Transcript Document

Charles Stafford
Many-body theory of electric and thermal transport in
single-molecule junctions
INT Program “From Femtoscience to Nanoscience: Nuclei, Quantum Dots and
Nanostructures,” July 31, 2009
1. Fundamental challenges of nanoelectronics
(a physicist’s perspective)
Fabrication:
Lithography → self-assembly?
For ultrasmall devices, even single-atom variations from device to
device (or in device packaging) could lead to unacceptable variations
in device characteristics → environmental sensitivity.
Contacts/interconnects to ultrasmall devices.
Switching mechanism:
Raising/lowering energy barrier necessitates dissipation of minimum
energy kBT per cycle → extreme power dissipation at ultrahigh
device densities.
Tunneling & barrier fluctuations in nanoscale devices.
Molecular electronics
Fabrication: large numbers of identical “devices” can be readily synthesized
with atomic precision. (Making the contacts is the hard part!)
But does not (necessarilly) solve fundamental problem of switching mechanism.
Single-molecule junction ≈ ultrasmall quantum dot
Similarities and differences:
Typically, π-orbitals of the carbon atoms are the itinerant degrees of freedom.
Charging energy of a single π-orbital: U ~ 9eV.
Charging energy of a benzene molecule: ‹U› ~ 5eV.
Nearest-neighbor π-π hopping integral: t ~ 2 – 3eV.
Lead-molecule coupling: Γ ~ 0.5eV (small parameter?).
Electronic structure unique for each molecule---not universal!
Alternative switching mechanism: Quantum interference
(a)
Phase difference of paths 1 and 2: kF 2d = π → destructive
interference blocks flow of current from E to C.
All possible Feynman paths cancel exactly in pairs.
(b) Increasing coupling to third terminal introduces new paths that do
not cancel, allowing current to flow from E to C.
David M. Cardamone, CAS & S. Mazumdar, Nano Letters 6, 2422 (2006);
CAS, D. M. Cardamone & S. Mazumdar, Nanotechnology 18, 424014 (2007);
U.S. Patent Application, Serial No. 60/784,503 (2007)
2. The nonequilibrium many-body problem
•Mean-field calculations based on density-functional theory are the dominant
paradigm in quantum chemistry, including molecular junction transport.
•They are unable to account for charge quantization effects (Coulomb blockade)
in single-molecule junctions!
•HOMO-LUMO gap not accurately described; no distinction of transport vs.
optical gap.
•Many-body effects beyond the mean-field level must be included for a quantitative
theory of transport in molecular heterojunctions.
•To date, only a few special solutions in certain limiting cases (e.g., Anderson
model; Kondo effect) have been obtained to the nonequilibrium many-body
problem.
•There is a need for a general approach that includes the electronic structure
of the molecule.
Nonequilibrium Green’s functions
Real-time Green’s functions
Molecular Junction Hamiltonian
Coulomb interaction (localized orthonormal basis):
Leads modeled as noninteracting Fermi gases:
Lead-molecule coupling (electrostatic coupling included in Hmol(1)):
Molecular Junction Green’s Functions
All (steady-state) physical observables of the molecular junction
can be expressed in terms of G and G<.
Dyson equation:
Tunneling self-energy:
Coulomb self-energy
must be calculated approximately.
Electric and Thermal Currents
Tunneling width matrix:
Elastic and inelastic contributions to the current
Elastic transport: linear response
3. Application to specific molecules:
Effective π-electron molecular Hamiltonian
 For the purpose of this talk we consider conjugated organic molecules.
• Transport due primarily to itinerant p-electrons.
• Sigma band is filled and doesn’t contribute appreciably to transport.
Hmol d n† , d n    n nˆn , n ,
 
nm ,

1
tnmdˆn† dˆm  H .c  U nmQˆ nQˆ m
2 nm
 Effective charge operator, including polarization charges induced by
lead voltages:
C 
Qn   dn† dn -  n  - 1
e

n
 Parameters from fitting electronic spectra of benzene, biphenyl, and trans-
stilbene up to 8-10eV:
 Accurate to ~1%
 U=8.9eV,t=2.64eV,ε=1.28
U nm   nmU  1 -  nm 
U
 1   ( Rnm /Ang)2
Castleton C.W.M., Barford W., J. Chem. Phys. Vol 17 No. 8 (2002)
Enhanced thermoelectric effects near transmission nodes
Effect of a finite minimum transmission
4. The Coulomb self-energy
Sequential-tunneling limit:
ΣC(0)
Nonequilibrium steady-state probabilities determined by detailed balance:
Correction to the Coulomb self-energy
Self-consistent Hartree-Fock correction to the
Coulomb self-energy of a diatomic molecule
•Narrowing of transmission resonances;
•No shift of transmission peak or node positions;
•No qualitative effect on transmission phase;
•Correction small in (experimentally relevant) cotunneling regime.
Coulomb blockade in a diatomic molecule
Higher-order corrections to the Coulomb self-energy: RPA
5. Results for 1,4-benzenedithiol-Au
junctions
Determining the lead-molecule coupling: thermopower
• Experimentally the BDT junction’s Seebeck coefficient is found to be
7.0.2mV/K
• Baheti et al, Nano Letters Vol 8 No 2 (2008)
• We can express the thermopower
in terms of the transmission
probability
 f 
T
E




  E - m  dE
1 -

E


S m  
eT
 f 
T
E


-   E  dE

Find that mAu- m0 =-3.22±.04eV, about 1.5eV above the HOMO level (hole dominated)
• Experimentally the linear-conductance of BDT is reported to be 0.011G0 (2e2/h)
•Xiaoyin Xiao, Bingqian Xu, and N.J Tao. Nano-letters Vol 4, No. 2 (2004)
• Comparison with calculated linear-response gives G=.63±.02eV
Differential conductance spectrum of a
benzene(1,4)dithiol-Au junction
•Junction charge quantized within ‘molecular diamonds.’
•Transmission nodes due to quantum interference.
•Resonant tunneling through molecular excited states at finite bias.
Justin P. Bergfield & CAS, Physical Review B 79, 245125 (2009)
Resonant tunneling through molecular excitons
Justin P. Bergfield & CAS, Physical Review B 79, 245125 (2009)
Conclusions
•Electron transport in single-molecule junctions is a key
example of a nanosystem far from equilibrium, and poses
a challenging nonequilibrium quantum many-body problem.
•Transport through single molecules can be controlled
by exploiting quantum interference due to molecular
symmetry.
•Large enhancement of thermoelectric effects predicted at
transmission nodes arising due to destructive quantum
interference.
•Open questions:
Corrections to Coulomb self-energy beyond RPA
Fabrication, fabrication, fabrication…