Dominik Suszalski and Adam RycerzAdiabatic quantum pumping in buckled graphene nanoribbon driven by...
Transcript of Dominik Suszalski and Adam RycerzAdiabatic quantum pumping in buckled graphene nanoribbon driven by...
Adiabatic quantum pumping in buckled graphenenanoribbon driven by a kink ∗
Dominik Suszalski and Adam Rycerz
Institute of Theoretical Physics, Jagiellonian University, ul. S. Łojasiewicza 11,PL–30348 Kraków, Poland
(Received February 18, 2020)
We propose a new type of quantum pump in buckled graphene nanorib-bon, adiabatically driven by a kink moving along the ribbon. From a prac-tical point of view, pumps with moving scatterers present advantages ascompared to gate-driven pumps, like enhanced charge transfer per cycleper channel. The kink geometry is simplified by truncating the spatial ar-rangement of carbon atoms with the classical φ4 model solution, includinga width renormalization following from the Su-Schrieffer-Heeger model forcarbon nanostructures. We demonstrate numerically, as a proof of concept,that for moderate deformations a stationary kink at the ribbon center mayblock the current driven by the external voltage bias. In the absence ofa bias, a moving kink leads to highly-effective pump effect, with a chargeper unit cycle dependent on the gate voltage.
1. Introduction
The idea of quantum pumping, i.e., transferring the charge between elec-tronic reservoirs by periodic modulation of the device connecting these reser-voirs [1], has been widely discussed in the context of graphene nanostructures[2, 3, 4, 5, 6, 7, 8]. Since early works, elaborating the gate-driven pumpingmechanism in graphene [2] and bilayer graphene [3], it becomes clear thatthe transport via evanescent modes may significantly enhance the effective-ness of graphene-based pumps compared to other quantum pumps. Otherpumping mechanisms considered involves laser irradiation [4], strain fields[5], tunable magnetic barriers [6], Landau quantization [7], or even slidingthe Moiré pattern in twisted bilayer graphene [8].
It is known that quantum pump with a shifting scatterer may showenhanced charge transfer per cycle in comparison to a standard gate-driven
∗ Presented at 45. Zjazd Fizyków Polskich, Kraków 13–18 września 2019.
(1)
arX
iv:2
002.
0850
7v1
[co
nd-m
at.m
es-h
all]
20
Feb
2020
2 werekink printed on February 21, 2020
µ0<latexit sha1_base64="qQzlwg91uy8qjKDB+2WNIz8w8OI=">AAAB4nicbVBdSwJBFL1rX2ZfVo+9DEnQk+yaUI9SLz0atCqoyOzsrA7Ozi4zdwMR/0AvEb0U9IP6C/2bRt0XtQMDh3POcO+5QSqFQdf9dQpb2zu7e8X90sHh0fFJ+fSsZZJMM+6zRCa6E1DDpVDcR4GSd1LNaRxI3g7GD3O//cK1EYl6xknK+zEdKhEJRtFKfi/OBu6gXHGr7gJkk3g5qUCO5qD80wsTlsVcIZPUmK7nptifUo2CST4r9TLDU8rGdMinixVn5MpKIYkSbZ9CslBXcjQ2ZhIHNhlTHJl1by7+53UzjO76U6HSDLliy0FRJgkmZN6XhEJzhnJiCWVa2A0JG1FNGdqrlGx1b73oJmnVqt5NtfZUrzTu8yMU4QIu4Ro8uIUGPEITfGAg4A0+4csJnVfn3flYRgtO/uccVuB8/wFrtYoo</latexit>
(a)
(c)
-3
-2
-1
0
1
2
3
-1 -0.5 0 0.5 1
Ener
gy,
E/t 0
Wavenumber, kya/ π
W = 10 a<latexit sha1_base64="jUHPl8b3eQhvIWJG/wer59K+leE=">AAAB5nicbVDLSgMxFL1TX7W+qi7dRIvgQspMFXQjFNy4rGAf0iklk2ba0MxkSO4IpfQX3Ii4UfBz/AX/xrSdTVsPBA7nnHDvuUEihUHX/XVya+sbm1v57cLO7t7+QfHwqGFUqhmvMyWVbgXUcCliXkeBkrcSzWkUSN4MhvdTv/nCtREqfsJRwjsR7cciFIyilZ6b/umd5/qXtFssuWV3BrJKvIyUIEOtW/zxe4qlEY+RSWpM23MT7IypRsEknxT81PCEsiHt8/FszQk5t1KPhErbFyOZqQs5GhkzigKbjCgOzLI3Ff/z2imGt52xiJMUeczmg8JUElRk2pn0hOYM5cgSyrSwGxI2oJoytJcp2OrectFV0qiUvaty5fG6VK1kR8jDCZzBBXhwA1V4gBrUgUEEb/AJX87AeXXenY95NOdkf45hAc73H9Gfis4=</latexit>
Ac<latexit sha1_base64="/8Irsrs+QOKCVGS+qDjKjeID1V8=">AAAB53icdVDLSgMxFL3js9ZX1aWbYBFclaSKbXdVNy4r2Ae2pWTSTBuayQxJRiil3+BGxI2Cf+Mv+Ddm2rqo6IHA4ZwT7j3Xj6UwFuMvb2V1bX1jM7OV3d7Z3dvPHRw2TJRoxusskpFu+dRwKRSvW2Elb8Wa09CXvOmPblK/+ci1EZG6t+OYd0M6UCIQjFonPUw6jEp0Ne2xXi6PCxhjQghKCSldYkcqlXKRlBFJLYc8LFDr5T47/YglIVeWSWpMm+DYdidUW8Ekn2Y7ieExZSM64JPZnlN06qQ+CiLtnrJopi7laGjMOPRdMqR2aH57qfiX105sUO5OhIoTyxWbDwoSiWyE0tKoLzRnVo4doUwLtyFiQ6ops+40WVf9px/6nzSKBXJeKN5d5KvXiyNk4BhO4AwIlKAKt1CDOjBQ8Axv8O4J78l78V7n0RVv8ecIluB9fAM4E4xq</latexit>
Re t
Im t
W<latexit sha1_base64="ORrK8ijuqLnqlZmWy3Q9jN82MYI=">AAAB3nicbVDLSgNBEOz1GeMr6tHLYBA8hd0o6DHoxWMC5gFJiLOT3mTI7IOZXiGEXL2IeFHwk/wF/8ZJspckFgwUVTV0V/uJkoZc99fZ2Nza3tnN7eX3Dw6Pjgsnpw0Tp1pgXcQq1i2fG1QywjpJUthKNPLQV9j0Rw8zv/mC2sg4eqJxgt2QDyIZSMHJSrVmr1B0S+4cbJ14GSlChmqv8NPpxyINMSKhuDFtz02oO+GapFA4zXdSgwkXIz7AyXy9Kbu0Up8FsbYvIjZXl3I8NGYc+jYZchqaVW8m/ue1UwruuhMZJSlhJBaDglQxitmsK+tLjYLU2BIutLQbMjHkmguyF8nb6t5q0XXSKJe861K5dlOs3GdHyME5XMAVeHALFXiEKtRBAMIbfMKX8+y8Ou/OxyK64WR/zmAJzvcfq9iIig==</latexit>
(b)
x
y
W 0<latexit sha1_base64="vcn+2p5ZXbXjfKOzLn8EV+jx9jE=">AAAB33icdVDLSgMxFL3xWeur6tJNsIiuhpnp0NZd0Y3LKvYBbSmZNNOGZh4kGaGUrt2IuFHwj/wF/8ZMWxcVPRA4nHPCvef6ieBK2/YXWlvf2Nzazu3kd/f2Dw4LR8dNFaeSsgaNRSzbPlFM8Ig1NNeCtRPJSOgL1vLHN5nfemRS8Th60JOE9UIyjHjAKdFGum9d9AtF27qqll2vjG3LtiuO62TErXglDztGyVCEJer9wmd3ENM0ZJGmgijVcexE96ZEak4Fm+W7qWIJoWMyZNP5fjN8bqQBDmJpXqTxXF3JkVCpSeibZEj0SP32MvEvr5PqoNqb8ihJNYvoYlCQCqxjnJXFAy4Z1WJiCKGSmw0xHRFJqDYnyZvqP/3w/6TpWk7Jcu+8Yu16eYQcnMIZXIIDFajBLdShARQCeIY3eEcEPaEX9LqIrqHlnxNYAfr4BmgpiP0=</latexit>
H<latexit sha1_base64="Gzt46qHE8//w1u4tkBoisnY8cPg=">AAAB3nicdVBLS0JBGP2uvcxet1q2GZKglcxYpO6kNi4V8gEqNnecq4NzH8zMDUTctoloU9BP6i/0b5qrtjDqwMDhnDN83/m8WAptMP5yMhubW9s72d3c3v7B4ZF7fNLSUaIYb7JIRqrjUc2lCHnTCCN5J1acBp7kbW9yl/rtR660iMJ7M415P6CjUPiCUWOlRm3g5nEBY0wIQSkhpRtsSaVSLpIyIqllkYcV6gP3szeMWBLw0DBJte4SHJv+jCojmOTzXC/RPKZsQkd8tlhvji6sNER+pOwLDVqoazkaaD0NPJsMqBnr314q/uV1E+OX+zMRxonhIVsO8hOJTITSrmgoFGdGTi2hTAm7IWJjqigz9iI5W/2nH/qftIoFclUoNq7z1dvVEbJwBudwCQRKUIUa1KEJDDg8wxu8Ow/Ok/PivC6jGWf15xTW4Hx8A+CMiLE=</latexit>
L<latexit sha1_base64="+0MvrjSaRWrF2qG9e/mkHrwhF64=">AAAB3nicbVDLSgNBEOyNrxhfUY9eBoPgKexGQY9BLx48JGAekIQ4O+lNhsw+mOkVQsjVi4gXBT/JX/BvnDwuSSwYKKpq6K72EyUNue6vk9nY3Nreye7m9vYPDo/yxyd1E6daYE3EKtZNnxtUMsIaSVLYTDTy0FfY8If3U7/xgtrIOHqiUYKdkPcjGUjByUrVx26+4BbdGdg68RakAAtUuvmfdi8WaYgRCcWNaXluQp0x1ySFwkmunRpMuBjyPo5n603YhZV6LIi1fRGxmbqU46Exo9C3yZDTwKx6U/E/r5VScNsZyyhJCSMxHxSkilHMpl1ZT2oUpEaWcKGl3ZCJAddckL1Izlb3Vouuk3qp6F0VS9XrQvlucYQsnME5XIIHN1CGB6hADQQgvMEnfDnPzqvz7nzMoxln8ecUluB8/wGbmoh/</latexit>
Wle
ad
<latexit sha1_base64="Tvf+WfvVFx0BOs+Cxuz0TXbbqXM=">AAAB6XicdVDLSgMxFL1TX7W+qi7dBIvgapipBV0W3bisYB/QKSWTudOGZh4kGaEM/Qg3Im4U/Bl/wb8x09ZFfRwIOZxzQu65fiq40o7zaZXW1jc2t8rblZ3dvf2D6uFRRyWZZNhmiUhkz6cKBY+xrbkW2Esl0sgX2PUnN4XffUCpeBLf62mKg4iOYh5yRrWRvO4w92REBNJgNqzWHLvhFCC/iWvPb6cGS7SG1Q8vSFgWYayZoEr1XSfVg5xKzZnAWcXLFKaUTegI8/mkM3JmpICEiTQn1mSuruRopNQ08k0yonqsfnqF+JfXz3R4Nch5nGYaY7b4KMwE0QkpapOAS2RaTA2hTHIzIWFjKinTZjkVU/27H/mfdOq2e2HX7xq15vVyCWU4gVM4BxcuoQm30II2MEjhCV7hzZpYj9az9bKIlqzlm2NYgfX+Bb5njUw=</latexit>
x
U(x)
U∞
U0
EF
Fig. 1. Buckled graphene nanoribbon as a quantum pump. (a) Typical pumpingcycles (schematic) for a single channel characterized by the complex transmissionamplitude, t = Re t + i Im t. Blue solid line enclosing the area Ac shows one-parameter cycle allowed for a standard gate-driven pump, red dashed line corre-sponds to the cycle involving a shift of a scatterer. (b) Flat ribbon with armchairedges attached to heavily-doped graphene leads (shaded areas). The ribbon widthW = 5 a (with a = 0.246 nm the lattice spacing) and length L = 11.5
√3 a are
chosen for illustrative purposes only. A cross section of buckled ribbon, charac-terized by the reduced width W ′ < W and the buckle height H > 0, and theelectrostatic potential energy profile U(x), are also shown. (c) Band structure ofthe infinite metallic-armchair nanoribbon of W = 10 a width, same as used in thecomputations.
pump [1], see also Fig. 1(a)1. Motivated by this conjecture we consider
1 In the simplest case of a one-parameter driven, one-channel pump, the charge percycle is given by Q = (e/π)Ac, with Ac being the ±area enclosed by the contour inthe (Re t, Im t) plane (where t is a parameter-dependent transmission amplitude).
werekink printed on February 21, 2020 3
a buckled nanoribbon, see Fig. 1(b), similar to the one studied numericallyby Yamaletdinov et. al. [9, 10] as a physical realization of the classical φ4
model and its topological solutions (kinks) connecting two distinct groundstates. Our setup is supplemented with two heavily-doped graphene leads,attached to the clamped edges of a ribbon, allowing to pass electric currentalong the system.
2. Model and methods
The analysis starts from the Su-Schrieffer-Heeger (SSH) model for theribbon, including the hopping-matrix elements corresponding to the nearest-neighbor bonds on a honeycomb lattice [11, 12, 13, 14]
Hribbon = −t0∑〈ij〉,s
exp
(−β δdij
d0
)(c†i,scj,s + c†j,sci,s
)+
1
2K∑〈ij〉
(δdij)2 , (1)
with a constrain∑〈ij〉 δdij = 0, where δdij is the change in bond length,
and d0 = a/√
3 is the equilibrium bond length defined via the lattice spacinga = 0.246 nm. The equilibrium hopping integral t0 = 2.7 eV, β = 3 is thedimensionless electron-phonon coupling, and K ≈ 5000 eV/nm2 is the springconstant for a C-C bond. The operator c†i,s (or ci,s) creates (or annihilates)a π electron at the i-th lattice site with spin s.
In order to determine the spatial arrangement of carbon atoms in a buck-led nanoribbon, {Rj = (xj , yj , zj)}, we first took the {xj} and {yj} coordi-nates same as for a flat honeycomb lattice in the equilibrium and set {zj}according to
z = H tanh
(y − y0
λ
)sin2
(πxW
), (2)
representing a topologically non-trivial solution of the φ4 model, with Hthe buckle height, W the ribbon width, y0 and λ the kink position and size(respectively). Next, x-coordinates are rescaled according to xj → xjW
′/W ,withW ′ adjusted to satisfy
∑〈ij〉 δdij = 0 in the y0 → ±∞ (“no kink”) limit.
In particular, H = 2a and W = 10a corresponds to W ′/W = 0.893. Wefurther fixed the kink size at λ = 3a (closely resembling the kink profileobtained from the molecular dynamics in Ref. [9]); this results in relativebond distorsions not exceeding |δdij |/d0 6 0.08. Full optimization of thebond lengths in the SSH Hamiltonian (1), which may lead to the alternatingbond pattern [14], is to be discussed elsewhere.
Heavily-doped graphene leads, x < 0 or x > W ′ in Fig. 1(b), are modeledas flat regions (dij = d0) with the electrostatic potential energy U∞ =−0.5t0 (compared to U0 = 0 in the ribbon), corresponding to 13 propagating
4 werekink printed on February 21, 2020
modes for W∞ = 20√
3a and the chemical potential µ0 = EF − U0 = 0.The scattering problem if solved numerically, for each value of the chemicalpotential µ0 and the kink position y0, using the Kwant package [15] allowingto determine the scattering matrix
S(µ0, y0) =
(r t′
t r′
), (3)
which contains the transmission t (t′) and reflection r (r′) amplitudes forcharge carriers incident from the left (right) lead, respectively.
The linear-response conductance can be determined from the S-matrixvia the Landuer-Büttiker formula [1], namely
G = G0Tr tt† =2e2
h
∑n
Tn, (4)
where G0 = 2e2/h is the conductance quantum and Tn is the transmissionprobability for the n-th normal mode. Similarly, in the absence of a volt-age bias, the charge transferred between the leads upon varying solely theparameter y0 is given by
∆Q = − ie2π
∑j
∫dy0
(∂S
∂y0S†)
jj
, (5)
where the summation runs over the modes in a selected lead. Additionaly,the integration in Eq. (5) is performed for a truncated range of −Λ 6 y0 6L+ Λ, with Λ = 250 a� λ for L = 75.5
√3 a.
3. Results and discussion
In Fig. 1(c) we depict the band structure for an infinite (and flat) metallic-armchair ribbon of W = 10 a width. It is remarkable that the second andthird lowest-lying subband above the charge-neutrality point (as well as thecorresponding highest subbands below this point) show an almost perfect de-generacy near their minima corresponding to E(2,3)
min ≈ 0.25 t0, which can beattributed to the presence of two valleys in graphene. For higher subbands,the degeneracy splitting due to the trigonal warping is better pronounced.
The approximate subband degeneracy has a consequence for the conduc-tance spectra of a finite ribbon, presented in Fig. 2: For both the flat ribbon(H = 0) and a buckled ribbon with no kink (H = 2a, y0 → − ∞) theconductance G ≈ G0 for |µ0| < E
(2,3)min and quickly rises to G/G0 ≈ 2.5−3
for µ > E(2,3)min . For µ < 0 the quantization is not so apparent, partly
werekink printed on February 21, 2020 5
0.0 0.2 0.4
0.0
1.0
2.0
3.0C
ondu
ctan
ce [
2e2 /
h]
0.00 0.04 0.08 0.12 0.160.0
0.4
0.8
1.2
flat ribbon, H= 0kink at the center, H= 2ano kink, H= 2a
Chemical potential, μ0/t0
Con
duct
ance
[2e
2 /h]
0.4 0.2
Fig. 2. Conductance of the ribbon with W = 10 a as a fuction of the chemicalpotential. The remaining parameters are L = 75.5
√3 a and Wlead = 20
√3 a.
Different lines (same in two panels) correspond to the flat ribbon geometry, H = 0,W ′ = W (black dashed), the buckled ribbon with H = 2a and y0 = L/2 (bluesolid) or y0 → −∞ (orange solid); see Eq. (2). Bottom panel is a zoom-in of thedata shown in top panel.
due the presence of two p-n interfaces (at x = 0 and x = W ′) leading tostronger-pronounced oscillations (of the Fabry-Perrot type), and partly dueto a limited number of propagating modes in the leads. For a kink posi-tioned at the center of the ribbon (H = 2a, y0 = L/2) the conductance
6 werekink printed on February 21, 2020
0.00 0.05 0.10 0.15
0
1
2
3
4
5
6C
harg
e pu
mpe
d pe
r kin
k, Q
/e H = 2.0 a
H = 1.5 a
H = 1.0 a
Chemical potential, μ0/t0
Fig. 3. Charge transferred upon adiabatic kink transition calculated from Eq.(5), for varying buckling amplitude H/a = 1, 2 (specified for each line in theplot), displayed as a function of the chemical potential. Dashed line shows theapproximation given by Eq. (6). Remaining parameters are same as in Fig. 2.
is strongly suppressed, G � G0, in the full |µ0| < E(2,3)min range (with the
exception from two narrow resonances at µ0 = ±0.1327 t0, which vanishesfor y0 6= L/2) defining a feasible energy window for the pump operation.
In Fig. 3 we display the charge transferred between the leads at zerobias, see Eq. (5), when slowly moving a kink along the ribbon (adiabatic kinktransition). In the energy window considered, the ribbon dispersion relationconsists of two subbands with opposite group velocities, E(1)
± = ±~vFky, seeFig. 1(c). Therefore, the total charge available for transfer at the ribbonsection of Leff = L−Winfty length can be estimated (up to a sign) as
Qtot
e≈ |µ0|
~vFLeff
π= 35.3
|µ0|t0
, (6)
where vF =√
3 t0a/(2~) ≈ 106 m/s is the Fermi velocity in graphene andwe put L−Winfty = 55.5
√3 a. For the strongest deformation (H = 2a) the
kink almost perfectly blocks the current flow, and the charge transferred isclose to the approximation given by Eq. (6); see Fig. 3.
4. Conclusion
We have investigated, by means of computer simulations of electrontransport, the operation of adiabatic quantum pump build within a topo-
werekink printed on February 21, 2020 7
logical defect moving in buckled graphene nanoribbon. We find that thepump effectiveness is close to the maximal (corresponding to a kink per-fectly blocking the current flow) for moderate bucklings, with relative bonddistortions not exceeding 8%. As topological defects generally move withnegligible energy dissipation, we hope our discussion will be a starting pointto design new graphene-based energy conversion devices.
Acknowledgments
We thank Tomasz Romańczukiewicz for discussions. The work was sup-ported by the National Science Centre of Poland (NCN) via Grant No.2014/14/E/ST3/00256.
REFERENCES
[1] Yu. V. Nazarov and Ya. M. Blanter, Quantum Transport: Introduction toNanoscience, (Cambridge University Press, Cambridge, 2009), Chapter 1.
[2] E. Prada, P. San-Jose, and H. Schomerus, Solid State Comm. 151, 1065 (2011).[3] M. Wakker and M. Blaauboer, Phys. Rev. B 82, 205432 (2010).[4] P. San-Jose, E. Prada, H. Schomerus, and S. Kohler, Appl. Phys. Lett. 101,
153506 (2012).[5] Y. Jiang, T. Low, K. Chang, M. I. Katsnelson, and F. Guinea, Phys. Rev. Lett.
110, 046601 (2013).[6] E. Grichuk and E. Manykin, Eur. Phys. J. B 86, 210 (2013).[7] B. Abdollahipour and E. Moomivand, Physica E 86 204 (2017).[8] M. Fujimoto, H. Koschke, and M. Koshino, Phys. Rev. B 101, 041112 (2020).[9] R. D. Yamaletdinov, V. A. Slipko, and Y. V. Pershin, Phys. Rev. B 96, 094306
(2017).[10] R. D. Yamaletdinov, T. Romańczukiewicz, and Y. V. Pershin, Carbon 141,
253 (2019).[11] K. Harigaya, Prog. Theor. Phys. Suppl. 113, 229 (1993).[12] J. Ma, S. Guan, and C.-H. Lai, Phys. Rev. B 74, 205401 (2006).[13] G. Dresselhaus, M. S. Dresselhaus, and R. Saito, Physical Properties Of Car-
bon Nanotubes, (World Scientific, Singapore, 1998), Chapter 11.[14] O. Gröning, S. Wang, X. Yao, et al., Nature 560, 209 (2018).[15] C. W. Groth, M. Wimmer, A. R. Akhmerov, and X. Waintal, New J. Phys.
16, 063065 (2014); see also: https://kwant-project.org.