Professional Documents
Culture Documents
+ 2009 Spin Routes in Organic Semiconductors
+ 2009 Spin Routes in Organic Semiconductors
Organic semiconductors are characterized by a very low spin–orbit interaction, which, together with their chemical flexibility
and relatively low production costs, makes them an ideal materials system for spintronics applications. The first experiments
on spin injection and transport occurred only a few years ago, and since then considerable progress has been made in improving
performance as well as in understanding the mechanisms affecting spin-related phenomena. Nevertheless, several challenges
remain in both device performance and fundamental understanding before organic semiconductors can compete with inorganic
semiconductors or metals in the development of realistic spintronics applications. In this article we summarize the main experi-
mental results and their connections with devices such as light-emitting diodes and electronic memory devices, and we outline
the scientific and technological issues that make organic spintronics a young but exciting field.
R
esearch in organic semiconductors (OSCs) has progressed or diffusive motion of spin-polarized carriers in delocalized band
impressively over the past few decades1. The greatest success states. The spin-flipping processes proceed mainly via spin–orbit
has undoubtedly been achieved in the field of optoelectronics. interaction, which are able to transfer the scattering of the angular
Display products based on hybrid light-emitting diodes with momentum into spin scattering. Organic semiconductors are not
organic emitters (OLEDs) have become available to consumers, and characterized by band conductivity, nor do they feature a significant
organic photovoltaic devices are challenging existing commercial spin–orbit interaction. The electrical carriers propagate in OSCs
applications. These accomplishments generate a strong demand for by incoherent hopping between strongly localized states that can
other mainstream products based on new hybrid organic–inorganic be approximated to very narrow bands only at low temperatures.
devices. Considerable improvements have already been achieved Unveiling the fundamental aspects of these materials is therefore
in the field of organic field-effect transistors (OFETs), and funda- what makes the field simultaneously challenging and exciting, with
mental research in the use of organic electric memories for data- the hope that improvements in understanding and technology will
storage applications has also progressed. allow OSCs to compete with and overtake other materials in the
Organic semiconductors have recently caught the attention of field of spintronics, or at least in some selected niche applications.
the spintronics community, and significant efforts are being made In this review we focus on the main achievements and questions
towards their integration in this field2,3. Spintronics is a branch of arising from spin injection and transport in OSC materials. We
electronics that takes full advantage of not only the charge, but also begin by presenting and discussing the concepts and facts that are
the spin of the electron. Spintronics encompasses different topics, widely accepted by the community, and conclude by addressing the
and it involves the generation of a non-equilibrium spin polariza- most controversial issues and open questions.
tion in various materials and devices, as well as its manipulation and To the best of our knowledge, no communication on fully organic
detection4–7. Among the most fascinating examples of spintronic spintronic devices has appeared yet, as only hybrid structures with
applications are giant magnetoresistance (MR) and tunnelling magnetic inorganic electrodes and organic active layers have been
magnetoresistance (TMR). In both cases, a considerable variation of investigated. Nevertheless, for the sake of simplicity, we will call
the electrical resistance can be achieved by switching the magnetiza- such devices organic spintronic devices (OSPDs).
tion of the device’s electrodes from parallel to antiparallel8–10.
The most attractive aspect for spintronic applications in OSCs electrical injection and detection
is the weakness of the spin-scattering mechanism, implying that In the following discussion on the electrically operated OSPDs, it
the spin polarization of the carriers can be maintained for a very is necessary to distinguish between two different cases: tunnelling
long time. This property is due to the very low spin–orbit coupling devices characterized by zero residence time of charges/spins in the
in OSCs: carbon has a low atomic number (Z), and the strength organic barrier, and injection devices in which there is a net flow of
of the spin–orbit interaction is proportional to Z4. Noticeably, spin-polarized carriers directly into the electronic levels of the OSC.
spin-relaxation times in excess of 10 μs were obtained by different Although tunnelling devices can feature large MR values and are
resonance techniques11,12 years before the first organic spintronic useful as 0–1 switching elements, or as magnetic sensors, injection
experiments; these values exceeded the characteristic times detected devices are characterized by a finite lifetime of the generated spin
in inorganic materials by orders of magnitude5,13. polarization in the OSC, and offer such possibilities as spin manipu-
It is difficult at this stage to predict whether research efforts in lation, control of the exciton statistics in the OSC, and so on. These
organic spintronics will lead to their implementation into actual options may well open the way to magnetically controlled OFETs
devices. At present, organic spintronics can be regarded as a fasci- or OLEDs.
nating puzzle in which many pieces are still missing. The electronic The electrical properties of OSCs are rather complex and in
properties of OSCs are radically different from those of such band principle may vary depending on the specific material used, but we
semiconductors as Si or GaAs. Consequently, their spin properties can nevertheless summarize most of the common features. OSCs are
are also different. In band semiconductors, the spin injection and mainly families of π-conjugated molecules, merging a considerable
transport are commonly described by considering either ballistic delocalization of carriers inside the molecule with weak van der Waals
ISMN-CNR, via Gobetti 101, 40129 Bologna, Italy. 2CIC nanoGUNE Consolider, Tolosa Hiribidea 76, E-20018 Donostia - San Sebastian and IKERBASQUE,
1
6T
LSMO LSMO –0.2
R(H)–R(0) (GΩ)
–0.4
–0.6 S S S
w = 70–500 nm S S S
V
–0.8
I
–1.0
100 200 300 400
Channel length (nm)
c d e
26.0
11 K 40 0
22.5 P P 30 P P
∆R/R (%)
Co –1
I∆R/RI (%)
20
OS
R (kΩ)
N
20.0 O O
Al
N N 10 LSMO
O
Substrate –2 T = 40 K
17.5 0
AP AP AP AP
–10
–1,500 -750 0 750 1,500 –3.0 –1.5 0 1.5 3.0
Magnetic field (Oe) Magnetic field (kOe)
Figure 1 | Organic spintronic devices. a, Lateral spin device as published by Dediu et al.20. In this device, two ferromagnetic LSMO electrodes patterned by
electron-beam lithography were separated by 70–500 nm and bridged by a 6T channel. b, Magnetoresistance (H = 0.3 T, where H is the magnetic field)
of the lateral spin device depicted in a as a function of the channel length. The error bars indicate the possible local deviations from the nominal channel
length. c, Magnetoresistance loop of a LSMO (100 nm)/Alq3 (130 nm)/Co (3.5 nm) spin-valve device measured at 11 K (ref. 23). As indicated in the figure,
an antiparallel (AP) configuration of the magnetization of the electrodes corresponds to a low-resistance state. d, Schematic of a typical vertical inorganic–
organic spin valve. In this case, LSMO is the bottom electrode and Co is the top one. A typical MR measurement is performed by applying a bias voltage to
the metallic ferromagnetic electrodes and the current is measured as a function of the in-plane external magnetic field. e, Magnetoresistance loop of a LSMO
(20 nm)/6T (100 nm)/Al2O3 (2 nm)/Co (20 nm) vertical spin-valve device measured at 40 K. Figure reproduced with permission: c, © 2004 NPG.
intermolecular interactions, which considerably limit the carrier velocity in the spin-polarized material, and Nv and Nv 2 are defined
mobility. The injection of carriers into OSCs is best described14 in by integration over the Brillouin zone. It has not yet been possible to
terms of thermal and field-assisted charge tunnelling across the inor- establish which particular definition is applicable for ferromagnetic/
ganic/organic interface, followed by carrier diffusion into the bulk of organic interfaces. The injection of charge/spins across such inter-
the OSC. Furthermore, the carriers propagate by random site-to-site faces corresponds to a transition from extended (band character)
hopping between pseudo-localized states that are distributed within states into almost fully localized ones, which accounts for a dra-
an energy interval of approximately 0.1 eV. Two conducting chan- matic symmetry breaking and leads to the failure of the common
nels are usually considered active: the lowest unoccupied molecular models used for purely inorganic interfaces17. Similarly, the exten-
level (LUMO) for n-type and the highest occupied molecular level sion of the band-symmetry model developed for TMR (inorganic)
(HOMO) for p-type carriers; however, defects and interface states devices should be treated with caution when applied to inorganic
may also have to be considered depending on the material and its spintronics18,19: the symmetry breaking mentioned above consti-
structural quality. Another important issue is the strong electron– tutes a serious barrier and requires a considerable modification of
phonon interaction leading to the carriers having a polaronic char- these models.
acter (for more extensive reading we suggest a recent review 15). The first report on experimental injection in OSPDs was
In principle, any discussion of spintronics effects requires an published as recently as 2002 (ref. 20). The authors designed a lateral
accurate definition of the spin polarization of the carriers. For device that combined ferromagnetic electrodes and OSC conducting
inorganic surfaces or interfaces featuring band-like behaviour, channels 100–500 nm in length (Fig. 1a). The selected ferromagnetic
it has been shown16 that the spin polarization may be defined in material was a highly polarized manganite (La0.7Sr0.3MnO3, or
different ways, depending on the specific experimental or theoreti- LSMO). This material was already well known for its spin-injection
cal approach adopted. The simplest definition, which considers just properties and has proved to be highly successful in diverse inor-
the density of states (N) at the Fermi level (EF), is that spin polari- ganic spintronic devices, such as tunnel junctions19 and artificial
zation is equal to (N↑ − N↓)/(N↑ + N↓), and it is mostly valid for grain-boundary devices21. Sexithiophene (6T), a rigid conjugated-
static cases such as, for example, spectroscopic characterization. The oligomer rod, and a pioneer in organic thin-film transistors22, was
symbol ↑(↓) indicates majority (minority) spin orientation in a fer- chosen for the spin-transport channel. A strong magnetoresistive
romagnetic material with respect to the direction of magnetization. response was recorded up to room temperature in 100-nm and
In experiments that involve current across interfaces, the definition 200-nm channels (Fig. 1b), and was explained as a result of the
of spin polarization may require the substitution of N by Nv (bal- conservation of the spin polarization of the injected carriers. Using
listic regime) or by Nv 2 (diffusive regime), depending on the specific the time-of-flight approach, a spin-relaxation time of the order of
interface and transport mechanism considered16. Here, v is the Fermi 1 μs was found. This work designated organic materials as extremely
Alq3 LSMO/Co (refs 25, 27–29) Ls ≈ 100 (ref. 27) ls ≈ 45 (ref. 23) 10−5 cm2 V−1 s−1 HOMO = 5.7 eV
Fe/Co (refs 32, 68) τs ≈ 26 × 10−6 n-type LUMO = 2.7 eV
N Co/Ni (ref. 76) τs ≈ 10−3 light emitter
O O
Al
N
N O
α-NPD LSMO/Co (ref. 25) 10−5 cm2 V−1 s−1 HOMO = 5.4 eV
p-type LUMO = 2.3 eV
N N
CVB LSMO/Co (ref. 25) 10−3 cm2 V−1 s−1 HOMO = 5.5 eV
LUMO = 2.5 eV
Blue dopant for OLED
N
RRP3HT LSMO/Co (ref. 28) Ls ≈ 80 (ref. 28) 10−1 cm2 V−1 s−1 HOMO = 5.1 eV
C 6H13 C 6H13 Fe50Co50 /Ni81Fe19 (ref. 31) p-type LUMO = 3.5 eV
S
S S n
C 6H13
Tetraphenylporphyrin (TPP) LSMO/Co (ref. 29) 10−5 cm2 V−1 s−1 Red emitter for OLED
n-type
N
NH HN
N
Pentacene Co:TiO2/Fe (ref. 77) 10−1 cm2 V−1 s−1 HOMO = 4.9 eV
p-type LUMO = 2.7 eV
Rubrene Co/Fe (ref. 39) Ls ≈ 13.3 1 cm2 V−1 s−1 HOMO = 5.2 eV
p-type LUMO = 3.0 eV
N N N
N Cu N
N N N
attractive candidates for long-distance spin transport, although Following this encouraging report, an important step forward
still leaving some important questions unanswered. For example, was the fabrication of a vertical spin-valve device consisting of
the experiment20 did not provide a straightforward demonstration LSMO and cobalt electrodes, sandwiching a thick (100–200 nm)
that the MR was related to the magnetization of the electrodes, and layer of tris(8-hydroxyquinoline)aluminium(iii) (Alq3) (Fig. 1c,d)23.
hence to its spin polarization, as no comparison between parallel Alq3 has been a popular OSC since the early stages of xerographic
and antiparallel magnetization of electrodes was available. and optoelectronics applications24. The spin valve showed MR up
a b
1.0 Interfacial dipole Interfacial dipole
R R R R R R
4.9 eV ∆ ~1 eV 5.0 eV ∆ ~1.4 eV
S S S S S S
0.5 R = C6H13
Normalized MR
LUMO
0.0 LUMO
EF ∆i EF ∆i
0.0
N HOMO
–0.5 O O HOMO
Al
N N
O
Figure 2 | magnetoresistance of lsmO/Co vertical OsPDs. a, Temperature dependence of the normalized MR of different vertical OSPD devices with
LSMO and Co as magnetic electrodes. Positive MR (upper panel) for a device with a 100-nm-thick RRP3HT interlayer28. Magnetoresistance for devices
with Alq3 as an organic interlayer as reported by different authors (bottom panel): (blue) 130-nm-thick Alq3 device28; (red) 160-nm-thick Alq3 device25;
(green) 100-nm-thick Alq3 device27. Arrows indicate the temperature at which the corresponding MR signal goes to zero. b, Schematic band diagram for
both LSMO/Alq3 and Co/Alq3 interfaces (Zhan et al.52). Δi represents the lowest injection barrier.
a c
Py
3
Alq
2
SiO
Co
ive
Nat
Si
b d
3
RJ(T)/RJ(T = 300 K)
8
300 K 40.0
77 K
2 600
MR (%)
4.2 K 400
6 1 20.0 200
0 100 200 300
T(K) 0
TMR (%)
I (µA)
–2 –1 0 1 2
0
4 N
0 Parallel configuration
AI –20.0 Antiparallel configuration
0 N
2 N
0
–40.0
–2 –1 0 1 2
0 V (V)
–100 –50 0 50 100 150
Magnetic field (Oe)
Figure 3 | Tunnelling mr. a, High-resolution transmission electron microscopy (HRTEM) image of a magnetic tunnel junction with a thin (few nm) Alq3
barrier. b, TMR of an organic tunnel junction similar to the one presented in a. The inset shows the temperature dependence of the resistance of the
junction35. The value for the spin polarization of the tunnel current across an Alq3 barrier is positive, with values around 30% for Co, Permalloy (Py) and Fe.
RJ represents the junction resistance. c, Schematic representation of a tricene molecule attached to a (001) Ni surface13. d, Theoretical calculation of the
current–voltage (I–V) characteristic of a tricene–nickel spin-valve. The inset shows the calculated MR ratio with values up to 600%13. Figures reproduced
with permission: a,b, © 2007 APS; c,d, © 2005 NPG.
100 Efinal
Normalized spin polarization (%)
Evac
80
LUMO
60 EF
HOMO
Cobalt CuPc
40
0 4 8 12 16
CuPc coverage (Monolayers)
b 40
V–
s
I– LiF
30 p
Spin-diffusion length (nm)
I+
FM2: NiFe
20
V+
Alq3
10
TPD
FM1: FeCo �10kV
Substrate
0 20 40 60 80 100 0V
Temperature (K)
Figure 4 | Non-electrical spin-injection experiments. a, Normalized spin polarization as a function of CuPc thickness on a Co/CuPc junction. The values,
obtained by two-photon photoemission spectroscopy42, show a spin-injection efficiency of 85% from the Co substrate into the CuPc unoccupied molecular
orbitals. In the two-photon photoemission experiment (shown right), CuPc electrons are excited from the Co by an initial light pulse in intermediate
states lying between the Fermi and the vacuum level of the heterojunction. A second photon gives to some of those excited electrons enough energy to be
photoemitted. The error bars show the standard deviation of the detected photoemission yield. b, Temperature dependence of the spin-diffusion length in
Alq3 extracted from the muon spin-relaxation experiments reported in ref. 43. Error bars represent one standard deviation for the spin-diffusion length are
an estimate based on the scatter in the data for the MR measurements. A schematic representation of the device is shown on the right. B, magnetic field;
p, muon momentum; s, initial muon polarization; FM1, FM2, ferromagnetic contacts. Figures reproduced with permission: a,b, © 2009 NPG.
Nature Materials 8, 707–716 (2009); published online: 1 September 2009; corrected after print: 11 September 2009.
In Table 1 on page 709 of the above Progress Article, the structures for the compounds poly(3-hexylthiophene) (RRP3HT), tetraphenyl
porphyrin (TPP), pentacene, rubrene and copper phthalocyanine (CuPc) should have appeared as shown below. These errors have been
corrected in the HTML and PDF versions.
C 6H13 C 6H13 N N
N N
S
S S NH HN N Cu N
n
C 6H13 N N N N