Letter
pubs.acs.org/NanoLett
Graphene As a Tunnel Barrier: Graphene-Based Magnetic Tunnel
Junctions
Enrique Cobas,*,† Adam L. Friedman,† Olaf M. J. van’t Erve, Jeremy T. Robinson, and Berend T. Jonker*
Naval Research Laboratory, Washington, DC 20375, United States
S Supporting Information
*
ABSTRACT: Graphene has been widely studied for its high
in-plane charge carrier mobility and long spin diffusion lengths.
In contrast, the out-of-plane charge and spin transport
behavior of this atomically thin material have not been well
addressed. We show here that while graphene exhibits metallic
conductivity in-plane, it serves effectively as an insulator for
transport perpendicular to the plane. We report fabrication of
tunnel junctions using single-layer graphene between two
ferromagnetic metal layers in a fully scalable photolithographic
process. The transport occurs by quantum tunneling
perpendicular to the graphene plane and preserves a net spin polarization of the current from the contact so that the
structures exhibit tunneling magnetoresistance to 425 K. These results demonstrate that graphene can function as an effective
tunnel barrier for both charge and spin-based devices and enable realization of more complex graphene-based devices for highly
functional nanoscale circuits, such as tunnel transistors, nonvolatile magnetic memory, and reprogrammable spin logic.
KEYWORDS: Graphene, tunnel barrier, spintronics, magnetic tunnel junction, magnetoresistance
E
lectrical transport in graphene has become one of the most
well-studied topics in materials science and condensed
matter physics since the first measurements were reported in
single-layer flakes.1 These studies have focused on graphene’s
extraordinary in-plane charge carrier mobility and long mean
free path,2,3 properties that suggest graphene may some day
replace indium tin oxide as a transparent conductor, metals as
chip interconnects, and serve as an alternate channel material in
complementary metal-oxide-semiconductor (CMOS) transistor
technology.4 The high mobility and low spin−orbit interaction
also make graphene an attractive medium for planar spin
transport,5,6 enabling realization of spin-based devices with new
performance and functionality.7−10 Several groups have
demonstrated graphene lateral spin-valve structures with long
spin lifetimes and diffusion lengths.5,6
In contrast, the out-of-plane charge and spin transport
behavior of this atomically thin material has not been well
addressed. Its parent compound, graphite, is known to have a
strong conductance anisotropy11 − the weak interlayer
coupling and wave function overlap produce relatively poor
conductivity perpendicular to the basal plane.12 Previous
studies of out-of-plane transport in graphene attributed their
data to space-charge limited effects,13 oxide layers that formed
on the metallic contacts,14 or to transport through defects14 or
graphene’s conductive edge states.15 The intrinsic out-of-plane
conductance has not been addressed to date. Spin transport of
hot electrons through 7−17 nm thick graphite flakes
perpendicular to the layer plane was recently demonstrated
using scanning tunneling microscopy based techniques.16
This article not subject to U.S. Copyright.
Published 2012 by the American Chemical
Society
The combination of excellent lateral transport and low outof-plane conductivity suggests that graphene could uniquely
serve as both a low loss medium for in-plane conduction as well
as a tunnel barrier for transport perpendicular to the plane,
providing a highly versatile single material platform for future
nanoscale devices. A step toward all-graphene circuits was
recently demonstrated by the fabrication of wafer-scale
inductor/transistor circuits monolithically integrated on a
single graphene/SiC wafer.17
Here we report the fabrication of tunnel junctions employing
single layer graphene as the tunnel barrier between two
ferromagnetic metal electrodes. We find that graphene serves
effectively as an insulator for transport perpendicular to the
plane; we show that the transport occurs by quantum tunneling
and preserves a net spin polarization of the current from the
contact so that the structures exhibit a tunneling magnetoresistance (TMR) to 425 K. Analysis of the bias and
temperature dependence further confirms that perpendicular
transport occurs by tunneling. These results demonstrate that
graphene functions effectively as a tunnel barrier, providing a
wide dynamic conductivity range for both charge and spinbased devices. Our results enable the realization of more
complex graphene-based devices for highly functional nanoscale
circuits, including tunnel transistors,18 nonvolatile memory,8
and reprogrammable logic based on spin tunnel junctions.9,10
Received: February 23, 2012
Revised: April 6, 2012
Published: May 11, 2012
3000
dx.doi.org/10.1021/nl3007616 | Nano Lett. 2012, 12, 3000−3004
Nano Letters
Letter
Tunnel barriers are the basis for many electronic and
spintronic device structures.7−10,18−20 Fabrication of ultrathin
and defect-free tunnel barriers is an ongoing challenge in
materials science. Typical tunnel barriers are based on metal
oxides (e.g., Al2O3 and MgO), and issues such as nonuniform
thicknesses, pinholes, defects and trapped charge compromise
their performance and reliability. Highly uniform single-atom
thick barriers like graphene provide the ultimate control over
the morphology of the barrier. In addition, graphene’s inert
chemical character minimizes interfacial reaction and interdiffusion, ensuring well-defined interfaces and robustness for
thermal processing, and preventing coupling through pinholes
in an oxide layer.14 Magnetic tunnel junctions (MTJs)
incorporate a tunnel barrier between two ferromagnetic metal
electrodes, enabling use of both charge and spin for information
storage and processing. They are currently used in hard drive
read heads and enable new emerging technologies including
magnetic random access memory and spin-transfer torque
devices.7−10,20 Theoretical studies of a graphite tunnel barrier
between two ferromagnetic metals have predicted a very large
magnetoresistance ratio for ideal, fully single crystal structures
with at least three layers of graphene due to spin filtering.21,22
However, such ideal structures are at present exceedingly
difficult to realize over a large scale, given the challenges of
producing defect-free multilayer graphene over even modest
lateral dimensions (∼100 um2), and of epitaxial growth of
suitable metals on graphene.
Our graphene was grown by chemical vapor deposition
(CVD) on copper foil23 and incorporated as the tunnel barrier
by physical transfer and standard lithographic processes. A
cross-sectional diagram and optical photographs of these
graphene-barrier MTJs are shown in Figure 1. The junction
stack structure is fabricated on a Si(100)/275 nm SiO2 wafer
and consists of 20 nm Ni0.9Fe0.1/graphene/20 nm Co/5 nm Ti/
50 nm Au. Two rings of insulation, one below and one above
the graphene mesa edge (8 nm SiN and 5 nm SiO2,
respectively), isolate the edges of the graphene from the
metal layers, preventing contact to conducting edge states.15
Reference samples omitting the graphene layer were fabricated
for comparison. The diameter of the junctions was varied
between 20 and 36 μm, much smaller than the typical grain size
of the CVD-graphene material used. This ensures a high
probability of obtaining continuous, single domain graphene
over the area of the tunnel junction, a critical consideration to
avoid conduction through defects14 or edge states.15 The crossbar geometry enables four-probe measurement of the local
junction resistance while avoiding other effects such as the
anisotropic magnetoresistance of the magnetic current leads.
Details of the fabrication procedure are found in the Methods
and Supporting Information.
Analysis of the current−voltage (I−V) characteristics as a
function of temperature confirmed that the electrical transport
across the graphene layer occurs by tunneling. The I−V curves
(Figure 2a) are nonlinear and symmetric, as expected for a
metal/insulator/metal tunnel junction.18 The zero bias
resistance (ZBR) of the tunnel barrier contact, defined as
R(T)/R(300 K), exhibits the modest temperature dependence
shown in Figure 2b. This has been shown to be a rigorous and
definitive indicator of tunneling through a pinhole free barrier
and more reliable than simple fits to the Brinkman−Dynes−
Rowell model or application of the usual Rowell criteria.24 Our
transport measurements thus provide evidence for this
prediction. Reference samples without the graphene layer
Figure 1. Graphene tunnel junction devices. (a) conceptual diagram of
the FM/graphene/FM junction, (b) cross-sectional diagram and
optical image of the junction area prior to top contact deposition, and
(c) photo of a completed four-probe device.
exhibited ohmic characteristics, confirming that any oxidation
of the bottom (Ni0.9Fe0.1) contact that may have occurred
during fabrication did not produce a tunnel barrier.
The spin-polarized tunneling process in MTJs depends upon
the spin-polarized density of states of the s- and d-orbital
electrons at the ferromagnetic metal/insulator interface.25 Spin
information is conserved in the single-step tunnel process, and
one can describe the transport as having two independent spin
channels. A low resistance state is observed for parallel
alignment of the two FMs (RP), when electrons with majority
spin in FM1 tunnel to the empty majority states in FM2 (Figure
3a). In the antiparallel alignment, to conserve spin the majority
spins in FM1 now tunnel from a large density of states to empty
minority states with a much lower density (Figure 3b). The
corresponding tunnel probability is low, and a high resistance
state (RAP) is observed. The associated tunneling magnetoresistance (TMR) ratio is defined as (RAP − RP)/RP.
The magnetoresistance data for a representative Ni0.9Fe0.1/
graphene/Co tunnel junction is shown in Figure 4. When a
magnetic field is applied in-plane, the magnetizations of the
NiFe and Co electrodes reverse at fields corresponding to their
respective coercivities with the NiFe switching at a much lower
field than the Co. Their magnetizations can thus be aligned
either parallel or antiparallel, and two distinct resistance states
are observed in the data, as described above. The TMR in the
3001
dx.doi.org/10.1021/nl3007616 | Nano Lett. 2012, 12, 3000−3004
Nano Letters
Letter
Figure 2. Tunnel junction transport characteristics. (a) Typical
current−voltage measurements of a graphene tunnel junction for
various temperatures. The curves are nonlinear and symmetric. (b)
Zero bias resistance (ZBR) vs temperature for four graphene tunnel
junction devices. The ZBR exhibits a modest temperature dependence,
confirming tunneling through pinhole free barriers.
Figure 3. Simplified spin-dependent density of states in the FM
electrodes for a FM/graphene/FM tunnel junction. (a) parallel
alignment, (b) antiparallel alignment.
significantly decrease the tunneling spin polarization P.26 Air
exposure of NiFe in particular produces antiferromagnetic
NiO,27 and the presence of such a material is known to produce
strong spin-scattering that reduces the tunneling spin polarization P and the TMR ratio. Future refinements to the
fabrication process may eliminate such interface contamination,
maximizing the TMR effect. For example, growth of multilayer
graphene directly on Ni surfaces has been demonstrated,28
which minimizes oxidation of the Ni surface even upon
exposure to atmosphere.29 However, the resultant graphene is
nonuniform, and the very small lateral dimensions of the
uniform regions (<100 um2) preclude device fabrication over
large areas. Future well-controlled growth of large area
multilayer graphene, together with graphene’s chemical inertness, will prevent oxidation as well as adsorption of
contaminants in subsequent processing.
The magnitude of the TMR decreases in a monotonic,
nonlinear fashion with increasing bias, as shown in Figure 4b,
where the peak value of the TMR is plotted as a function of
bias. This behavior is typical for MTJs and is a key signature
because it is caused by the basic physics of the spin tunneling
process. At a finite bias, the electrons tunnel into empty states
of the receiving electrode with an excess energy, generating
phonons and magnons that increase the spin relaxation rate.30
The receiving empty states are hot electron states for which the
spin polarization is significantly reduced.31 These two effects
lead to the reduction of TMR with bias,30,31 and observation of
this characteristic excludes extrinsic contributions such as
anisotropic magnetoresistance that increases with bias. Note
that the bias dependence is asymmetric, as expected, reflecting
the different density-of-states for the two different ferromag-
graphene MTJs reaches 2% at the lowest temperatures and
biases measured (Figure 4a), while reference junctions without
graphene exhibit no magnetoresistance. For comparison, theory
predicts a value of 25% for a perfectly ordered single crystal
Ni(111)/graphene/Ni(111) junction with a single graphene
layer21 (note that we have converted the MR values from the
“pessimistic” definition used in ref 21 to the one used here).
This value drops to 10% if the layer of Ni atoms adjacent to the
graphene monolayer at just one of the Ni/graphene interfaces
contains disorder or roughness (with the rest of the structure
and the other Ni/graphene interface assumed to be perfectly
ordered). The interfaces in our polycrystalline Ni0.9Fe0.1/
graphene/Co samples are likely to exhibit a (111) texture
with disorder or roughness at both graphene/metal interfaces.
Thus the 2% TMR value we measure experimentally is
comparable to the theoretical value.
A simple relation between the TMR and the polarization of
the electrodes for an ideal junction is given by TMR = 2P1P2/(1
− P1P2), where P1 and P2 are the tunnel spin polarizations of
the FM1/insulator and FM2/insulator interfaces, respectively.25
Using this model, a TMR of 2% corresponds to a tunnel spin
polarization of P1 ∼ P2 = P ∼ 0.1. For comparison, the NiFe/
Al2O3 interface typically exhibits P ∼ 0.3. We attribute the
lower value obtained for the graphene MTJs to air exposure of
our NiFe surface prior to application of the graphene layer,
allowing contaminants to adsorb on the surface as well as
oxidation to occur. TMR heterostructures are typically grown
completely in situ with no air exposure of the interfaces,
because contaminants at the FM/tunnel barrier interface
3002
dx.doi.org/10.1021/nl3007616 | Nano Lett. 2012, 12, 3000−3004
Nano Letters
Letter
Figure 4. TMR as a function of applied bias. (a) TMR curves for low
bias (125 and 250 uV) at T = 4 K. The arrows indicate the relative
orientation of the NiFe (lower arrow) and Co contacts. (b) TMR
magnitude versus applied bias for higher biases at T = 20 K. The TMR
decreases with bias and is asymmetric, reflecting the two different FM
metals used.
Figure 5. Temperature dependence of the TMR magnitude. (a) TMR
curves for selected temperatures to 400 K. (b) Temperature
dependence of the TMR magnitude for four graphene tunnel
junctions compared with the model by Shang et al. (ref 32) for
selected values of the fitting parameter α. For NiFe, α ∼ 5 × 10−5
K−3/2.
netic metals used and providing further support for spinpolarized tunneling.
All devices that exhibited TMR at low temperature retained
an easily measurable TMR well above 300 K (Figure 5a); the
TMR signature is clearly visible with good signal-to-noise in the
400 K curve. The shift of the peaks with temperature is due to
the temperature dependence of the magnetic coercivity of Co
(see Supporting Information). The decrease of TMR with
increasing temperature is typical for MTJs, and attributed to the
thermal excitation of spin waves in the FM material which
decreases the tunnel spin polarization P.32 This is a characteristic of the ferromagnetic metal. The temperature dependence
was compared to the model of Shang et al.32 (Figure 5b), where
the tunnel spin polarization is given by the Bloch law, P(T) =
P0(1 − αT3/2). Substituting P1 and P2 in Julliere’s model25
described above with P(T), we obtain excellent fits to the data
using an α ranging from 5 × 10−5 to 1.1 × 10−4 K−3/2 . The
lower value is comparable to αNiFe in ref 32, and higher values
needed to fit a weaker temperature dependence are an
indication of contamination at the interface.32 The fact that
our data are well fit by this model over a wide temperature
range (5−425 K) provides further strong evidence for spin
polarized tunneling.
In summary, we have shown that while graphene exhibits
metallic conductivity in-plane, it serves effectively as an
insulator for transport perpendicular to the plane. Our results
demonstrate the feasibility of spin-dependent tunneling
employing inexpensive single-layer materials, in this case
graphene, as tunnel barriers. The TMR effect is easily
measurable with good signal-to-noise ratio up to 425 K in
our graphene MTJ structures despite air exposure of the
bottom FM metal electrode. We anticipate that much higher
TMR ratios can by obtained through fabrication improvements
resulting in cleaner interfaces, incorporation of additional
graphene layers to enhance spin filtering21 or chemically
functionalized graphene, or use of other monolayer materials
such as hexagonal boron-nitride. Our results and future
improvements in fabrication will enable development of new
graphene-based nanoscale charge- and spin-based devices.
Methods. The bottom electrode and bottom insulator ring
were patterned using standard lift-off photolithography using a
lift-off resist (LOR) layer. The metal and insulator consist of 20
nm NiFe and 8 nm SiN deposited via DC and RF sputtering
respectively. The more conformal nature of sputter deposition
(compared with evaporation) avoids sharp edges that can later
damage the graphene during transfer and also insulates the sideedge of the bottom electrode. Graphene was synthesized via
low-pressure chemical vapor deposition on a copper substrate
using methane gas.23 Poly(methyl methacrylate) (PMMA) was
used to support the graphene while the copper substrate was
dissolved in Transene APS 100 etchant. The resulting
graphene-PMMA film was transferred onto the prefabricated
bottom electrode and insulator ring, after which the PMMA
was removed in acetone.
To avoid photoresist contamination of the graphene surface,
all subsequent lithography steps used a PMMA (50 nm) as a
3003
dx.doi.org/10.1021/nl3007616 | Nano Lett. 2012, 12, 3000−3004
Nano Letters
Letter
(13) Lee, Y.-H.; Kim, Y.-J.; Lee, J. H. Vertical conduction behavior
through atomic graphene device under transverse electric field. Appl.
Phys. Lett. 2011, 98, 133112.
(14) Mohiuddin, T. M. G.; Hill, E.; Elias, D.; Zhukov, A.; Novoselov,
K.; Geim, A. Graphene in multilayers CPP spin valves. IEEE Trans.
Magn. 2008, 44, 2624−2627.
(15) Acik, M.; Chabal, Y. J. Nature of Graphene Edges: A Review.
Jpn. J. Appl. Phys. 2011, 50, 070101.
(16) Banerjee, T.; van der Wiel, W. G.; Jansen, R. Spin injection and
perpendicular spin transport in graphite nanostructures. Phys. Rev. B
2010, 81, 214409.
(17) Lin, Y.-M.; Valdes-Garcia, A.; Han, S.-J.; Farmer, D. B.; Meric, I.;
Sun, Y.; Wu, Y.; Dimitrakopoulos, C.; Grill, A.; Avouris, P.; Jenkins, K.
A. Wafer scale graphene integrated circuit. Science 2011, 332, 1294.
(18) Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices; John
Wiley & Sons Inc.: Hoboken, New Jersey, 2007.
(19) Bibes, M.; Villegas, J. E.; Barthelemy, A. Ultrathin oxide films
and interfaces for electronics and spintronics. Adv. Phys. 2011, 60, 5−
84.
(20) Deac, A. M.; Fukushima, A.; Kubota, H.; Maehara, H.; Suzuki,
Y.; Yuasa, S.; Nagamine, Y.; Tsunekawa, K.; Djayaprawira, D. D.;
Watanabe, N. Bias-driven high-power microwave emission from MgObased tunnel magnetoresistance devices. Nat. Phys. 2008, 4, 803−809.
(21) Karpan, V. M.; Giovanetti, G.; Khomyakov, P. A.; Talanana, M.;
Starikov, A. A.; Zwierzycki, M.; van den Brink, J.; Brocks, G.; Kelly, P.
J. Graphite and Graphene as Perfect Spin Filters. Phys. Rev. Lett. 2007,
99, 176602;(a) Phys. Rev. B 2008, 78, 195419.
(22) Yazyev, O. V.; Pasquarello, A. Magnetoresistive junctions based
on epitaxial graphene and hexagonal boron nitride. Phys. Rev. B 2009,
80, 035408.
(23) Li, X.; Magnuson, C. W.; Venugopal, A.; Tromp, R. M.;
Hannon, J. B.; Vogel, E. M.; Colombo, L.; Ruoff, R. S. Large-Area
Graphene Single Crystals Grown by Low-Pressure Chemical Vapor
Deposition of Methane on Copper. J. Am. Chem. Soc. 2011, 133,
2816−2819.
(24) Jonsson-Akerman, B. J.; Escudero, R.; Leighton, C.; Kim, S.;
Schuller, I. K.; Rabson, D. A. Reliability of normal-state current-voltage
characteristics as an indicator of tunnel-junction barrier quality. Appl.
Phys. Lett. 2000, 77, 1870−1872.
(25) Julliere, M. Tunneling between ferromagnetic films. Phys. Lett.
1975, 54a, 225.
(26) LeClair, P.; Swagten, H. J. M.; Kohlhepp, J. T.; de Jonge, W. J.
M. Apparent spin polarization decay in Cu-dusted Co/Al2O3/Co
tunnel Junctions. Phys. Rev. Lett. 2000, 84, 2933.
(27) Fitzsimmons, M. R.; Silva, T. J.; Crawford, T. M. Surface
oxidation of Permalloy thin films. Phys. Rev. B 2006, 73, 014420.
(28) Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, K. S.;
Ahn, J. H.; Kim, P.; Choi, J. Y.; Hong, B. H. Large-scale pattern growth
of graphene films for stretchable transparent electrodes. Nature 2009,
457, 706−710.
(29) Chen, S.; Brown, L.; Levendorf, M.; Cai, W.; Ju, S.-Y.;
Edgeworth, J.; Li, X.; Magnuson, C. W.; Velamakanni, A.; Piner, R. D.;
Kang, J.; Park, J.; Ruoff, R. S. Oxidation Resistance of GrapheneCoated C and Cu/Ni Alloy. ACS Nano 2011, 5, 1321−1327.
(30) Zhang, S.; Levy, P. M.; Marley, A. C.; Parkin, S. S. P. Quenching
of Magnetoresistance by Hot Electrons in Magnetic Tunnel Junctions.
Phys. Rev. Lett. 1997, 79, 3744.
(31) Valenzuela, S. O.; Monsma, D. J.; Marcus, C. M.;
Narayanamurti, V.; Tinkham, M. Spin polarized tunneling at finite
bias. Phys. Rev. Lett. 2005, 94, 196601.
(32) Shang, C. H.; Nowak, J.; Jansen, R.; Moodera, J. S. Phys. Rev. B
1998, 58, R2917.
deep UV photoresist. An oxygen reactive ion etch was used to
remove all graphene not over the designated junctions. The top
layers consisted of a 5 nm SiO2 ring, and top electrode of 20
nm Co, 5 nm Ti, and 50 nm Au all deposited by electron beam
evaporation. The resistance-area product of the junctions is
between 35 and 75 kΩ μm2.
■
ASSOCIATED CONTENT
S Supporting Information
*
Description of the sample fabrication and measurements. This
material is available free of charge via the Internet at http://
pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: (E.C.) enrique.cobas@nrl.navy.mil; (B.T.J.) jonker@
nrl.navy.mil
Notes
The authors declare no competing financial interests.
†
NRL Karle Fellows.
■
ACKNOWLEDGMENTS
This work was supported by core programs at NRL and the
Office of Naval Research. E.C. and A.F. gratefully acknowledge
support through the NRL Karles Fellow program. The authors
gratefully acknowledge use of facilities in the NRL Nanoscience
Institute. E.C., A.L.F. and B.T.J. conceived the experiments.
J.T.R. grew the CVD graphene and transferred layers to the
device structures. E.C. fabricated the tunnel barrier structures.
E.C. and O.M.J.v.E. acquired and analyzed the transport data.
All authors provided insight and expertise in interpretation of
the data and in writing the manuscript.
■
REFERENCES
(1) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang,
Y.; Dubonos, S. V.; Griorieva, I. V.; Firsov, A. A. Electric Field Effect in
Atomically Thin Carbon Films. Science 2004, 306 (5696), 666−669.
(2) Zhang, Y.; Tan, Y.-W.; Stormer, H. L.; Kim, P. Experimental
Observations of the Quantum Hall Effect and Berry’s Phase in
Graphene. Nature 2005, 438, 201−204.
(3) Castro Neto, A. H.; Guinea, F.; Novoselov, K. S.; Geim, A. K.
The electronic properties of graphene. Rev. Mod. Phys. 2009, 81 (1),
109−162.
(4) International Technology Roadmap for Semiconductors, www.
itrs.org (2009).
(5) Tombros, N.; Jozsa, C.; Popincuic, M.; Jonkman, H. T.; van
Wees, B. J. Electronic spin transport and spin precession in single
graphene layers at room temperature. Nature 2007, 448, 571−575.
(6) Han, W.; Kawakami, R. Spin Relaxation in Single-Layer and
Bilayer Graphene. Phys. Rev. Lett. 2011, 107, 047207.
(7) Fert, A. Nobel Lecture: Origin, development and future of
spintronics. Rev. Mod. Phys. 2008, 80, 1517−1530.
(8) Chappert, C.; Fert, A.; Nguyen van Dau, F. The emergence of
spin electronics in data storage. Nat. Mater. 2007, 6, 813−823 and
references therein..
(9) Dery, H.; Dalal, P.; Cywiński, L.; Sham, L. J. Spin-based logic in
semiconductors for reconfigurable large-scale circuits. Nature 2007,
447, 573 DOI: .
(10) Behin-Aein, B.; Datta, D.; Salahuddin, S.; Datta, S. Proposal for
an all-spin logic device with built-in memory. Nat. Nanotechnol. 2010,
5, 266−270.
(11) Krishnan, K. S.; Ganguli, N. Large anisotropy of the electrical
conductivity of graphite. Nature 1939, 144, 667−667.
(12) Wallace, P. R. The Band Theory of Graphite. Phys. Rev. 1947, 71
(9), 622−634.
3004
dx.doi.org/10.1021/nl3007616 | Nano Lett. 2012, 12, 3000−3004