Transport Calculations with

TranSIESTA
T SIESTA

Pablo Ordejón
Instituto de Ciencia de Materiales de Barcelona - CSIC,, Spain
p
M B
M. Brandbyge,
db K St
K. Stokbro
kb and
d JJ. T
Taylor
l
Mikroelectronik Centret - Technical Univ. Denmark

Jose L. Mozos and Frederico D. Novaes

Instituto de Ciencia de Materiales de Barcelona - CSIC, Spain

The SIESTA team

E. Anglada, E. Artacho, A. García, J. Gale. J. Junquera, D. Sánchez-Portal,
J. M. Soler, ....
 Outline

1 Electronic transport in the nanoscale: Basic theory

1.
2. Modeling: the challenges and our approach:
• The problem at equilibrium (zero voltage)
• Non-equilibrium (finite voltage)
3. Practicalities
1. Electronic Transport
p in the Nanoscale:
Basic Theory
d
L R

• Scattering in nano-scale systems:

– electron-electron interactions
d << Lm
– phonons
– impurities, defects
d≈λ
– elastic scattering by the potential of the contact

• Semiclassical theory breaks down – QM solution needed

• Landauer formulation: Conductance as transmission probability
S. Datta, Electronic transport in mesoscopic systems (Cambridge)
Narrow constriction (meso-
(meso-nanoscopic)
E

kx

Transversal confinement QUANTIZATION

ik x x = 2k 2x
ψ (x, y) = ψ m (y ) e Em ((k x ) = ε m +
2
2m
 Landauer formulation - no scattering
E
µ+δµ
µ

δµ = − eV
k
bands
2e 2
I = − 2e ∑ ∫ d k f ( k ) v g ( k , i) ≈ V N bands (E F )
i BZ h
I 2e 2
G = = N bands = N bands G0
V h
2e 2
QUANTUM OF CONDUCTANCE GO = ⇒ 12.9
12 9 kΩ
h
Landauer formulation - scattering
• Transmission probability of an incoming electron at energy ε:

T (ε ) =Tr [t t ] (ε )
†

• Current:
transmission matrix: ψout = t ψin

2e 2
eV I = ∫ dε (fL (ε ) − fR (ε ))T (ε )
I h

2e 2
• P f t conductance
Perfect d t ((one channel):
h l) T=1
T 1 GO =
h
 2. Modeling:
g
The Challenges and Our Approach

• Model the molecule-electrode system from first principles:

No parameters fitted to the particular system Æ DFT

• Model a molecule coupled to bulk (infinite) electrodes

• Electrons out of equilibrium (do not follow the thermal Fermi occupation)
• Include finite bias voltage/current and determine the potential profile
• Calculate the conductance (quantum transmission through the molecule)
• Determine geometry: Relax the atomic positions to an energy minimum
Restriction: Ballistic conduction
• Ballistic conduction: consider only the scattering of the
incoming electrons by the potential created by the contact

d
L R

• Two terminal devices (three terminals in progress)

• Effects not described: inelastic scattering

– electron-electron interactions (Coulomb blockade)
– phonons
h ((current-induced
ti d d phonon
h excitations)
it ti )
First Principles: DFT
• Manyy interacting-electrons
g problem mapped
p pp to a one
particle problem in an external effective potential
(Hohemberg-Kohn, Kohn-Sham)

Veff =Vext +Vps +VH +VXC

• Charge density as basic variable:

Veff =Veff [ρ ( r )]
• Self-consistency: Veff ↔ ρ
• Ground
G d state
t t th
theory: VXC
 SIESTA http://www uam es/siesta
http://www.uam.es/siesta

Soler, Artacho, Gale, García, Junquera, Ordejón and Sánchez-Portal

J. Phys.: Cond. Matt. 14, 2745 (2002)

• Self consistent DFT code (LDA, GGA, LSD)

Self-consistent
• Pseudopotentials (Kleinman-Bylander)
• LCAO approximation:
B i set:
Basis t
Confined Numerical Atomic Orbitals

φµ (r )
Sankey’s “fireballs”

• Order N methodology (in the calculation and the solution of the

Order-N
DFT Hamiltonian)
 TRANSIESTA
Implementation of non-equilibrium electronic transport in SIESTA

• Atomistic description (both contact and electrodes)

• Infinite electrodes
• Electrons out of equilibrium
• Include finite bias and determine the potential profile
• Calculates the conductance (both linear and non
non-linear)
linear)
• Forces and geometry

Brandbyge, Mozos, Ordejón, Taylor and Stokbro

Phys. Rev. B. 65, 165401 (2002)

Mozos, Ordejón
Mozos Ordejón, Brandbyge
Brandbyge, Taylor and Stokbro
Nanotechnology 13, 346 (2002)
The problem at Equilibrium
(Zero Bias)
Challenge:
Coupling the finite contact to infinite electrodes

Solution: G (z ) = (z - H )−1
Green’s Functions
1
ρ (ε ) = − Im G (ε + iδ )
π
⎡∞ ⎤
ρ (r ) = ∑ ⎢ ∫ dε ρ µν (ε ) nF (ε − µ F )⎥ φµ (r ) φν (r )
µν ⎣ −∞ ⎦

Dµν
Setup ((zero bias)) G (ε ) = (ε − H ) =
−1

−1
⎡ ... -VL + ⎤
⎢ − ⎥
C ⎢-VL ε - H L -VLC ⎥
L R ⎢ -V
VCL ε - HC -V
VCR ⎥
⎢ ⎥
⎢ -VRC ε - HR -VR+ ⎥
⎢ -VR− ... ⎥⎦
⎣
Contact:
• Contains the molecule, and part of the Right and Left electrodes
• Sufficiently large to include the screening

Solution in finite system:

C −1
B L R B ⎡ε − H L − Σ L −VLC ⎤
G (ε ) = ⎢⎢ −VCL ε − HC −VCR ⎥
⎥
⎢⎣ −VRC ε − HR − Σ R ⎥⎦

Σ (ε ) = Selfenergies
Selfenergies. Can be obtained from the bulk Greens functions
Lopez-Sancho et al. J. Phys. F 14, 1205 (1984)
Calculations ((zero bias):
)
• Bulk Greens functions and self-energies (unit cell calculation)
• Hamiltonian of the Contact region:

SCF

PBC
• Solution of GF’s equations ⇒ ρ(r)

• Landauer-Büttiker: transmission probability: [ ]

T (ε ) =Tr t †t (ε )

t (ε ) = (Im [ΣR (ε )]) G (ε ) (Im [ΣL (ε )])

1/2 1/2
The problem at Non-
Non-Equilibrium
(Finite Bias)
C
L R
µL
e-
µR
2 add
additional
to a p problems:
ob e s

• Non-equilibrium situation:
– current flow
– two different chemical potentials

• Electrostatic potential with boundary conditions

Non--equilibrium formulation:
Non

• Scattering states (from the left)

Lippmann-Schwinger
Lippmann Schwinger Eqs.: ψ l = ψ lo + G (εo ) VL ψ lo

• Non-equilibrium Density Matrix:

∞
Dµν = ∫ dε [ρ µνL (ε ) nF (ε − µ L ) + ρ µνR (ε ) nF (ε − µ R )]
−∞
1
ρ (ε ) = ( G (ε + i δ ) Im [ ΣL (ε + i δ )]G t (ε + i δ ))µν
L
µν
π
 Electrostatic Potential
Given ρ(r), VH(r) is determined
except up to a linear term:
VH (r ) = φ (r ) + a ⋅ r + b
12 au
φ( r): particular solution of +V/2 -V/2
P i
Poisson’s
’ equation
ti Au Au
a and b: determined imposing BC:
the shift V between electrodes

• φ ((r)) computed
p using
g FFT’s

V ⎛ L⎞
• Linear term: − ⎜ z − ⎟
L ⎝ 2 ⎠
 3. TranSIESTA Practicalities

3 Step process:

1. SIESTA calculation of the bulk electrodes, to get H, ρ, and Self-energies

Self energies
2. SIESTA calculation for the open system
• reads the electrode data
• builds H from ρ
• solves the open problem using Green’s Functions (TranSIESTA)
• builds
b ild new ρ
3. Postprocessing: compute T(E), I, ...
 −1
⎡ε − H L − Σ L −VLC ⎤
G (ε ) = ⎢⎢ −VCL ε − HC −VCR ⎥
⎥
⎢⎣ −VRC ε − HR − Σ R ⎥⎦

Supercell - PBC

• H, DM fixed to bulk in L and R

• DM computed in C from Green’s functions
• HC, VLC and VCR computed in a supercell
approach (with potential ramp)
• B (buffer) does not enter directly in the
calculation (only in the SC calc. for VHartree)
Contour integration
g
Contour
Integration
SolutionMethod Transiesta

# GENGF OPTIONS
TS.ComplexContour.Emin -3.0 Ry
TS.ComplexContour.NPoles 6
TS.ComplexContour.NCircle 20
TS.ComplexContour.NLine 3
TS.RealContour.Emin -3.0 Ry
TS.RealContour.Emax 2.d0 Ry
TS.TBT.Npoints 100

# TS OPTIONS
TS.Voltage
S l 1 000000 eV
1.000000
TS.UseBulkInElectrodes .True.
TS.BufferAtomsLeft 0
TS.BufferAtomsRight 0

# TBT OPTIONS
TS.TBT.Emin -5.5 eV
TS TBT Emax +0
TS.TBT.Emax +0.5
5 eV
TS.TBT.NPoints 100
TS.TBT.NEigen 3