Generalized Breit-Wigner treatment of molecular transport: Charging effects in a

single decanedithiol molecule

Hugo Andres Cabrera-Tinoco, Augusto C. L. Moreira, and Celso P. de Melo

Citation: The Journal of Chemical Physics 148, 194304 (2018); doi: 10.1063/1.5016284
View online: https://doi.org/10.1063/1.5016284
View Table of Contents: http://aip.scitation.org/toc/jcp/148/19
Published by the American Institute of Physics

Articles you may be interested in

Effect of molecular conformations on the electronic transport in oxygen-substituted alkanethiol molecular
junctions
The Journal of Chemical Physics 148, 184703 (2018); 10.1063/1.5025190

Perspective: Theory of quantum transport in molecular junctions

The Journal of Chemical Physics 148, 030901 (2018); 10.1063/1.5003306

Perspective: How to understand electronic friction

The Journal of Chemical Physics 148, 230901 (2018); 10.1063/1.5035412

On simulation of local fluxes in molecular junctions

The Journal of Chemical Physics 148, 204103 (2018); 10.1063/1.5029252

Perspective: Ring-polymer instanton theory

The Journal of Chemical Physics 148, 200901 (2018); 10.1063/1.5028352

Nonequilibrium thermodynamics and boundary conditions for reaction and transport in heterogeneous media
The Journal of Chemical Physics 148, 194114 (2018); 10.1063/1.5031761
 THE JOURNAL OF CHEMICAL PHYSICS 148, 194304 (2018)

Generalized Breit-Wigner treatment of molecular transport:

Charging effects in a single decanedithiol molecule
Hugo Andres Cabrera-Tinoco, Augusto C. L. Moreira, and Celso P. de Melo
Departamento de Fı́sica, Universidade Federal de Pernambuco, 50670-901 Recife, PE, Brazil
(Received 17 November 2017; accepted 26 April 2018; published online 18 May 2018)

We examine the relative contribution of ballistic and elastic cotunneling mechanisms to the charge
transport through a single decanedithiol molecule linked to two terminal clusters of gold atoms.
For this, we first introduced a conceptual model that permits a generalization of the Breit-Wigner
scattering formalism where the cation, anion, and neutral forms of the molecule can participate with
different probabilities of the charge transfer process, but in a simultaneous manner. We used a density
functional theory treatment and considered the fixed geometry of each charge state to calculate the
corresponding eigenvalues and eigenvectors of the extended system for different values of the external
electric field. We have found that for the ballistic transport the HOMO and LUMO of the neutral species
play a key role, while the charged states give a negligible contribution. On the other hand, an elastic
cotunneling charge transfer can occur whenever a molecular orbital (MO) of the cation or anion
species, even if localized in just one side of the molecule-gold clusters complex, has energy close
to that of a delocalized MO of the neutral species. Under these conditions, a conduction channel is
formed throughout the entire system, in a process that is controlled by the degree of resonance between
the MOs involved. Our results indicate that while different charge transfer mechanisms contribute
to the overall charge transport, quantum effects such as avoided-crossing situations between relevant
frontier MOs can be of special importance. In these specific situations, the interchange of spatial
localization of two MOs involved in the crossing can open a new channel of charge transfer that
otherwise would not be available. Published by AIP Publishing. https://doi.org/10.1063/1.5016284

I. INTRODUCTION to the overall charge transport,18–21 mainly in situations

Molecular electronics represent the promise of a new
technological paradigm, in which active logic components
would be based on the direct connection of single molecules
to metallic electrodes.1 However, while the understanding
and control of charge transport through isolated molecules
is a necessary step for this, at the present stage, even for
the paradigmatic family of alkane molecules, results at the
theoretical2–7 and experimental4,8–11 levels are not always
consistent. Although the observed disagreement can be in
part attributed to the still insufficient control of relevant
experimental parameters (such as the orientation of the indi-
vidual molecules relative to the electrodes and the type of
electrode-molecule bond12,13 ), a better insight on the fun-
damentals of the molecular charge transport remains direly
needed.
The usual theoretical description of electron transfer
through a single molecule assumes the prevalence of coher-
ent (“ballistic”) charge transport mechanisms. And yet, there
are several cases where a fully coherent transport model is not
sufficient to explain the observed charge transfer,14–16 with
the contribution of cotunneling processes (i.e., higher-order
tunneling events in which two or more electrons participate)
becoming important.17
Consequently, in more recent years, several propos-
als have been advanced to account for the simultane-
ous contributions of coherent and non-coherent processes
where a change in the oxidation state of the molecule is
observed. In fact, recent experimental observation of hys-
teresis14 and of molecular charge switching features during
charge transfer 22,23 demand a consideration of possible redox
contributions.16,19,20,24
In the present work, we first introduce a conceptual
model that permits elastic coherent and non-coherent contri-
butions25 to compete in the molecular transport, by allowing
changes in the oxidation state of the molecule of interest.
Based on this model, we extend the Breit-Wigner formula
to describe the mixing of three different molecular charge
states (cation, neutral, and anion) that with varying probabili-
ties simultaneously participate in the charge transfer process.
In this manner, not only coherent elastic (ballistic) processes
involving a single charge state are considered (and remain,
in fact, dominant) but also changes in the oxidation num-
ber of the molecular species (e.g., from the single cation
to the neutral form) are explicitly accounted for. We first
apply these ideas to benchmark our method against the stan-
dard solution for the transport through an Anderson impu-
rity. Then, we investigate the contribution of charging effects
on the transport through an isolated decanedithiol molecule,
a representative example of saturated molecules commonly
used as test-bed cases of molecular transport. The electronic
structure of the complex formed by the organic molecule
plus the metal atoms in its closest proximity (the “extended
molecule”—see later) is determined at increasing values of

0021-9606/2018/148(19)/194304/10/$30.00 148, 194304-1 Published by AIP Publishing.

194304-2 Cabrera-Tinoco, Moreira, and de Melo J. Chem. Phys. 148, 194304 (2018)

the external field. A peculiar feature that we have identified the actual device [Fig. 1(a)] by a conceptual system of a dual
is that avoided-crossing events between neighboring molecu- character [Fig. 1(b)]: while source leads S1 and S2 connect the
lar orbitals (MOs) can occur at specific values of the applied
bias, causing an interchange of their spatial localization. As
a result, such events are associated with the opening of con-
duction channels, with a consequent increase in the overall
current.
II. THEORETICAL FORMALISM
A. Ballistic and sequential charge transfer
molecular states N  1 and N to the drain leads D1 and D2,
in the other sub-system, the N and N + 1 charge states con-
nect the S2 and S3 source leads to the D2 and D3 drain leads.
The conceptual leads will be labeled by β x (with β = S, D
and x = 1, 2, 3). In addition, we admit that each possible
pathway can contribute to the overall transport process with
a probability γx2 , with the condition γ12 + γ22 + γ32 = 1 being
observed. We specifically assign the subscript x = 1, 2, and
3 to indicate the individual probability of participation of the
Ballistic transport occurs whenever the strength of the cation (N  1), neutral (N), and anion (N + 1) charge states,
interaction between the molecule and the metallic leads is respectively.
large. In this strong coupling limit, sometimes referred to as Since all charge states can contribute to the transport,
the self-consistent field (SCF) regime, the molecular levels in the description of the conceptual device, we must con-
are broadened by their interaction with the contacts. Under sider the complete set of eigenvalues (of the real device)
those conditions, the Landauer approach can be used to ε x ,s of all three charge states [Fig. 1(c)], which we will
describe the molecular transport, with the current calculated refer to as the eigenvalues ζ n of the complete system [see
as Fig. 1(d)]. Hence, if two (zero-temperature) eigenvalues ε x ,s

2e +∞ belonging to different charge states have identical energies,
I= T (E)(fL (E) − fR (E))dE, (1) the corresponding conceptual eigenvalue ζ n will be doubly
h −∞
degenerate.
where the transmission T (E) is given by the well-known Breit- Let us consider that at T = 0 we have S x = {ε x ,s } and
Wigner formula.26 In fact, although originally introduced for S full = {ζ n } as, respectively, the set of eigenvalues in the
the description of scattering processes in nuclear physics, the x-charge state of the real device and the complete set of dis-
Breit-Wigner expression has been widely used within the Lan- tinct eigenvalues. As an example, in Fig. 1(c) we represent the
dauer approximation for the treatment of molecular charge case where S 1 = {ε 1,1 , ε 1,2 , ε 1,3 }, S 2 = {ε 2,1 , ε 2,2 , ε 2,3 }, and
transport 4,27 in the form
S 3 = {ε 3,1 , ε 3,2 , ε 3,3 } so that S full = {ζ1(1) , ζ2(2) , ζ3(3) , ζ4(2) , ζ5(1) },
4Γn,L (E)Γn,R (E) with the superscripts indicating the “degeneracy” of each level,
Tn (E) =
2 , (2)
(E − ε n )2 + Γn,L (E) + Γn,R (E) as shown in Fig. 1(d). As we will see in a moment, this degener-
acy is a necessary but not sufficient condition for the existence
where of non-coherent (i.e., non-ballistic) elastic transport contribu-
2
Γn,β (E) = π ρβ (E)τn,β  , β = L, R (3)
are the scattering rates through the nth channel.25
These rates can be viewed as the imaginary part of the
self-energies Im{Σn (E)} = Γn,L (E) + Γn,R (E) which arise due
to the coupling of the molecule to the electrodes. The total
transmission T = n Tn is the sum of the contributions from
P
all orbitals whose energies lie in the specific Fermi window of
the problem.
However, this picture is not valid if the coupling between
the molecule and the metallic leads is small. In this limit, the
charge transport can be sequential and we must allow for the
contribution of charged states of the molecular species con-
sidered. One faces then the weak coupling regime, where
charging effects such as the Coulomb blockade (CB), for
instance, may occur. In this case, one should either use the
multi electron master equation (MEME) approach or resort to
Non-Equilibrium Green Function (NEGF) methods, but now
adding a Coulomb electronic repulsion energy term to the
Hamiltonian and employing the full expression for the elec-
tric current (see, for instance, Ref. 25). Our purpose is to deal
with molecular transport without any a priori hypothesis with
respect to the prevailing coupling regime.
B. The conceptual system
To analyze the charge transport through a single molecule
connected to two terminal electrodes, we begin by substituting
tions. Thus, at zero temperature, S full represents the collection
of all distinct eigenvalues in all charge states of the real device,
and we will assume that each level will maintain its original
occupancy fixed.
C. Extending the Breit-Wigner formula
for the description of the conceptual system
In this conceptual model of the molecular device, more
than a single charge state can participate in the transport
process. We will use this model to generalize the Breit-
Wigner expression to describe situations where charging
effects become relevant. For this, we define the (vector)
coupling strength
γ1 Υn,β1
~ n,β = *.γ2 Υn,β2 +/,
Λ (4a)
,γ3 Υn,β3 -
where
N 
X εx,s +η
Υn,βx = Γx,s,β (E 0)FT (E 0 − ζn )dE 0 (4b)
s=1 εx,s −η
is the corresponding (thermal broadened) coupling strength of
the βth conceptual lead (see Sec. SI-1 of the supplementary
material), with the subscripts referring to the charge state (x)
and to the nth ζn channel eigenvalue.
The parameter η is a measure of the degree of resonance
between two eigenvalues of the real device. Together with the
194304-3 Cabrera-Tinoco, Moreira, and de Melo J. Chem. Phys. 148, 194304 (2018)

FIG. 1. (a) Schematic view of a “real” device (RD). (b) Conceptual device (CD) considered in this work. Each pair of leads [S1, D1], [S2, D2], and [S3, D3]
couples channels with [N  1], [N], and [N + 1] electrons, respectively. (c) Eigenvalues of the “real” device in three different (N  1, N, and N + 1) charge states
and their occupation. (d) The set of (occupied and unoccupied) eigenvalues of the conceptual device. The double line indicates degenerate eigenvalues and the
occupation distribution is analog to that of a small island (or quantum dot) with 3N particles and 3S states.

spatial localization of the corresponding MOs, the value of η ~ T is the transpose of Λ

where Λ ~ n,β and the matrix Ωn is defined
↔

n,β
determines how relevant would be the simultaneous contribu- as
tion of these states to the overall charge transport (see Sec.
SI-1 of the supplementary material). 1 a1,2,n 0
↔
Here, Γx ,s,β , which corresponds to the strength of the cou- Ωn = *.−a1,2,n 1 a2,3,n +/, (6)
pling of lead β to the sth channel eigenvalue in the xth charge , 0 −a2,3,n 1 -
state of the real device, is given by
2 for each conceptual device eigenvalue ζ n . In Eq. (6),
Γx,s,β (E 0) = Γβ (E 0, ε x,s ) = π ρβ (E 0)τx,s,β  , (4c) ax ,y,n = δfx ,n,Ay,n + δAx ,n,fy,n is a sum of Kronecker deltas
where ρβ (E 0) is the density of states of the electrode con- containing the Fermi distribution describing the degree in
nected to the βth lead and τ x ,s,β is the coupling between the which the nth “MO” of the conceptual device is occupied
sth MO in the xth charge state of the real device and the β
metallic cluster. Hence, by construction, we assure that more
than one charge state can participate in an elastic scattering
event through the conceptual device, if they possess eigen-
states with energy ε x ,s in common. In this manner, one can
(f x ,n = f (ζ n  µx )) or unoccupied (Ax ,n = 1  f x ,n ). Here, µx
is the corresponding Fermi level of the real device in the xth
charge state, which we will assume to be localized midway
between the lowest unoccupied level and the highest occu-
pied level of the charge state considered. Since no transitions
rewrite the Breit-Wigner expression in a more general form between the N  1 to N + 1 states are allowed, the Ω13 and
as Ω31 elements are taken as equal to zero to comply with the
condition that the initial and final charge states can differ at
↔
~ T · Ωn · Λ
4Λ ~ n,D most by one electron. The factor ax ,y,n = δfx ,n,Ay,n + δAx ,n,fy,n
n,S
Tn (E) =   ↔  , (5) is zero if the nth “MO” of the conceptual device is both occu-
(E − ζn )2 + Λ~T +Λ
n,S
~ T · Ωn · Λ
n,D
~ n,S + Λ
~ n,D pied or unoccupied, in such a way that only off-diagonal terms
194304-4 Cabrera-Tinoco, Moreira, and de Melo J. Chem. Phys. 148, 194304 (2018)

(such as Ax,n ,f y,n , for example in δAx ,n,fy,n ) that mix different In Sec. III, we first apply these ideas to illustrate how our
charge states will survive (see Sec. SI-2 of the supplementary method would describe the charge transport through an Ander-
↔
material for more details on the Ωn matrix). son impurity, the simplest model that allows for Coulomb
↔
If we expand the product Λ ~ T · Ωn · Λ
n,S
~ n,D in the numer- charging effects.25,28–30 Then, in Secs. IV and V we use the
ator of Eq. (5), the generalized transmission function can be generalized Breit-Wigner (GBW) approach to calculate the
written as composed by five terms. Three of them, with the electrical current traversing a decanedithiol molecule, a mem-
form ber of the frequently investigated alkane family. In doing this,
we will pay special attention to identifying the relative con-
4Υn,Sx Υn,Dx tributions of the different pathways that involve more than a
bal
Tn,x (E) =  ↔ 

(E − ζn )2 + Λ
~T +Λ ~ T · Ωn · Λ~ n,S + Λ
~ n,D
 single charge state. To better illustrate the effect of generaliz-
n,S n,D
ing the Breit-Wigner expression, we will compare the overall
(7) charge transport that occurs when just the neutral species par-
ticipate in the process (i.e., γ22 = 1) to that in the case where
(for x = 1, 2, 3), will be associated with the coherent transport. γ12 = γ22 = γ32 = 1/3, i.e., when the three charge states
In addition, one can observe that the nth MO will not par- contribute on an equal footing.
ticipate in the ballistic transport if its corresponding (source
or drain) coupling strength vanishes (i.e., Λ ~ n,{S/D} → 0),
III. TRANSPORT THROUGH AN ANDERSON IMPURITY
implying that the wave function does not extend to that
region. The charge transport through an isolated Anderson-
The remaining two terms have the form impurity A connected to two electrodes (left and right,
r = L, R) can be described by the Hamiltonian H D=H DL + HDR
†
Here, Hr = ε kσr akσr akσr refers
 
4γx γy Υn,Sx Υn,Dy − Υn,Sy Υn,Dx 29,30 P
+ HA + HT ,L + HT ,R .
D D D D
cot
Tn,(x/y) (E) = kσ
DA = †
ε σ cσ
 ↔  P
to the electrodes, H cσ + Un↑ n↓ describes the
 
(E − ζn )2 + Λ ~T +Λ
n,S
~ T · Ωn · Λ
n,D
~ n,S + Λ
~ n,D
σ
(8) impurity A composed by a single-level of energy ε σ with
an intra-site Coulomb repulsion U for electrons of different
and will correspond to cotunneling transport mechanisms. In spin orientations (n↑ and n↓ being
 the corresponding
 num-
these processes, two different charge states must be involved, DT ,r = P tr a† cσ + h.c. is associated
ber operators), and H kσr
kσ
and only two allowed possibilities exist (x = 1, y = 2 and with the tunneling between A and the electrodes, whose cou-
x = 2, y = 3) so that the restriction ax ,y,n = δfx ,n,Ay,n + δAx ,n,fy,n pling strength is Γr = 2π ρe |t r |2 , where ρe is the density of
will apply. Equations (7) and (8) express in an explicit man- states of the leads. Since the singly degenerate energy level
ner the ballistic and cotunneling contributions to the trans- can be occupied by two electrons of opposite spin orienta-
mission function, and based on this, we can calculate the tions, the corresponding Fock space of the impurity comprises
total current passing through the molecule as composed by four states of energies E 1 = 0, E 2 = E 3 = ε 0 , and E 4 = 2ε 0
two additive terms, in the form I = I bal + I cot . In Sec. SI-2 + U [Fig. 2(a)]. In this strongly correlated system, the many-
of the supplementary material, we discuss further how the body energy levels must correspond to differences between the
↔
matrix Ωn can be associated with the distinct types of transport total energies of the neutral (N), cationic (N  1), and anionic
mechanism. (N + 1) species [Fig. 2(b)].31
  ↔  
The Λ ~T +Λ ~ T · Ωn · Λ ~ n,S + Λ
~ n,D term that appears In the standard solution of the Anderson impurity
n,S n,D
in the denominator of Eq. (5) can be understood as represent- model,25,28–30 characteristic peaks in the differential conduc-
ing a multi-terminal device, with each conceptual lead being tance curve must appear at the degeneracy points between
associated with the broadening of the corresponding energy neighboring levels. Hence, the I-V curve would consist in an
level. In the total transmission T, we will consider contribu- intermediate plateau separated by two current jumps at the bias
tions from all MOs of the conceptual device whose energies values ε 0 and ε 0 + U which correspond to the opening of new
lie in the appropriate Fermi window. transport channels at the crossing points.25,28,30

FIG. 2. (a) Fock space for the differ-

ent charge states of a single Anderson
impurity. (b) Many-body level structure
of an Anderson impurity connected to
two electrodes.
194304-5 Cabrera-Tinoco, Moreira, and de Melo J. Chem. Phys. 148, 194304 (2018)

FIG. 3. The profile of the I-V curve of

an Anderson impurity at 30 K (a) and
120 K (b) as calculated in Ref. 25 is
reproduced when the GBW treatment is
used with γN2 = 0.55, γN−1
2 = 0.0, and
γN+1 = 0.45.
2

For the corresponding GBW treatment, we will assume the electric field in the (0–1.1 V) range. However, since the Gaus-
3-level structure of Fig. 2(b). Adopting the same parameters sian 03 program allows forming a grid of electric field values
ΓL = ΓR = 1 meV, ε 0 = 50 meV, and U = 100 meV as those
used in Ref. 25, we have found that when the set γN=1 2 = 0.55,
γN=0
2 = 0.0, and γN=2
2 = 0.45 of probabilities is used, the GBW
formalism is able to reproduce the profile of the I-V curve
of the standard treatment of the Anderson model at different
temperatures (Fig. 3).
There are a few points worth noticing that, such as, (i)
due to the sparse nature of the level structure of the problem
[Fig. 2(b)], at low temperatures the calculated GBW current
is dominated by the ballistic contributions; (ii) for a proper
comparison of the two current profiles, in Fig. 3, they have been
normalized to their respective maximum value; and (iii) while
in the standard Anderson model treatment the probabilities ~pst
vary with the bias value [Eq. (5)], in the GBW method here
introduced the probabilities γ are chosen a priori according
to the physical-chemical characteristics of the system studied.
Even though, it is reassuring to observe that the GBW method
allows for the correct description of the general features of
the exact solution for the charge transport through an isolated
Anderson impurity.
IV. CHARGE TRANSPORT THROUGH
A DECANEDITHIOL MOLECULE
of separated by no less than 104 a.u. steps, we have resorted
to a linear interpolation of the relevant parameters (such as
the energy and molecule-electrode couplings of the relevant
MOs) for four intermediate values between two consecutive
grid points. (In the Normalized Current vs. Voltage figures
below, the voltage values used for the Gaussian 03 calcula-
tion are indicated by dashed vertical lines, while the curves for
the current passing through these points and the interpolated
intermediate values are just guides for the eyes.)
The final mesh of calculated values will help us to better
follow the participation of the different MOs in the over-
all transport. The interpolation procedure should not be a
matter of concern, unless a special situation as avoided-
crossing of neighboring energy levels occurs (see later). In all
cases, we assumed the fixed geometry of the neutral species
since the changes in the energy spectrum due to the pres-
ence of an extra electron (or hole) are expected to occur in
a much shorter time than the subsequent geometry adjust-
ments. From these results, we extracted the energy spectrum
and the coupling parameter τ s,β as defined in Eq. (4c). The
latter, which is strongly dependent of the spatial of localiza-
tion of the corresponding molecular orbital, can be written
as
X X D E X X
τs,β = ε s cβ∗ cη φβ φη = ε s cβ∗ cη Sβη ,
Based on the GBW expression, we will model the charge β ∈(L,R) η ∈(M) β ∈(L,R) η ∈(M)
transport through a single decanedithiol molecule submitted to (9)
a potential difference applied between two terminal electrodes. ( E)
We will examine the simultaneous contributions of ballistic or
coherent processes, when the charge of the molecule remains
unaltered, and of non-coherent cotunneling mechanisms, when
one electron or hole can be exchanged with the electrodes and
where ε s , cβ ,η , and φβ , η are the energy of the molec-
ular orbital, the projections of the molecular orbital on the
atomic orbital, and the elements of the basis of the atomic
orbitals, respectively.
D E Due to the presence of the overlap
the charge state changes.
We will adopt an extended molecule model where the ter-
minal ends of the organic molecule are coupled to two 12 gold
atoms clusters that play the role of electrode fragments. The
calculations were performed at the density functional theory
(DFT) level using the Gaussian 03 program,32 with the B3LYP
hybrid functional with the 6-31G basis (d, p) adopted for the
organic molecule and the Los Alamos National Laboratory 2-
double-z (LANL2DZ) basis adopted for the gold atoms. For
a better description of the electronic rearrangements resulting
from the varying voltage, we have implemented the calcula-
tions for the electronic structure of the neutral, single-cation,
and single-anion forms of the molecule for several values of the
term Sβη = φβ φη , the coupling τ s,β is directly related
to the degree in which a given wave function will be shared
between the molecule and the metallic clusters, with its value
representing an indirect measure of the amount of spatial
localization of the sth molecular orbital. The Fermi level,
which lies midway between the HOMO and LUMO ener-
gies, will be assumed to vary linearly with the applied field
intensity = in the form ∆V = e=d, where e and d are, respec-
tively, the elementary charge and the length of the extended
molecule. Finally, since we are more interested in the gen-
eral behavior of the estimated current than on their absolute
magnitudes, in each case we will discuss the results in terms
of the normalized electrical current, i.e., the ratio between
194304-6 Cabrera-Tinoco, Moreira, and de Melo
the varying intensity of the current and its maximum value
(I max ).
V. RESULTS
A. Pure transport through the neutral form
Let us first examine the case where just the neutral form
participates in the transport process (i.e., γ22 = 1). As the data
presented in Fig. 4(a) reveal, the symmetry of some frontier
MOs of the neutral species is broken once the external electric
field is switched on. While in the absence of an applied voltage
these molecular orbitals are localized on both sides of the sys-
tem and non-vanishing couplings between the molecule and
the metallic clusters exist, only the HOMO and the LUMO
remain delocalized when the external electric field is switched
on. All other frontier MOs become localized in just one side
of the system [in Fig. 4(a), we only show the HOMO-1, as a
typical example of this behavior]. Hence, even if their ener-
gies were within the Fermi window, these MOs would not
contribute to the field-induced ballistic electron transport. In
fact, as shown in Fig. 4(b), only the energy of the delocalized
HOMO and LUMO lies between the electrode Fermi levels
(FL1 and FL2), and hence, these frontier orbitals will entirely
respond for the ballistic transport. To certify that this is indeed
the case, we artificially suppressed first the HOMO contribu-
tion (curve I (1) ) to the normalized electrical current I and then
both HOMO and LUMO contributions (I (2) ), in a procedure
that involves zeroing the corresponding coupling strengths
between the molecule and the gold clusters. As shown in
Fig. 4(c), while the current decreases considerably in the first
case, it vanishes in the latter. It is noteworthy to observe that our
generalized treatment corresponds to the usual Breit-Wigner
expression in this γ22 = 1 limiting case (see Sec. SI-3 of the
supplementary material).
B. Identical probabilities of participation
of the three charge states
Let us now examine the case where the three possible
charge states participate in the overall transport on an
equal footing, i.e., when γ12 = γ22 = γ32 = 1/3. In
Sec. SI-1 of the supplementary material, we show that
within a good approximation this situation corresponds to a
FIG. 4. (a) HOMO-1, HOMO, and LUMO electron density maps (isovalue = 0.02) for 0.00 and 0.0001 a.u. electric fields applied in the positive x-direction. (b)
HOMO-1, HOMO, and LUMO energies as a function of the voltage. The FL1 and FL2 lines represent the variation of the Fermi levels of the ideal electrodes.
(c) Normalized current for C10 when γ22 = 1. The effect of the suppression of LUMO (I (1) ) and HOMO and LUMO (I (2) ) contributions is shown.
J. Chem. Phys. 148, 194304 (2018)
Boltzmann-distribution of the energies of the three differ-
ent charge states. In this case, the maximum value of the
total electrical current (I tot ) is 1.2195 nA, a value attained at
V = 0.349 V. As expected, the overall transport has now a mixed
character, being composed by contributions from both ballistic
and co-tunneling processes (I bal and I cot , respectively)—[see
Fig. 5(a)].
1. Ballistic transport
We will first analyze the degree of participation of each
one of the three charge states in the ballistic transport through
the C10 molecule. For the neutral species, we have found that
the conclusions reached in the γ22 = 1 case remain valid; i.e.,
the spatial localization of some frontier MOs is dramatically
changed when the external electric field is switched on. Once
more, while both HOMO and LUMO continue to be delo-
calized, while the field intensity increases, all other frontier
MOs become localized in just one of the sides of the sys-
tem. Consequently, one can expect that also in this case the
ballistic transport through the neutral species would be dom-
inated by the HOMO and LUMO contributions. In fact, as
the data shown in Fig. 5(b) reveal, the ballistic current dra-
matically decreases when we artificially suppressed first the
(1) (2)
LUMO (Ibal ) and then both the LUMO and HOMO (Ibal ) par-
ticipation in the process. (In fact, as we discuss below, these
frontier MOs also play a determinant role in the cotunneling
processes.)
We observed that a similar field-induced break of sym-
metry occurs for the relevant frontier MOs of both anion
and cation forms of the C10 molecule. However, for these
charged species only the localized MOs have energies sit-
uated within the Fermi window. Hence, in spite our initial
assumption of γ12 = γ22 = γ32 = 1/3, the ballistic transport
through the C10 molecule continues to involve only the neutral
form.
2. Cotunneling contributions
One can also observe in Fig. 5(a) that for the investi-
gated range of applied bias the cotunneling terms contribute
at most with 13% of the total current at 0.349 V (although
exploratory calculations have revealed that this participation
could even become dominant for higher bias values). We
194304-7 Cabrera-Tinoco, Moreira, and de Melo
assume that two electrons are involved in a cotunneling pro-
cess. For this to occur, different charge states must exhibit
thermal-broadened resonant levels with properly disposed spa-
tial localizations. In fact, at the end, the effects of spatial
localization at a given energy range will be responsible for
the contribution of the cotunneling pathways, as a result from
the increased degree of resonance of the relevant MOs due to
thermal broadening effects (see Sec. SI-4 of the supplementary
material).
To effectively contribute to the charge transfer process,
the two resonant levels must have energies that lie inside the
pertinent Fermi window and, in addition, the corresponding
MOs must be localized in the appropriate spatial region. In
fact, while the occupied MOs in the neutral species and in
the anionic species must have a significant presence in the
left-hand side of the molecule, the unoccupied MOs of the
cationic and neutral species must be localized in its right-hand
side. Since the HOMO and the LUMO of the neutral species
are delocalized through the extended molecule, we just need to
examine the spatial localization and the overlap S between the
left [right] cluster and the (alpha and beta) MOs of the anionic
[cationic] species.
The first non-zero bias value directly obtained from our
quantum chemistry calculations corresponds to 0.159 V. In
Figs. 6(a) and 6(b), we show that at this field intensity the den-
sity of states of the anion alpha and beta molecular orbitals are
resonant (at T = 300 K) with the LUMO of the neutral species.
FIG. 6. Density of states and electron density map (isovalue = 0.02) of the anion alpha (a) and beta (b) molecular orbitals for a 0.159 V applied bias. Some
relevant values of the overlap S between the left cluster and the alpha and beta molecular orbitals of the anion form of the decanedithiol molecule are also
indicated as (a) and (b).
J. Chem. Phys. 148, 194304 (2018)
FIG. 5. Normalized electrical current
for C10 in the γ12 = γ22 = γ32 case. (a)
Total I total , ballistic (I bal ), and cotun-
neling (I cot ) currents. (b) The ballistic
current decreases when the contribution
of the LUMO of the neutral species is
(1)
suppressed (Ibal ) and vanishes if both
contributions of LUMO and HOMO of
the neutral species are simultaneously
(2)
neglected (Ibal ).
In these figures, we also indicate the corresponding electron
density maps (for an isovalue equal to 0.2) and the values of
their overlap S with neighboring gold atoms of the left cluster.
One can observe that the anions HOMOα-15, HOMOα-12,
HOMOβ-14, and HOMOβ-11 are those MOs to present a
higher spatial localization in that region, and therefore, they
appear as more likely candidates to contribute to the cotun-
neling process. A similar analysis for the cation alpha and
beta frontier MOs (see Sec. SI-4 of the supplementary mate-
rial) reveals that even after the T = 300 K broadening they
exhibit less resonance with the HOMO of the neutral species.
Hence, we can anticipate that the cation MOs should con-
tribute less to the overall transport through the decanedithiol
molecule.
In Fig. 7, we show how the cotunneling current is affected
by the suppression of the contributions of selected MOs of the
(1)
anion species: Icot , when the HOMOα-15 and HOMOβ-14
(2)
contributions are neglected, and Icot , when the HOMOα-15,
HOMOβ-14, HOMOα-12, and HOMOβ-11 contributions are
not included. In Fig. 7(a), one can observe that the anion
HOMOα-12 and HOMOβ-11 are less important to the cotun-
neling process than the anion HOMOα-15 and HOMOβ-14. In
addition, there is an essential participation of the HOMO and
LUMO of the neutral species in the non-ballistic charge trans-
fer pathways, as anticipated. The co-tunneling current van-
ishes when the participation of these two MOs is not included
(3)
(Icot ). Proceeding in a similar manner, we have found that the
194304-8 Cabrera-Tinoco, Moreira, and de Melo
elimination of the contribution of the anion MOs affects the
(4)
cotunneling current (Icot ) more than the suppression of the
cation MOs [Fig. 7(b)], a result that seems to confirm our previ-
ous analysis based on the different degrees in which the energy
of the anion and cation frontier MOs satisfies the condition of
resonance with the neutral MOs. The vanishing of the cotun-
neling current when the contribution of all anion and cation
(5)
MOs (Icot ) is eliminated just expresses the fact that at least
two different charge states must participate in a non-ballistic
process.
However, when we examined in more detail the con-
tribution of the anion MOs to the observed increase in the
co-tunneling current in the 0.159 V–0.317 V region, we
found that for this energy range avoided-crossing situations
occur for some of the neighboring energy levels of the anion
species.
3. Importance of avoided-crossing mechanisms
for the effective charge transport
When analyzing the cotunneling processes in the C10
molecule, we have seen that for voltages lower than 0.159 V the
J. Chem. Phys. 148, 194304 (2018)
FIG. 7. (a) The total cotunneling cur-
rent (I cot ) is affected by the suppression
of the contributions from specific MOs
of the anion form and from the HOMO
and LUMO of the neutral species—
(1) (2) (3) (2)
(Icot , Icot and Icot ). (b) Change of Icot
after suppression of contributions from
(4)
all anion (Icot ) MOs and from all cation
(5)
and anion MOs (Icot ).
most important contributions come from the anion HOMOα-
15, HOMOβ-14, HOMOα-12, and HOMOβ-11. However, a
peculiar feature can be identified in the curve in Figs. 5(a)
and 5(b) (where the contributions of these MOs are sup-
pressed): increments of the calculated currents are observed
in the region comprised between the 0.159 V and 0.317 V
consecutive Gaussian 03 grid points. We have found that the
explanation for this observed behavior resides in changes in
the spatial localization of the relevant MOs that occur in
that specific window of applied bias. As shown schemati-
cally in Fig. 8, where we have used a finer grid of interpo-
lated results in this bias region, the increase in the current
could be understood as resulting from avoided-crossing pro-
cesses between neighboring MOs that lead to the interchange
of their spatial localization. In Fig. 8(a), we show that the
anion HOMOβ-11 and HOMOβ-10 energies first approach
each other and then distance themselves, with a consequent
switch of the corresponding spatial localization of the MOs.
A double avoided-crossing sequence of events occurs for the
anion HOMOβ-14, HOMOβ-13, and HOMOβ-12 [Fig. 8(b)].
These spatial modifications of the charge density favor the
FIG. 8. A single (a) and a double (b)
avoided-crossing between the energies
of relevant anion MOs. Note that in each
case there is an interchange of the spatial
localization of the MOs involved.
194304-9 Cabrera-Tinoco, Moreira, and de Melo
observed charge transfer and the resulting increment in the
cotunneling current. By altering not only the localization but
also the degree of resonance with the energy levels of the neu-
tral species, these non-crossing events respond for the observed
increase of the cotunneling current increase with progressive
voltage increments.
The larger degree of resonance between the anion and
neutral eigenvalues leads to an increase in the corresponding
coupling strength [see Eq. (4b)], resulting in a higher co-
tunneling contribution. The alpha MOs have a similar behavior
to that of the beta MOs, and a single and a double avoided-
crossing also occur for them. However, since their contribution
to the overall transport is slightly smaller, we opted for dis-
cussing only the crossing situations for the beta MOs. The
occurrence of avoided crossing situations is indicative of the
degree of complexity of the “dynamics” of molecular orbital
localization when an applied external electric field acts on
the system, with the corresponding changes in behavior of
relevant MOs changes affecting the details of the transport
process.
VI. CONCLUSIONS
In this work, we first introduced a conceptual model
to treat coherent and non-coherent elastic molecular trans-
port that allows for the possibility that changes in energy of
the molecule of interest not necessarily follow a single sur-
face associated with a given valence state. By construction,
this conceptual system permits a generalization of the Breit-
Wigner formalism where three different molecular charge
states (cation, neutral, and anion) can simultaneously partici-
pate (with varying probabilities) in the charge transfer process.
Hence, ballistic and elastic cotunneling mechanisms of elec-
tron transport can contribute concomitantly to the transmission
function. To assess the general reliability of the GBW method,
we have first shown that it correctly reproduces the main phys-
ical features of the charge transport through a single Anderson
impurity. Then, as an illustrative and more realistic exam-
ple, we applied this treatment by considering the case of a
decanedithiol molecule linked to a small cluster of gold atoms
on both sides. We used a DFT treatment and considered the
fixed geometry of each charge state to calculate the eigenval-
ues and eigenvectors of the extended system at each one of the
charge states for different values of the external electric field.
In this manner, we were able to closely follow the changes
in energy and spatial localization of the relevant molecular
orbitals of all three different charged species as the external
bias is varied.
Our results indicate that while different charge transfer
mechanisms contribute to the overall charge transport, quan-
tum effects such as avoided-crossing situations between rel-
evant frontier MOs can be of special relevance. In specific
situations, the interchange of spatial localization of two MOs
involved in the crossing can open a new channel of charge
transfer that would not be otherwise available.
The charge transfer between the two electrodes may
involve ballistic and co-tunneling mechanisms. We have
shown that the charged states give a negligible contribution
for the first type of transport, in which the HOMO and LUMO
J. Chem. Phys. 148, 194304 (2018)
of the neutral species play the key role. In the pertinent liter-
ature,3,4,33 charge transport in alkane molecules is attributed
mainly to the HOMO contribution, due to its proximity in
energy to the Fermi level of gold. Our results indicate that
an important feature of the ballistic charge transfer is that in
these molecules, the spatial localizations of both the HOMO
and LUMO are not strongly affected by the applied bias, differ-
ent to what is observed for the other MOs. Consequently, both
frontier molecular orbitals could respond for the calculated
ballistic current.
Our results also indicate that co-tunneling processes have
a non-negligible contribution to the charge transport through
a single decanedithiol molecule. Our treatment corresponds to
a full quantum chemistry calculation of the molecule of inter-
est. With this procedure, we can consider the participation
of the single charged anion and cation species. Two different
charge states must be involved in an elastic co-tunneling pro-
cess. We have shown that this becomes possible whenever a
MO of the cation or anion species, even if localized in just
one side of the extended molecule, is resonant with a delo-
calized MO of the neutral species. Under these conditions, a
conduction channel is formed throughout the entire system, in
a process that is controlled by the degree of resonance between
the MOs involved. We should point out that charging effects
are seldom considered in theoretical descriptions of molecular
transport.
It is in this regard that in specific situations the occur-
rence of avoided-crossing situations seems crucial in allowing
an increase in the degree of resonance between the energies
of MOs belonging to different charge states as the inten-
sity of the applied field is progressively increased. For the
decanedithiol molecule, we have shown that the main con-
tribution results from the interaction between energy levels
of the anion and neutral forms. In spite of the dramatic
effect of the symmetry swapping of the involved MOs that
accompanies the avoided-crossing event, this quantum phe-
nomenon has been only sparsely discussed in the pertinent
literature.34
We have presented here an intuitive method of gen-
eralizing the Breit-Wigner equation. In future studies, we
intend to present a more formal discussion of the generalized
Breit-Wigner treatment and examine charge transport through
distinct types of organic molecules.
SUPPLEMENTARY MATERIAL
See supplementary material for more details of the con-
tribution of the thermal broadening to the increase of the total
↔
scattering rate, the physical interpretation of the matrix Ωn ,
the limiting behavior the generalized transmission function,
and the conditions that the cation and anion MOs must satisfy
for effectively contributing to the cotunneling charge transfer
processes.
ACKNOWLEDGMENTS
This work was supported by CNPq and by the INFO
National Institute. H. A. Cabrera-Tinoco would like to thank
CNPq for a Ph.D. fellowship.
194304-10 Cabrera-Tinoco, Moreira, and de Melo
1 L. Sun, Y. A. Diaz-Fernandez, T. A. Gschneidtner, F. Westerlund,
S. Lara-Avila, and K. Moth-Poulsen, Chem. Soc. Rev. 43(21), 7378–7411
(2014).
2 G. Fagas, P. Delaney, and J. C. Greer, Phys. Rev. B 73(24), 241314 (2006).
3 K. H. Müller, Phys. Rev. B 73(4), 045403 (2006).
4 C. Li, I. Pobelov, T. Wandlowski, A. Bagrets, A. Arnold, and F. Evers,
J. Am. Chem. Soc. 130(1), 318–326 (2008).
5 M. Paulsson, C. Krag, T. Frederiksen, and M. Brandbyge, Nano Lett. 9(1),
117–121 (2009).
6 A. Sen and C.-C. Kaun, ACS Nano 4(11), 6404–6408 (2010).
7 R.-W. Yan, X. Jin, S.-Y. Guan, X.-G. Zhang, R. Pang, Z.-Q. Tian, D.-Y. Wu,
and B.-W. Mao, J. Phys. Chem. C 120(22), 11820–11830 (2016).
8 M. T. González, J. Brunner, R. Huber, S. Wu, C. Schönenberger, and
M. Calame, New J. Phys. 10(6), 065018 (2008).
9 M. M. Thuo, W. F. Reus, C. A. Nijhuis, J. R. Barber, C. Kim, M. D. Schulz,
and G. M. Whitesides, J. Am. Chem. Soc. 133(9), 2962–2975 (2011).
10 J. R. Niskala, W. C. Rice, R. C. Bruce, T. J. Merkel, F. Tsui, and W. You,
J. Am. Chem. Soc. 134(29), 12072–12082 (2012).
11 E. Pires, J. E. Macdonald, and M. Elliott, Nanoscale 5(19), 9397–9403
(2013).
12 W. R. French, C. R. Iacovella, and P. T. Cummings, ACS Nano 6(3),
2779–2789 (2012).
13 K. Wang, J. Hamill, J. Zhou, C. Guo, and B. Xu, Faraday Discuss. 174,
91–104 (2014).
14 S. W. Wu, N. Ogawa, G. V. Nazin, and W. Ho, J. Phys. Chem. C 112(14),
5241–5244 (2008).
15 L. Xiang, J. L. Palma, C. Bruot, V. Mujica, M. A. Ratner, and N. Tao, Nat.
Chem. 7(3), 221–226 (2015).
16 F. Schwarz, G. Kastlunger, F. Lissel, C. Egler-Lucas, S. N. Semenov,
K. Venkatesan, H. Berke, R. Stadler, and E. Lörtscher, Nat. Nanotechnol.
11(2), 170–176 (2016).
17 S. De Franceschi, S. Sasaki, J. M. Elzerman, W. G. van der Wiel, S. Tarucha,
and L. P. Kouwenhoven, Phys. Rev. Lett. 86(5), 878–881 (2001).
18 M. Buttiker, IBM J. Res. Dev. 32(1), 63–75 (1988).
19 A. Migliore and A. Nitzan, ACS Nano 5(8), 6669–6685 (2011).
20 A. Migliore and A. Nitzan, J. Am. Chem. Soc. 135(25), 9420–9432
(2013).
21 W.-H. Zhu, G.-H. Ding, and B. Dong, Nanotechnology 25(46), 465202
(2014).
22 S. Jan van der Molen and P. Liljeroth, J. Phys.: Condens. Matter 22(13),
133001 (2010).
J. Chem. Phys. 148, 194304 (2018)
23 R. Tuerhong, S. Kezilebieke, B. Barbara, and J.-P. Bucher, Nano Res. 9(10),
3171–3178 (2016).
24 A. Migliore, P. Schiff, and A. Nitzan, Phys. Chem. Chem. Phys. 14(40),
13746–13753 (2012).
25 J. C. Cuevas and E. Scheer, Molecular Electronics: An Introduction to
Theory and Experiment (World Scientific, Singapore; Hackensack, NJ,
2010).
26 G. Breit and E. Wigner, Phys. Rev. 49(7), 519–531 (1936).
27 C. Brooke, A. Vezzoli, S. J. Higgins, L. A. Zotti, J. J. Palacios, and
R. J. Nichols, Phys. Rev. B 91(19), 195438 (2015).
28 M. R. Wegewijs, M. H. Hettler, C. Romeike, A. Thielmann, K. Nowack
and J. König, in Introducing Molecular Electronics, edited by G. Cuniberti,
K. Richter, and G. Fagas (Springer Berlin Heidelberg, Berlin, Heidelberg,
2005), pp. 207–228.
29 A. Thielmann, M. H. Hettler, J. Konig, and G. Schon, Phys. Rev. B 68(11),
115105 (2003).
30 A. Thielmann, M. H. Hettler, J. Konig, and G. Schon, Phys. Rev. Lett.
95(14), 146806 (2005).
31 B. Muralidharan, L. Siddiqui, and A. W. Ghosh, J. Phys.: Condens. Matter
20(37), 374109 (2008).
32 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb,
J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J.
C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci,
M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada,
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima,
Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski,
P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V.
G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D.
K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz,
Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu,
A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith,
M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M.
W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople,
gaussian 03, Revision C.02, Gaussian, Inc., Wallingford CT, 2004.
33 W. Haiss, S. Martin, L. E. Scullion, L. Bouffier, S. J. Higgins, and
R. J. Nichols, Phys. Chem. Chem. Phys. 11(46), 10831–10838
(2009).
34 C. P. de Melo and A. C. L. Moreira, J. Phys. Chem. C 116(4), 3122–3131
(2012).