense static electric field or by

an electrical current. In particular this can alter the morpho-

logy of the transport medium and change its resistance. Such a

principle has been used to manipulate the spatial distribution

of oxygen vacancies in conducting oxides leading to the

discovery of the memristor.106 A memristor is a essentially a

resistor showing hysteresis, i.e. it is the fourth basic circuit

element.107

A recent experiment has taken these concepts and

applied them to organic spintronics.48 Again the device is a

La0.7Sr0.3MnO3/Alq3/Co spin-valve. The main feature however

is that in addition to the standard spin-valve MR effect,

there is an irreversible switch of the I–V curve both at positive

and negative bias (see Fig. 6). The crucial point is that the

MR is different along the different irreversible branches

of the I–V curve,108 meaning that the junction is program-

mable in different conducting states, each one of them

presenting a different magnetic response. The potential for

this device platform is quite vast since both multi-state

memories and memristive logic devices can be envisioned.

Again this line of investigation is still at an early stage, but

progress is expected to arrive in a near future. In particular a

potentially revolutionary research line may open following the

recent development of both room temperature ferroelectric

molecular crystals16,109 and of magnetically controllable

organic ferroelectrics.110

Fig. 6 Current/voltage room temperature characteristics for a

La0.7Sr0.3MnO3/Alq3/Co spin-valve displaying voltage memory effects.

Note the irreversible switching both at positive and negative bias. The

spin-valve presents different MR over the different irreversible branches

of the I–V. L. E. Hueso, I. Bergenti, A. Riminucci, Y. Zhan and

V. Dediu, Adv. Mater., 2007, 19, 2639–2642. Copyright (2007) by

WILEY-VCH Verlag GmbH & Co. KGaA.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3346 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

3. From organic spintronics to single-molecule

spintronics

The discussion presented up to this point, concerning mainly

macro- and mesoscopic devices, has revealed the potentialities

but also the complexity of organic spintronics. One clear

message is that the study of spin-transport in organic devices

suffers from the lack of accurate microscopic characterization

tools. Partially for this reason and partially for creating a

natural extension of the blossoming field ofmolecular electronics111

to spin phenomena, the last few years have witnessed a growing

interest in the study of spin-transport in single molecules.

These include both the implementation of standard spintronics

devices (for instance the spin-valve) at the single molecule

level, but also the study of electron transport through three

terminal molecular devices112 and through molecular objects

possessing internal spin degrees of freedom. This field, parallel

to that of organic spintronics, takes the collective name of

Molecular Spintronics.10

Interestingly, organic and molecular spintronics nicely

complement each other, with the advantages of one field being

the drawbacks of the other. Thus, while single molecule

measurements are intrinsically local, so that they are easy to

relate directly to the geometry and the electronic structure

of a device, they have usually low yield, since the electrical

contact between the molecules and the electrodes is difficult to

establish. In contrast in large-area devices many molecules are

contacted at the same time so that the device properties are

relatively uniform from device to device, but the microscopic

understanding of the device operation is often incomplete.

This holds true also for theory and modeling. Large-area

devices are usually described by macroscopic transport theories

implemented on parametric model Hamiltonians.57,85,113–115

These require a number of adjustable parameters, which can

often be well inferred from the experimental data. In contrast

experiments conducted with single molecules are more suitable

for first principles transport theory, which provides a quantitative

description without any free parameters.116

In this final part of the review I have taken a rather

unorthodox approach in presenting the field. Firstly, I have

looked at STM measurements for molecules on surfaces.

These are extremely important since they provide spatially

local information about the spin-transport, so that one can

understand fully how spins cross not just the molecule, but

its constituent parts. As such STM measurements may be

extremely informative on the microscopic physics/chemistry of

large scale devices. Then I have considered devices made with

single molecule magnets. Note that to date the experiments on

these have been quite limited so that an exhaustive treatment is

not possible. For this reason I have preferred simply to

introduce the most relevant concepts. This is indeed a personal

perspective. In doing so I have not discussed experi-

ments relating to the Kondo effect117,118 for which extensive

reviews have been already written.119 Finally I have decided to

introduce spin crossover molecules for which little published

experimental work addressing electron transport exists.

Therefore the last part concerns mostly theoretical considera-

tions and somehow has the goal of stimulating experimental

activity.

3.1 SP-STM and Spinterface

One of the most direct ways to measure the spin-polarization

of the electron current emerging from a magnetic surface

through a single molecule is provided by SP-STM.27 This

essentially consists in performing a spin-valve experiment

where one of the electrodes is the scanning tip and the second

one is the magnetic substrate hosting the molecule. State of the

art SP-STM has a spatial resolution finer than the typical

molecule bond-length so that the atomic details of the spin-

injection can be investigated. The only disadvantage of

SP-STM with respect to other transport techniques for single

molecules is that the samples need to be prepared in ultra-high

vacuum (UHV), i.e. it is poorly compatible with wet chemistry.

Fortunately many planar molecules are compatible with UHV

deposition (porphyrins, phthalocyanines, salens, etc.). These

can be prepared in a large variety of configurations and can

incorporate transition metal centers. As such they form an

important class of materials attracting a rapidly growing

interest.120–126

SP-STM can provide a direct proof of the manipulation of a

spin-current by chemical bond tailoring, i.e. it is a preferential

tool of Spinterface science. As in a standard magnetic tunnel

junction the current in SP-STM depends on the mutual

orientation between the magnetization of the tip and that of

the substrate. SP-STM however possesses additional spatial

and spectral resolution, i.e. the tunneling current changes with

bias and with the position of the tip with respect to the

molecule. As such it allows one to obtain a spatial and energy

mapping of the spin tunneling process. It is then possible to

determine which particular molecular orbital contributes the

most to the net spin polarization of the current. Intriguingly,

since different molecular orbitals may present DOS with

opposite spin-polarization (in particular if hybrid with the

electronic states of the substrate) and they may be spatially

distributed over different portions of the molecule, opposite

spin-polarization may be detected for the same molecule at

different tip positions. This was theoretically predicted some

time ago127 and recently demonstrated experimentally.128,129

Indeed the recent data from Atodiresei et al.128 and

Brede et al.129 show clearly not only that the spin polarization

of the current emerging from an organic molecule (either

Phthalocyanine128 or Co-Phthalocyanine129) can be opposite

to that of the magnetic substrate on which the molecule is

adsorbed [two monolayers of Fe on W(110)], but also that

different regions of the molecule can sustain different current

spin-polarization. The orbitals responsible for both the chemical

bond and the conductivity of the Fe surface have dz2 symmetry.

These are spin split with the current being dominated by

minority electrons. The bond with the molecule is between

the Fe dz2 and the Phthalocyanine pz orbitals and brings

additional majority DOS at the Fermi level (via the energy

level shift and broadening mechanism described in section 2.4.3).

As a consequence the spin-polarization of the composite

molecule+substrate is reversed with respect to that of the

Fe substrate (see Fig. 7).

The ability of STM to investigate surface magnetism however

does not stop here, as STM can also be used to extract detailed

information on the magnetic excitations. This is enabled

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3347

by spin-flip inelastic tunnel spectroscopy (SF-IETS).130,131

SF-IETS is based on a principle common to all inelastic

spectroscopy techniques, namely that a change in the electrical

conductance occurs every time the bias voltage reaches the

critical value characteristic of a given excitation. At that

particular bias a second channel adds to the elastic tunneling.

This is given by the inelastic process, where an electron

exchanges energy with the molecule, exciting it. Whether

or not the opening of an inelastic channel increases the current

depends sensibly on the details of the junction. In the case

the molecular excitations are phonons, propensity rules have

been established correlating the symmetry of the various

vibrational modes and of their coupling to the electrodes to

the transmission.132,133

In SF-IETS the molecular excitations have spin origin and

they involve the flipping of the spin direction of the current-

carrying electrons. At variance with the phonon case, a spin-

transition must satisfy some selection rules, so that the theory

becomes more informative. However the energy scale of the

magnetic excitations is typically in the sub-meV range, i.e. it is

more than one order of magnitude lower than that of the

molecular vibrations. As such SF-IETS requires severely

low temperatures. SF-IETS was pioneered at IBM with

experiments conducted on magnetic ions deposited over

metallic surfaces covered by a thin insulating film.130 Landmark

results include the measurement of the magnetic coupling134

and of the magnetic anisotropy135 of single atom chains. Most

recently the same technique has been employed for planar

magnetic molecules deposited on surfaces. The investigation of

the details of the super-exchange mechanism in Co phthalo-

cyanine atomic layers deposited on Pb136 and of the charging

state of the same molecules137 are some examples of this

research. It is important to note in this context that SF-IETS

works best in the limit of weak electronic coupling between the

magnetic center and the electrodes. In the opposite situation

the transport, at least at low bias, is dominated by the Kondo

effect. This should deserve a review in itself,119 here we just

wish to mention that the Kondo effect in magnetic molecules is

well established and can be tuned by STM manipulation of

the molecule itself.138 Also it is important to remark that a

quantitative parameter-free theory of the Kondo effect has

begun to emerge.139

3.2 Single molecular magnets devices

Although STM is a powerful tool for understanding the basic

mechanism of the magnetic interaction between a molecule

and a substrate, it is not a device fabrication platform. In

contrast two- and three-terminal junctions incorporating

magnetic molecules are closer to real devices since both the

spin and charging state of the molecule can be altered. These

are usually fabricated by combining breaking junction type

technologies with wet chemistry and, as such, one does not

need any longer UHV growth conditions. However, when

moving away from planar molecules and UHV, other pro-

blems emerge.

A rather fundamental one is the fact that, despite many

classes of magnetic molecules can be chemically synthesized,140

most of them are extremely fragile away from solution and

often react on a metallic surface.141,142 Furthermore some

molecules can lose entirely their magnetic moment by

coupling with the electrodes, even if they remain intact.143

This creates a substantial uncertainty since it is difficult to

establish whether the molecule entering a device is the same

one that was intentionally designed for that device. Rapid

progress however has been made in constructing robust

magnetic molecules and it has been already demonstrated

that members of the Fe4 family can survive on surfaces and

preserve both their spin-state and most of the magneto-crystalline

anisotropy.144,145

Let us now assume that two- and three-terminal devices

incorporating magnetic molecules which preserve their single

molecular magnet (SMM) properties, can be made. The next

important question is: how should these devices operate? In

particular one has to understand which among the molecule

degrees of freedom must be used for operating the device

(reading, writing and manipulating information in logic

devices or reacting to an external stimulus in sensors). For

magnetic molecules the spin is clearly the degree of freedom of

choice, but then the question becomes, which property to use.

This is of course a crucial choice, since different properties rely

Fig. 7 The geometry and electronic structure of a free Co-Phthalocyanine

(a) and of the same molecule adsorbed on an Fe surface (b). Results in

(c) are for Co-Phthalocyanine adsorbed on Fe when the calculations

include van der Waals interaction (this situation describes better the

experimental finding). Note that, while in (a) the spin-polarization of

the DOS is dominated by the substrate minority dz2 orbitals, an hybrid

state forms in (c) whose spin polarization is opposite, i.e. it is mainly

due to majority spins. J. Brede, N. Atodiresei, S. Kuck, P. Lazic,

V. Caciuc, Y. Morikawa, G. Hoffmann, S. Blugel and R.Wiesendanger,

Phys. Rev. Lett., 2010, 105, 047204. Copyright (2010) by The American

Physical Society.Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3348 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

on different interactions, which in turn set the energy scale

(and thus the temperature) of the device operation.

Consider for instance the case of a spin-valve. The two key

ingredients to make the device functioning are the ability of

the electrodes to produce a spin-polarized current and the

possibility of altering the mutual orientation of the electrodes’

magnetizations vectors. The first property depends on the

exchange interaction and therefore survives up to tempera-

tures close to the material’s Curie temperature. The second

one relies on the magneto-crystalline anisotropy, hence on the

spin–orbit coupling. If one now wishes to scale the spin-

valve concept down to the molecular level a problem will

immediately appear, namely that the magneto-crystalline

anisotropy in magnetic molecules is usually small (the typical

energy barriers are in the few Kelvin range), mostly because

only a handful of atoms carry the magnetic moment. As a

consequence the molecule is likely to behave as a paramagnetic

object at any reasonable temperatures and the resulting

current will not be spin-polarized.

Also in this case, however, recent progress seems to indicate

a more positive outlook. For instance many theoretical studies

have pointed to transition metal multidecker clusters (TM–P

with TM= Sc, Ti, V, Ni, and P = cyclopentadienyl, benzene)

as possible sources of spin polarized electrons.143,146–149 An

alternative and intriguing prospect is also that of constructing

devices with magnetic molecules whose magnetic anisotropy

produces little or no magnetic moment but a definite toroidal

moment.150 There are predictions for such molecules to be able

to switch the spin-polarization of an injected current151 and

still to remain protected from dipolar-interaction, one of

the main sources of spin-relaxation in molecular crystals

(note that here I refer to the magnetic spin-relaxation of the

molecules and not to the spin-diffusion of current carrying

electrons).

A second and possibly more promising strategy for fabricating

devices by using SMMs is that of exploiting the robust

exchange interaction instead of the tiny magneto-crystalline

anisotropy. This essentially means to use the spin state of the

molecule as the physical property to read, write and manipulate.

Clearly one then has to be able of addressing (reading, writing

and manipulating) the spin state of the molecule without

the need of maintaining the spin-quantization axis fixed,

i.e. without the need of strong magneto-crystalline anisotropy.

I will first consider the problem of reading the magnetic state,

while a discussion over the possible strategies for writing and

manipulating will be reviewed in the remaining sections. The

general idea is to convert spin information into molecular

orbital information. These are then detected by an electrical

current. In practice one wants to demonstrate that the different

spin-states of a molecule present different frontier molecular

orbitals or different coupling to the electrodes, both features

that might affect an electrical current. Again, because of

the unlikely possibility of fixing the molecule anisotropy axis,

the electrical read out of the molecules’ state needs to be

done without using a spin-polarized current, i.e. by using

non-magnetic electrodes.

The possibility for such a non-spin-polarized electrical read

out of the magnetic state of a single molecule magnet was

recently explored theoretically for a prototypical two terminal

device incorporating a Mn12 molecule.152 In particular it

was demonstrated that the S = 10 ground state (GS) can

be distinguished from a spin-flip state (SFS) obtained by

reversing the magnetic moment of both a Mn3+ and a

Mn4+ ion relatively to their GS orientation (the total spin

projection of the SFS is 9). This is possible because the frontier

molecular orbitals of a particular magnetic state re-hybridize

differently under-bias, so that the I–V presents low-bias

negative differential resistances (NDRs) characteristic of the

molecule spin state.

In order to understand this concept in some more detail

consider Fig. 8(a) where the DOS for the two magnetic states

are presented. If one neglects the spin polarization, as for a

paramagnetic molecule, the DOS of the GS and the SFS

appear rather similar, so that one may expect similar I–Vs.

This however is not the case, as demonstrated by the calculated

I–V curve shown in Fig. 9.

The presence of NDR in the I–V curve is the result of orbital

polarization under bias. Consider for instance the spatial

Fig. 8 DOS around the Fermi level (vertical line at 0) for

[Mn12O12(CH3COO)16(H2O)4] in the ground state (a) and the spin-flip

state (b). The three small plots show isosurfaces of the HOMO wave-

function for the GS at different bias: top = 300 mV, center = 0,

bottom=�300 mV. Note the drastic polarization of the wave-function

in the electric field. C. D. Pemmaraju, I. Rungger and S. Sanvito, Phys.

Rev. B, 2009, 80, 104422. Copyright (2009) by The American Physical

Society.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3349

distribution of the HOMO wave-function as a function of the

bias, presented in the three small panels of Fig. 8. As the bias

is swiped from positive to negative the HOMO electronic

wave-function polarizes in order for the molecule to maintain

its local charge neutrality. Such a molecular orbital polari-

zation changes the degree of overlap between the HOMO and

the extended states of the electrodes, thus changing the

electronic coupling and hence the molecular state life-time. If

such a polarization is significant and the electronic coupling

with one of the two electrodes is drastically reduced, then the

current will decrease despite an increase in the applied bias.

This mechanism generates the NDR.

The quantum mechanical way to polarize a given molecular

orbital is to hybridize it with other available molecular orbitals.

These should be sufficiently close in energy and have an

appropriate orbital and spin-symmetry. Since in the GS the

HOMO multiplet is composed by states having the same spin,

while in the SFS some states have the opposite spin direction,

the number of molecular orbitals available for re-hybridization

is different and hence the NDR appears at a different bias. This

means that the response of a spin-state to an electric field is

driven by the dielectric response of its molecular orbitals, i.e.

spin information is translated into orbital information. Note

moreover that such an argument does not apply only to SMMs,

but more generally to situations where the HOMO is charac-

terized by a multiplet of closely spaced orbitals. This is for instance

the case of the non-magnetic dithienylethene molecules.105

Is there any experimental proof of the above mechanism?

It is quite difficult to answer to this question at the moment

since only a few transport experiments on SMMs have been

conducted to date. Still early measurements for Mn12incorporated in a three-terminal device153,154 reveal the

presence of low energy features (NDRs satellite to Coulomb

blockade peaks not involving different charging states), which

might be ascribed to orbital effects. At present however the

evidence is too little to call for a definitive explanation

and alternative theories based on selection-rule forbidden

transitions between different spin states have been brought

forward and are at present equally possible.155,156 In summary,

it appears that the demonstration of the electrical read out of

spin-states of a molecule has been provided both experi-

mentally and theoretically. Whether or not one will be able

to assign with certainty a given spin-state to a specific finger-

print in the transport and whether this will be controllable still

remains an open question.

In closing this section I wish to briefly mention another

device protocol different from the SMM-based three-terminal

junction discussed so far. This is based on grafting SMMs on

either graphene or carbon nanotubes and in investigating their

magneto-transport response.21 The concept underpinning such

a device is that the interaction between the molecule and the

transport channel (either the nanotube or graphene) produces

an electrical response which is sensitive to the magnetic state of

the molecule (spin and spin orientation with respect to the

channel). Also in this case one has to identify the strength of

the interactions available. In general both the spin of the

molecules used and its magneto-anisotropy should be as large

as possible, and this is why bis(phthalocyaninato)terbium(III)

complexes have been utilized so far.157 This however is enough

only to stabilize the molecule against thermal fluctuation, then

one has to make the molecule interacting with the channel.

Proposals to date include the detection of the magnetic

flux originating from the molecule stray field21 and the use

of robust chemical bonds able to produce spin-dependent

Fano-like resonances in the channel conductance spectrum.158

The first case may provide a technological platform for ultra-

sensitive magnetometry while the second one for magnetic

field sensors.

3.3 Manipulation of the spin state of a molecule

The last step in demonstrating the validity of the concept of

single molecule spintronics consists in designing a strategy

for writing and eventually manipulating the possible states

of a device. In the case of SMMs devices this translates

into the ability of altering the spin of a SMM, i.e. in the

ability of manipulating the exchange interaction in a control-

lable way. Note that this rather fundamental aspect underpins

not only the potential for new logic devices and sensors but

also the use of magnetic molecules as elements for quantum

computation.24,159

A general, always available, strategy for altering the magnetic

properties of a molecule is that of acting chemically.

For instance it was recently demonstrated that the magnetic

exchange between two Cr7Ni rings can be modified by

changing the molecular linker coupling the rings.160 An intriguing

aspect of this protocol is that the electronic structure of

the constituent magnetic elements (the Cr7Ni rings) is little

Fig. 9 The I–V curves for both the GS and the SFS of a Mn12two-probe device. Note the spin-state characteristic NDRs. C. D.

Pemmaraju, I. Rungger and S. Sanvito, Phys. Rev. B, 2009, 80,

104422. Copyright (2009) by The American Physical Society.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3350 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

affected by the linker, so that the freedom in the design of the

desired compound is rather large. One can then envision the

use of active linkers, i.e. of linkers whose electronic structure

can be changed by a local probe such as a gate. Examples of

these are the one-electron reducible [Ru2+Ru3+(OCCMe3)4]+

linker still for Cr7Ni rings160 and [PMo12O40(VO)2]q�molecules.161

The chemical strategy has also been successfully employed for

manipulating the spin state 162 and the magnetic anisotropy of

planar magnetic molecules on surfaces.124

Such a chemical approach, which may be a valuable option

for sensor applications, is not well suited for the electronic

ones. These in fact require a reversible switching mechanism

with relatively fast access times, two requirements hardly

matched by a chemical reaction. As a consequence one has

to look at reversible spin-manipulation triggered by an easily

accessible external stimulus, for instance electromagnetic

radiation or a static electric field. In both case the transition

between two different states should have electronic origin

(to be fast), although it might be also accompanied by con-

figurational changes (to be stable). The next sections will

discuss some of the present possibilities for switching magnetic

molecules and the effects that such switch produces on the

electron transport.

3.3.1 Spin-crossover compounds and valence tautomerism.

Demonstrations that the spin-state of a molecule can be

changed by an external stimulus are abundant. In fact there

exists an entire class of molecules, named spin-crossover

compounds, whose magnetic ground state can be altered by

light, temperature or pressure.163 These molecules incorporate

a single transition metal ion and display an entropy driven

low-spin to high-spin transition accompanied by a geometrical

relaxation of the first coordination shell around the ion. The

microscopic mechanism driving the spin-crossover is illustrated

in the top panel of Fig. 10 for the prototypical case of FeII. In

the low-spin configuration the six 3d-electrons occupy the

t2g levels (note that the Fe coordination is approximately

octahedral) in a spin-zero state (1A1g). By increasing the

temperature the competing high-spin state, where two electrons

are promoted to the eg levels, (5T2g), becomes thermodynamically

more stable. Such a phase transition is driven by the larger

entropy of the high-spin state. In particular there are two

contributions to the entropy: one is provided by the spin and

the second by the molecule vibrations. The t2g orbitals have

bonding nature and the eg are non-bonding, so that the

promotion of electrons to the eg shell weakens the chemical

bond and softens the phonon mode between the magnetic ion

and the ligands.

A special case among spin-crossover compounds is represented

by molecules presenting valence tautomerism.164 These are

complexes in which the crossover is obtained by an inter-

conversion between redox isomers.165,166 The lower panel of

Fig. 10 schematically describes the inter-conversion process in

Co-dioxolene. In the low-spin state the eight valence electrons

are distributed to give a diamagnetic Co(III)-catecholate

configuration. The entropy driven phase transition leads

to paramagnetic high-spin Co(II)-semiquinonate, which is

obtained from the low-spin configuration by transferring one

electron from the o-dioxolene group to Co.

These two broad classes of molecules are extremely interesting

as potential materials platform for molecular spintronics for a

number of reasons. Firstly, usually the two states accessible by

the compound are both long-living (at least at low temperature),

so that the molecules can be maintained in the desired spin

state for relatively long times. This essentially means that such

molecules have an intrinsic non-volatile nature. Secondly, since

the crossover always involves an internal charge re-arrangement

and an atomic relaxation, the two different molecular states

present rather different electronic structures. As such a two-

terminal device incorporating crossover compounds is expected

to display different I–V curves for different spin-state, i.e. the

spin-state is likely to be electrically readable. Finally, the cross-

over transition appears to be sensitive to the local electrostatic

environment,167,168 so that an electric field may be used to

induce the transition.169 This means that a potential device

might be electrically switchable.

What are the challenges for spin crossover compounds as

active materials for single molecule spintronics? These in

general are similar to those facing SMMs. Again the molecule

stability on the surface is a primary concern and there is no

fundamental reason to believe that spin crossover molecules

will be less fragile than SMMs. Certainly the crossover activity

Fig. 10 Energy level schemes for the low-spin to high-spin transition

in various spin crossover molecules. In the top panel the conventional

spin-crossover of FeII is presented, where the transition is driven by the

transfer of two electrons from the t2g to the eg levels. The lower panel

describes valence tautomerism in Co-dioxolene complexes. In this case

an electron is transferred from the o-dioxolene group to Co.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3351

is strongly dependent on the environmental conditions

(chemical and electrostatic conditions, wetting etc.), so that

a certain fragility appears intrinsic to these compounds.

This means that a molecule displaying spin crossover activity

in the single crystal phase may lose this property when

deposited on the surface, because one of the two states will

be no longer accessible. Secondly as the hysteresis in the

switching temperature is the result of the collective inter-

action among the molecules in a crystal, it is a property that

does not transfer to the single molecule world. Whether the

hysteresis will be possible for two-dimensional arrangement

of molecules on a surface, so that spin crossover compounds

may be used in large-area devices, is certainly not clear at

present.

To date the experimental activity on spin crossover

compounds as active elements for nano-junctions has been

quite limited although a few key results have already emerged.

However all the experiments showing spin crossover activity

do not involve single molecules but instead crystals or

nanoparticles, where the stability issues mentioned above

are less relevant. As such we still lack of an experimental

demonstration for a device made with spin crossover single

molecules. In any case it was first demonstrated that the

conductivity of a [Fe(qsal)2][Ni(dmit)2]3�CH3CN�H2O single

crystal170 has an hysteretic behaviour with temperature,

suggesting that the two spin states across the crossover

transition have a different resistivity. Secondly it was proved

that the spin state of FeII complexes containing pairs of planar

terdentate N ligands and immobilized on highly oriented

pyrolytic graphite is detectable by STM.171 Finally, very

recently it was shown that nanoparticles made of spin cross-

over molecules present a temperature hysteresis in their

conductivity, which also in this case is attributed to the

spin crossover transition.172 Intriguingly this last experiment

also demonstrated that the crossover can be induced by a

potential bias, i.e. by a static electric field. This finding paves

the way for electrically controlled spin devices at the molecular

level.

3.3.2 Electrostatic spin crossover effect. Another possibility

for manipulating electrically the spin of a molecule, alternative

to changing the spin of a single magnetic ion, is that of acting

on the exchange interaction between two or more magnetic

centers. This is the concept behind the electrostatic spin-

crossover effect (ESCE), first proposed for high electron

density molecular wires173 and then for polar molecules.174

The general idea of the ESCE is that the high-spin and the

low-spin state of a molecule can Stark shift differently if

they have different polarizabilities and, in the case of polar

molecules, also different permanent electrical dipoles. One can

then speculate that there exists a particular condition where a

high-spin to low-spin crossover is possible. The fundamental

question is how large is the crossover field.

In molecules with inversion symmetry the first order contri-

bution to the Stark effect vanishes and the difference between

the polarizabilities alone induces the crossover.173 Realistic

crossover electric fields, Ecross, are obtained only for molecules

with a large spin-contrast in the polarizability (i.e. the

polarizability of different spin-states must be very different).

This translates into a large charge density and a small

HOMO–LUMO gap, and brings two serious drawbacks.

First, one needs to pursue an extremely challenging strategy

for chemical synthesis in order to produce the desired molecules

(the ones proposed in ref. 173 were 10p benzene) and it is not

clear whether such chemical route will be ever available.

Secondly and most importantly, the small HOMO–LUMO

gap prevents the on-set of a large electric field in a real device,

so that even if 10p-benzene are made, one will probably never

be able to produce an electric field intense enough to switch the

molecule.

The use of polar molecules circumvents these problems. The

crucial idea is that a permanent electrical dipole can effectively

‘‘bias’’ the crossover field to smaller fields. In fact (see Fig. 11)

the energy change as a function of field, DEGS, is a parabolic

function, centered at E = 0 for non-polar molecules

(the second order Stark shift is proportional to 12EiaijEj, where

aij is the polarizability tensor). Then, Ecross is determined

by the interception between the parabolas associated to

the different spin-states (a spin singlet and a spin triplet in

Fig. 11.). The addition of a linear contribution to DEGS

(the first order Stark shift ~p�~E, with ~p the permanent electrical

dipole) shifts the center of the parabolas bringing their

interception closer to E = 0 at least for one of the two field

polarities. Hence the molecule electric dipole effectively

introduces a bias field Ebias, which reduces the external electric

field needed for the crossover. This mechanism brings a much

more promising strategy since one can first engineer the

magnetic molecule and then the specific electrical dipole, as

demonstrated for acetylene-bridged di-Cobaltocene(CoCp2)

molecules functionalized with different substituents.174

Such a field-induced crossover has important consequences

for the electrical transport. If one of these molecules is sand-

wiched in a two-terminal device in such a way that there is a

potential drop between the magnetic centers, then the strength

Fig. 11 Stark energy gain, DEGS, for the singlet and triplet state of a

magnetic molecule as a function of the applied electric field, E. In

panel (a) we represent a molecule with inversion symmetry, while in

(b) one where the symmetry is broken by an electrical dipole. Note that

the shift of the DEGS(E) parabola by Ebias generates a shift of the

crossover field to lower fields. J0ST is the exchange interaction energy at

zero field. N. Baadji, M. Piacenza, T. Tugsuz, F. D. Sala, G. Maruccio

and S. Sanvito, Nature Materials, 2009, 8, 813–817. Copyright (2009)

by the Nature Publishing Group.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3352 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

of the magnetic coupling will change with bias. Interestingly

for biases corresponding to the crossover field the different

spin-states become degenerate, i.e. on average the molecule

spends an identical amount of time in any one of them. This

essentially means that there is no magnetic energy scale at

Ecross and the current is predicted to become temperature

independent.175 Intriguingly the electrical control of a high-

spin to low-spin transition has been recently demonstrated

experimentally in a single Mn2+ ion coordinated by two

tetrapyridine ligands in a three-terminal device geometry.176

We wish to close this section by mentioning that some

theoretical work has been devolved to understand how the

magnetization direction of a magnetic molecule can be

manipulated with an external current, by essentially translating

the concept of spin-transfer torque to the molecular magnets

world.177 Importantly was shown that a spin-polarized current

can switch the magnetic moment of the molecule despite the

molecule intrinsic spin-relaxation.

4. Conclusions

In this contribution I have reviewed the most recent advances

in the emerging and fascinating fields of organic and single

molecule spintronics. In particular I have highlighted the main

difference between spintronics in extended organic semi-

conductors and that in single molecules. The picture emerging

is encouraging although the challenges ahead are still significant.

Devices made with extended organic semiconductors appear

now reproducible to a good degree and may show room

temperature spin-valve magnetoresistance. Controlling the

quality and the nature of the interfaces between the organic

and the inorganic elements of the device appears to be the

critical aspect in the fabrication process. This sensitivity however

can be turned into a strength and examples of improved

magnetoresistance tuned by chemical doping of the interfaces

are rapidly appearing.

In contrast single molecule devices remain highly unstable

and difficult to reproduce. However, when made, these may

show a range of intriguing properties not shared by large area

devices. In particular the use of functional molecules where

switching activity is present may give access to electrically

controlled single molecule spin devices.

In moving forward organic andmolecular spintronics have also

to face an additional challenge beyond the technical issues

mentioned above. This is the need of finding a range of specific

applications where the unique characteristics of organic materials,

namely the long spin diffusion times, are fully exploited.

Acknowledgements

This work is sponsored by Science Foundation of Ireland (grant

No. 08/ERA/I1759 and 07/IN.1/I945) and by CRANN. I would

like to thankNadjib Baadji for a critical review of the manuscript.

References

1 N. F. Mott, Proc. R. Soc. London, Ser. A, 1936, 153, 699–717.2 M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau,F. Petroff, P. Etienne, G. Creuzet, A. Friederich andJ. Chazelas, Phys. Rev. Lett., 1988, 61, 2472–2475.

3 G. Binasch, P. Grunberg, F. Saurenbach and W. Zinn, Phys. Rev.B, 1989, 39, 4828–4830.

4 I. Zutic, J. Fabian and S. Das Sarma, Rev. Mod. Phys., 2004, 76,323–410.

5 Semiconductor Spintronics and Quantum Computation, ed. D. D.Awschalom, D. Loss and N. Samarth, Springer-Verlag, BerlinHeidelberg, Germany, 2002.

6 D. D. Awschalom and M. E. Flatte, Nat. Phys., 2007, 3,153–159.

7 M. Gajek, M. Bibes, S. Fusil, K. Bouzehouane, J. Fontecuberta,A. Barthelemy and A. Fert, Nat. Mater., 2006, 6,296–302.

8 J. S. Moodera, T. S. Santos and T. Nagahama, J. Phys. Condens.Matter, 2007, 19, 165202.

9 V. Dediu, M. Murgia, F. C. Matacotta, C. Taliani andS. Barbanera, Solid State Commun., 2002, 122, 181–184.

10 A. R. Rocha, V. Garcia Suarez, S. W. Bailey, C. J. Lambert,J. Ferrer and S. Sanvito, Nat. Mater., 2005, 4, 335–339.

11 D. D. Awschalom and M. E. Flatte, Nat. Phys., 2007, 3,153–159.

12 J. M. Kikkawa and D. D. Awschalom, Nature, 1999, 397,139–141.

13 S. Datta and B. Das, Appl. Phys. Lett., 1990, 56, 665–667.14 C. K. Chiang, C. R. Fincher, Jr., Y. W. Park, A. J. Heeger,

H. Shirakawa, E. J. Louis, S. C. Gau and A. G. MacDiarmid,Phys. Rev. Lett., 1977, 39, 1098–1101.

15 J. M. Manriquez, G. T. Yee, S. Mclean, A. J. Epstein andJ. S. Miller, Science, 1991, 252, 1415–1417.

16 S. Horiuchi and Y. Tokura, Nat. Mater., 2008, 7, 357–366.17 W. J. M. Naber, S. Faez and W. G. van der Wiel, J. Phys. D:

Appl. Phys., 2007, 40, R205–R228.18 V. Alek Dediu, L. E. Hueso, I. Bergenti and C. Taliani,

Nat. Mater., 2009, 8, 707–716.19 S. Sanvito and A. R. Rocha, J. Comput. Theor. Nanosci., 2006, 3,

624–642.20 S. Sanvito, J. Mater. Chem., 2007, 17, 4455–4459.21 L. Bogani and W. Wernsdorfer, Nat. Mater., 2008, 7, 179–186.22 J. Lehmann, A. Gaita-Arino, E. Coronado and D. Loss, J. Mater.

Chem., 2009, 19, 1672–1677.23 P. C. E. Stamp and A. Gaita-Arino, J. Mater. Chem., 2009, 19,

1718–1730.24 M. Affronte, J. Mater. Chem., 2009, 19, 1731–1737.25 J. Camarero and Eugenio, J. Mater. Chem., 2009, 19,

1678–1684.26 M. Mas-Torrent, N. Crivillers, V. Mugnaini, I. Ratera, C. Rovira

and J. Veciana, J. Mater. Chem., 2009, 19, 1691–1695.27 R. Wiesendanger, Rev. Mod. Phys., 2009, 81, 1495–1550.28 Optical Orientation, ed. F. Meier and B. P. Zachachrenya,

North-Holland, Amsterdam, The Netherlands, Modern Problemsin Condensed Matter Science series, vol. 8 edn, 1984.

29 Spin Electronics, ed. M. Ziese and M. J. Thornton, Springer-Verlag, Heidelberg, Germany, Lecture Notes in Physics series,vol. 569 edn, 2001.

30 A. Kirilyuk, A. V. Kimel and T. Rasing, Rev. Mod. Phys., 2010,82, 2731–2784.

31 B. Dieny, Phys. Rev. B, 1991, 43, 1297–1300.32 T. Valet and A. Fert, Phys. Rev. B, 1993, 48, 7099–7113.33 I. I. Mazin, Phys. Rev. Lett., 1999, 83, 1427–1430.34 J. M. D. Coey and S. Sanvito, J. Phys. D: Appl. Phys., 2004, 37,

988–993.35 A. Abragam, The principles of nuclear magnetism, Oxford

University Press, Oxford, UK, 14th edn, 2002.36 G. Schmidt, D. Ferrand, L. W. Molenkamp, A. T. Filip and

B. J. van Wees, Phys. Rev. B, 2000, 62, R4790–R4793.37 E. I. Rashba, Phys. Rev. B, 2000, 62, R16267–R16270.38 M. Julliere, Phys. Lett. A, 1975, 54, 225–226.39 W. H. Butler, X.-G. Zhang, T. C. Schulthess and

J. M. MacLaren, Phys. Rev. B, 2001, 63, 054416.40 J. Mathon and A. Umerski, Phys. Rev. B, 2001, 63, 220403.41 I. Rungger, O. Mryasov and S. Sanvito, Phys. Rev. B, 2009, 79,

094414.42 T. S. Santos, J. S. Lee, P. Migdal, I. C. Lekshmi, B. Satpati and

J. S. Moodera, Phys. Rev. Lett., 2007, 98, 016601.43 B. N. Limketkai, P. Jadhav and M. A. Baldo, Phys. Rev. B, 2007,

75, 113203.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3353

44 G. Szulczewski, S. Sanvito and J. M. D. Coey, Nat. Mater., 2009,8, 693–695.

45 J. Bass and W. P. Pratt, J. Phys.: Condens. Matter, 2007, 19,183201.

46 B. Huang, D. Monsma and I. Appelbaum, Phys. Rev. Lett., 2007,99, 177209.

47 I. Appelbaum, B. Huang and D. Monsma, Nature, 2007, 447,295–298.

48 L. E. Hueso, J. M. Pruneda, V. Ferrari, G. Burnell, J. P.Valdes-Herrera, B. D. Simons, P. B. Littlewood, E. Artacho,A. Fert and N. D. Mathur, Nature, 2007, 445, 410–413.

49 K. Tsukagoshi, B. W. Alphenaar and H. Ago, Nature, 1999, 401,572–574.

50 N. Tombros, C. Jozsa, M. Popinciuc, H. T. Jonkman andB. J. van Wees, Nature, 2007, 448, 571–574.

51 Z. H. Xiong, D. Wu, Z. Valy Vardeny and J. Shi, Nature, 2004,427, 821–824.

52 S. Pramanik, C. G. Stefanita, S. Patibandla, S. Bandyopadhyay,K. Garre, N. Harth and M. Cahay, Nat. Nanotechnol., 2007, 2,216–219.

53 J. H. Shim, K. V. Raman, Y. J. Park, T. S. Santos, G. X. Miao,B. Satpati and J. S. Moodera, Phys. Rev. Lett., 2008, 100,226603.

54 B. Huang, H.-J. Jang and I. Appelbaum, Appl. Phys. Lett., 2008,93, 162508.

55 V. I. Krinichnyi, Synth. Met., 2000, 108, 173–222.56 S. Sanvito, Nat. Mater., 2007, 6, 803–804.57 P. A. Bobbert, T. D. Nguyen, F. W. A. van Oost,

B. Koopmans and M. Wohlgenannt, Phys. Rev. Lett., 2007, 99,216801.

58 O. Mermer, G. Veeraraghavan, T. L. Francis, Y. Sheng,D. T. Nguyen, M. Wohlgenannt, A. Kohler, M. K. Al-Suti andM. S. Khan, Phys. Rev. B, 2005, 72, 205202.

59 F. Kuemmeth, S. Ilani, D. C. Ralph and P. L. McEuen, Nature,2008, 452, 448–452.

60 J. Takeya, M. Yamagishi, Y. Tominari, R. Hirahara,Y. Nakazawa, T. Nishikawa, T. Kawase, T. Shimoda andS. Ogawa, Appl. Phys. Lett., 2007, 90, 102120.

61 H. O. H. Churchill, A. J. Bestwick, J. W. Harlow, F. Kuemmeth,D. Marcos, C. H. Stwertka, S. K. Watson and C. M. Marcus,Nat. Phys., 2009, 5, 321–326.

62 T. D. Nguyen, G. Hukic-Markosian, F. Wang, L. Wojcik,X.-G. Li, E. Ehrenfreund and Z. V. Vardeny, Nat. Mater.,2010, 9, 345–352.

63 J. M. Lupton, D. R. McCamey and C. Boehme, ChemPhysChem,2010, 11, 3040–3058.

64 H. Agarwal, S. Pramanik and S. Bandyopadhyay, New J. Phys.,2008, 10, 015001.

65 A. M. Bratkovsky, Phys. Rev. B, 1997, 56, 2344–2347.66 F. J. Wang, C. G. Yang, Z. V. Vardeny and X. G. Li, Phys. Rev.

B, 2007, 75, 245324.67 V. Dediu, L. E. Hueso, I. Bergenti, A. Riminucci, F. Borgatti,

P. Graziosi, C. Newby, F. Casoli, M. P. De Jong, C. Taliani andY. Zhan, Phys. Rev. B, 2008, 78, 115203.

68 S. Yuasa, T. Nagahama, A. Fukushima, Y. Suzuki and K. Ando,Nat. Mater., 2004, 3, 868–871.

69 S. S. P. Parkin, C. Kaiser, A. Panchula, P. M. Rice,B. Hughes, M. Samant and S.-H. Yang, Nat. Mater., 2004, 3,862–867.

70 J. M. De Teresa, A. Barthelemy, A. Fert, J. P.Contour, F. Montaigne and P. Seneor, Science, 1999, 286,507–509.

71 S. Majumdar, H. S. Majumdar, R. Laiho and R. Osterbacka,J. Alloys Compd., 2006, 423, 169–171.

72 W. Xu, G. J. Szulczewski, P. LeClair, I. Navarrete, R. Schad,G. Miao, H. Guo and A. Gupta, Appl. Phys. Lett., 2007, 90,072506.

73 H. Vinzelberg, J. Schumann, D. Elefant, R. B. Gangineni,J. Thomas and B. Buchner, J. Appl. Phys., 2008, 103,093720.

74 C. Barraud, P. Seneor, R. Mattana, S. Fusil, K. Bouzehouane,C. Deranlot, P. Graziosi, L. Hueso, I. Bergenti, V. Dediu,F. Petroff and A. Fert, Nat. Phys., 2010, 6, 615–620.

75 J. Hwang, A. Wan and A. Kahn, Mater. Sci. Eng., R, 2009, 64,1–31.

76 I. Rungger, X. Chen, U. Schwingenschlogl and S. Sanvito, Phys.Rev. B, 2010, 81, 235407.

77 C. Toher and S. Sanvito, Phys. Rev. B, 2008, 77, 155402.78 T. Maruyama, Y. Shiota, T. Nozak, K. Ohta, N. Toda,

M. Mizuguchi, A. A. Tulapurkar, T. Shinjo, M. Shiraishi,S. Mizukami, Y. Ando and Y. Suzuki, Nat. Nanotechnol., 2009,4, 158–161.

79 W. F. Brinkman, R. C. Dynes and J. M. Rowell, J. Appl. Phys.,1970, 41, 1915–1921.

80 I. G. Hill, A. Rajagopal, A. Kahn and Y. Hu, Appl. Phys. Lett.,1998, 73, 662–664.

81 K. Marumoto, N. Arai, H. Goto, M. Kijima, K. Murakami,Y. Tominari, J. Takeya, Y. Shimoi, H. Tanaka, S. Kuroda,T. Kaji, T. Nishikawa, T. Takenobu and Y. Iwasa, Phys. Rev.B, 2011, 83, 075302.

82 J.-C. Tai, J.-C. Huang, Y.-M. Chang, K.-S. Li, J.-Y. Hong,S.-S. Wong, W.-C. Chiang and M.-T. Lin, Appl. Phys. Lett.,2010, 96, 262502.

83 S. Krause, M. and. A Scholl and E. Umbach, New J. Phys., 2008,10, 085001.

84 J. B. Neaton, M. S. Hybertsen and S. G. Louie, Phys. Rev. Lett.,2006, 97, 216405.

85 J. J. H. M. Schoonus, P. G. E. Lumens, W. Wagemans,J. T. Kohlhepp, P. A. Bobbert, H. J. M. Swagten andB. Koopmans, Phys. Rev. Lett., 2009, 103, 146601.

86 J.-W. Yoo, H. W. Jang, V. N. Prigodin, C. Kao, C. B. Eom andA. J. Epstein, Phys. Rev. B, 2009, 80, 205207.

87 J.-W. Yoo, H. W. Jang, V. N. Prigodin, C. Kao, C. B. Eomand A. J. Epstein, Synth. Met., 2010, 160, 216–222.

88 K. V. Raman, S. M. Watson, J. H. Shim, J. A. Borchers,J. Chang and J. S. Moodera, Phys. Rev. B, 2009, 80,195212.

89 G. Szulczewski, H. Tokuc, K. Oguz and J. M. D. Coey, Appl.Phys. Lett., 2009, 95, 202506.

90 J. S. Jiang, J. E. Pearson and S. D. Bader, Phys. Rev. B, 2008, 77,035303.

91 M. Cinchetti, K. Heimer, J.-P. Wustenberg, O. Andreyev,M. Bauer, S. Lach, C. Ziegler, Y. Gao and M. Aeschlimann,Nat. Mater., 2009, 8, 115–119.

92 A. J. Drew, J. Hoppler, L. Schulz, F. L. Pratt, P. Desai,P. Shakya, T. Kreouzis, W. P. Gillin, A. Suter, N. A.Morley, V. K. Malik, A. Dubroka, K. W. Kim, H. Bouyanfif,F. Bourqui, C. Bernhard, R. Scheuermann, G. J. Nieuwenhuys,T. Prokscha and E. Morenzoni, Nat. Mater., 2009, 8,109–114.

93 L. Schulz, L. Nuccio, M. Willis, P. Desai, P. Shakya, T. Kreouzis,V. K. Malik, C. Bernhard, F. L. Pratt, N. A. Morley, A. Suter,G. J. Nieuwenhuys, T. Prokscha, E. Morenzoni, W. P. Gillin andA. J. Drew, Nat. Mater., 2011, 10, 39–44.

94 S. Sanvito, Nat. Nanotechnol., 2007, 2, 204–206.95 S. Bandyopadhyay, Phys. Rev. B, 2010, 81, 153202.96 R. J. Elliott, Phys. Rev., 1954, 96, 266–279.97 H. M. McConnell, J. Chem. Phys., 1956, 24, 764–766.98 P. A. Bobbert, Nat. Mater., 2010, 9, 288–290.99 D. R. McCamey, K. J. van Schooten, W. J. Baker, S. Y. Lee,

S. Y. Paik, J. M. Lupton and C. Boehme, Phys. Rev. Lett., 2010,104, 017601.

100 N. J. Rolfe, M. Heeney, P. B. Wyatt, A. J. Drew, T. Kreouzis andW. P. Gillin, Phys. Rev. B, 2009, 80, 241201.

101 S. Sanvito, Nat. Phys., 2010, 6, 562–564.102 L. Fang, K. Deniz Bozdag, C.-Y. Chen, P. A. Truitt, A. J.

Epstein and E. Johnston-Halperin, Phys. Rev. Lett., 2011, 106,156602.

103 J.-W. Yoo, C.-Y. Chen, H. W. Jang, C. W. Bark, V. N.Prigodin, C. B. Eom and A. J. Epstein, Nat. Mater., 2010, 9,638–642.

104 S. J. van der Molen and P. Liljeroth, J. Phys.: Condens. Matter,2010, 22, 133001.

105 A. Odell, A. Delin, B. Johansson, I. Rungger and S. Sanvito, ACSNano, 2010, 4, 2635–2642.

106 D. B. Strukov, G. S. Snider, D. R. Stewart and R. S. Williams,Nature, 2008, 453, 80–83.

107 L. Chua, IEEE Trans. Circuit Theory, 1971, 18, 507–519.108 M. Prezioso, A. Riminucci, I. Bergenti, P. Graziosi, D. Brunel

and V. A. Dediu, Adv. Mater., 2011, 23, 1371–1375.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

3354 Chem. Soc. Rev., 2011, 40, 3336–3355 This journal is c The Royal Society of Chemistry 2011

109 S. Horiuchi, Y. Tokunaga, G. Giovannetti, S. Picozzi, H. Itoh,R. Shimano, R. Kumai and Y. Tokura, Nature, 2010, 463,789–792.

110 F. Kagawa, S. Horiuchi, M. Tokunaga, J. Fujioka andY. Tokura, Nat. Phys., 2010, 6, 169–172.

111 F. Chen, J. Hihath, Z. Huang, X. Li and N. J. Tao, Annu. Rev.Phys. Chem., 2007, 58, 535–564.

112 A. Osorio, T. Bjørnholm, J.-M. Lehn, M. Ruben and H. S. J. vander Zant, J. Phys.: Condens. Matter, 2008, 20, 374121.

113 S. J. Xie, K. H. Ahn, D. L. Smith, A. R. Bishop and A. Saxena,Phys. Rev. B, 2003, 67, 125202.

114 P. P. Ruden and D. L. Smith, J. Appl. Phys., 2004, 95,4898–4904.

115 M. Yunus, P. P. Ruden and D. L. Smith, Synth. Met., 2010, 160,204–209.

116 N. Baadji, S. Kuck, J. Brede, G. Hoffmann, R. Wiesendanger andS. Sanvito, Phys. Rev. B, 2010, 82, 115447.

117 J. Park, A. N. Pasupathy, J. I. Goldsmith, C. Chang, Y. Yaish,J. R. Petta, M. Rinkoski, J. P. Sethna, H. D. Abruna,P. L. McEuen and D. C. Ralph, Nature, 2002, 417, 722–725.

118 W. Liang, M. P. Shores, M. Bockrath, J. R. Long and H. Park,Nature, 2002, 417, 725–729.

119 G. D. Scott and D. Natelson, ACS Nano, 2010, 4, 3560–3579.120 A. Scheybal, T. Ramsvik, R. Bertschinger, M. Putero, F. Nolting

and T. A. Jung, Chem. Phys. Lett., 2005, 411, 214–220.121 H. Wende, M. Bernien, J. Luo, C. Sorg, N. Ponpandian,

J. Kurde, J. Miguel, M. Piantek, X. Xu, P. Eckhold, W. Kuch,K. Baberschke, P. M. Panchmatia, B. Sanyal, P. M. Oppeneerand O. Eriksson, Nat. Mater., 2007, 6, 516–520.

122 C. Iacovita, M. V. Rastei, B. W. Heinrich, T. Brumme, J. Kortus,L. Limot and J. P. Bucher, Phys. Rev. Lett., 2008, 101,116602.

123 M. Bernien, J. Miguel, C. Weis, M. E. Ali, J. Kurde, B. Krumme,P. M. Panchmatia, B. Sanyal, M. Piantek, P. Srivastava,K. Baberschke, P. M. Oppeneer, O. Eriksson, W. Kuch andH. Wende, Phys. Rev. Lett., 2009, 102, 047202.

124 P. Gambardella, S. Stepanow, A. Dmitriev, J. Honolka,F. M. F. de Groot, M. Lingenfelder, S. S. Gupta, D. D. Sarma,P. Bencok, S. Stanescu, S. Clair, S. Pons, N. Lin, A. P. Seitsonen,H. Brune, J. V. Barth and K. Kern, Nat. Mater., 2009, 8,189–193.

125 S. Javaid, M. Bowen, S. Boukari, L. Joly, J. B. Beaufrand,X. Chen, Y. J. Dappe, F. Scheurer, J. P. Kappler, J. Arabski,W. Wulfhekel, M. Alouani and E. Beaurepaire, Phys. Rev. Lett.,2010, 105, 077201.

126 S. Stepanow, A. Mugarza, G. Ceballos, P. Moras, J. C. Cezar,C. Carbone and P. Gambardella, Phys. Rev. B, 2010, 82,014405.

127 A. R. Rocha and S. Sanvito, J. Appl. Phys., 2007, 101,09B102–5.

128 N. Atodiresei, J. Brede, P. Lazicacute, V. Caciuc, G. Hoffmann,R. Wiesendanger and S. Blugel, Phys. Rev. Lett., 2010, 105,066601.

129 J. Brede, N. Atodiresei, S. Kuck, P. Lazicacute, V. Caciuc,Y. Morikawa, G. Hoffmann, S. Blugel and R. Wiesendanger,Phys. Rev. Lett., 2010, 105, 047204.

130 A. J. Heinrich, J. A. Gupta, C. P. Lutz and D. M. Eigler, Science,2004, 306, 466–469.

131 S. Loth, C. P. Lutz and A. J. Heinrich, New J. Phys., 2010, 12,125021.

132 A. Troisi and M. A. Ratner, Nano Lett., 2006, 6, 1784–1788.133 M. Paulsson, T. Frederiksen, H. Ueba, N. Lorente and

M. Brandbyge, Phys. Rev. Lett., 2008, 100, 226604.134 C. F. Hirjibehedin, C. P. Lutz and A. J. Heinrich, Science, 2006,

312, 1021–1024.135 C. F. Hirjibehedin, C.-Y. Lin, A. F. Otte, M. Ternes, C. P. Lutz,

B. A. Jones and A. J. Heinrich, Science, 2007, 317, 1199–1203.136 X. Chen, Y.-S. Fu, S.-H. Ji, T. Zhang, P. Cheng, X.-C. Ma,

X.-L. Zou, W.-H. Duan, J.-F. Jia and Q.-K. Xue, Phys. Rev.Lett., 2008, 101, 197208.

137 Y.-S. Fu, T. Zhang, S.-H. Ji, X. Chen, X.-C. Ma, J.-F. Jia andQ.-K. Xue, Phys. Rev. Lett., 2009, 103, 257202.

138 A. Zhao, Q. Li, L. Chen, H. Xiang, W. Wang, S. Pan, B. Wang,X. Xiao, J. Yang, J. G. Hou and Q. Zhu, Science, 2005, 309,1542–1544.

139 P. Lucignano, R. Mazzarello, A. Smogunov, M. Fabrizio andE. Tosatti, Nat. Mater., 2009, 8, 563–567.

140 D. Gatteschi, R. Sessoli and J. Villain, Molecular Nanomagnets,Oxford University Press, Oxford, 2006.

141 S. Voss, M. Fonin, U. Rudiger, M. Burgert, U. Groth andY. S. Dedkov, Phys. Rev. B, 2007, 75, 045102.

142 M. Mannini, P. Sainctavit, R. Sessoli, C. Cartier dit Moulin,F. Pineider, M.-A. Arrio, A. Cornia and D. Gatteschi, Chem.–Eur.J., 2008, 14, 7530–7535.

143 Z. Yi, X. Shen, L. Sun, Z. Shen, S. Hou and S. Sanvito, ACSNano, 2010, 4, 2274–2282.

144 M. Mannini, F. Pineider, P. Sainctavit, C. Denieli, E. Otero,C. Sciancalepore, A. M. Talarico, M. A. Arrio, A. Cornia,D. Gatteschi and R. Sessoli, Nat. Mater., 2009, 8, 194–197.

145 M. Mannini, F. Pineider, C. Danieli, F. Totti, L. Sorace,P. Sainctavit, M. A. Arrio, E. Otero, L. Joly, J. C. Cezar,A. Cornia and R. Sessoli, Nature, 2010, 468, 417–421.

146 J. Wang, P. H. Acioli and J. Jellinek, J. Am. Chem. Soc., 2005,127, 2812–2813.

147 H. Xiang, J. Yang, J. G. Hou and Q. Zhu, J. Am. Chem. Soc.,2006, 128, 2310–2314.

148 V. V. Maslyuk, A. Bagrets, V. Meded, A. Arnold, F. Evers,M. Brandbyge, T. Bredow and I. Mertig, Phys. Rev. Lett., 2006,97, 097201.

149 L. Wang, Z. Cai, J. Wang, J. Lu, G. Luo, L. Lai, J. Zhou, R. Qin,Z. Gao, D. Yu, G. Li, W. N. Mei and S. Sanvito, Nano Lett.,2008, 8, 3640–3644.

150 A. Soncini and L. F. Chibotaru, Phys. Rev. B, 2008, 77,220406.

151 A. Soncini and L. F. Chibotaru, Phys. Rev. B, 2010, 81,132403.

152 C. D. Pemmaraju, I. Rungger and S. Sanvito, Phys. Rev. B, 2009,80, 104422.

153 H. B. Heersche, Z. de Groot, J. A. Folk, H. S. J. van der Zant,C. Romeike, M. R. Wegewijs, L. Zobbi, D. Barreca, E. Tondelloand A. Cornia, Phys. Rev. Lett., 2006, 96, 206801.

154 M.-H. Jo, J. E. Grose, K. Baheti, M. M. Deshmukh, J. J. Sokol,E. M. Rumberger, D. N. Hendrickson, J. R. Long, H. Park andD. C. Ralph, Nano Lett., 2006, 6, 2014–2020.

155 C. Romeike, M. R. Wegewijs and H. Schoeller, Phys. Rev. Lett.,2006, 96, 196805.

156 C. Romeike, M. R. Wegewijs, M. Ruben, W. Wenzel andH. Schoeller, Phys. Rev. B, 2007, 75, 064404.

157 M. Lopes, A. Candini, M. Urdampilleta, A. Reserbat-Plantey,V. Bellini, S. Klyatskaya, L. Marty, M. Ruben, M. Affronte,W. Wernsdorfer and N. Bendiab, ACS Nano, 2010, 4,7531–7537.

158 P. Wei, L. Sun, E. Benassi, Z. Shen, S. Sanvito, S. Hou, Preprint,2011.

159 M. N. Leuenberger and D. Loss, Nature, 2001, 410, 789–793.160 G. A. Timco, S. Carretta, F. Troiani, F. Tuna, R. J. Pritchard,

C. A. Muryn, E. J. L. McInnes, A. Ghirri, A. Candini, P. Santini,G. Amoretti, M. Affronte andR. E. P.Winpenny,Nat. Nanotechnol.,2009, 4, 173–178.

161 J. Lehmann, A. Gaita-Arino, E. Coronado and D. Loss, Nat.Nanotechnol., 2007, 2, 312–317.

162 C. Wackerlin, D. Chylarecka, A. Kleibert, K. Muller, C. Iacovita,F. Nolting, T. A. Jung and N. Ballav, Nat. Commun., 2010,1, 94.

163 P. Gutlich and H. A. Goodwin, Spin Crossover in TransitionMetal Compounds, Springer-Verlag, Berlin Heidelberg, Germany,2004.

164 O. Sato, J. Tao and Y.-Z. Zhang, Angew. Chem., Int. Ed., 2007,46, 2152–2187.

165 P. Gutlich, H. A. Goodwin, D. N. Hendrickson and C. G. Pierpont,in Valence Tautomeric Transition Metal Complexes, SpringerBerlin/Heidelberg, 2004, vol. 234, pp. 786–786.

166 A. Beni, C. Carbonera, A. Dei, J.-F. Letard, R. Righini,C. Sangregorio and L. Sorace, J. Braz. Chem. Soc., 2006, 17,1522.

167 M. Kepenekian, B. L. Guennic and V. Robert, J. Am. Chem. Soc.,2009, 131, 11498–11502.

168 M. Kepenekian, B. Le Guennic and V. Robert, Phys. Rev. B,2009, 79, 094428.

169 A. Droghetti and S. Sanvito, 2011, arXiv:1101.4777v1.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Chem. Soc. Rev., 2011, 40, 3336–3355 3355

170 K. Takahashi, H.-B. Cui, Y. Okano, H. Kobayashi, Y. Einagaand O. Sato, Inorg. Chem., 2006, 45, 5739–5741.

171 M. S. Alam,M. Stocker, K. Gieb, P.Muller,M.Haryono, K. Studentand A. Grohmann, Angew. Chem., Int. Ed., 2010, 49, 1159–1163.

172 F. Prins, M. Monrabal-Capilla, E. A. Osorio, E. Coronado andH. S. J. van der Zant, Adv. Mater., 2011, 23, 1545–1549.

173 M. Diefenbach and K. S. Kim, Angew. Chem., 2007, 119,7784–7787.

174 N. Baadji, M. Piacenza, T. Tugsuz, F. D. Sala, G. Maruccio andS. Sanvito, Nat. Mater., 2009, 8, 813–817.

175 S. K. Shukla and S. Sanvito, Phys. Rev. B, 2009, 80,184429.

176 E. A. Osorio, K. Moth-Poulsen, H. S. J. van der Zant, J. Paaske,P. Hedegard, K. Flensberg, J. Bendix and T. Bjørnholm,Nano Lett.,2009, 10, 105–110.

177 M. Misiorny and J. Barnas, Phys. Rev. B, 2008, 77, 172414.

Publ

ishe

d on

09

May

201

1. D

ownl

oade

d by

IN

DIA

N I

NST

ITU

TE

OF

TE

CH

NO

LO

GY

BO

MB

AY

on

20/1

0/20

15 1

2:19

:08.

View Article Online