◯Motoi Kimata1, Daisuke Nozaki1, Yasuhiro Niimi1, Hiroyui Tajima2, YoshiChika Otani1, 3

1Institute for Solid State Physics, University of Tokyo2Graduate School of Materials Science, University of Hyogo

3Center for Emergent Matter Science, RIKEN

Spin transport and relaxation mechanism in disordered organic film

Organic spintronics• Organic semiconductors consist of relatively light elements.(For example, C, H, O, S, ...)

Long spin lifetime & diffusion length

Small spin-orbit interaction

NATURE MATERIALS | VOL 8 | SEPTEMBER 2009 | www.nature.com/naturematerials 709

PROGRESS ARTICLENATURE MATERIALS DOI: 10.1038/NMAT2510

attractive candidates for long-distance spin transport, although still leaving some important questions unanswered. For example, the experiment20 did not provide a straightforward demonstration that the MR was related to the magnetization of the electrodes, and hence to its spin polarization, as no comparison between parallel and antiparallel magnetization of electrodes was available.

Following this encouraging report, an important step forward was the fabrication of a vertical spin-valve device consisting of LSMO and cobalt electrodes, sandwiching a thick (100–200 nm) layer of tris(8-hydroxyquinoline)aluminium(iii) (Alq3) (Fig. 1c,d)23. Alq3 has been a popular OSC since the early stages of xerographic and optoelectronics applications24. The spin valve showed MR up

N

NH

N

HN

N

N

N

NN

N

N

N

Cu

SSS

C6H13

C6H13

C6H13

n

SSS S

SS

NAl

O

NO

N

O

NN

N

N

Table 1 | Main properties of organic semiconductors investigated in organic spintronics.

Organic semiconductors Spin-polarized electrodes Spin-diffusion length ls (nm) and time τs (s)

Carrier mobility Electronic and optical properties

6T LSMO/LSMO

Ls ≈ 70 (refs 20, 75) τs ≈ 10−6

10−1 cm2 V−1 s−1

p-typeHOMO = 4.9 eVLUMO = 2.3 eV

Alq3 LSMO/Co (refs 25, 27–29)Fe/Co (refs 32, 68)Co/Ni (ref. 76)

Ls ≈ 100 (ref. 27) ls ≈ 45 (ref. 23) τs ≈ 26 × 10−6

τs ≈ 10−3

10−5 cm2 V−1 s−1

n-typeHOMO = 5.7 eV LUMO = 2.7 eVlight emitter

α-NPD LSMO/Co (ref. 25) 10−5 cm2 V−1 s−1

p-typeHOMO = 5.4 eVLUMO = 2.3 eV

CVB LSMO/Co (ref. 25) 10−3 cm2 V−1 s−1 HOMO = 5.5 eV LUMO = 2.5 eV Blue dopant for OLED

RRP3HT LSMO/Co (ref. 28)Fe50Co50/Ni81Fe19 (ref. 31)

Ls ≈ 80 (ref. 28) 10−1 cm2 V−1 s−1

p-typeHOMO = 5.1 eVLUMO = 3.5 eV

Tetraphenylporphyrin (TPP) LSMO/Co (ref. 29) 10−5 cm2 V−1 s−1

n-typeRed emitter for OLED

Pentacene Co:TiO2/Fe (ref. 77) 10−1 cm2 V−1 s−1

p-typeHOMO = 4.9 eV LUMO = 2.7 eV

Rubrene Co/Fe (ref. 39) Ls ≈ 13.3 1 cm2 V−1 s−1

p-typeHOMO = 5.2 eV LUMO = 3.0 eV

CuPc Co (ref. 42) 10−2 cm2 V−1 s−1

p-typeHOMO = 5.3 eV LUMO = 3.6 eV

nmat_2510_SEP09.indd 709 9/9/09 15:11:18

V. Dediu et al., Solid State Commun. 122 (2002) 181.

Alq3Xiong, Z. H., Wu, D., Vardeny, Z. V. & Shi, J. Nature 427, 821 (2004).

Organic spintronics

For a review, G. Szulczewski, et.al., Nature Mat, 8 (2009) 693.

NATURE MATERIALS | VOL 8 | SEPTEMBER 2009 | www.nature.com/naturematerials 695

commentary

valves3— have work functions between 4.5 and 5 eV. Even with the pillow effect, it is unsafe to assume that electrons are injected rather than holes13. In fact, a two-terminal device can be unipolar or bipolar. There is an urgent need to study three-terminal devices, where the sign of the carriers is defined by the bias on the gate, and the mobility and charge density can be deduced independently.

From a practical point of view, although most work has been done on the LSMO/Alq3 couple, there are good reasons to explore new combinations. Advantages of LSMO are its high spin polarization at low temperature and stable surface on exposure to air. A big disadvantage is its low Curie temperature, ≈ 340 K in thin films, which probably precludes room-temperature device operation. Alq3 is the archetypical light-emitting organic semiconductor, widely used in OLEDs. It is commercially available and inexpensive, but Alq3 has only low mobility (10–5 cm2 V–1 s–1) at room temperature. It exists in many polymorphs and isomers, and can even conduct holes. It reacts with evaporated Co (ref 14), and commercial-grade Alq3 tends to contain paramagnetic impurities15. There are opportunities to investigate other highly spin-polarized, high-Curie-temperature electrodes such as Fe3O4 or Co2MnSi in organic spin valves, although difficulties in the growth and in the definition of the terminations in these materials have to be overcome first. A good guess would be to use Co-based alloys that work well with metallic and insulating interlayers. In any case, the quality of the interfaces is critical, as the experience of the OLED and organic-electronics communities teach us.

What is it all good for?There are accepted metrics for organic-semiconductor devices, namely light output and quantum efficiency for OLEDs, and mobility, turn-on voltage and on/off ratios for OFETs. Similarly, inorganic spintronics also has its own set of metrics: room-temperature magnetoresistance, bias-dependence and spin-torque switching current. What should the metrics be for organic spin valves? We need to have some robust working devices, but once a thorough electrical characterization has established the charge-carrier regime, resistance changes at magnetic fields corresponding to the coercivity of the electrodes are needed to provide proof of spin transport across the entire junction. The variation of magnetoresistance with the organic layer thickness will then give an idea of the spin-diffusion length ls. Extracting the transport spin lifetime τs is more difficult. One needs to measure the layer mobility and spin

lifetime in a working device, something only recently achieved for unpolarized electrons16. A mobility value may be plucked from the literature, but the typical experimental range varies across several orders of magnitude.

Optical methods of measuring spin lifetime in inorganic semiconductors, based on Kerr or Faraday rotation, depend on spin–orbit interaction. Spin–orbit coupling is very weak in organic materials, so we need to look elsewhere. Other possibilities are noise measurements, electron paramagnetic resonance, muon spin resonance17, electrically detected magnetic resonance and spin-polarized two-photon photoemission18, which could help us follow the fate of spin-polarized electrons in organic semiconductors.

Once a well-characterized organic device is achieved, the fundamental question that still remains is what can this device do that is not already done by conventional spintronics? The plot of spin lifetime versus diffusion length for spin-polarized electrons in Fig. 4 is based on published data. When only one quantity is available, the other is calculated by using literature values for the mobility. Despite the uncertainties, organic materials cluster in the top-left corner of the plot — they possess long spin lifetimes, but the spins do not travel far. This suggests that potential applications will rely on interfacial properties.

We can therefore look forward to a phase of ‘spinterface engineering’ as spin electronics learns from organic electronics and organic chemistry on how best to get the spins moving into these structures. There are prospects for spin optics, quantum computing and non-volatile memory, but organic synthesis will not really affect this field until we can control the experimental conditions sufficiently to make reproducible

devices. A concerted effort is needed to understand how vacuum conditions, growth rates, film morphology, surface roughness and impurities influence the performance of organic-based magnetic devices. Some encouraging pointers are beginning to emerge from theory19,20. However, evidence for spin injection and transport is scarce and the first reproducible results have only recently been obtained17,18. It is certainly safe to say that organic spintronics is still a young field that needs more working hands and thinking heads for its potential to be fulfilled.

Greg Szulczewski†, Stefano Sanvito and Michael Coey are at the School of Physics and CRANN, Trinity College Dublin, Ireland. †Permanent address, Department of Chemistry, University of Alabama, Tuscaloosa, USA. e-mail: gjs@bama.ua.edu; sanvitos@tcd.ie; jcoey@tcd.ie

References1. Coey, J. M. D. Magnetism and Magnetic Materials (Cambridge

Univ. Press, 2009).2. Awschalom, D. A. & Flatté, M. E. Nature Phys. 3, 153–159 (2007).3. Dediu, V. A., Hueso, L. E., Bergenti, I. & Taliani, C. Nature Mater.

8, 707–716 (2009).4. Xiong, Z. H., Wu, D., Vardeny, Z. V. & Shi, J. Nature

427, 821–824 (2004).5. Santos, T. S. et al. Phys. Rev. Lett. 98, 016601 (2007).6. Wagemans, W., Bloom, F. L., Bobbert, P. A., Wohlgenannt, M. &

Koopmans, B. J. App. Phys. 103, 07F303 (2008).7. Parkin, S. S. P. in Handbook of Magnetism and Advanced Magentic

Materials Vol. 5 (eds Kronmüller, H. & Parkin, S. S. P.) Ch. 1 (Wiley, 2008).

8. Smidt, G. J. Phys. D. 38, R107–R122 (2005).9. Rashba, E. I. Phys Rev B 62, R16262 (2000).10. Hwang, J., Wan, A. & Kahn, A. Mater. Sci. Eng. R 64, 1–31 (2009). 11. Maruyama, T. et al. Nature Nanotech. 4, 158–161 (2009).12. Krause, S., Casu, M. B., Schöll, A. & Umbach, E. New J. Phys.

10, 085001 (2008).13. Jiang, J. S., Pearson, J. E. & Bader, S. D. Phys. Rev. B

77, 035303 (2008).14. Xu, W., Brauer, J., Szulczewski, G., Diver, M. S. & Caruso, A. N.

App. Phys. Lett. 94, 233302 (2009).15. Burke, F., Abid, M., Stamenov, P. & Coey, J. M. D.

J. Magn. Magn. Mater. (in the press).16. Matsui, H., Hasegawa, T., Tokura, Y., Hiraoka, M. & Yamada, T.

Phys. Rev. Lett. 100, 122601 (2008).17. Drew, A. J. et al. Nature Mater. 8, 109–114 (2009).18. Cinchetti, M. et al. Nature Mater. 8, 115–119 (2009).19. Bobbert, P. A., Wagemans, W., van Oost, F. W. A. & Koopmans, B.

Phys. Rev. Lett. 102, 156604 (2009).20. Xie, S. J., Ahn, K. H., Smith, D. L., Bishop, A. R. & Saxena, A.

Phys. Rev. B 67, 125202 (2003).21. Bass, J. & Pratt, W. P. Jr J. Phys. Condens. Matter 19, 183201 (2007).22. Kikkawa, J. & Awschalom, D. D. Nature 397, 139–141 (1999).23. Tsukagoshi, K., Alphenaar, B. W. & Ago, H. Nature

401, 572–574 (1999).24. Hueso, L. E. et al. Nature 445, 410–413 (2007).25. Tombros, N., Jozsa, C., Popinciuc, M., Jonkman, H. T. &

van Wees, B. J. Nature 448, 571–574 (2007).26. Dediu, V. A., Murgia, M., Matacotta, F. C., Taliani, C. &

Barbanera, S. Solid State Commun. 122, 181–184 (2002).27. Pramanik, S., Stefanita, C.-G., Patibandla, S. &

Bandyopadhyay, S. et al. Nature Nanotech. 2, 216–218 (2007).28. Shim, J. H. et al. Phys. Rev. Lett. 100, 226603 (2008).29. Limketkai, D. M., Jadhav, P. & Baldo, M. A. Phys. Rev. B

75, 113203 (2007).30. Appelbaum, I., Huang, B. & Monsma, D. Nature

447, 295–298 (2007).

Corrected after print: 24 August 2009

Figure 4 | Spin-diffusion length lS versus spin-diffusion time τs, for various materials. The organic semiconductors appear in the top-left corner. They have a long spin lifetime but, owing to their low mobilities, spin-diffusion lengths are short. The data are taken from the literature as indicated.

6T (ref. 26)

100 101 102 103 104 105 106

ls (nm)

10–16

10–12

10–8

10–4

100

s (s)

GaAs(ref. 22)

CNT (ref. 24)

Graphene (ref. 25)

CNT (ref. 23)Cu (ref. 21)

Co (ref. 21)

Rubrene (ref. 28)

Si (ref. 30)

Alq3(refs 4,27)

nmat_2518_SEP09.indd 695 20/8/09 10:48:03

Alq3 6TRubrene

CoCu

CNT

GrapheneCNT

Si

GaAs

λS (nm)

τ S (s

)

100 101 102 103 104 105 10610-16

10-12

10-8

10-4

100

• Pure spin current transport properties are not fully understood.

cf.) Recently reported works, K. Ando et.al., Nature Mat. 12 (2013) 622. S. Watanabe et. al., Nature Phys. 10 (2014) 308.

• Spin transport is strongly limited by disorder.

vsBand transport

Inorganic

Hopping transportOrganic

(How is the spin relaxation mechanism in the hopping regime? )

• Comprehensive study of pure spin transport mechanism in organic semiconductors with strong disorder

Motivation

λS = √DS×τSSpin pumping & inverse spin Hall effect

Charge transport

ESR(Electron spin resonance)

• Experimental investigation of characteristic spin-transport parameters

Conducting polymer PEDOT:PSS

Water-based solution of PEDOT: PSS

• Dopant density: ~1020-1021 /cm3

• Resistivity ~1 Ωcm (in-plane)~103 Ωcm (out-of-plane)

Spin coating

PSS

PEDOT

PEDOT is doped with PSS.

• Carrier: Hole of PEDOT

Conducting polymer PEDOT:PSS• Nano-scale core-shell structure of PEDOT:PSS

structure of the micelle, assuming an unbent chain of PEDOTin the solid film.22 This consistency indicates that thenanocrystal along the c-axis is composed not by severalfragments of PEDOT but by a single polymer chain from oneend to the other. This structure is difficult to be confirmed bythe present WAXS measurements owing to only one polymermolecule unit expected in the PEDOT nanocrystal. To someextent, however, the linear structure of PEDOT may besupported by the observation of a tiny peak observed at q =16.8 nm−1 corresponding to a real space periodicity of 0.374nm, which is comparable to a periodicity of the monomer unitof PEDOT (∼0.385 nm). This observation on the periodicitysuggests that PEDOT chain does not have a compact coil but alinear structure in the solid film as has been reported so far.22

For the a-axis, there must be two stacked PEDOT layersbecause the short axis length, dν, of ∼2.4 nm in the oblateellipsoidal micelle is almost equal to double the thickness of thePEDOT layer stacking. It is noted that, in this model, PSSshould be coordinated outside of the PEDOT nanocrystal andnot in between the PEDOT layers. Therefore, a stablecrystallite can be composed solely of two PEDOT layers, thatis, the single PEDOT bilayer, along the a-axis and not ofalternative stacks with inserted PSS. In the proposed PEDOTbilayer structure, an electrostatic instability between PEDOTlayers can be suppressed owing to the charge balance betweenPEDOT cation and anionic PSS. The orientation of the b- andc-axes between two PEDOT layers is reasonably expected to beordered considering from overlapping the π-orbital betweenPEDOT layers. This bilayer structure, however, if it exists, ishard to be detected in the WAXS profiles because of too broadscattering structure expected from single bilayer unit in onePEDOT nanocrystal.We discuss briefly the origin of the crystallization of PEDOT

and the enhancement of crystallinity by the additive solvents.The PEDOT has an inclination to be crystallized essentiallywith doping small anion molecules as p-toluenesulfonate(tosylate),18 PF6

−,19 etc., because of the intrinsic nature ofthe linear and planar molecular characteristics of PEDOT. Inthe case of PSS polymer as the charge balancing anion,however, the crystallinity of PEDOT is inhibited by strongelectrostatic coupling to the entangled PSS. The additive polarsolvent like as EG may weaken electrostatic interactions by theshielding effect. Then the transfer integrals, i.e., π−πinteractions, among PEDOT works to orient themselves andthe resultant crystallization occurs at the moment of theevaporation of water and additive solvent. In this sense, theeffective solvent for the enhancement of the crystallizationshould have an advantage of larger polarization and higherboiling temperature than water. This scenario is consistent withthe additive solvent effect of ethanol. Adding ethanol to thewater dispersion before preparing the films is not effective toincrease the conductivity,12 but the post-treatment for thepristine film gives the high conductivity and enhances thecrystallinity as presented in this paper. These different resultswith using the same solvent are attributed to whether theethanol still exists at the moment of the solidification or not. Inthe former case, ethanol disappears before evaporation of water,while ethanol remains to the last moment of the solidificationin the latter case.Based on these considerations, the structural model of

PEDOT:PSS in the solid film with EG post-treatment is shownin Figure 4. Nanometer-size PEDOT crystal is found to begrown in almost whole PEDOT core region. In this figure,

however, the two-dimensional configuration of PEDOT:PSS isnot represented to scale, and the alignment of eachPEDOT:PSS is not accurately shown although the structureand size of the PEDOT nanocrystal is drawn quantitatively onthe basis of the observations. From the present WAXS studies,we do not mention the PSS outer-shell part because thestructure consisting of complex and entangled PSS soft chains isdifficult to be analyzed from X-ray scattering experiments. Inorder to supplement the present structure model, weperformed a dynamic light scattering (DLS) measurement forthe water dispersion of PH1000 without EG. The observedparticle size was in a range of 10−15 nm assuming a sphericalshape, which was consistent with previous report on PH500 byDLS measurements.23 This diameter obtained by DLS shouldbe a whole size of the PEDOT:PSS particle. For comparison,diameter of a spherical particle formed by single self-entangledPSS chain with PEDOT segments has been estimated to be ∼6nm by simple calculations on the basis of the weight ratio(1:2.5) of PEDOT to PSS and the molecular weight (ca. 1000−2500 for PEDOT and ca. 400 000 for PSS).10 This value mightbe ideal lower bound for the PEDOT:PSS particle size becauseonly single PEDOT chain was considered to form theminimum particle. Considering such rough estimation,10 theparticle size obtained by DLS is reasonably consistent with thecalculation for the minimum size. Therefore, a possiblethickness of the PSS shell is evaluated to be roughly 3−5 nmfrom DLS results although the micelle may become swellingwith water particularly in the PSS part. The estimated thicknessof the PSS shell is consistent with previous reports.10,21−25 Inaddition, it is interesting to compare the estimated size for thePEDOT:PSS monolayer with the minimum thickness (∼16nm) achieved in the thin film fabrication.4,23 Based on thepresent estimation, the PSS shell parts are depicted rathersimply as an arrangement of isolated ellipsoidal PEDOT:PSS inFigure 4. But the complex overlap and entanglement of PSSbetween the adjacent shells likely occur in a real sample.In previous studies on the morphology of the film,9 the

oblate ellipsoidal PEDOT:PSS aligns on the substrate lining theflattened directions; that is, the a-axis of the PEDOTnanocrystals may be aligned perpendicular to the film. Theorientation of the b- and c-axes of the PEDOT bilayernanocrystals, however, is considered to be random along the

Figure 4. Proposed structural model of PEDOT:PSS in the solid film.Nanocrystals of PEDOT are formed throughout the entire PEDOTcore region by adding EG or post-treatment with EG.

Macromolecules Article

dx.doi.org/10.1021/ma300120g | Macromolecules 2012, 45, 3859−38653863

Highly doped OSC with strong disorder

T. Takano et.al., Macromolecules 45 (2012) 3859.

Insulating PSS shellConducting PEDOT core

Spin pumping & ISHE measurement in Py/PEDOT:PSS/Pt trilayers

MicrowavePt

PEDOT:PSS

Py

V

φH JS

Py

PEDOT:PSS

Pt

Hdc

hac

V

Py/PEDOT:PSS/Pt trilayer

The pure spin current through the PEDOT:PSS is detected as a voltage signal at the Pt layer.

Spin injector

Spin detector

S. Mizukami, et.al., PRB 66, 104413 (2002). E. Saitoh, et.al., APL 88, 182509 (2006).

Spin pumping & ISHE measurement in Py/PEDOT:PSS/Pt trilayers

PEDOT:PSS: 60nmf = 9.5 GHz

JS

Py

PEDOT:PSS

Pt

Hdchac

V 01020304050

-10

Volta

ge (n

V)FM

R in

tens

ity

(arb

. uni

ts)

Symmetric

Asymmetric

VISHE

Symmetric contribution of the voltage signal Inverse spin Hall voltage

FMR

Voltage

Spin diffusion length of PEDOT:PSS = 140 ± 20 nm

PEDOT:PSS thickness dependence of VISHE

0 50 100 150 200PEDOT:PSS thickness (nm)

Norm

alize

d V I

SHE

(arb

. uni

ts)

λS = 160 ± 8 nm

λS = 120 ± 40 nm

VISHE ∝ JS(tPE) ≈ JS(0)exp( )[1-tanh( )]tPE

λPE

tPE

λPE

Spin absorption by the Pt layer is considered.

JS(0)

Py PEDOT:PSS Pt

JS

x

0 t PE Z0

JS(t PE)

JS(x)

• 1D diffusion equation for trilayer

Comprehensive study of spin transport

Spin pumping

Charge transport

ESR(Electron spin resonance)

λS = 140 nm

λS = √DS×τS

Charge transport measurement

Insulating behavior below room temperature

T (K)

PEDOT thickness = 148 nmρ P

E⊥ (Ω

cm)

103

104

105

106

Charge transport measurement

• Almost linear with T -1/4

T -1/4 (K -1/4)

103

104

105

106

ρ PE⊥

(Ωcm

)

300K 100K 50K 20K

Charge transport measurement

ρPE ∝ exp⊥ T0T[ ]1/4

Variable range hopping (VRH) conduction

β : numerical factor (18.1 for 3D-VRH)

ξ: localization length

N(EF): DOS at EF

T0 = 1.4×105 K

T0 = βkBN(EF)ξ3

with characteristic temperature

• Almost linear with T -1/4

T -1/4 (K -1/4)

103

104

105

106

ρ PE⊥

(Ωcm

)

300K 100K 50K 20K

Einstein relation for VRH conduction• Electron transport is dominated by tunneling process between metallic localized states. • Hopping probability (∝conductivity) is proportional to the N(EF) of the localized states.

σ = e2N(EF)DSG. Paasch, et. al., Synth. Met. 132 (2002) 97.

ξLmkBT

R

EF

Einstein relation for VRH conduction• Electron transport is dominated by tunneling process between metallic localized states. • Hopping probability (∝conductivity) is proportional to the N(EF) of the localized states.

N(EF) ≈ 1×1018 [eV-1cm-3]

DS = 7×10-3 cm2/s

A.M. Nerdes et. al., Adv. Mater. (2007), 19, 1196.

Cross section AFM

case PEDOT:PSS, generally leads to a phase contrast in theTM-AFM. Therefore, dark and bright features in the X-AFMimages are interpreted as PEDOT-rich and PSS-rich regions,respectively. With this in mind, Figure 2B can be interpretedas a side view of PEDOT-rich “pancakes” with a thickness ofa few nanometers and a diameter of a few tens of nanometersseparated by PSS lamellas. In the normal direction, the sepa-rating barriers, that is, the PSS lamellas are quasi-continuous,whereas the separations in the lateral direction do not seemto be fully closed. Clearly, a top view of these lasagna-likestructures should look like Figure 2A. Moreover, this lasagna-type morphology is fully consistent with the lamellar structureproposed previously.[4] These STM and X-AFM measure-ments can be combined into the schematic morphologicalmodel depicted in Figure 3.

Now, the differences in electrical transport properties inthe ! and ⊥ directions can be understood. In the ! direction,that is, within the PEDOT-rich lamella, conduction can takeplace by 3D VRH because in this direction the PEDOT-richdomains are only separated by the not-completely closedconstrictions. These are likely to either form a thin barrier,or no barrier at all, thereby allowing carriers to hop to non-

nearest-neighbor sites. Moreover, theabsence of thick barriers in the con-duction path will lead to relativelyhigh r values (ca. 10–3 S cm–1 at roomtemperature). In the ⊥ direction, thePEDOT-rich domains are separatedby thick barriers, formed by the PSS la-mella, which enforce nearest-neighborhopping and cause a reduction of r(ca. 2 × 10–6 S cm–1 at room tempera-ture). In a hopping system, an increasein barrier thickness has a similar effectas an increase in temperature in thesense that it promotes nearest-neighborhopping.[13] A transition to a VRH re-gime at low T was however not ob-served in the investigated temperatureregime.

A striking feature that is often ob-served[16] on small-scale STM images is

a pronounced alignment of the PEDOT-rich domains (seeFig. 4B and C). Such alignment could easily result from thecentrosymmetric forces in the spin coating process and wouldlikely cause in-plane anisotropy in the electrical properties ofthe resulting film. On larger scale images like Figure 4A, aunidirectional alignment is not always clearly observable.Moreover, no statistically relevant correlation was foundbetween the alignment direction and the position on the3 cm × 3 cm substrate that was used for spin coating. In orderto shed further light on the alignment issue, the conductivitywas measured in different in-plane directions. For this, a shad-ow mask as depicted in Figure 5 A was made. It allows r to beprobed in different orthogonal directions on equivalent posi-tions on the substrate. The results of the measurements areshown in Figure 5B.

Figure 5B shows that the spin coated films of PEDOT:PSSform a reasonably homogeneous material in terms of electri-cal conductivity. Moreover, in multiple data sets, no correla-tion between the electrode alignment and the conductivitywas found. From this and from larger scale topographic im-ages (Fig. 4A) we conclude that the observed alignment ofPEDOT-rich domains is most likely confined to randomly ori-ented sub-micrometer-sized domains. Note, also, that theaverage r in Figure 5B is about one order of magnitude lowerthan of that in Figure 1A at room temperature. Because themeasurements of Figure 5B were performed in air andPEDOT:PSS films are hygroscopic, such differences are dueto rapid water uptake.[6] Further studies on the influence ofwater and post-treatments on the r of PEDOT:PSS is on-going.

In summary, we studied the conductivity of PEDOT:PSSthin films by temperature-dependent conductivity measure-ments in the perpendicular and lateral directions with respectto the substrate surface. The surprising anisotropy, both inconductivity magnitude and in the conduction mechanismcould successfully be correlated to a detailed morphological

COM

MUN

ICATI

ON

1198 www.advmat.de © 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2007, 19, 1196–1200

Figure 2. A) 200 nm × 200 nm topographic STM image of PEDOT:PSS on indium tin oxide (ITO)at 2.3 V, tunneling current 10 pA, and vertical scale 15 nm, the inset shows a line section.B) 200 nm × 200 nm cross-sectional AFM phase image of cleaved PEDOT:PSS on glass, verticalscale is 8°. The glass substrate is on the bottom side of the image, as shown by the inset of530 nm × 580 nm and vertical scale 70°. A pancake-like particle is highlighted by the ellipse.

Figure 3.Cross-sectional view of the schematic morphological model forPEDOT:PSS thin films derived from combined STM and X-AFM measure-ments. PEDOT-rich clusters (dark) are separated by lamellas of PSS(light). The PEDOT-rich lamella is composed of several pancake-like par-ticles as pictured by the dotted lines. The typical diameter d of the parti-cles is about 20–25 nm and the height h is about 5–6 nm.

200

nmT0 = 18.1kB N(EF) ξ3

≈10 nm1.4×105 K

N(EF) ≈ 1×1018 [eV-1cm-3]

PEDOT:PSS cluster: 10-20 nm

σ = e2N(EF)DS

• Characteristic temperature of hopping conduction

cf) DS ≈ 2×102 cm2/sfor pure Cu and Ag

Comprehensive study of spin transport

Charge transport

ESR(Electron spin resonance)

λS = √DS×τS

DS = 7×10-3 cm2/s

λS = 140 nmSpin pumping

ESR experiment of PEDOT:PSS film

ΔHESR = 24 Oe

3300 3350 3400 3450H (Oe)

dIES

R/dH

(arb

. uni

ts)

Fit

IESR ∝ mz

I ESR

Microwave magnetic field

Long T1

Short T1

Long T1

Short T1 m

PEDOT:PSS

IESR

Spin lifetime for dc spin current

• Microwave strength dependence of ESR intensity ∂mz

∂t = γ[m×H]z + m0 - mzT1

Longitudinal spin relaxation time T1

ESR experiment of PEDOT:PSS film

• IESR does not saturate up to the highest hac.

0 0.2 0.4 0.6 0.8 1hac (Oe)

ESR

inte

nsity

(arb

. uni

ts)

PEDOT:PSSR.T.

Experiment

T1 = 5 - <100 ns

ΔHESR = 24 Oe

3300 3350 3400 3450H (Oe)

dIES

R/dH

(arb

. uni

ts)

Fit

IESR

Spin lifetime for dc spin current

• Microwave strength dependence of ESR intensity ∂mz

∂t = γ[m×H]z + m0 - mzT1

Longitudinal spin relaxation time T1

Simulation

IESR =

T1=100 ns

T1 = 5 ns

hac1+hacγ2T2T1 2

ESR

inte

nsity

(arb

. uni

ts)

0 0.2 0.4 0.6 0.8 1hac (Oe)

T1 = 50 ns

T1=1 μs

PEDOT:PSS

Comprehensive study of spin transport

ESR(Electron spin resonance)T1 = 5 - <100 ns

Spin pumpingλS = 140 nm

Charge transportDS = 7×10-3 cm2/s

λS = √DS×τS

How is the relation between τStransport and T1 ?.

τStransport = 28 ns

Comparison between τStransport and T1

1/⌧ transportS

= 1/⌧hopS

+ 1/⌧ trapS

1

τStransport: trap and hopping process

λPE ≈ 140 nm, τStransport ≈ 28 ns

Trapping process

Hopping processhopping length Lm = 25 nm

1/⌧ transportS

= 1/⌧hopS

+ 1/⌧ trapS

1/T1

⇡ 1/⌧ trapS

1/⌧ transportS

⇡ 1/⌧hopS

+ 1/T1

1/⌧ transportS

⇡ 1/T1

1

Now,

1/⌧ transportS

= 1/⌧hopS

+ 1/⌧ trapS

1/T1

⇡ 1/⌧ trapS

1/⌧ transportS

⇡ 1/⌧hopS

+ 1/T1

1/⌧ transportS

⇡ 1/T1

1/⌧hopS

⌧ 1/⌧ trapS

1

Spin relaxation mostly occurs in the trapping process.

T1: trapping state only1/⌧ transportS

= 1/⌧hopS

+ 1/⌧ trapS

1/T1

⇡ 1/⌧ trapS

1

ESR

1/⌧ transportS

= 1/⌧hopS

+ 1/⌧ trapS

1/T1

⇡ 1/⌧ trapS

1/⌧ transportS

⇡ 1/⌧hopS

+ 1/T1

1

~ξ

~Lm

~λPES

Py Pt

PEDOT-rich corePSS insulating shell

Spin transport and relaxation mechanism in PEDOT:PSS

Lm = 25 nm < λS =140 nm

• Spin angular momentum is almost preserved in the hopping event.• Spin relaxation mostly occurs in the trapping process.

M. Kimata, et. al., PRB accepted.

Summary• Comprehensive study of spin transport in highly doped disordered polymer film PEDOT:PSS was performed.

ESRT1 = 5 - <100 ns

Spin pumpingλS = 140 nm

Charge transportDS = 7×10-3 cm2/s

λS = √DS×τS

τStransport = 28 ns τStransport ≈ T1

• Spin relaxation mostly occurs in the trapping process.

• Spin angular momentum is almost preserved in the hopping event.

~ξ

~Lm

~λPES

Py Pt

PEDOT-rich corePSS insulating shell

M. Kimata, et. al., PRB accepted.