The fullerene C60, C70, and C84 molecules, that are composed of ∼99% naturally abundant 12C having spinless nuclei, are considered to have miniature hyperfine interaction and also weak intrinsic spin-orbit coupling (SOC) due to the light carbon atoms. However, it has been theoretically predicted that the curvature of the fullerene molecules may increase the SOC due to the induced hybridization of the π and σ electrons on the carbon atoms that reside on the fullerene molecule surface. In this work, we have measured the spin diffusion length in films of C60, C70, and C84 in NiFe/fullerene/Pt trilayer devices, where pure spin current is injected into the fullerene film at the NiFe/fullerene interface via spin pumping induced by microwave absorption at ferromagnet resonance conditions, and subsequently detected at the fullerene/Pt interface as electrical current via the inverse spin-Hall effect. The obtained spin diffusion lengths in the fullerene films are of the order of 10 nm and increase from C60 to C84 in which the fullerene molecule’s curvature decreases; this finding validates the existence of curvature-induced SOC in the fullerenes. Our results deepen the understanding of spin transport in fullerene films and may benefit the design of molecular spintronic devices.

The electron spin degree of freedom has been utilized in spintronic devices, where information is carried out by the spin sense.1 In the past decade, organic semiconductors have attracted increasing attention in the field of spintronics due to their inherent long spin lifetime.2,3 The organic semiconductors are mainly composed of light atoms such as carbon and hydrogen and are thus expected to have weak spin-orbit coupling (SOC). Because fullerene molecules are composed of carbon atoms with ∼99% naturally abundant 12C atoms having spinless nuclei, the hyperfine interaction (HFI) is also expected to be weak in these molecules. The weak SOC and HFI should lead to a long spin relaxation time in the fullerenes. Spin related processes in the fullerenes have been recently studied in various optoelectronic and spintronic devices4–8 such as organic light emitting diodes (OLED),9,10 spin valves,4,11,12 spin-pumping devices,13 and devices based on spin optical conversion.8 However, no direct comparison of the spin related properties among the various fullerene molecules has been done so far.

The spin diffusion length in organic semiconductors has been traditionally measured by electrical means using spin-valve devices,2 i.e., La2/3Sr1/3MnO3 (LSMO)/organic/Co. In such devices, spin polarized electric current is injected into the organic layer from one ferromagnetic (FM) electrode (LSMO) and detected by the other FM electrode (Co). Indeed, giant magnetoresistance (GMR) has been obtained in such organic spin valves, which originates from the spin scattering rate change when the magnetizations of the two FM electrodes vary between parallel and antiparallel configurations. Furthermore, C60 molecules based spin valves have been successfully fabricated using a variety of FM electrodes. Also, it has been shown that GMR occurs in C60 based spin valves in a wide range of temperatures.4,5,12 However, the obtained spin diffusion length, λS60, in such devices has been in the range of several to hundreds of nanometers.4,5 One possible reason for the obtained broad λS60 distribution is that the spin polarized electrical current injection in C60 spin valves is limited by the conductivity mismatch between the FM electrodes and the C60 interlayer. Consequently, it crucially depends on the FM/C60 spinterface states that may help the injection of spin aligned carriers into the C60 interlayer.14–18 

Recently, however, thanks to the discovery of spin pumping,19 pure spin current transport without the need of a charge carrier may be achieved at FM/organic interfaces.20,21 The pure spin current is generated in this case by microwave (MW) induced magnons at the FM/organic layer that are at ferromagnetic resonance (FMR) conditions. At these conditions, the FM layer absorbs the resonant microwaves and induces pure spin current in the organic layer through the FM/organic interface via the spin-pumping process. Subsequently, the induced spin current in the organic layer may be detected either via the inverse spin-Hall effect (ISHE) in the same layer or by reaching another interface with heavy atom metals having a large SOC, say Pt, where it is detected electrically again by the ISHE.21,22 Upon varying the organic layer thickness, the latter method may be effectively used to determine the spin diffusion length in the organic interlayer, since the induced spin current is attenuated as it reaches the Pt overlayer. The advantage of this method is that the well-known FM/organic conductivity mismatch and spinterface problems are overcome, since no spin polarized electric current is involved.

Most notably, it has been shown that the SOC plays a key role in the attenuation of spin polarized electric current in the fullerene molecules, which cannot be attributed to the very weak intrinsic SOC and HFI in these compounds.5,10 One explanation of this finding may be based on theoretical models claiming that the SOC strength, ξ, in fullerene molecules should be larger than thought before because it is composed of three terms: intrinsic, curvature-induced, and Rashba:23,24ξ=ξin+ξcurv+ξRashba. The Rashba SOC, ξRashba, is related to the inversion symmetry breaking and should be negligible in the fullerene films. Since the other terms are small, the SOC strength in the fullerene mainly comes from ξcurv, which originates from the curvature related hybridization between the π and σ electrons.10 

In the present work, we have measured the spin diffusion lengths of several fullerene molecules including C60, C70, and C84 by the method of spin-pumping pure spin current transport, excluding the inaccuracy related to the FM/fullerene conductivity mismatch and spinterface states. We found that the spin diffusion length in the fullerenes actually increases from C60 to C84, in agreement with the decrease in the molecular curvature, which is consistent with the theory.23,24

Figure 1(a) shows the molecular structure of the fullerene C60, C70, and C84, and the corresponding surface morphologies in fullerene films with ∼30 nm capped on the Pt layer, which were measured by atomic force microscopy (AFM). We note that the effective molecular curvature decreases from C60 to C84. In addition, all the fullerene films show extremely low roughness on a large scale, which indicates comparable deposition conditions for the subsequent NiFe overlayer, needed to complete the trilayer device structure. Figure 1(b) shows the schematics of our spin current transport measurement geometry. The Ni80Fe20 /fullerene/Pt trilayer (protected by a SiO2 layer) was placed on a coplanar microwave guide. Magnon excitations in the NiFe were generated by microwave absorption at FMR conditions. The microwave-generated spin waves transfer spin angular momentum to the fullerene film via spin pumping at the NiFe/fullerene interface, inducing pure spin current in the fullerene layer. The spin current is attenuated while being transported through the fullerene film, reaching the fullerene/Pt interface at the other side of the trilayer device. The attenuated spin current is then detected as an electric current in the Pt film via the ISHE, measured parallel to the interface.

FIG. 1.

(a) The molecular structure of C60, C70, and C84 fullerenes and corresponding surface morphologies characterized by AFM measurements. The roughness of the fullerene films is ∼4 nm in scans having a range of ∼100 μm. (b) Schematics of the pure spin current transport used in our experiments. The NiFe ferromagnetic substrate absorbs MW at a magnetic field resonance condition that generates nonequilibrium magnons. Spin angular momentum carried by the generated magnons is transferred to the carriers in the fullerene film at the NiFe/fullerene interface by the spin-pumping process. The induced spin current in the fullerene film diffuses from the NiFe/fullerene interface and reaches the fullerene/Pt interface, where it is converted into electrical current via the ISHE in the Pt layer. (c) Cross-sectional SEM of a typical microstructure in the NiFe/fullerene/Pt trilayer device used for the spin diffusion measurements.

FIG. 1.

(a) The molecular structure of C60, C70, and C84 fullerenes and corresponding surface morphologies characterized by AFM measurements. The roughness of the fullerene films is ∼4 nm in scans having a range of ∼100 μm. (b) Schematics of the pure spin current transport used in our experiments. The NiFe ferromagnetic substrate absorbs MW at a magnetic field resonance condition that generates nonequilibrium magnons. Spin angular momentum carried by the generated magnons is transferred to the carriers in the fullerene film at the NiFe/fullerene interface by the spin-pumping process. The induced spin current in the fullerene film diffuses from the NiFe/fullerene interface and reaches the fullerene/Pt interface, where it is converted into electrical current via the ISHE in the Pt layer. (c) Cross-sectional SEM of a typical microstructure in the NiFe/fullerene/Pt trilayer device used for the spin diffusion measurements.

Close modal

The metallic NiFe and Pt layers and the SiO2 protection layer in the device were deposited by electron beam evaporation. All depositions were done in high vacuum at a base pressure of 1 × 10−7 mbar. The C60/C70/C84 powders (American Dye Source) were thermally evaporated at a slow evaporation rate of ∼0.05 nm/s on top of the Pt film. The fullerene film thickness was determined by profilometry (KLA Tencor) measurements. The device size was typically 0.7 × 3 mm2 for the Pt layer and 0.7 × 1 mm2 for the NiFe layer designed to be at the center of the Pt layer. The FM/C60 interface was characterized by cross-section view scanning electron microscopy (SEM) using a FEI Helios NanoLab 650 setup [see Fig. 1(c)]. The sharp interfaces seen in the SEM picture indicate relatively low fullerene surface roughness.

The microwave (MW) radiation for the spin pumping into the fullerene layer was generated by an Agilent N5173B amplifier at a frequency ranging from 9 kHz to 20 GHz.22 The MW guide consisted of a 250 μm wide transmission line with the radio frequency magnetic field perpendicular to the DC magnetic field. For the FMR measurements, we used a DC magnetic field, H, a 10 dBm attenuator for C60 and C70, and a 20 dBm attenuator for the C84 device, and the MW was detected by a zero-bias Schottky detector (Krytar 201B). The ISHE voltage generated in the Pt overlayer was detected by the phase sensitive technique, where the MW intensity was modulated at 17 kHz. The electric current direction in the Pt layer is perpendicular to both spin current and spin polarization directions, EISHEJS×σ, where JS and σ are the spin current and spin polarization, respectively [see Fig. 1(b)]. For each device with a certain fullerene film thickness, we measured two nominally same devices in order to estimate the experimental error bars for the ISHE measurements.

Figures 2(a) and 2(c) show the FMR(H) response spectra of the NiFe/Fullerene/Pt trilayer for the three fullerene films of thickness ∼65 nm, measured by sweeping the DC magnetic field, H, subjected to MW irradiation at a fixed frequency, f = 3 GHz. The resonance fields (HR = 8.1 mT) and linewidths are comparable for all three fullerene-based trilayers; this shows that the fullerene interlayer does not influence the NiFe magnetization properties. The obtained HR can be fit using the Kittel equation25f=γHR(HR+4πMS), where γ is the gyromagnetic ratio, and MS=1.1×106A/m is the saturation magnetization of the NiFe film. The relatively large NiFe obtained magnetization value indicates high quality FM films grown onto the fullerene films, having relatively flat surfaces. The very similar resonance fields, combined with the AFM measurements shown in Fig. 1(a), demonstrate that the NiFe films qualities do not significantly change upon deposition on the various fullerene films used here. We note that the different signal to noise (S/N) levels of the FMR responses are due to different attenuations for the MW detections, as noted above.

FIG. 2.

The FMR(H) and VISHE(H) responses obtained in NiFe/fullerene/Pt trilayer devices. (a)–(c) Derivative of the microwave absorption in the trilayer devices based on fullerene C60, C70, and C84, respectively. The film thickness of all the three fullerenes was ∼65 nm. (d)–(f) The corresponding ISHE(H) response obtained in the Pt layer at the resonance conditions of NiFe films.

FIG. 2.

The FMR(H) and VISHE(H) responses obtained in NiFe/fullerene/Pt trilayer devices. (a)–(c) Derivative of the microwave absorption in the trilayer devices based on fullerene C60, C70, and C84, respectively. The film thickness of all the three fullerenes was ∼65 nm. (d)–(f) The corresponding ISHE(H) response obtained in the Pt layer at the resonance conditions of NiFe films.

Close modal

Figures 2(d)2(f) show the H-response of the corresponding ISHE voltages, VISHE(H), measured at the Cu contacts attached to the Pt layer in the three fullerene devices. It is clear that VISHE increases with an increase in the fullerene size from C60 to C84, in which the molecular effective curvature decreases, indicating that the spin current in the related fullerene film that survives the spin polarization attenuation increases from C60 to C84. This indicates that the smaller the attenuation of the spin polarization, the smaller is the molecular curvature, which, according to the theory, is consistent with a smaller SOC.23,24

Figures 3(a)3(c) show the VISHE(H) responses in the NiFe/fullerene/Pt trilayers with the three fullerene molecules (C60, C70, or C84) measured at various thicknesses, d. It is clear that VISHE decays with d for all three fullerene-based trilayers. The decay with d may be fit by an exponential function,21 namely, VISHEed/λs, where λs is the spin diffusion length in the fullerene film [see Fig. 3(d)]. From the fits, we obtain the value of the room temperature spin diffusion lengths in the C60, C70, and C84 films. These are λS60 = 13 ± 2 nm, λS70 = 18 ± 3 nm, and λS84 = 25 ± 3 nm, respectively.

FIG. 3.

Spin diffusion length measurement in C60, C70, and C84 films using the pure spin current transport method in NiFe/fullerene/Pt trilayers measured at room temperature. (a)–(c) VISHE(H) responses obtained in the trilayer structures with various fullerene layer thickness, d, at FMR conditions. (d) Thickness dependence of the maximum VISHE(H) response for C60, C70, and C84 films, which is fitted by an exponential decay. The respective extracted values of the spin diffusion length for the different fullerene films are shown.

FIG. 3.

Spin diffusion length measurement in C60, C70, and C84 films using the pure spin current transport method in NiFe/fullerene/Pt trilayers measured at room temperature. (a)–(c) VISHE(H) responses obtained in the trilayer structures with various fullerene layer thickness, d, at FMR conditions. (d) Thickness dependence of the maximum VISHE(H) response for C60, C70, and C84 films, which is fitted by an exponential decay. The respective extracted values of the spin diffusion length for the different fullerene films are shown.

Close modal

The spin-orbit coupling in conventional organic semiconductors dominates the spin relaxation time, τS, which is linked to the spin diffusion length via the carrier mobilities μ:26λS=τSμe/kBT. The reported carrier mobilities of the fullerene films range from 0.02 to 0.65 cm2/V s for C60, from 0.003 to 0.066 cm2/V s for C70, and 0.003 cm2/V s for C84, respectively.27,28 We can conclude that the obtained variation in the spin diffusion lengths does not originate from the variation of the carrier mobilities in the different fullerene films. We, therefore, infer that the spin relaxation time dominates the spin diffusion length in the fullerenes. Thus, the measured increase of the spin diffusion length from C60 to C84 is due to a decrease of the corresponding spin relaxation time, τS, with the fullerene molecular size. The spin relaxation mechanism in the fullerene films is mainly attributed to the SOC, since the HFI is negligibly small in these compounds.3 Under this assumption, the spin diffusion length is nearly proportional to the square root of the SOC strength. Our results are therefore in agreement with the assumption that the SOC in the fullerenes decreases upon increasing the molecular size or equivalently decreasing the effective molecular curvature.29,30 The π-electrons alone do not possess a substantial SOC, but the molecular curvature in the fullerene induces hybridization between the π and the σ electrons, and this, in turn, leads to an enhanced SOC in the fullerene molecules.

In summary, we have systematically investigated the spin diffusion length in films based on three different fullerene molecules, namely, C60, C70, and C84, using a pure spin current transport method. In this technique, magnons in the NiFe film are excited by microwave absorption at resonance conditions and subsequently transfer spin angular momentum to the fullerene film via the FM/fullerene interface; this spin-pumping method does not require charge injection into the fullerene semiconductor. The induced spin current in the fullerene layer is attenuated and converted by the ISHE into charge current in the Pt overlayer at the fullerene/Pt interface. We found that the room temperature spin diffusion length in the fullerenes is of the order of ∼10 nm and increases with a smaller molecular curvature. We attribute the mechanism of spin attenuation in the fullerene films to the curvature-induced SOC that stems from the hybridization between the π and the σ electrons of the carbon atoms on the fullerene molecule surface.

We thank M. Groesbeck and C. Zhang for help with the experiments. This work was supported by the National Science Foundation (NSF) (Grant No. DMR-1701427).

1.
S. A.
Wolf
,
D. D.
Awschalom
,
R. A.
Buhrman
,
J. M.
Daughton
,
S.
von Molnár
,
M. L.
Roukes
,
A. Y.
Chtchelkanova
, and
D. M.
Treger
, “
Spintronics: A spin-based electronics vision for the future
,”
Science
294
,
1488
1495
(
2001
).
2.
Z. H.
Xiong
,
D.
Wu
,
Z.
Valy Vardeny
, and
J.
Shi
, “
Giant magnetoresistance in organic spin-valves
,”
Nature
427
,
821
824
(
2004
).
3.
V.
Dediu
,
M.
Murgia
,
F. C.
Matacotta
,
C.
Taliani
, and
S.
Barbanera
, “
Room temperature spin polarized injection in organic semiconductor
,”
Solid State Commun.
122
,
181
184
(
2002
).
4.
X.
Zhang
,
S.
Mizukami
,
T.
Kubota
,
Q.
Ma
,
M.
Oogane
,
H.
Naganuma
,
Y.
Ando
, and
T.
Miyazaki
, “
Observation of a large spin-dependent transport length in organic spin valves at room temperature
,”
Nat. Commun.
4
,
1392
(
2013
).
5.
T. D.
Nguyen
,
F.
Wang
,
X.-G.
Li
,
E.
Ehrenfreund
, and
Z. V.
Vardeny
, “
Spin diffusion in fullerene-based devices: Morphology effect
,”
Phys. Rev. B
87
,
075205
(
2013
).
6.
C. D.
Dimitrakopoulos
and
P. R. L.
Malenfant
, “
Organic thin film transistors for large area electronics
,”
Adv. Mater.
14
,
99
117
(
2002
).
7.
X.
Sun
,
S.
Vélez
,
A.
Atxabal
,
A.
Bedoya-Pinto
,
S.
Parui
,
X.
Zhu
,
R.
Llopis
,
F.
Casanova
, and
L. E.
Hueso
, “
A molecular spin-photovoltaic device
,”
Science
357
,
677
680
(
2017
).
8.
M. C.
Wheeler
,
F. A.
Ma’Mari
,
M.
Rogers
,
F. J.
Gonçalves
,
T.
Moorsom
,
A.
Brataas
,
R.
Stamps
,
M.
Ali
,
G.
Burnell
,
B. J.
Hickey
, and
O.
Cespedes
, “
Optical conversion of pure spin currents in hybrid molecular devices
,”
Nat. Commun.
8
,
926
(
2017
).
9.
B. R.
Gautam
,
T. D.
Nguyen
,
E.
Ehrenfreund
, and
Z. V.
Vardeny
, “
Magnetic field effect spectroscopy of C60-based films and devices
,”
J. Appl. Phys.
113
,
143102
(
2013
).
10.
H.
Liu
,
J.
Wang
,
M.
Groesbeck
,
X.
Pan
,
C.
Zhang
, and
Z. V.
Vardeny
, “
Studies of spin related processes in fullerene C60 devices
,”
J. Mater. Chem. C
6
,
3621
3627
(
2018
).
11.
S.
Liang
,
R.
Geng
,
B.
Yang
,
W.
Zhao
,
R.
Chandra Subedi
,
X.
Li
,
X.
Han
, and
T. D.
Nguyen
, “
Curvature-enhanced spin-orbit coupling and spinterface effect in fullerene-based spin valves
,”
Sci. Rep.
6
,
19461
(
2016
).
12.
M.
Gobbi
,
F.
Golmar
,
R.
Llopis
,
F.
Casanova
, and
L. E.
Hueso
, “
Room-Temperature spin transport in C60-based spin valves
,”
Adv. Mater.
23
,
1609
1613
(
2011
).
13.
D.
Sun
,
K. J.
van Schooten
,
M.
Kavand
,
H.
Malissa
,
C.
Zhang
,
M.
Groesbeck
,
C.
Boehme
, and
Z.
Valy Vardeny
, “
Inverse spin hall effect from pulsed spin current in organic semiconductors with tunable spin-orbit coupling
,”
Nat. Mater.
15
,
863
869
(
2016
).
14.
Q.
Zhang
,
L.
Yin
,
W.
Mi
, and
X.
Wang
, “
Large spatial spin polarization at benzene/La2/3Sr1/3MnO3 spinterface: Toward organic spintronic devices
,”
J. Phys. Chem. C
120
,
6156
6164
(
2016
).
15.
F.
Djeghloul
,
F.
Ibrahim
,
M.
Cantoni
,
M.
Bowen
,
L.
Joly
,
S.
Boukari
,
P.
Ohresser
,
F.
Bertran
,
P.
Le Fèvre
,
P.
Thakur
,
F.
Scheurer
,
T.
Miyamachi
,
R.
Mattana
,
P.
Seneor
,
A.
Jaafar
,
C.
Rinaldi
,
S.
Javaid
,
J.
Arabski
,
J. P.
Kappler
,
W.
Wulfhekel
,
N. B.
Brookes
,
R.
Bertacco
,
A.
Taleb-Ibrahimi
,
M.
Alouani
,
E.
Beaurepaire
, and
W.
Weber
, “
Direct observation of a highly spin-polarized organic spinterface at room temperature
,”
Sci. Rep.
3
,
1272
(
2013
).
16.
V. A.
Dediu
,
L. E.
Hueso
,
I.
Bergenti
, and
C.
Taliani
, “
Spin routes in organic semiconductors
,”
Nat. Mater.
8
,
707
(
2009
).
17.
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
, “
Unravelling the role of the interface for spin injection into organic semiconductors
,”
Nat. Phys.
6
,
615
620
(
2010
).
18.
M.
Gobbi
,
A.
Bedoya-Pinto
,
F.
Golmar
,
R.
Llopis
,
F.
Casanova
, and
L. E.
Hueso
, “
C60-based hot-electron magnetic tunnel transistor
,”
Appl. Phys. Lett.
101
,
102404
(
2012
).
19.
E.
Saitoh
,
M.
Ueda
,
H.
Miyajima
, and
G.
Tatara
, “
Conversion of spin current into charge current at room temperature: Inverse spin-Hall effect
,”
Appl. Phys. Lett.
88
,
182509
(
2006
).
20.
K.
Ando
,
S.
Watanabe
,
S.
Mooser
,
E.
Saitoh
, and
H.
Sirringhaus
, “
Solution-processed organic spin–charge converter
,”
Nat. Mater.
12
,
622
(
2013
).
21.
S.
Watanabe
,
K.
Ando
,
K.
Kang
,
S.
Mooser
,
Y.
Vaynzof
,
H.
Kurebayashi
,
E.
Saitoh
, and
H.
Sirringhaus
, “
Polaron spin current transport in organic semiconductors
,”
Nat. Phys.
10
,
308
313
(
2014
).
22.
H.
Liu
,
C.
Zhang
,
H.
Malissa
,
M.
Groesbeck
,
M.
Kavand
,
R.
McLaughlin
,
S.
Jamali
,
J.
Hao
,
D.
Sun
,
R. A.
Davidson
,
L.
Wojcik
,
J. S.
Miller
,
C.
Boehme
, and
Z. V.
Vardeny
, “
Organic-based magnon spintronics
,”
Nat. Mater.
17
,
308
312
(
2018
).
23.
G. A.
Steele
,
F.
Pei
,
E. A.
Laird
,
J. M.
Jol
,
H. B.
Meerwaldt
, and
L. P.
Kouwenhoven
, “
Large spin-orbit coupling in carbon nanotubes
,”
Nat. Commun.
4
,
1573
(
2013
).
24.
H.
Min
,
J. E.
Hill
,
N. A.
Sinitsyn
,
B. R.
Sahu
,
L.
Kleinman
, and
A. H.
MacDonald
, “
Intrinsic and Rashba spin-orbit interactions in graphene sheets
,”
Phys. Rev. B
74
,
165310
(
2006
).
25.
C.
Kittel
, “
On the theory of ferromagnetic resonance absorption
,”
Phys. Rev.
73
,
155
161
(
1948
).
26.
Z. G.
Yu
, “
Suppression of the Hanle effect in organic spintronic devices
,”
Phys. Rev. Lett.
111
,
016601
(
2013
).
27.
J. N.
Haddock
,
X.
Zhang
,
B.
Domercq
, and
B.
Kippelen
, “
Fullerene based n-type organic thin-film transistors
,”
Org. Electron.
6
,
182
187
(
2005
).
28.
K.
Shibata
,
Y.
Kubozono
,
T.
Kanbara
,
T.
Hosokawa
,
A.
Fujiwara
,
Y.
Ito
, and
H.
Shinoha
, “
Fabrication and characteristics of C84 fullerene field-effect transistors
,”
Appl. Phys. Lett.
84
,
2572
(
2004
).
29.
J. W.
Arbogast
and
C. S.
Foote
, “
Photophysical properties of C70
,”
J. Am. Chem. Soc.
113
,
8886
8889
(
1991
).
30.
D.
Huertas-Hernando
,
F.
Guinea
, and
A.
Brataas
, “
Spin-orbit coupling in curved graphene, fullerenes, nanotubes, and nanotube caps
,”
Phys. Rev. B
74
,
155426
(
2006
).