Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
CHAPTER 14
Superoperator many-body theory of
molecular currents: non-equilibrium
Green functions in real time
Upendra Harbola and Shaul Mukamel
Department of Chemistry, University of California, Irvine, CA 92697-2025, USA
Abstract
The electric conductance of a molecular junction is calculated by recasting the Keldysh
formalism in Liouville space. Dyson equations for non-equilibrium many-body Green
functions (NEGF) are derived directly in real (physical) time. The various NEGFs appear
naturally in the theory as time-ordered products of superoperators, while the Keldysh
forward/backward time loop is avoided.
14.1 INTRODUCTION
Recent advances in the fabrication and measurements of nanoscale devices have led to a
considerable interest in non-equilibrium current carrying states of single molecules. The
tunneling of electrons between two metals separated by a thin oxide layer was first
observed experimentally by Giaever [1] and later by others [2]. Vibrational resonances
can be observed for molecules absorbed at the metal–oxide interface by analyzing the
tunneling current as a function of the applied bias [3,4]. More recent development of
scanning tunneling microscopy (STM) led to a direct, real space determination of surface
structures. A metal tip is brought near the surface so that tunneling resistance is
measurable. A contour map of the surface is obtained by recording the tunneling
resistance as the tip scans the surface. The tunneling electrons interact and may exchange
energy with the nuclear degrees of freedom of the absorbed molecule. This opens up
inelastic channels for electron transmission from the tip to the surface, leading to inelastic
electron tunneling (IET). IET may play an important role in manipulating molecules with
STM [5,6]. Recently, IET was combined with STM for the chemical analysis of a single
absorbed molecule with atomic spatial resolution [7,8]. The atomic scale images of STM
q 2005 Elsevier B.V. All rights reserved.
Theory and Applications of Computational Chemistry
Edited by C. Dykstra et al. 373
ARTICLE IN PRESS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
References pp. 395–396
have been useful in analyzing different binding configurations of the molecule
chemisorbed on a semiconductor surface [9,10]. Recent advances in the theory of
STM are reviewed in Ref. [11].
Electron tunneling was first analyzed by Bardeen [12] and Cohen et al. [13] using the
perturbative transfer Hamiltonian (TH) approach and more recently by many other
authors [14–16]. Although the TH gives, in most cases, a good description of the
observed effects, it lacks a firm first principles theoretical basis and does not account
properly for many-body effects [17]. An improved form of TH [18] that involved energy
dependent transfer matrix elements was used to incorporate many-body effects. However,
this model does not describe the electron–phonon interaction properly [19].
A many-body non-equilibrium Green functions (NEGF) formulation of electron
tunneling was proposed by Caroli et al. [20]. The NEGF theory was originated by
Schwinger [21] and Kaddanof and Baym [22], and developed further by Keldysh [23] and
Craig [24]. This formalism involves the calculation of four basic Green functions, time
ordered ðGT Þ; anti-time ordered ðG~TÞ; lesser ðG,Þ and greater ðG.Þ: Additional retarded
(Gr) and advanced (Ga) Green functions are defined as specific combinations of these
basic functions. At equilibrium suffice it to know only the retarded or advanced Green
functions; all other Green functions simply follow from the fluctuation–dissipation
theorem that connects the ‘lesser’ and ‘greater’ with the retarded Green function through
the equilibrium Fermi distribution function ðf0ðEÞÞ [25]. However, for non-equilibrium
Q1
measurements, where the distribution function is not known a priori, one needs to solve
for the various NEGFs self-consistently.
Electronic transport in molecular wires and STM currents of single molecules have
received considerable attention [26–31]. Electron transport through a single molecule
[32–34] or a chain of several atoms [35] was studied. From a theoretical point of view,
this is very similar to the electron tunneling in semiconductor junctions and various
theories developed for STM [36,37] can directly be applied to molecular wires. The
NEGF technique developed for tunneling currents has been used to analyze the electron
conduction through a single molecule attached to electrodes [26,36,38–42]. The method
has also been combined with density functional theory for the modeling of transport in
molecular devices [43,44].
In this chapter, we develop a non-equilibrium superoperator Green function theory
[23,25,45] (NESGF) of molecular currents [46]. A notable advantage of working with
superoperators in the higher dimensional Liouville space [44,47] is that we need to
consider only time-ordered quantities in real (physical) time; all NEGFs show up
naturally without introducing artificial time variables. Observables can be expressed in
terms of various Liouville space pathways (LSP) [46]. The ordinary (causal) response
function which represents the density response to an external field is one particular
combination of these LSPs. Other combinations represent the spontaneous density
fluctuations and the response of these fluctuations to the external field [47,48]. A simple
time ordering operation of superoperators in real time is all it takes to derive the non-
equilibrium theory, avoiding the Keldysh loop or Matsubara imaginary time. The NESGF
theory provides new physical insights into the mechanism of the current. It can also be
more naturally used to interpret time domain experiments involving external pulses.
Chapter 14374
ARTICLE IN PRESS
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
In Section 14.2, we give a brief introduction to the superoperator formalism and recast
the NEGF theory in terms of the superoperator Green functions. Starting from the
microscopic definitions for various NESGF, we construct dynamical equations of motion
and obtain the Dyson matrix equation of Keldysh which couples the various NESGFs. In
Section 14.3, we apply the NESGF theory to the conduction through a molecular
junction. In Section 14.4 we end with a discussion.
14.2 DYSON EQUATIONS FOR SUPEROPERATOR GREEN FUNCTIONS
We consider a system of externally driven electrons and phonons described by the
Hamiltonian [16,23,26,38],
H ¼ H0 þ Hep þ Hex ð1Þ
where H0 represents the non-interacting electrons and phonons,
H0 ¼ð
drc†ðrÞh0ðrÞcðrÞ þð
drf†ðrÞV0ðrÞfðrÞ ð2Þ
h0ðrÞ ¼ ð2"2=2mÞ72 is the kinetic energy and m is the electron mass. cðc†Þ represent the
annihilation (creation) operators which satisfy the Fermi anticommutation relations,
cðrÞc†ðr0Þ þ c†ðr0ÞcðrÞ ¼ dðr 2 r0Þ
c†ðrÞc†ðr0Þ þ c†ðr0Þc†ðrÞ ¼ 0
cðrÞcðr0Þ þ cðr0ÞcðrÞ ¼ 0
ð3Þ
and fðf†Þ are boson operators with the commutation relations,
fðrÞf†ðr0Þ2 f†ðr0ÞfðrÞ ¼ dðr 2 r0Þ
f†ðrÞf†ðr0Þ2 f†ðr0Þf†ðrÞ ¼ 0
fðrÞfðr0Þ2 fðr0ÞfðrÞ ¼ 0
ð4Þ
The second term in Eq. (1) denotes the electron–phonon interaction,
Hep ¼ð
drlðrÞ½f†ðrÞ þ fðrÞ�c†ðrÞcðrÞ ð5Þ
where lðrÞ is the coupling strength. Finally, Hex represents the coupling to a time-
dependent external potential jðr; tÞ;
Hex ¼ð
drjðr; tÞc†ðrÞcðrÞ ð6Þ
We next briefly survey some properties of Liouville space superoperators that will be
useful in the following derivations [49]. The elements of the Hilbert space N £ N density
matrix, rðtÞ; are arranged as a Liouville space vector (bra or ket) of length N2: Operators
of N2 £ N2 dimension in this space are denoted as superoperators. With any Hilbert space
operator A; we associate two superoperators AL (left) and AR (right) defined through their
Superoperator many-body theory of molecular currents 375
ARTICLE IN PRESS
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
References pp. 395–396
action on another operator X as
ALX ; AX and ARX ; XA ð7Þ
We further define symmetric and antisymmetric combinations of these superoperators
Aþ ¼1
2ðAL þ ARÞ and A2 ¼ ðAL 2 ARÞ ð8Þ
The commutator and anticommutator operations in Hilbert space can thus be
implemented with a single multiplication by a ‘2’ and ‘þ’ superoperator, respectively.
We further introduce the Liouville space–time ordering operator T . This is a key
ingredient for extending NEGF to superoperators: when applied to a product of
superoperators it reorders them so that time increases from right to left. We define
kAðtÞl ; Tr{AðtÞreq} where req ¼ rðt ¼ 0Þ represents the equilibrium density matrix of
the interacting system. It is straightforward to see that for any two operators A and B we
have
kT AþðtÞB2ðt0Þl ¼ 0; t0 . t ð9Þ
kT AþðtÞB2ðt0Þl is thus always a retarded function. This follows from the definitions (8).
Since a ‘2’ superoperator corresponds to a commutator in Hilbert space, this implies that
for t , t0; kAþðtÞB2ðt0Þl becomes a trace of a commutator and must vanish, i.e.
kT AþðtÞB2ðt0Þl ¼ Tr{B2ðt
0ÞAþðtÞreq} t , t0
¼1
2Tr{½Bðt0Þ;AðtÞreq þ reqAðtÞ�} ¼ 0
Similarly, it follows that the trace of two ‘minus’ operators always vanishes
kT A2ðtÞB2ðt0Þl ¼ 0 for all t and t0 ð10Þ
We shall make use of Eqs. (9) and (10) for discussing the retarded and advanced Green
functions in Appendix 14D. Superoperator algebra was surveyed in Ref. [49].
In Liouville space the density matrix, r(t) is a vector whose time dependence is given by
rðtÞ ¼ Gðt; t0Þrðt0Þ ð11Þ
with the Green function,
Gðt; t0Þ ¼ T exp 2i
"
ðt
t0
H 2ðtÞdt
� �ð12Þ
and H 2 is the superoperator corresponding to the Hamiltonian (Eq. (1)). Note that unlike
Hilbert space, where time dependence of the ket and the bra is governed by forward and
backward time-evolution operators, respectively, in Liouville space one keeps track of
both bra and ket simultaneously and the density matrix needs only to be propagated
forward in time (Eq. (11)).
To introduce the interaction picture in Liouville space we partition H 2 ¼
H 02 þH 02 corresponding to the non-interacting and interaction Hamiltonians.
Chapter 14376
ARTICLE IN PRESS
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
With this partitioning, Eq. (12) can be written as
Gðt; t0Þ ¼ G0ðt; t0ÞGIðt; t0Þ ð13Þ
where G0 represents the time evolution with respect to H0;
G0ðt; t0Þ ¼ uðt 2 t0Þ exp 2i
"H 02ðt 2 t0Þ
� �ð14Þ
GIðt; t0Þ is the time evolution operator in the interaction picture,
GIðt; t0Þ ¼ T exp 2i
"
ðt
t0
~H 02ðtÞdt
� �ð15Þ
and ~H 02 is the interaction picture representation of H 0
2: We shall denote superoperators
in the interaction picture by a (,),
~AaðtÞ ; G†0ðt; t0ÞAaðt0ÞG0ðt; t0Þ ð16Þ
where a ¼ þ;2 or L, R. Superoperators in the Heisenberg picture will be represented
by a caret
AaðtÞ ; G†ðt; t0ÞAaðt0ÞGðt; t0Þ ð17Þ
By adiabatic switching of the interaction H 02 starting at t0 ¼ 21 we have
rðtÞ ¼ r0 2i
"
ðt
21dtG0ðt; tÞH
02ðtÞrðtÞ ð18Þ
where r0 ¼ rð21Þ is the equilibrium density matrix of the non-interacting system
r0 ¼expð2bH0Þ
Tr{expð2bH0Þ}ð19Þ
An iterative solution of Eq. (18) yields
rðtÞ ¼ GIðt;21Þr0 ð20Þ
which can also be obtained by applying the time evolution operator (15) to r(t0) and
setting t0 ¼ 21: Using Eq. (20), the equilibrium density matrix of the interacting system
can be generated from the non-interacting one by switching on the interactions
adiabatically, starting at t ¼ 21: The external potential is constant in time for t , 0 and
is assumed to be time dependent only for t . 0: We then get
req ¼ GIð0;21Þr0 ð21Þ
This adiabatic connection formula has been shown [49] to be very useful for calculating
expectation values using the interaction picture. In the corresponding Gellman-law
expression in Hilbert space [50] there is an extra denominator that takes care of the phase
Superoperator many-body theory of molecular currents 377
ARTICLE IN PRESS
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
References pp. 395–396
of the wavefunction. This is not necessary in Liouville space since the density matrix
does not acquire such a phase.
In the Heisenberg picture, the expectation value of an operator AaðtÞ is given by
kAaðtÞl ; Tr{AaðtÞreq} ð22Þ
where req ¼ rðt ¼ 0Þ: Using Eqs. (16), (17) and (20), this can be recast in the interaction
picture as
kAaðtÞl ¼ Tr{ ~AaðtÞGIðt;21Þr0} ; k ~AaðtÞGIðt;21Þl0 ð23Þ
Eq. (23) is a good starting point for developing a perturbation theory around the non-
interacting system. Through Eqs. (22) and (23) we also define the expectation values k· · ·land k· · ·l0. While the former represents the trace with respect to the interacting density
matrix, the latter is defined with respect to the non-interacting density matrix. This will be
used in the following.
Corresponding to the Hilbert space electron and phonon operators, c; c†; f and f† we
define ‘left’ (a ¼ L) and ‘right’ (a ¼ R) superoperators, ca; c†a; fa and f†
a: The
dynamics of a superoperator, ca; is described by the generalized Liouville equation,
2i"›caðtÞ
›t¼ ½H 2ðtÞ; caðtÞ� ¼ H 2ðtÞcaðtÞ2 caðtÞH 2ðtÞ ð24Þ
where H 2 is the superoperator corresponding to the Hamiltonian given in Eq. (1). A
similar equation can be written down for the phonon superoperators. In order to evaluate
the commutator appearing in the RHS of Eq. (1), we need the commutation relations of
superoperators [51]. The ‘left’ and the ‘right’ operators always commute. Thus, for
a – b we have
½caðrÞ;cbðr0Þ� ¼ ½c†
aðrÞ;c†bðr
0Þ� ¼ ½c†aðrÞ;cbðr
0Þ� ¼ 0
½faðrÞ;fbðr0Þ� ¼ ½f†
aðrÞ;f†bðr
0Þ� ¼ ½f†aðrÞ;fbðr
0Þ� ¼ 0
ð25Þ
For Fermi superoperators we have
caðrÞcaðr0Þ þ caðr
0ÞcaðrÞ ¼ 0
c†aðrÞc
†aðr
0Þ þ c†aðr
0Þc†aðrÞ ¼ 0
caðrÞc†aðr
0Þ þ caðr0Þc†
aðrÞ ¼ dðr 2 r0Þ
ð26Þ
Similarly for the boson operators
f†aðrÞf
†aðr
0Þ2 f†aðr
0Þf†aðrÞ ¼ 0
faðrÞfaðr0Þ2 faðr
0ÞfaðrÞ ¼ 0
faðrÞf†aðr
0Þ2 f†aðr
0ÞfaðrÞ ¼ kadðr 2 r0Þ
ð27Þ
Here ka ¼ 21 for a ¼ R and unity for a ¼ L:
Chapter 14378
ARTICLE IN PRESS
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
Using the commutation relations (25) and (26) and the identity
ðXY…ZÞa ¼ XaYa…Za; a ¼ L;R ð28Þ
we can recast H 2 in terms of the elementary field superoperators,
H 2 ¼ H 02 þHep2 þH
ex2 ð29Þ
with
H 02 ¼X
a¼L;R
ka
ðdrðc†
aðrÞh0ðrÞcaðrÞ þ f†aðrÞV0ðrÞfaðrÞÞ
He2p2 ¼
Xa¼L;R
ka
ðdrlðrÞFaðrÞc
†aðrÞcaðrÞ
Hex2 ¼
Xa¼L;R
ka
ðdrc†
aðrÞjðr; tÞcaðrÞ
ð30Þ
where Fa ¼ fa þ f†a:
We next define electron and phonon superoperator Green functions
Gabðrt;r0t0Þ¼2i
"kT caðr;tÞc
†bðr
0;t0Þl
Dabðrt;r0t0Þ¼2i
"kT Faðr; tÞF
†bðr
0; t0Þlð31Þ
As shown in Ref. [44] (see Appendix 14A), GLL, GRR, GLR and GRL, respectively,
coincide with the standard Hilbert space–time ordered GT ; antitime ordered G~T; lesser
G, and greater G. Green functions defined on a closed time loop.
Using the commutation relations (3), the Heisenberg equations of motion for
superoperator caðtÞ reads
i"ka›caðr; tÞ
›t¼ hðr; tÞcaðr; tÞ þ lðrÞFaðr; tÞcaðr; tÞ ð32Þ
where hðr; tÞ ¼ h0ðrÞ þ jðr; tÞ: By taking the time derivative of the electron Green
function in Eq. (31) and using Eq. (32), we obtain the equation of motion for Gab;
i"›
›t2 kahðr; tÞ
� �Gabðrt; r0; t0Þ ¼ dabdðx 2 x0Þ2
i
"kalðrÞ
� kT Faðr; tÞcaðr; tÞc†bðr
0; t0Þl ð33Þ
In order to derive the equation of motion for the phonon Green function Da; we add the
following coupling term ðdrJðr; tÞFðrÞ ð34Þ
to the Hamiltonian. Here Jðr; tÞ is some artificial field that will be set to zero at the end of
calculations. This new term does not effect the electron Green function since F and c
Superoperator many-body theory of molecular currents 379
ARTICLE IN PRESS
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
References pp. 395–396
commute. Thus, superoperator Hex2 in Eq. (30) takes the form
Hex2 ¼
Xa¼L;R
ka
ðdr½c†
aðrÞjðr; tÞcaðrÞ þ Jaðr; tÞFaðrÞ� ð35Þ
where fields JL and JR couple to the ‘left’ and ‘right’ boson operators, respectively. Using
the boson commutation (Eq. (26)), the Heisenberg equation for superoperator fa is
obtained as
2i"ka›faðr; tÞ
›t¼ V0ðrÞfaðr; tÞ þ lðrÞc†
aðr; tÞcaðr; tÞ þ Jaðr; tÞ ð36Þ
Using Eq. (36) we next obtain the equation of motion for operator Fa
2"2ka
V0ðrÞ
›2Faðr; tÞ
›t2¼ V0ðrÞFaðr; tÞ þ 2lðrÞc†
aðr; tÞcaðr; tÞ þ 2Jaðr; tÞ ð37Þ
Taking trace with respect to the density matrix rðt ¼ 0Þ we obtain
21
2
"2ka
V0ðrÞ
›2
›t2þV0ðrÞ
!kFaðr; tÞl ¼ lðrÞkc†
aðr; tÞcaðr; tÞlþ Jaðr; tÞ ð38Þ
Using the interaction picture representation (Eq. (23)) with Hex2 given by Eq. (35) we can
write
kFaðr; tÞl ¼ k ~Fðr; tÞGIðt;21Þl0 ð39Þ
By taking the functional derivative with respect to Jb; and setting JL ¼ JR ¼ 0; we obtain
d
dJbðr0; t0Þ
kFaðr; tÞl
JL¼JR¼0
¼ 2i
"ebkT ~Faðr; tÞ ~F†
bðr0; t0ÞGIðt;21Þl0
¼ kbDabðrt; r0t0Þ ð40Þ
Using Eqs. (38)–(40), the equation of motion for the phonon Green function is obtained as
21
2
"2kb
V0ðrÞ
›2
›t2þV0ðrÞ
!Dabðrt; r0; t0Þ ¼ dabdðx 2 x0Þ þ
i
"lðrÞka
� kT c†aðr; tÞcaðr; tÞF†
bðr0; t0Þl ð41Þ
We shall denote the space and time coordinates collectively by x ¼ r; t; thus in Eqs. (33)
and (41) dðx 2 x0Þ ; dðr 2 r0Þdðt 2 t0Þ:Following Keldysh, we shall rearrange the superoperator Green functions in a 2 £ 2
matrix �G;
�Gðx; x0Þ ¼GLLðx; x0Þ GLRðx; x0Þ
GRLðx; x0Þ GRRðx; x0Þ
!ð42Þ
Chapter 14380
ARTICLE IN PRESS
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
and similarly the phonon Green function matrix �D with elements Dab: The corresponding
Green functions of the non-interacting system described by the Hamiltonian (2) are
denoted by �G0 and �D0; respectively. These are given by
G0abðrt; r0t0Þ ¼ i"
›
›t2 kahðr; tÞ
� �21
dabdðx 2 x0Þ
D0abðrt; r0t0Þ ¼ 2
1
2
"2kb
V0ðrÞ
›2
›t2þV0ðrÞ
!21
dabdðx 2 x0Þ
ð43Þ
Using our matrix notation, we can recast Eqs. (33) and (41) in the form of Dyson
equations
�G ¼ �G0 þ �G0 �S �G;
�D ¼ �D0 þ �D0 �P �Dð44Þ
The effect of all interactions is now included in the electron ð �SÞ and phonon ð �PÞ self-
energies. Exact expressions for the self-energies are obtained by comparing Eqs. (33) and
(41) with Eq. (44)
Sabðrt; r0t0Þ ¼ 2i
"kalðrÞ
Xb0
ðdtð
dr1kT F†aðr; tÞcaðr; tÞc†
b0 ðr1; tÞlG21b0bðr1t; r0t0Þ
Pabðrt; r0t0Þ ¼i
"kalðrÞ
Xb0
ðdtð
dr1kT c†aðr; tÞcaðr; tÞF†
b0 ðr1; tÞlD21b0bðr1t; r0t0Þ
ð45Þ
Eqs. (43)–(45) are exact and constitute the NESGF formalism.
In order to evaluate the self-energies perturbatively, we rewrite the Green functions,
Eq. (31), in the interaction picture
Gabðrt; r0t0Þ ¼ 2i
"kT ~caðr; tÞ ~c†
bðr0; t0ÞGIðt;21Þl0
Dabðrt; r0t0Þ ¼ 2i
"kT ~Fðr; tÞ ~F†
bðr0; t0ÞGIðt;21Þl0
ð46Þ
where GIðt;21Þ is given by Eq. (15) with t0 ¼ 21: Using Eqs. (13), (16) and (20), the
self-energies (45) can also be expressed in the interaction picture as
Sabðrt;r0t0Þ¼2i
"kalðrÞ
Xb0
ððdtdr1kT ~F†
aðr;tÞ ~caðr;tÞ ~c†b0 ðr1;tÞGIðt;21Þl0G21
b0bðr1t;r0t0Þ
Pabðrt;r0t0Þ¼i
"kalðrÞ
Xb0
ððdtdr1kT ~c†
aðr;tÞ ~caðr;tÞ ~F†b0 ðr1;tÞGIðt;21Þl0D21
b0bðr1t;r0t0Þ
ð47Þ
Eq. (47) together with Eq. (46) constitutes closed form equations for the self-energies
where all the averages are given in the interaction picture, k· · ·l0, with respect to
Superoperator many-body theory of molecular currents 381
ARTICLE IN PRESS
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
References pp. 395–396
the non-interacting density matrix. By expanding GI (Eq. (15)) perturbatively in ~H 02 we
can obtain perturbative expansion for the self-energies. Each term in the expansion can
be calculated using Wick’s theorem for superoperators [49] which is given in
Appendix 14E. This results in a perturbative series in terms of the zeroth order Green
functions.
14.3 THE CALCULATION OF MOLECULAR CURRENTS
We have applied the NESGF to study the charge conductivity of a molecular wire
attached to two perfectly conducting leads. In the simplest approach the leads ‘a’ and ‘b’
are treated as two free electron reservoirs. Nuclear motions in the molecular region are
described as harmonic phonons which interact with the surrounding electronic structure
and the environment (secondary phonons) [26]. We first recast the general Hamiltonian,
Eq. (1), in a single electron local basis and partition it as
H ¼ Hf þ Hint ð48Þ
where Hf represents the free, non-interacting electrons and phonons and with no coupling
between molecule and leads
Hf ¼Xi; j
Ei; jc†i cj þ
Xk[a;b
ekc†kck þ
Xl
Vlf†l fl þ
Xm
vmf†mfm ð49Þ
The indices ði; jÞ represent the electronic basis states corresponding to the molecule, k
labels the electronic states in the leads (a and b), l denotes primary phonons which
interact with the electrons and m denotes the secondary phonons which are coupled to the
primary phonons and constitute a thermal bath. The applied external voltage V maintains
a chemical potential difference, ma 2 mb ¼ eV ; between the two leads and also modifies
the single electron energies. In addition it provides an extra termP
i Vic†i ci which is
included in the zeroth order Hamiltonian, Hf ; by modifying the single electron energies.
The interaction Hamiltonian is given by
Hint ¼X
k[a;b;i
ðVkic†kci þ h:c:Þ þ
Xl;i
lliFlc†i ci þ
Xl;m
UlmFlFm ð50Þ
The three terms represent the molecule/lead interaction, coupling of primary phonons
with the molecule and the interaction of primary and secondary phonons, respectively.
The total current passing through the junction can be expressed in terms of the electron
Green functions and the corresponding self-energies. At steady state it is given by (see
Appendix 14B, Eq. (B22))
IT ¼2e
"
Xij0
ð dv
2p
hS
ij 0
LRðvÞGj 0iRLðvÞ2 S
ij 0
RLðvÞGj 0iLRðvÞ
ið51Þ
The electron Green functions G0LR and G0
RL correspond to the free Hamiltonian, Hf ; and
the self-energies SLR and SRL represent the effects of all interactions (Eq. (50)).
Chapter 14382
ARTICLE IN PRESS
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
Sijab has contributions from the electron-lead (s) and electron-phonon (J ) interactions
SijabðvÞ ¼ s
ijabðvÞ þJ
ijabðvÞ ð52Þ
These are given in Eqs. (C29) and (C38). The self-energy expressions (C38) and (C40)
are calculated perturbatively to second order in the electron–phonon coupling in terms of
the zeroth order Green functions (Eq. (55)). The simplest expression for current is
obtained by substituting Eqs. (55), (C29) and (C38) in Eq. (51). This zeroth order result
can be improved by using the renormalized Green functions obtained from the self-
consistent solution of the Dyson equation (44).
In order to solve self-consistently for the electron Green functions that appear in
the current formula, Eq. (51), the self-energy is calculated under the Born
approximation by replacing the zeroth order Green functions, G0ab and D0
ab with
the corresponding renormalized Green functions, Gab and Dab; as is commonly done
in mode-coupling theories [52,53]. This approximation sums an infinite set of non-
crossing diagrams [54,55] that appear in the perturbation expansion of the many-body
Green function, Gab:Since the electron self-energy (Eq. (C38)) also depends on the phonon Green function,
the phonon self-energy, Pll0
ab; is also required for a self-consistent solution of the electron
Green functions. The phonon self-energy calculated in Appendix 14C is given by
PijabðvÞ ¼ g
ijabðvÞ þ L
ijabðvÞ ð53Þ
where gijabðvÞ (Eq. (C30)) and L
ijabðvÞ (Eq. (C40)) represent the contributions from the
phonon–phonon and the electron–phonon interactions, respectively.
Computing the renormalized electron and phonon Green functions and the
corresponding self-energies involves the self-consistent solution of the following
coupled equations for the Green functions:
GLRðvÞ ¼ G0LLðvÞSLLðvÞGLRðvÞ þ G0
LLðvÞSLRðvÞGRRðvÞ
GRLðvÞ ¼ G0RRðvÞSRLðvÞGLLðvÞ þ G0
RRðvÞSRRðvÞGRLðvÞ
GLLðvÞ ¼ G0LLðvÞ þ G0
LLðvÞSLLðvÞGLLðvÞ þ G0LLðvÞSLRðvÞGRLðvÞ
GRRðvÞ ¼ G0RRðvÞ þ G0
RRðvÞSRLðvÞGLRðvÞ þ G0RRðvÞSRRðvÞGRRðvÞ
ð54Þ
Similarly the equations for the phonon Green functions are obtained by replacing Gab
with Dab and Sab with Pab: Here Green functions corresponding to the free Hamiltonian
G0ijab and D0ll0
ab are given by
G0ijabðvÞ ¼
dab
vdij 2 kaEij þ ih; D0ll0
ab ðvÞ ¼2Vldabdll0
ka"2v2 2V2
l þ ihð55Þ
Superoperator many-body theory of molecular currents 383
ARTICLE IN PRESS
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
References pp. 395–396
where we set " ¼ 1 and h! 0: Eij ¼ Ei 2 Ej is the energy difference between single
electron ith and jth states. Vl denotes the molecular phonon eigenstates.
Once the Green functions, GLR; GRL and the corresponding self-energies SLR; SRL;are obtained from the self-consistent solution of Eq. (54) together with Eq. (52),
formula (51) can be used to calculate the total current through the molecular junction.
14.4 DISCUSSION
In this chapter, we have developed the NESGF formalism and applied it to the
computation of molecular current. The Liouville space–time ordering operator provides
an elegant way for performing calculations in real time, thus avoiding the artificial
backward and forward time evolution required in Hilbert space (Keldysh loop). Wick’s
theorem for superoperators is used to compute the self-energies perturbatively to the
second order in phonon–electron coupling. Eq. (54) have been derived earlier by many
authors [19,43,55]. Recently, Galperin et al. [26] have used a fully self-consistent
solution to study the influence of different interactions on molecular conductivity for a
strong electron–phonon coupling. The main aim of the present work is to demonstrate
that by doing calculations in Liouville space one can avoid the backward/forward time
evolution (Keldysh loop) required in Hilbert space. This originates from the fact that in
Liouville space both ket and bra evolve forward in time. Thus, one can couple the
system with two independent fields, ‘left’ and ‘right’. This property of Liouville space
can be used to construct real (physical) time generating functionals for the non-
perturbative calculation of the self-energies.
The present model [37–39] ignores electron–electron interactions. These may be
treated using the GW technique [56–58] formulated in terms of the superoperators and
extended to non-equilibrium situations. All non-equilibrium observables can be
obtained from a single generating functional in terms of ‘left’ and ‘right’ operators.
The retarded (advance) Green function that describes the forward (backward) motion of
the system particle can also be calculated in terms of the basic Green functions, Gab
(see Appendix 14D).
The NESGF formulation can also be recast in terms of the þ and 2 (rather than L/R)
superoperators which are more directly related to observables. This is done in
Appendix 14D. We focused on the primary quantities that are represented in terms of
the ‘left’ and ‘right’ superoperators and all other quantities are obtained as the linear
combination of these basic operators.
14.5 ACKNOWLEDGEMENTS
The support of the National Science Foundation (Grant No. CHE-0132571) and NIRT
(Grant No. EEC 0303389) is gratefully acknowledged. We wish to thank Prof. Wilson Ho
for useful discussions and Prof. Abraham Nitzan for sending us the preprint of his paper
(Ref. [23]).
Chapter 14384
ARTICLE IN PRESS
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
APPENDIX 14A: SUPEROPERATOR EXPRESSIONS FOR
THE KELDYSH GREEN FUNCTIONS
The standard NEGF theory formulated in terms of the four Hilbert space Green functions:
time ordered ðGT Þ; anti-time ordered ðG~TÞ; greater (G.) and lesser (G,) [23,25]. These
are defined in the Heisenberg picture as
GT ðx; x0Þ ; 2i
"kTcðxÞc†ðx0Þl
¼ 2i
"uðt 2 t0ÞkcðxÞc†ðx0Þlþ uðt0 2 tÞkc†ðx0ÞcðxÞl
G~Tðx; x0Þ ; 2
i
"k ~TcðxÞc†ðx0Þl
¼ 2i
"uðt0 2 tÞkcðxÞc†ðx0Þlþ uðt 2 t0Þkc†ðx0ÞcðxÞl
G.ðx; x0Þ ; 2i
"kcðxÞc†ðx0Þl
G,ðx; x0Þ ;i
"kc†ðx0ÞcðxÞl
ðA1Þ
These are known as T ð ~TÞ is the Hilbert space–time (anti-time) ordering operator: when
applied to a product of operators, it reorders them in ascending (descending) times from
right to left.
The four Green functions that show up naturally in Liouville space are defined as
GLLðx; x0Þ ¼ 2i
"kT cLðxÞc
†Lðx
0Þl
GRRðx; x0Þ ¼ 2i
"kT cRðxÞc
†Rðx
0Þl
GLRðx; x0Þ ¼ 2i
"kT cLðxÞc
†Rðx
0Þl
GRLðx; x0Þ ¼ 2i
"kT cRðxÞc
†Lðx
0Þl
ðA2Þ
T is the Liouville space–time ordering operator, which rearranges all superoperators in
increasing order of time from right to left.
Superoperator many-body theory of molecular currents 385
ARTICLE IN PRESS
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
References pp. 395–396
To establish connection between Liouville space and Hilbert space Green functions we
shall convert superoperators back to ordinary operators [44]. For GLR, and GRL, we obtain
GLRðx; x0Þ ; 2i
"Tr{T cLðxÞc
†Rðx
0Þreq}
¼ 2i
"Tr{cðxÞreqc
†ðx0Þ}
¼ 2i
"kc†ðx0ÞcðxÞl ¼ G,ðx; x0Þ
GRLðx; x0Þ ; 2i
"Tr{T cRðxÞc
†Lðx
0Þreq}
¼ 2i
"Tr{c†ðx0ÞreqcðxÞ}
¼ 2i
"kcðxÞc†ðx0Þl ¼ G.ðx; x0Þ
ðA3Þ
where req is the fully interacting many body equilibrium density matrix.
For GLL and GRR we have two cases
(i) For t . t0; we get
GLLðx; x0Þ ; 2i
"Tr{T cLðxÞc
†Lðx
0Þreq}
¼ 2i
"Tr{cðxÞc†ðx0Þreq} ¼ 2
i
"kcðxÞc†ðx0Þl
GRRðx; x0Þ ; 2i
"Tr{T cRðxÞc
†Rðx
0Þreq}
¼ 2i
"Tr{reqc
†ðx0ÞcðxÞ} ¼ 2i
"kc†ðx0ÞcðxÞl
ðA4Þ
(ii) For the reverse case, t , t0; we get
GLLðx; x0Þ ; 2i
"Tr{T cLðxÞc
†Lðx
0Þreq}
¼ 2i
"Tr{c†ðx0ÞcðxÞreq} ¼ 2
i
"kc†ðx0ÞcðxÞl
GRRðx; x0Þ ; 2i
"Tr{T cRðxÞc
†Rðx
0Þreq}
¼ 2i
"Tr{reqcðxÞc
†ðx0Þ} ¼ 2i
"kcðxÞc†ðx0Þl
ðA5Þ
Chapter 14386
ARTICLE IN PRESS
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
Combining Eqs. (A4) and (A5) we can write
GLLðx; x0Þ ; 2i
"Tr{T cLðxÞc
†Lðx
0Þreq}
¼ 2i
"½uðt 2 t0ÞkcðxÞc†ðx0Þl2 uðt0 2 tÞkc†ðx0ÞcðxÞl�
¼ GT ðx; x0Þ
GRRðx; x0Þ ; 2i
"Tr{T cRðxÞc
†Rðx
0Þreq}
¼ 2i
"½uðt 2 t0Þkc†ðx0ÞcðxÞl2 uðt0 2 tÞkcðxÞc†ðx0Þl�
¼ G~Tðx; x0Þ
ðA6Þ
Eqs. (A3) and (A6) establish the equivalence of Hilbert and Liouville space Green
functions and they can be summarized as
GLLðx; x0Þ ¼ GT ðx; x0Þ; GRRðx; x0Þ ¼ G�Tðx; x0Þ
GLRðx; x0Þ ¼ G,ðx; x0Þ; GRLðx; x0Þ ¼ G.ðx; x0ÞðA7Þ
APPENDIX 14 B: SUPEROPERATOR GREEN FUNCTION
EXPRESSION FOR THE CURRENT
In this appendix, we present a formal microscopic derivation for the current flowing
through a conductor. The conductor could be a molecule or a metal or any conducting
material attached to two electrodes held at two different potentials.
In Hilbert space the charge current–density is given by
jðr; tÞ ¼ 2ie"
2mk½c†ðr; tÞ7cðr; tÞ2 ð7c†ðr; tÞÞcðr; tÞ�l ðB8Þ
where e and m are the electron charge and mass, respectively. Eq. (B8) can also be
expressed in a slightly modified form as
jðr; tÞ ¼ie"
2m½kð72 70Þc†ðr; tÞcðr0; t0Þl�x0¼x ðB9Þ
where 70 represents the derivative with respect to r0.
Using relation (A7) the current density can be expressed in terms of the superoperator
Green function as
jðr; tÞ ¼ 2e"2
2m½ð72 70ÞGLRðrt; r0t0Þ�x0¼x ðB10Þ
Superoperator many-body theory of molecular currents 387
ARTICLE IN PRESS
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
References pp. 395–396
At steady state, the Green functions only depend on the time difference ðt 2 t0Þ and the
total current density (JT) becomes time independent. Transforming to the frequency
(energy) domain, the current density per unit energy is
jðr;EÞ ¼ 2e"
2m½ð72 70ÞGLRðrr0;EÞ�r0¼r ðB11Þ
and the total current density
JTðrÞ ¼ð dE
2pjðr;EÞ ðB12Þ
Eq. (B12) provides a recipe for calculating the current profile across the conductor once
the Green function GLR is known from the self-consistent solution of the Dyson equation.
For computing the total current passing through the conductor, Eq. (B11) can be
expressed in the form of Eq. (51). In order to get the total current per unit energy ðITðEÞÞ
passing between electrode and conductor we need to integrate the current density over the
surface area of the conductor–electrode contact
ITðEÞ ¼ð
sjðr;EÞ · ndS ¼
ð7 · jðr;EÞdr ðB13Þ
where n is the unit vector normal to surface S: Substituting into Eq. (B13) from Eq. (B11),
we get
ITðEÞ ¼ 2e"
2mTr½ð72 2 702ÞGLRðrr0;EÞ� ðB14Þ
In general, a conductor–electrode system can be described by the Hamiltonian
H ¼ H0 þ Hint ðB15Þ
where H0 represents the non-interacting part
H0 ¼ð
drc†ðrÞh0ðrÞcðrÞ ðB16Þ
where h0ðrÞ ¼ 2 "2
2m72 and all the interaction terms (conductor–electrode, electron–
phonon) are included in Hint: The total current per unit energy, Eq. (B14), is
ITðEÞ ¼ 2e
"Tr½ðh0ðrÞ2 hp
0ðr0ÞÞGLRðrr0;EÞ� ðB17Þ
The Dyson equations for the retarded Green function (see Appendix 14D, Eq. (D48)) in
frequency (energy) can be expressed in the matrix form as
h0Gr ¼ EGr 2 I 2 srGr ðB18Þ
where I is the identity matrix and Sr is the retarded self-energy, Eq. (D50). E ¼ "v is a
number. Henceforth, we write all the expressions in the matrix notation. Taking
Chapter 14388
ARTICLE IN PRESS
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
the complex conjugate of Eq. (B18), we obtain the Dyson equation for the advanced
Green function,
Gah0 ¼ EGa 2 I 2 Gasa ðB19Þ
with the corresponding advanced self-energy, Sa: From the matrix Dyson equation
(D48), we also have the relation
GLR ¼ GrSLRGa ðB20Þ
Using the relations (B18)–(B20), it is easy to see that
h0GLR 2 GLRh0 ¼ GLRSa þ GrSLR 2 SrGLR 2 SLRGa ðB21Þ
Substituting this in Eq. (B17), the total current per unit energy becomes
ITðEÞ ¼2e
"Tr½SLRðEÞGRLðEÞ2 SRLðEÞGLRðEÞ� ðB22Þ
where a factor of 2 is introduced to account for the spin degeneracy.
We have calculated the total current in real space. In practice, the Green functions and
the self-energy matrices are calculated in an electronic basis ði; jÞ: The total current
through the conductor is obtained by integrating Eq. (B22) over energy resulting in
Eq. (51).
APPENDIX 14C: SELF-ENERGIES FOR SUPEROPERATOR GREEN
FUNCTIONS
The basic quantities required for describing the coupled molecule–lead system are the
one particle electron and the phonon Green functions. Following the steps outlined in
Section 14.2, the time development for various superoperators (Heisenberg equations) is
(all primed indices should be summed over)
i"ka›
›tciaðtÞ ¼ Eij0 cj0aðtÞ þ Vk0ick0aðtÞ þ ll0ifl0aðtÞciaðtÞ
2i"ka›
›tflaðtÞ ¼ li0lc
†i0aðtÞci0aðtÞ þVlflaðtÞ þ Ulm0fm0aðtÞ
i"ka›
›tckaðtÞ ¼ ekckaðtÞ þ Vki0 ci0 ðtÞ
2 i"ka›
›tfmaðtÞ ¼ vmfmaðtÞ þ Ul0mfl0 ðtÞ
ðC23Þ
Using Eq. (C23) and following the procedure described in Section 14.2, it is
straightforward to write the matrix Dyson equation (44) for the electron and phonon
Superoperator many-body theory of molecular currents 389
ARTICLE IN PRESS
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
References pp. 395–396
Green functions defined as
Gijabðx;x
0Þ¼2i
"kT ciaðr;tÞc
†jbðr
0; t0Þl
Dll0
abðx;x0Þ¼2
i
"kT Flaðr; tÞF
†l0bðr0;t0Þl
ðC24Þ
with the corresponding self-energy matrix elements
Sijabðt; t
0Þ ¼2i
"kaXb0; j0
ðdt
Xl0
ll0ikT Fl0aðtÞciaðtÞc†j0b0 ðtÞlþ
Xk0
Vk0ikT ck0aðtÞc†j0b0 ðtÞl
" #
�Gj0j21
b0bðt; t0Þ;J
ijabðt; t
0Þþsijabðt; t
0Þ
Pll0
abðt; t0Þ ¼
i
"
ðdtXb0;l00
Xm0
Ulm0kT Fm0aðtÞF†l00b0 ðtÞlþ
Xi0
lli0kT c†i0aðtÞci0aðtÞF
†l00b0 ðtÞl
" #
£Dl00l021b0b ðt; t0Þ; gll0
abðt; t0ÞþLll0
abðt; t0Þ ðC25Þ
The two terms in the electron self-energy represent the contributions from the phonon–
electron (J) and molecule–lead (s) interactions. Similarly, the phonon self-energy has
contributions from the electron–phonon (L) and the primary–secondary phonon (g)
couplings. The self-energy due to the molecule–lead coupling can be calculated exactly.
To this end we need to obtain the quantity kT ~ck0aðtÞc†j0b0 ðtÞl: By multiplying the third
equation in Eq. (C23) by c†j0b0 ðtÞ from the left and from the right, taking trace and
subtracting, we get (here primed indices are not summed over)
i"ka›
›t2 ek0
� �kT ck0aðtÞc
†j0b0 ðtÞl¼
Xi0
Vk0i0 kT ci0aðtÞc†j0b0 ðtÞl
) kT ck0aðtÞc†j0b0 ðtÞl¼ i"
Xi0
Vk0i0gk0 ðtÞGi0j0
ab0 ðt;tÞðC26Þ
where gkðtÞ ¼ ði"ka››t2 ek0 Þ
21: Substituting expression (C26) in Eq. (C25) gives for the
molecule–lead self-energy
sijabðt; t
0Þ ¼kadabX
k0[a;b
Vk0iVk0jgk0 ðtÞdðt2 t0Þ ðC27Þ
Similarly, the contribution to the phonon self-energy from the interaction with secondary
phonons can be calculated exactly
gll 0
ab ¼2kadabXm0
Ulm0Ul0m0g0m0 ðtÞdðt2 t0Þ ðC28Þ
where g0m0 ðtÞ ¼ ði"ka
››tþvm0 Þ21: At steady state all Green functions and self-energies
depend only on the time difference ðt1 2 t2Þ and it is very convenient to express them in
the frequency space. The self-energy contributions due to molecule–lead ðsijabÞ and
phonon–phonon ðg l0
abÞ interactions, Eqs. (C27) and (C28), can be represented in
Chapter 14390
ARTICLE IN PRESS
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
frequency space as
sijabðvÞ ¼ kadab
Xk0
Vk0iVk0j
kav2 ek0 þ ihðC29Þ
g ll0
abðvÞ ¼2kadabXm0
Ulm0Ul0m0
kavþvm0 þ ihðC30Þ
where h!0: However, in real calculations it is a common practice to calculate self-
energies sab and gab in the wide band approximation, implying that the real parts of the
self-energies can be ignored and the imaginary parts are considered as frequency
independent. Eqs. (C29) and (C30) then reduce to simpler forms
sijab ¼ kadab
i
2G ij; g ll0
ab ¼2kadabi
2~G ll0 ðC31Þ
where G ijð2pP
k0 Vk0iVk0j and ~G ll0 ¼ 2pP
m0 Um0lVm0l0 :The phonon contribution to the electronic self-energy is obtained perturbatively in the
phonon–electron coupling. We recast the phonon contribution (first term on the RHS of
Eq. (C25) for Sijab) in the interaction picture by writing
kT FlaðtÞciaðtÞc†jbðt
0Þl ¼ kT ~FlaðtÞ ~ciaðtÞ ~c†jbðt
0ÞGIðt;21Þl0 ðC32Þ
where
GIðt;21Þ ¼ exp 2i
"
ðdtXi0a0
ka0
Xl0
ll0i0~Fl0a0 ðtÞ ~c†
i0a0 ðtÞ ~ci0a0 ðtÞ
"(
þXk0
Vk0i0 ð ~c†k0a0 ðtÞ ~ci0a0 ðtÞþ ~c†
i0a0 ðtÞ ~ck0a0 ðtÞÞ
# )ðC33Þ
Substituting Eq. (C33) in Eq. (C32), expanding the exponential to first order in lli and
using Wick’s theorem for superoperators [49] we obtain
kT FlaðtÞciaðtÞc†jbðt
0Þl¼2"2Xl0i0a0
ka0ll0i0
ðdtD0ll0
aa0 ðt;tÞhG
0ijabðt; t
0ÞG0i0i0
a0a0 ðt;tþÞ
þG0ii0
aa0 ðt;tÞG0i0j
a0bðt; t0Þ
iðC34Þ
Here the superscript ‘0’ represents the trace with respect to the non-interacting density
matrix. The zeroth order Green functions are given in Eq. (55). The terms coming from
the lead-molecule coupling (Vki) vanish because they are odd in creation and annihilation
operators. Substituting Eq. (C34) in Eq. (C25) gives for the phonon contribution
Superoperator many-body theory of molecular currents 391
ARTICLE IN PRESS
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
References pp. 395–396
to the self-energy
Jijabðt; t
0Þ ¼ i"Xl1l2
kall1i kbll2jD0l1l2ab ðt; t0ÞG0ij
abðt; t0Þ
h
þ dijdabdðt2 t0ÞXi1a
0
ll2i1ka0
ðdtD
0l1l2aa0 ðt;tÞG
0i1i1a0a0 ðt;t
þÞi
ðC35Þ
In the derivation of Eq. (C35), we have used the identity
ðdtXa0j0
G0ij0
ab0 ðt;tÞG21 0j0j
b0bðt; t0Þ ¼ dabdijdðt2 t0Þ ðC36Þ
Similarly the contribution of the electron–phonon interaction to the phonon self-energy
(second term in Eq. (C25) for Pijab) can be obtained perturbatively. To the second order in
phonon–electron coupling, we obtain
Lll0
abðt; t0Þ ¼2i"
Xij
kakbllill0j½G0jibaðt
0; tÞG0ijabðt; t
0Þ þG0iiaaðt; t
þÞG0jjbbðt
0; t0þÞ� ðC37Þ
To second order in electron–phonon coupling, the electronic self-energy depends on both
the electron and phonon green functions while the phonon self-energy contains only the
electron Green functions.
At steady state we shift to the frequency domain and obtain
JijabðvÞ ¼ i"
Xl1l2
kakbll1ill2j
ð dv0
2pD
0l1l2ab ðv0ÞG
0ijabðv2 v0Þ
þ dijdabX
l1;l2;i1;a0
ka0ll1ill2i1r0
i1i1D
0l1l2aa0 ðv ¼ 0Þ ðC38Þ
where
r0ii ; i"Gii
aaðt ¼ 0Þ ¼ ið dE
2pGii
aaðEÞ ðC39Þ
The phonon self-energy becomes
Lll0
abðvÞ ¼ 2i"X
ij
kakbllill0j
ð dv0
2pG
0ijabðv
0ÞG0jibaðv
0 2 vÞ
þi
"
Xij
kaebllill0jr0iir
0jjdðv ¼ 0Þ ðC40Þ
Chapter 14392
ARTICLE IN PRESS
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
APPENDIX 14D: DYSON EQUATIONS IN THE 1 /2 REPRESENTATION
In this appendix, we define the retarded and advance Green’s functions and the
corresponding self-energies and relate them to the basic Green functions and self-
energies obtained in Appendix 14C. From definition (9), the Liouville space retarded (Gr)
and advance (Ga) Green functions are defined as
Gijr ðt; t0Þ ; 2
i
"kT ciþðtÞc
†j2ðt
0Þl ðD41Þ
Gija ðt; t0Þ ; 2
i
"kT ci2ðt
0Þc†jþðtÞl ðD42Þ
We further introduce the correlation function
Gijc ðt; t0Þ ; 2
2i
"kT ciþðt
0Þc†jþðtÞl ðD43Þ
It follows from Eq. (10) that there are only three Green functions in the þ /2
representation. These are given by Eqs. (D41)–(D43). Using Eq. (8) these can be
represented in terms of the basic Green functions (31) as
Gijr ðt; t0Þ ¼
1
2½G
ijLLðt; t0Þ2 G
ijLRðt; t0Þ þ G
ijRLðt; t0Þ2 G
ijRRðt; t0Þ�
¼ GijLLðt; t0Þ2 G
ijLRðt; t0Þ
Gija ðt; t0Þ ¼
1
2½G
ijLLðt; t0Þ2 G
ijRRðt; t0Þ2 G
ijRLðt; t0Þ þ G
ijLRðt; t0Þ�
¼ 2GijRRðt; t0Þ þ G
ijLRðt; t0Þ ¼ G
ijLLðt; t0Þ2 G
ijRLðt; t0Þ
Gijc ðt; t0Þ ¼
1
2½G
ijLLðt; t0Þ þ G
ijRRðt; t0Þ þ G
ijLRðt; t0Þ þ G
ijRLðt; t0Þ�
¼ GijLLðt; t0Þ þ G
ijRRðt; t0Þ
ðD44Þ
where we have used the identity GLL þ GRR ¼ GLR þ GRL which can be verified using
Eq. (10). A Dyson equation corresponding to Gr; Ga and Gc can be obtained from Eq. (44)
using unitary transformation
G ¼ S �GS21 ðD45Þ
where G represents the matrix
G ¼0 Ga
Gr Gc
!ðD46Þ
and
S ¼1ffiffi2
p1 21
1 1
!ðD47Þ
Superoperator many-body theory of molecular currents 393
ARTICLE IN PRESS
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
References pp. 395–396
The transformed Dyson equation (44) reads
G ¼ G0 þ G0 ~SG ðD48Þ
and the corresponding self-energy matrix reduces to
~S ¼Sc Sr
Sa 0
!ðD49Þ
with the matrix elements given by
Sijr ðt; t
0Þ ¼ SijLLðt; t0Þ þ S
ijLRðt; t0Þ
Sija ðt; t
0Þ ¼ SijRRðt; t0Þ þ S
ijLRðt; t
0Þ
Sijc ðt; t
0Þ ¼ SijRRðt; t0Þ þ S
ijLLðt; t0Þ
ðD50Þ
Similar relations also hold for the phonon Green functions and self-energies.
Using Eqs. (D44) and (D50), the retarded self-energies for electron and phonon Green
functions (retarded) coming from the electron–phonon coupling are obtained as
Jijr ðvÞ ¼ i"
Xll0
ll0illj
ð dv0
2p½D0ll0
r ðv0ÞG0ijr ðv2 v0Þ þ D0ll0
r ðv0ÞG0ijLRðv2 v0Þ
þ D0ll0
LRðv0ÞG0ij
r ðv2 v0Þ�
Lll0
r ðvÞ ¼ 2i"X
ijllill0j
ð dv0
2p
hG
0ijLRðv
0ÞG0ija ðv2 v0Þ
þ G0ijr ðv0ÞðG
0ijRLðv2 v0Þ þ Gaðv2 v0ÞÞ
iðD51Þ
Similarly the retarded self-energies due to the lead and secondary phonons can be written
in the wide band limit as
s ijr ¼
i
2G ij and g ll0
r ¼ 2i
2~G ll0 ðD52Þ
where Gij includes contributions from both the leads a and b, i.e. G ij ¼ Gija þ G
ijb :
APPENDIX 14E: WICK’S THEOREM FOR SUPEROPERATORS
Wick’s theorem for superoperators was formulated in Ref. [49]. Using Eqs. (8) and (27),
it can be shown that similar to the L and R superoperators, the commutator of ‘þ ’ and
‘2 ’ boson superoperators are also numbers. Thus, boson superoperators follow Gaussian
statistics and Wick’s theorem holds for both the L, R and ‘þ ’, ‘2 ’ representations.
However, for Fermi superoperators life is more complicated. The anticommutator
corresponding to only the ‘left’ or the ‘right’ Fermi superoperators are numbers but that
for the ‘left’ and ‘right’ superoperators, in general, is not a number. Thus, the Fermi
superoperators are not Gaussian. However, since the left and right superoperators always
commute, the following Wick’s theorem [49] can be applied to the time-ordered product
Chapter 14394
ARTICLE IN PRESS
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
of any number of ‘left’ and ‘right’ superoperators, e.g.
kT Ai1n1ðt1ÞAi2n2
ðt2Þ· · ·AinnnðtnÞl0 ¼
Xp
kT AianaðtaÞAibnb
ðtbÞl0 · · · kT AipnpðtpÞAiqnq
ðtqÞl0ðE53Þ
Here Ainnn; nn ¼ L;R; represents either a boson or a fermion superoperator. iana…iqvq is a
permutation of i1n1…innn and sum on p runs over all possible permutations, keeping the
time ordering. In case of fermions, each term should be multiplied by ð21ÞP; where P is
the number of permutations of superoperators required to put them into a particular order.
Only permutations among either ‘left’ or among ‘right’ superoperators count in P:The permutations among ‘L’ and ‘R’ operators leave the product unchanged.
14.6 REFERENCES
1 I. Giaever, Phys. Rev. Lett., 5 (1960) 147.
2 J. Nicol, S. Shapiro and P.H. Smith, Phys. Rev. Lett., 5 (1960) 461.
3 R.C. Jaklevik and J. Lambe, Phys. Rev. Lett., 17 (1966) 119.
4 W. Wang, T. Lee, I. Kretzschmar and M.A. Reed, Nano Lett., 4 (2004) 643; J. Chen and M.A. Reed,
Chem. Phys., 281 (2002) 127.
5 R.E. Walkup, D.M. Newns and Ph. Avouris, Phys. Rev. B, 48 (1993) 1858.
6 S. Gao, M. Persson and B.I. Landqvist, Solid State Commun., 84 (1992) 271.
7 W. Ho, J. Chem. Phys., 117 (2002) 11033.
8 B.C. Stipe, M.A. Rezaei and W. Ho, Science, 280 (1998) 1732.
9 F. Wang, D.C. Sorescu and K.D. Jordan, J. Phys. Chem., 106 (2002) 1316.
10 P. Sautet, Chem. Rev., 97 (1997) 1097.
11 W.A. Hofer, A.S. Foster and A.L. Shluger, Rev. Mod. Phys., 75 (2003) 1287.
12 J. Bardeen, Phys. Rev. Lett., 6 (1961) 57.
13 M.H. Cohen, L.M. Felicov and J.C. Phillips, Phys. Rev. Lett., 8 (1962) 31.
14 J. Tersoff and D.R. Hamann, Phys. Rev. Lett., 50 (1983) 1998; Phys. Rev. B 31 (1985) 805.
15 N.D. Lang, Phys. Rev. Lett., 55 (1985) 230.
16 B.N.J. Persson and A. Baratoff, Phys. Rev. Lett., 59 (1987) 339; N. Lorent and M. Persson, Phys. Rev.
Lett., 85 (2000) 2997; N. Lorent, M. Persson, L.J. Lauhon and W. Ho, Phys. Rev. Lett., 85 (2001) 2997.
17 J.A. Appelbaum and W.F. Brinkman, Phys. Rev., 183 (1969) 553.
18 J.A. Appelbaum and W.F. Brinkman, Phys. Rev., 186 (1969) 464.
19 C. Caroli, R. Combescot, P. Nozieres and D.S. James, J. Phys. C: Solid State Phys., 5 (1972) 21.
20 C. Caroli, R. Combescot, P. Nozieres and D.S. James, J. Phys. C: Solid State Phys., 4 (1971) 916.
21 J. Schwinger, J. Math. Phys., 2 (1961) 407.
22 L.P. Kaddanof and G. Baym, Quantum statistical mechanics, Benjamin, New York, 1992.
23 L.V. Keldysh, Sov. Phys. J., 20 (1965) 1018.
24 R.A. Craig, J. Math. Phys., 9 (1968) 407.
25 H. Haug and A.-P. Jauho, Quantum kinetics in transport and optics of semiconductors, Springer,
Heidelberg, 1996.
26 M. Galperin, A. Nitzan, S. Sek and M. Majda, J. Electroanal. Chem., 550 (2003) 337; M. Galperin,
M.A. Ratner and A. Nitzan, Nano Lett., 4 (2004) 1605.
27 P.S. Damle, A.W. Ghosh and S. Datta, Phys. Rev. B, 64 (2001) 201403.
28 M. Di Ventra, S.-G. Kim, S.T. Pantelides and N.D. Lang, Phys. Rev. Lett., 86 (2001) 288; Z. Yang and
M. Di Ventra, Phys. Rev. B, 67 (2003) 161311.
29 W. Tian, S. Datta, S. Hong, R. Reifenberger, J.I. Henderson and C.P. Kubiak, J. Chem. Phys., 109 (1998)
2874.
30 R. Baer, T. Seideman, S. Llani and D. Neuhauser, J. Chem. Phys., 120 (2004) 3387.
31 H. Ness, H.A. Shevlin and A.J. Fisher, Phys. Rev. B, 63 (2001) 125422.
Superoperator many-body theory of molecular currents 395
ARTICLE IN PRESS
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
References pp. 395–396
32 L.A. Bumm, J.J. Arnold, M.T. Cygan, T.D. Dunbar, T.P. Burgin, L. Jones, D.L. Allara, J.M. Taur and
P.S. Weiss, Science, 271 (1996) 1705.
33 X.D. Cui, A. Primak, X. Zarate, J. Tomfohr, O.F. Sankey, A.L. Moore, T.A. Moore, D. Gust, G. Harris and
S.M. Lindsay, Science, 294 (2001) 571.
34 J. Park, A.N. Pasupathy, J.I. Goldsmith, A.V. Soldatov, C. Chang, Y. Yaish, J.P. Sethna, H.D. Abruna,
D.C. Ralph and P.L. McEuen, Thin Solid Films, 438 (2003) 457.
35 N. Agrait, C. Untiedt, G. Rubio-Bollinger and S. Vieira, Phys. Rev. Lett., 88 (2002) 216803.
36 S. Datta, Electronic transport in mesoscopic systems, Cambridge University Press, Cambridge, 1997.
37 A. Nitzan, Annu. Rev. Phys. Chem., 52 (2001) 681.
38 S. Tikhodeev, M. Natario, K. Makoshi, T. Mii and H. Ueba, Surf. Sci., 493 (2001) 63.
39 T. Mii, S. Tikhodeev and H. Ueba, Surf. Sci., 502 (2002) 26.
40 A. Di Carlo, Physica B, 314 (2002) 211.
41 P. Hanggi, M. Ratner and S. Yaliraki (Eds.) Chem. Phys. (special issue) 281 (2002).
42 T. Mii, S.G. Tikhodeev and H. Ueba, Phys. Rev. B, 68 (2003) 205406.
43 Y. Xue, S. Datta and M.A. Ratner, Chem. Phys., 281 (2002) 151.
44 O. Berman and S. Mukamel, Phys. Rev. B, 69 (2004) 155430.
45 K. Chau, Z. Su, B. Hao and L. Yu, Phys. Rep., 118 (1985) 1; B. Hao, Physica A, 109 (1981) 221.
46 S. Mukamel, Principles of nonlinear optical spectroscopy, Oxford University Press, New York, 1997.
47 A.E. Cohen and S. Mukamel, Phys. Rev. Lett., 91 (2003) 233202.
48 E. Wang and U. Heinz, Phys. Rev. D, 66 (2002) 025008.
49 S. Mukamel, Phys. Rev. E, 68 (2003) 021111.
50 J.W. Negele and H. Orland, Quantum many particle systems, Addison-Wesley, New York, 1988.
51 T.L.C. Jansen and S. Mukamel, J. Chem. Phys., 119 (2003) 7979.
52 J.-P. Bouchaud, L. Cagliandolo, J. Kurchan and M. Mezard, Physica A, 226 (1996) 243.
53 J. Schofield, R. Lim and I. Oppenheim, Physica A, 181 (1992) 89.
54 N.E. Bickers, Rev. Mod. Phys., 59 (1987) 845.
55 N.S. Wingreen and Y. Meir, Phys. Rev. B, 49 (1994) 11040.
56 L. Hedin, Phys. Rev., 139 (1965) 796; J. Phys.: Condens. Matter 11, (1999) 489.
57 G. Onida, L. Reining and A. Rubio, Rev. Mod. Phys., 74 (2002) 601.
58 M. Rohlfing and S.G. Louie, Phys. Rev. B, 62 (2000) 4927.
Chapter 14396
ARTICLE IN PRESS
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
Author QueriesJOB NUMBER: 9585
JOURNAL: Superoperator many-body theory of molecular currents: non-equilibrium
Green functions in real time
Q1 Please check sense of the sentence ‘At equilibrium suffice it to know only the….
ARTICLE IN PRESS
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
References pp. 395–396