Topological junction states in graphene

nanoribbons: A route to topological chemistry

arXiv:2412.17949v1 [cond-mat.mes-hall] 23 Dec 2024

Hazem Abdelsalam,∗,†,‡ Domenico Corona,∗,¶ Renebeth B. Payod,∗,§ Mahmoud

A. S. Sakr,∗,∥ Omar H. Abd-Elkader,∗,⊥ Qinfang Zhang,∗,† and Vasil A.

Saroka∗,¶,#,@

†School of Materials Science and Engineering, Yancheng Institute of Technology, Yancheng

224051, P.R. China
‡Theoretical Physics Department, National Research Centre, Giza12622 Dokki, Egypt
¶Department of Physics, University of Rome Tor Vergata and INFN, Via della Ricerca
Scientifica 1, 00133 Roma, Italy
§Institute of Physics, University of the Philippines, College, Baños, Laguna 4031
Philippines
∥Center of Basic Science (CBS), Misr University of Science and Technology (MUST), 6th
October City, Egypt
⊥Department of Physics and Astronomy, College of Science, King Saud University, P.O.
Box 2455, Riyadh 11451, Saudi Arabia
#TBpack Ltd., 27 Old Gloucester Street, London, WC1N 3AX, United Kingdom
@Institute for Nuclear Problems, Belarusian State University, Bobruiskaya 11, 220030
Minsk, Belarus

E-mail: hazemabdelhameed@gmail.com; domenico.corona@roma2.infn.it; rbpayod@up.edu.ph;

mahmoud.sakr@must.edu.eg; omabdelkader7@ksu.edu.sa; qfangzhang@gmail.com;
vasil.saroka@roma2.infn.it

1
 Abstract

Two-dimensional topological insulators with propagating topological edge states

are promising for dissipationless transport, while their one-dimensional analogs are

capable of hosting localized topological junction states that are mainly envisaged for

quantum computing and spintronics. Here, in contrast, we propose to use localized

nature of topological junction states for sensing applications. We report a systematic

topological classification of a wide class of graphene nanoribbons represented by already

synthesized extended chevron species. By using this classification, we theoretically

model a double-junction transport device that shows enhanced interaction with NO2

molecule. Our results show that topological junction states of nanoribbons can open

an avenue for topological sensing and junction-assisted chemistry applications.

Introduction

The exploration of topological states in condensed matter physics has revolutionized our un-
derstanding of electronic materials, 1–3 opening new pathways for advanced technological ap-
plications. 4 Among the myriads of topologically intriguing systems, 5–7 carbon-based nanos-
tructures played a key role from the very beginning. 8,9 In recent years, particularly graphene
nanoribbons (GNRs) have garnered significant attention. 10–22 These one-dimensional (1D)
nanostructures, derived from graphene, exhibit unique electronic properties due to their
distinct edge terminations and narrow widths. 23,24
Bulk-boundary correspondence principle is at the heart of the topological band theory of
solids. It states that, at the interfaces between regions of different topological orders, topo-
logical interface states occur as highly localized electronic modes that are resilient to disorder
or perturbations. 1 The topological stability of these interface states can be controlled by sev-
eral topological invariants. In the case of the integer quantum Hall effect 25,26 or Haldane
model, 27 the invariant is the Chern number, while it is Z2 invariant instead for topological
insulators. 8,9 The Z2 invariant provides the binary classification of topological phases with

2
values of 0 or 1. A wider classification based on the Chern number utilizes the whole set
of integer numbers Z. In both cases, a zero value stands for a trivial topological order. An
interface state appears, whenever two regions with different Z or Z2 values meet each other.
This basic principle works in all dimensions and has recently attracted a lot of attention
for 1D structures, such as graphene nanoribbons, where topological junction states (TJS)
have been revealed and classified. 10,11,17,18 The classification of TJS in GNRs can be carried
out base on Z2 invariant 10–12 and also can be generalized by incorporating chiral symmetry
and spin corrections. 17 Some of these states have already been experimentally shown and
investigated through an atomically precise on-surface bottom-up engineering. 14,15,21,28
The envisaged applications of topologically protected states are (i) disorder-resilient low-
dissipation transport and (ii) robust long-coherence spin states. 2,4 The latter has become a
roadmap for topological junction states as potential candidates for designing customizable
spintronics 10,13,20 and quantum computing 13,20 nanodevices. Notably, this widely accepted
roadmap overlooks chemical sensing, which is arguably a more natural and immediate ap-
plication. Chemical sensing relies critically on the interaction between the sensor material
and the target analyte, where sensitivity and selectivity are paramount. 29 The 1D nature of
GNRs, combined with topological junctions providing a high density of localized electronic
states to facilitate interaction with chemical species, potentially offers enhanced sensitivity
and selectivity that is accessible for readout in a standard transport device similar to a
field-effect transistor. 30,31
This paper presents a theoretical investigation of topological junction states in GNRs,
with a particular focus on their application in gas sensing. The gas sensing application
has already been proposed and investigated both theoretically 32–35 and experimentally 36,37
for the edge and end states of GNRs; see also a review of more general sensing schemes in
Ref. 31. Some experimental studies have demonstrated the detection of volatile compounds
such as ethanol and methanol 36 as well as nitrogen dioxide NO2 . 37 However, the sensitiv-
ity of TJS in GNRs remains unknown. Here, we explore the electronic properties of these

3
states, their formation mechanisms, and their interaction with NO2 as the target molecule.
We have noticed that some of the ribbons that we have studied previously 34 (we refer to
those as A60 within a more general class of zigzag-shaped GNRs 38–41 ), are, in fact, topolog-
ically non-trivial. For some combinations of their unit cell structural parameters, they are
characterized by the Z2 invariant derived from the intercellular Zak phase 42,43 calculated in
the periodic gauge of the tight-binding (TB) Hamiltonian. 3 Thus, they can form TJSs via
seamless combination with a topologically trivial unit cells of armchair GNRs (AGNRs). 10
Performing density functional theory (DFT) calculations to study sensing, we focus on ni-
trogen dioxide (NO2 ) gas, which is a free-radical and a moderate Lewis acid. NO2 is an
air pollutant affecting the environment and health of people 44,45 and a prototype of a vast
range of substances belonging to nitrocompound group, e.g. nitroparaffins and nitroarenes.
By leveraging the unique properties of topological junctions in GNRs, we aim to elucidate
their potential in creating highly sensitive and selective chemical sensors, paving the way for
advancements in environmental monitoring, medical diagnostics, and beyond.

Results and discussion

Topological properties

We begin by verifying the TJS theory at the level of nearest-neighbor TB model and by
constructing novel junctions. Then we will proceed to the electronic properties investigation
within cluster approach of the TB model for the range of constructed junctions.
The TB Hamiltonian of π-electron network is

X X  
H= εc†i ci + t1 c†i cj + h.c (1)
i i,j

where c†i and ci are electron creation and annihilation operators, respectively, ε = 0 is the
on-site energy equal for all lattice sites, t1 = 3.12 eV 46 is the nearest-neighbor hopping

4
integral. We note here that t1 varies for different structures, 47,48 therefore, we will use E/t1
dimensionless units in what follows. By going into reciprocal k-space in Eq. (1) via a Fourier
transform, c†i = √1N k e−ikRi c†k (and similar for ci ) with N being the number of unit cells
P

and Ri being the site position, we solve the eigenproblem of n × n matrix Hamiltonian
for a given unit cell of a GNR with n lattice sites in the cell. Then k-space Hamiltonian
eigenvectors, Ci (k) = (Ci,1 (k), Ci,2 (k), . . . , Ci,n (k))T , can be employed for the calculation of
the intercellular Zak phase by Eq. (31) in Ref. 42:

( Nk
)
Y  † 
γ2 = −Im log Det Cp (kj )Cq (kj+1 ) , (2)
j=1

where Nk is the number of k-point sampling the Brillouin zone of the GNR, and the indices
p and q run from 1 to n/2 so that dot products of eigenvectors in Eq. (2) form a matrix
from all occupied states. Here we present the explicit ready-to-use formula for the TB
R
calculations instead of a more common total Zak phase: 10–12,49 γ = i BZ ⟨uik |∂k uik ⟩ with
uik being periodic parts of the Bloch functions. The common definition of the total Zak
phase has several practical issues such as irregularity of the complex phase in numerical
eigenvectors, peculiarities of valence band crossings, and dependence on the real space origin
position and gauge choice.
A few comments are worth making before we proceed. First, Eq. (2) must be evaluated
on a closed path in k-space, i.e. on a loop, leading to the same values of k1 and kNk +1 .
Second, it is also important to impose in Eq. (2) a periodic gauge on the eigenvectors of the
Hamiltonian: 42 Ci (k + 2π/T ) = Ci (k), where T is the translation period of a GNR. Most
naturally, this periodic gauge condition is achieved by constructing a periodic Hamiltonian
in k-space, i.e. by employing a periodic gauge Hamiltonian based on so-called basis I, see
Sec. 4.2.5 in Ref. 3 and Ref. 50. Satisfying the two conditions above, the intercellular Zak
phase in Eq. (2) is guaranteed to be a well-defined modular 2π and independent of the
position of the real space origin. This Zak phase, however, depends on the geometry of GNR

5
unit cell. In this study, the actual Hamiltonian is constructed using TBpack Mathematica
application 51 updated with the periodic gauge. We have verified that Eq. (2) reproduces the
proposed classification 10 of semiconducting AGNR(Np ), where Np is the number of carbon
atom pairs in the unit cell, when all conditions mentioned are met.
Having equipped with the tools presented above, we can now define Z2 = abs[mod(γ2 , 2π)],
where ‘abs’ stands for the absolute value, which is needed because the sign of γ2 is not fixed.
The bulk-boundary correspondence for TJS can be verified by constructing finite-size sys-
tems that host junctions between (i) two topological unit cells both characterized by Z2 = 1,
and (ii) one topological unit cell Z2 = 1 and another trivial unit cell Z2 = 0. Taking a
unit cell of an AGNR(Np ), we can seamlessly connect it with some other ribbon. Let us
take a chevron GNR (cGNR) and AGNR(7) that both have been synthesized by bottom-up
self-assembling method. 52 As shown in Refs. 10,11, the unit cells of cGNR and AGNR(7)
can be chosen so that they are either topologically non-trivial, Z2 = 1, or not, Z2 = 0. We
can take cGNR unit cell to be topological, then its connection with a trivial AGNR(7) unit
cell must result in a TJS, while a combination with a non-trivial AGNR(7) unit cell must
lead to the absence of a TJS. Having connected the two unit cells, we can translate them in
opposite directions away from the junction to obtain a finite size system, see Cases 1 and 2
in Table 1. The electronic properties of such a system can be derived from the real space
Hamiltonian given by Eq. (1). By employing this cluster approach within the TB model,
we are able to track what is happening not only at the junction between the two ribbons
but also at the system ends which interface with the vacuum characterized by Z2 = 0. For
example, when two topological unit cells are linked in the junction, a pair of topological end
states (TESs) can arise at the border of each of the ribbons with a trivial vacuum.
Next we consider combinations of AGNR(Np ) with less known ribbons of A60 class. 38,41
Generically, these ribbons are characterized by a set of structural parameters (ℓ1 , ℓ2 ; w1 , w2 ),
where ℓ1,2 stands for arm indices, and w1,2 sets the width vector. 41 We can stick to the
mirror symmetric case such that ℓ1 = ℓ2 = ℓ and w1 = w2 = w, see Scheme 1. Within

6
Table 1: Junctions between various types of GNRs. Light green marks topological struc-
tures, while light blue and green-blue - trivial and undefined, i.e. “metallic”, structures,
respectively. Black dot is the alignment point for a pair of unit cells forming a junction.
Crimson dots show removed dangling atoms from the structure.

Case Unit cell 1 Unit cell 2 Junction

cGNR AGNR(7) cGNR-AGNR(7)
1a Z2 = 1 Z2 = 0

cGNR AGNR(7) cGNR-AGNR(7)

a
2 Z2 = 1 Z2 = 1

A60(2, 2, 4, 4) AGNR(7) A60(2, 2, 4, 4)-AGNR(7)

3 Z2 = 1 Z2 = 0

A60(2, 2, 4, 4) AGNR(7) A60(2, 2, 4, 4)-AGNR(7) modified ends

4 Z2 = 1 Z2 = 0

A60(2, 2, 3, 3) AGNR(5) A60(2, 2, 3, 3)-AGNR(5)

5 Z2 = 1 Z2 = {0, 1}b

AGNR(5) AGNR(5) AGNR(5)-AGNR(5)

6c Z2 = 0; tedge < t1 Z2 = 1; tedge > t1

a
Junctions previously reported, see Fig. 3(b) in Ref. 11; b Ambiguity of Z2 with respect to
applied strain has been reported in Ref. 53; see, for example, Fig. 3(c); Here the structure
is considered to be gapless, i.e. metallic one; c This junction is obtained by reducing model
of Ref. 53, that accounts up to third 7nearest neighbor hopping interactions.
 6
5 4

4
3
3
2 2

1
1

Scheme 1: The main structural parameters of a mirror symmetric A60 class of GNRs: ℓ1 =
ℓ2 = ℓ and w1 = w2 = w. A60(ℓ1 , ℓ2 , w1 , w2 ) =⇒ A60(4, 4, 6, 6) is taken as an example. It
was synthesized in Ref. 36.

this classification, the extended chevron GNR 36 shall be referred to as A60(4, 4; 6, 6). In
Table 2, we present the topological properties of A60 ribbons for various values of their
structural parameters ℓ and w. Table 2 provides references, where few species of this class
were studied experimentally or theoretically. From Table 2, one sees that A60(2, 2; 4, 4) with

Table 2: Z2 topological invariant of A60(ℓ1 , ℓ2 ; w1 , w2 ) GNRs as a function of their structural

parameters ℓ = ℓ1 = ℓ2 and w = w1 = w2 . Ligthgray: A60 GNRs used further in this study.
Footnotes: species previously studied in some aspects.

ℓ \w 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
2 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1
3 1c 1 0 0d 0 1 1 1 0 0 0 1 1 1 0 0 0 1
4 1 1a 0 0b 0 1 1 1 0 0 0 1 1 1 0 0 0 1
5 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1
6 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0 0 0 1
7 0 1 0 0 0 1 1 1 0 0 0 0 1 1 0 0 0 1
8 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
9 0 1 0 0 0 1 0 1 0 0 0 0 1 0 0 0 0 0
10 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
a b
Fig. 2 in Ref. 38; Fig. 2(b) in Ref. 11 and Fig. 1 in Ref. 36; c Fig. 2(a,d) in Ref. 34 ; d

Fig. 2(b,e) in Ref. 34.

Z2 = 1 can be combined with a trivial AGNR(7) with Z2 = 0 to construct a topological

junction. Translating these unit cells into opposite directions, we obtain a finite-size system,
as presented by Case 3 in Table 1. Note, however, that ends of the finite system possess
carbon atoms with a coordination number equal to 1. In on-surface synthesis, such atoms

8
will most likely turn into methyl groups with sp3 hybridization, in contrast to other atoms
preserving the sp2 hybridization of graphene sheet. In this case, the C-C single bond linking
the methyl group to the rest of the structure does not contribute to the π-electrons network
captured by Eq. (1). That is why we also investigated a structure that is striped from
dangling carbon atoms as shown in Case 4 of Table 1.
Within the nearest neighbor TB model, AGNRs described by Np = 3p + 2, with p being
an integer, are metallic . 53,54 Strictly speaking Eq. (2) does not give reliable results in this
case since is ill-defined for metallic AGNRs. Nevertheless, geometrically nothing prevents us
from connecting a metallic AGNR with a topological A60, and this is indeed an instructive
example to consider without any reference to the topological band theory. The Case 5 in
Table 1 shows a junction between the metallic AGNR(5) and the topological A60(2, 2; 3, 3).
Beyond the simplest TB model the AGNR(5) has a tiny gap due to edge relaxation. 53,54
Once such relaxation is modeled by setting the hopping integrals at the edges tedge > t1 ,
Eq. (2) gives Z2 for gapped AGNR(5) in agreement with the analytical formulae reported
in Ref. 10. When relaxation accounted for as small compression of the C-C edge bond is
replaced by stretch of the edge bond, tedge < t1 , the Z2 invariant changes its value according
to findings of Ref. 53. The same happens for other metallic and gapped AGNRs. Thus, by
varying the edge strain, it is possible to construct topological junctions between unit cells of
the same shape, as presented by Case 6 in Table 1. The minimal edge strain could probably
be achieved even by functionalization of metallic AGNRs with different chemical species
along the edges, but for now metamaterial engineering 55–57 seems to be a more viable route
towards the edge-strain-based topological junctions.
Figure 1 summarizes the results for the main types of nanoribbon junctions presented in
Table 1 and the states observed for them. As one can see from Figure 1a, describing the
finite-size structure of Case 1 in Table 1, the zero energy hosts four states, two of which, i.e.
No. 195 and 198, are localized at the junction region marked with a vertical dashed black line.
These states also have some weight at the trivial end of the structure. This feature is missing

9
in periodic structure calculations and never mentioned in Ref. 11. The two other states, the
highest occupied molecular orbital (HOMO) No. 196 and lowest unoccupied molecular orbital
(LUMO) No. 197, are the end states. The state No. 196 is a TES localized at the border
of the topological cGNR unit cell and the trivial vacuum, because it can be explained by
the bulk-boundary correspondence. In contrast, the state No. 197 unexpectedly forms at
the interface between trivial AGNR(7) unit cell (light blue color) and the trivial vacuum
which must be an ordinary end state related to dangling carbon atoms as we shall see later.
Figure 1b shows the energy levels and the wave functions of the cGNR-AGNR(7) junction,
which is Case 2 in Table 1. In this structure, there are only two zero-energy states, which are
end states. These HOMO and LUMO states No. 196 and 197, respectively, are both localized
at the interface between a topological unit cell and a trivial vacuum. Thus, these states are
predicted by the bulk-boundary correspondence principle so that they are both TESs. For
comparison, the two bulk states are also presented in Figure 1b. Analogous relations can be
noticed in Figure 1c corresponding to Case 3 in Table 1. Here we show electronic energy levels
and wave functions for A60(2, 2, 4, 4)-AGNR(7) junction between topological and trivial unit
cells, respectively. Similar to Figure 1a, along with the two TJSs, No. 165 and 168, having
some weight at the end of the structure, there are states No. 166 and 167 localized entirely
at the ends of the structure. To verify the hypothesis that they are merely related to the
peculiar dangling C atoms, we constructed Case 4 structure with 5 atoms removed as shown
in Table 1. Figure 1d confirms that only one TJS remains after all the dangling C atoms
have been removed. Finally, we present the results for the controversial Case 5. We note
that Eq. (2) applied to the AGNR(5) yielded Z2 = 1. Despite the appeal to interpret the
results in Figure 1e with the bulk-boundary correspondence as in the Case 2 discussed with
Figure 1b, this approach will not work for other “metallic” structures. Due to this reason,
the unit cell of AGNR(5) is highlighted with a blend between light green and light blue
colors. On the other hand, when the metallic structure has a well-defined gap due to the
edge strain as in Case 6, then a localized TJS is clearly identifiable, as one can see from

10
Figure 1(f).
In this section, we have seen that topological band theory allows us to predict and reliably
construct localized junction states. From a chemistry point of view, these nucleophilic regions
with excess electrons due to TJSs are similar to radicals and they must be perfect reaction
sites. The geometry of the junction compatible with a standard transport experiment and
radical-like nature of the TJS must make them useful for sensing application. Next we
investigate this idea at an advanced quantum chemistry level.

Topological sensing

The above presented junctions are all asymmetric. The asymmetry of the junctions may pose
a significant problem for potential transport measurements by requiring differing fabrication
approaches of the left and right leads. This problem could be bypassed in a mirror symmetric
double junction device (DJD). As shown in Scheme 2, a DJD has a few topological unit cells
of an A60 GNR sandwiched between trivial AGNRs.

(iii)
(i)
(ii)

Scheme 2: The double junction device scheme. Following notation used in Table 1, the light
green unit cells of A60(2, 2, 4, 4) are topological ones, while light blue unit cells of AGNR(7)
are trivial. The alignment points for the two junctions are depicted with black dots, while the
junctions are marked with two dashed black lines. The dangling C atoms that are excluded
from π-electron network owing to their conversion into methyl groups are shown as crimson
dots. The three circles highlight the chosen site for investigating adsorption properties: (i)
violet - AGNR(7), (ii) dark blue - A60(2, 2, 4, 4), and (iii) orange - one of the two topological
junctions. Hydrogen atoms are not displayed.

Using DFT calculations as implemented in Gaussian 16 58 and post-processing with Multi-

wfn software 59 for partial density of states (PDOS) extraction, we scrutinized the electronic

11
 (a) (b) (c)

(d) (e) (f)

Figure 1: Electronic properties of topological junctions. (a) Energy levels for cGNR-
AGNR(7) topological-trivial junction. Vertical orange and violet lines mark HOMO and
LUMO, respectively. Dashed black circles highlight TJSs, gray circles highlight end states.
The wave functions of the numbered states are presented below, where the scaling factors
together with the state energies in terms of E/t1 are given. Dark red and green show the
positive and negative phase of the wave function, respectively. Vertical dashed black lines
denote the junction. Black dot is the alignment point for the two underlined unit cells.
Light green and light blue highlight topological and trivial unit cells, respectively. (b) Same
as (a) but for cGNR-AGNR(7) topological-topological junction. (c) Same as (a) but for
A60(2, 2, 4, 4)-AGNR(7) topological-trivial junction. (d) Same as (a) but for A60(2, 2, 4, 4)-
AGNR(7) junction with modified ends. (e) Same as (a) but for A60(2, 2, 3, 3)-AGNR(5)
ill-defined junction. The green blue color highlights the “metallic” unit cell. (f) Same as (a)
but for the edge-strain-induced junction in AGNR(5): tedge = 0.5t1 to the left and 1.5t1 to
12
the right of the junction.
and adsorption properties of the DJD. The B3LYP hybrid functional, 60,61 combined with
the 6-31G basis set, 62,63 is chosen for modeling due to its efficiency and accuracy with re-
spect to the electronic properties of carbon-based materials. 64–66 Grimme’s Van der Waals
correction 67 is applied to the B3LYP functional to account for long-range interactions with
the target NO2 molecule. In the constructed device, the dangling C atoms are properly
hydrogenated to form methyl groups, therefore, we expect one TJS per junction similar to
Figure 1d. Since two junctions are placed close to each other in our DJD, the two TJS
can slightly overlap. This interaction between TJSs shall result in their symmetric splitting
around the Fermi energy, Ef = 0 eV, similar to what happens with the ground state of a
double quantum well. 68
As one can see from Figure 2(a), the DFT modeling confirms the expected ground state
splitting. The DJD HOMO and LUMO energies are separated by the gap Eg = 1.08 eV.
The distributions of the HOMO and LUMO are clearly different from those of the bulk
orbitals. While bulk states extend over a large number of trivial AGNR(7) unit cells, the
HOMO and LUMO representing TJSs are sharply localized at the two A60(2, 2, 4, 4) unit
cells in the region between the two junctions. These topological states are expected to be
more interactive than the bulk ones. To compare TJSs to other energy states in the DJD,
the reactivity is tested by studying the adsorption properties of the NO2 gas molecule on
different sites of the DJD. In this testing, the NO2 is initially positioned at 3 Å above
the DJD at three sites: (i) the AGNR, (ii) the A60 GNR, and (iii) the junction. Then
the whole structure is optimized so that NO2 ends up at the sites illustrated in Scheme 2.
The three chosen adsorption configurations cover reduced electron densities away from the
junction and increased electron density due to TJS in the junction region. The adsorption
energies in the first two cases are comparable, namely −0.27 and −0.30 eV, respectively. In
contrast, the adsorption energy at the junction site is greater than double of these values,
i.e. Ea = −0.66 eV. This indicates that TJSs indeed boost interaction with the target
molecule. The bond distance analysis suggests that the interaction is mediated by oxygen

13
atoms. Adsorption of NO2 is accompanied by the electron charge transfer of −0.54e Mulliken
charge from the junction to NO2 molecule. Before the adsorption, the O-atom charge in NO2
molecule is −0.19e for both oxygen atoms, while N-atom charge is 0.38e. However, these
charges decrease to −0.41e and −0.35e for oxygen atoms and 0.22e for the nitrogen atom
after adsorption at the junction site. Interatomic N-O distances are 1.19 Å and the bond
angle is 134.5◦ before the adsorption. These distances elongate to 1.30 Å and 1.27 Å after
adsorption and the angle reduces to 118.7◦ . For comparison, adsorption of NO2 on the
A60 and AGNR sites results in −0.19e and −0.11e Mulliken charge transfers, respectively.
Therefore, physical adsorption at the junction site is stronger than on the other two sites.
Figure 2(b) shows spin-polarized energy levels for the system, wherein NO2 is adsorbed
on the junction spot of the DJD. The main effect of the NO2 adsorption is induced spin-
splitting of the TJSs, both HOMO and LUMO, and the resulting band gap reduction for
spin-down component: Eg,↓ = 0.51 eV, as compared to Eg,↑ = 1.04 eV and Eg = 1.08 eV
of a non-magnetic case in Figure 2(a). As seen from the HOMO spin-down component, the
contribution of NO2 into the spin-down component of the bonding HOMO is significant in the
case of adsorption at the junction which is compelling evidence of the localized TJSs being
more reactive than extended bulk states. The partial densities of states plotted in Figure 2(c)
and (d) further demonstrate that NO2 contributions (dark blue and violet) into HOMOs are
almost vanishing when the adsorption takes place at A60(2, 2, 4, 4) and AGNR(7) sites even
though the energy gaps are comparable to the junction case: Eg,↓ = 0.71 eV and 0.38 eV and
Eg,↑ = 1.07 eV and 1.09 eV, respectively. This proves that TJSs are suitable for facilitating
sensing and catalytic applications.

Sensor read-out with coherent transport

The constructed symmetric DJD has a favorable configuration for connecting leads to the
device and passing a coherent current through it. This current must be sensitive to the
electronic properties of the scattering region represented by two junctions between topological

14
 (a) (b)

HOMO-1 HOMO-UP

HOMO LUMO-UP

LUMO HOMO-DOWN

LUMO+1 LUMO-DOWN

(c) (d)

Figure 2: Interaction with NO2 gas molecule. (a) The electronic energy levels of pristine
DJD together with molecular orbitals of TJS and extended bulk states. Isovalue: 0.02.
Similar to Fig. 1 vertical orange and violet lines mark HOMO and LUMO, while dashed
black circles highlight TJSs. (b) The spin-polarized electronic energy levels for DJD after
adsorption of NO2 molecule at the A60(2, 2, 4, 4)-AGNR(7) junction between topological and
trivial unit cells. (c) The spin-resolved partial density of states for DJD and NO2 molecule
adsorbed at A60(2, 2, 4, 4). The NO2 data are plotted by the same dark blue color that is
used to denote NO2 adsoption site in Scheme 2. (d) Same as (c), but for NO2 adsorbed at
AGNR(7). Broadening: 0.1 eV. The frame colors in (b), (c), and (d) correspond to those of
15
circles showing the adsoption sites in Scheme 2.
and trivial nanoribbon unit cells. The changing current shall be a signature of the physical
adsorption happening at the junction. The geometry of the leads is crucial here. In order
to facilitate electron injection into the semiconducting structure, we use a combination of
topological and chemical engineering. Specifically, we terminate the scattering region with
topologically non-trivial AGNR(7) unit cell that has been used in Case 2 of Table 1. In
addition, we dope the left and right leads with N-atoms as shown in Figure 3a. The latter
may be a bit challenging step from a technological point of view, though the atomically
precise doping of GNRs has been demonstrated 69 and N-C interfaces have already been
studied experimentally. 70
The quantum transport through the lead-DJD-lead system is explored using DFT and
non-equilibrium Green’s function methods implemented in NanoDCAL software. 71,72 The
local density approximation (LDA) is used for the exchange-correlation functional with an
energy cutoff of 160 Ry. The k-points grid is set to 1 × 1 × 100 for the leads and 1 × 1 × 30
for the Brillouin zone. The current passing through the system is determined by Landauer-
Büttiker formalism. 73 Thus, we sum the transmission probabilities, T = i Ti , for electron
P

states, i, from one electrode to the other within the energy window subjected to the applied
voltage V :
2e
I= T (µL − µR ) , (3)
h

where µL − µR = |eV |, and the µL and µR are the electrochemical potentials of the left and
right electrodes, respectively.
In Figure 3b, the current-voltage (I-V ) characteristics for the DJD-based sensor are
presented within a reasonable range of voltages from 0 to 1.6 V. The modeled sensor demon-
strates measurable currents of about tens of µA’s. Due to the optimal ratio between the
AGNR(7) and A60(2, 2, 4, 4) fragments and the lead engineering, the device is conducting.
Longer segments of AGNR(7) show lower conductance in our quantum transport simula-
tions. The adsorption of NO2 at the favorable junction site results in a twofold increase in
current. Simultaneously, the NO2 adsorption at energetically less favorable sites of AGNR(7)

16
and A60(2, 2, 4, 4) blocks the current through the sensor. The presented detection mecha-
nism differs from that of the previously proposed junction-free AGNR-based sensors. 74 In
our case, the adsorption of the molecule at the junction increases the current, which results
in an on-off ratio that is greater than unity, whereas in other schemes the current is usually
blocked.

(a)

(iii)
(i)
(ii)

(b)

(iii)

(i) (ii)

Figure 3: Quantum transport readout of gas molecules. (a) The scheme of the quantum
transport sensor. Notations are the same as in Table 1 and Scheme 2. The red dots de-
note nitrogen atoms. ‘L’, ‘S’, and ‘R’ are the left lead, scattering region, and right lead,
respectively. (b) The current-voltage characteristics of the device before and after NO2 ad-
sorption: (i) at AGNR(7), (ii) at A60(2, 2, 4, 4), and (iii) the topological junction between
A60(2, 2, 4, 4) and AGNR(7) unit cells. The red [disks] are data for pristine sensor, while
violet [rhombuses], dark blue [triangles] and orange [squares] are data for the sites that are
marked with the same colors in panel (a).

17
Conclusions

In summary, we report a systematic topological classification for an A60 class of graphene

nanoribbons, some of which have already been synthesized. 36 It is shown that, using this
classification, topological junction states can be engineered at the junctions between topolog-
ical and trivial ribbons. Since these states facilitate charge transfer to NO2 gas molecule and
formation of a polar bond, they can be used in a mirror symmetric double junction configura-
tion as active elements of a gas sensor that is read out by quantum transport measurements.
This envisages a niche for sensor applications of topological materials, in addition to the
quantum computing perspective suggested in the current literature. In view of the recent
advance in an experimental study of topological end states in germanene nanoribbons, 75 we
infer that our calculations can also be extended to inorganic chemistry. Another future line
of research may be a modeling of homogeneous topological junctions that are induced by
edge strain and, therefore, difficult to realize by chemical methods. This could be especially
interesting for metamaterials, whereby the localization of junction states could be used to
enhance optical and acoustic detections.

Acknowledgement

The authors thank T. Giovannini for useful discussions and help with some data visualiza-
tion. This work is supported by the National Natural Science Foundation of China (No.
12274361, No. 12474276). This work is also supported by Researchers Supporting Project
(No. RSP2025R468), King Saud University, Riyadh, Saudi Arabia. R.B.P. acknowledges
the support from the Institute of Physics, University of the Philippines – Los Baños. V.A.S.
was partly supported by HORIZON EUROPE MSCA-2021-PF-01 (Project No. 101065500,
TeraExc).

18
References

(1) Hasan, M. Z.; Kane, C. L. Colloquium: Topological insulators. Rev. Mod. Phys. 2010,
82, 3045–3067.

(2) Ren, Y.; Qiao, Z.; Niu, Q. Topological phases in two-dimensional materials: a review.
Rep. Prog. Phys 2016, 79, 066501.

(3) Cayssol, J.; Fuchs, J. N. Topological and geometrical aspects of band theory. J. Phys.
Mater. 2021, 4, 034007.

(4) Weber, B. et al. 2024 Roadmap on 2D Topological Insulators. J. Phys. Mater. 2024,
7, 022501.

(5) Qi, X.-L. L.; Zhang, S.-C. C. Topological insulators and superconductors. Rev. Mod.
Phys. 2011, 83, 1057–1110.

(6) Ando, Y. Topological insulator materials. J. Phys. Soc. Jpn. 2013, 82, 102001.

(7) Wehling, T.; Black-Schaffer, A.; Balatsky, A. Dirac materials. Adv. Phys. 2014, 63,
1–76.

(8) Kane, C. L.; Mele, E. J. Quantum spin Hall effect in graphene. Phys. Rev. Lett. 2005,
95, 226801.

(9) Kane, C. L.; Mele, E. J. Z2 topological order and the quantum spin Hall effect. Phys.
Rev. Lett. 2005, 95, 146802.

(10) Cao, T.; Zhao, F.; Louie, S. G. Topological phases in graphene nanoribbons: Junction
States, spin centers, and quantum spin shains. Phys. Rev. Lett. 2017, 119, 076401.

(11) Lee, Y.-l.; Zhao, F.; Cao, T.; Ihm, J.; Louie, S. G. Topological phases in cove-edged
and chevron graphene nanoribbons: Geometric structures, Z2 invariants, and junction
states. Nano Lett. 2018, 18, 7247–7253.

19
(12) Lin, K.-S.; Chou, M.-Y. Topological properties of gapped graphene nanoribbons with
spatial symmetries. Nano Lett. 2018, 18, 7254–7260.

(13) Ortiz, R.; Garcı́a-Martı́nez, N. A.; Lado, J. L.; Fernández-Rossier, J. Electrical spin
manipulation in graphene nanostructures. Phys. Rev. B 2018, 97, 195425.

(14) Rizzo, D. J.; Veber, G.; Cao, T.; Bronner, C.; Chen, T.; Zhao, F.; Rodriguez, H.;
Louie, S. G.; Crommie, M. F.; Fischer, F. R. Topological band engineering of graphene
nanoribbons. Nature 2018, 560, 204–208.

(15) Gröning, O.; Wang, S.; Yao, X.; Pignedoli, C. A.; Borin Barin, G.; Daniels, C.;
Cupo, A.; Meunier, V.; Feng, X.; Narita, A.; Müllen, K.; Ruffieux, P.; Fasel, R. Engi-
neering of robust topological quantum phases in graphene nanoribbons. Nature 2018,
560, 209–213.

(16) Joost, J. P.; Jauho, A. P.; Bonitz, M. Correlated topological states in graphene nanorib-
bon heterostructures. Nano Lett. 2019, 19, 9045–9050.

(17) Jiang, J.; Louie, S. G. Topology Classification using Chiral Symmetry and Spin Corre-
lations in Graphene Nanoribbons. Nano Lett. 2021, 21, 197–202.

(18) Arnold, F. M.; Liu, T.-J.; Kuc, A.; Heine, T. Structure-imposed electronic topology in
cove-edged graphene nanoribbons. Phys. Rev. Lett. 2023, 129, 216401.

(19) Kuo, D. M. T. Effects of coulomb blockade on the charge transport through the topolog-
ical states of finite armchair graphene nanoribbons and heterostructures. Nanomaterials
2023, 13, 1757.

(20) Perkins, D. T. S.; Ferreira, A. Ultrafast all-electrical universal nanoqubits. Phys. Rev.
B 2024, 109, L041411.

(21) Wang, Z.; Yin, R.; Tang, Z.; Du, H.; Liang, Y.; Wang, X.; Deng, Q.-S.; Tan, Y.-Z.;

20
 Zhang, Y.; Ma, C.; Tan, S.; Wang, B. Topologically localized vibronic excitations in
second-layer graphene nanoribbons. Phys. Rev. Lett. 2024, 133, 036401.

(22) Zhao, C.; Catarina, G.; Zhang, J.-J.; Henriques, J. a. C. G.; Yang, L.; Ma, J.; Feng, X.;
Gröning, O.; Ruffieux, P.; Fernández-Rossier, J.; Fasel, R. Tunable topological phases in
nanographene-based spin-1/2 alternating-exchange Heisenberg chains. Nat. Nanotech.
2024,

(23) Yano, Y.; Mitoma, N.; Ito, H.; Itami, K. A quest for structurally uniform graphene
nanoribbons: Synthesis, properties, and applications. J. Org. Chem. 2020, 85, 4–33.

(24) Chen, Z.; Narita, A.; Müllen, K. Graphene nanoribbons: On-Surface synthesis and
integration into electronic devices. Adv. Mater. 2020, 32, 2001893.

(25) Simon, B. Holonomy, the quantum adiabatic theorem, and Berry’s phase. Phys. Rev.
Lett. 1983, 51, 2167–2170.

(26) Niu, Q.; Thouless, D. J.; Wu, Y.-S. Quantized Hall conductance as a topological in-
variant. Phys. Rev. B 1985, 31, 3372–3377.

(27) Haldane, F. D. M. Model for a quantum hall effect without landau levels: Condensed-
matter realization of the ”parity anomaly”. Phys. Rev. Lett. 1988, 61, 2015–2018.

(28) Wang, S.; Talirz, L.; Pignedoli, C. A.; Feng, X.; Müllen, K.; Fasel, R.; Ruffieux, P.
Giant edge state splitting at atomically precise graphene zigzag edges. Nat. Commun.
2016, 7, 11507.

(29) Gao, N.; Gao, T.; Yang, X.; Dai, X.; Zhou, W.; Zhang, A.; Lieber, C. M. Specific
detection of biomolecules in physiological solutions using graphene transistor biosensors.
Proc. Natl. Acad. Sci. U.S.A. 2018, 113, 14633–14638.

21
(30) Adamu, B. I.; Chen, P.; Chu, W. Role of nanostructuring of sensing materials in per-
formance of electrical gas sensors by combining with extra strategies. Nano Ex. 2021,
2, 042003.

(31) Johnson, A. P.; Sabu, C.; Swamy, N. K.; Anto, A.; Gangadharappa, H.; Pramod, K.
Graphene nanoribbon: An emerging and efficient flat molecular platform for advanced
biosensing. Biosens. Bioelectron 2021, 184, 113245.

(32) Abdelsalam, H.; Saroka, V. A.; Younis, W. O. Edge functionalization of finite graphene
nanoribbon superlattices. Superlattices Microstruct. 2019, 129, 54–61.

(33) Čerņevičs, K.; Pizzochero, M.; Yazyev, O. V. Even–odd conductance effect in graphene
nanoribbons induced by edge functionalization with aromatic molecules: Basis for novel
chemosensors. Eur. Phys. J. Plus 2018, 135, 681.

(34) Abdelsalam, H.; Saroka, V.; Teleb, N.; Ali, M.; Osman, W.; Zhang, Q. Electronic
and adsorption properties of extended chevron and cove-edged graphene nanoribbons.
Physica E Low Dimens. 2021, 126, 114438.

(35) Sakr, M. A.; Abdelsalam, H.; Teleb, N. H.; Abd-Elkader, O. H.; Saroka, V. A.; Zhang, Q.
Investigating adsorption characteristics and electronic properties of Clar’s goblet and
beyond. Chem. Phys. Lett. 2024, 849, 141428.

(36) Mehdi Pour, M.; Lashkov, A.; Radocea, A.; Liu, X.; Sun, T.; Lipatov, A.; Kor-
lacki, R. A.; Shekhirev, M.; Aluru, N. R.; Lyding, J. W.; Sysoev, V.; Sinitskii, A.
Laterally extended atomically precise graphene nanoribbons with improved electrical
conductivity for efficient gas sensing. Nat. Commun. 2017, 8, 820.

(37) Cho, K. M.; Cho, S.-Y.; Chong, S.; Koh, H.-J.; Kim, D. W.; Kim, J.; Jung, H.-T.
Edge-functionalized graphene nanoribbon chemical sensor: Comparison with carbon
nanotube and graphene. ACS Appl. Mater. Interfaces 2018, 10, 42905–42914.

22
(38) Saroka, V. A.; Batrakov, K. G.; Chernozatonskii, L. A. Edge-modified zigzag-shaped
graphene nanoribbons: Structure and electronic properties. Phys. Solid State 2014, 56,
2135–2145.

(39) Saroka, V. A.; Batrakov, K. G.; Demin, V. A.; Chernozatonskii, L. A. Band gaps
in jagged and straight graphene nanoribbons tunable by an external electric field. J.
Condens. Matter Phys 2014, 27, 145305.

(40) Saroka, V. A.; Batrakov, K. G. In Physics, Chemistry and Applications of Nanostruc-

tures; Borisenko, V. E., Gaponenko, S. V., Gurin, V. S., Kam, C. H., Eds.; World
Scientific: Singapore, 2015; pp 240–243.

(41) Saroka, V. A.; Batrakov, K. G. Zigzag-shaped Superlattices on the basis of graphene

nanoribbons: Structure and electronic properties. Russ. Phys. J. 2016, 59, 633–639.

(42) Kudin, K. N.; Car, R.; Resta, R. Berry phase approach to longitudinal dipole moments
of infinite chains in electronic-structure methods with local basis sets. J. Chem. Phys.
2007, 126, 234101.

(43) Rhim, J.-W.; Behrends, J.; Bardarson, J. H. Bulk-boundary correspondence from the
intercellular Zak phase. Phys. Rev. B 2017, 95, 035421.

(44) Last, J. A.; Sun, W. M.; Witschi, H. Ozone, NO, and NO2: oxidant air pollutants and
more. Environ. Health Perspect. 1994, 102, 179–184.

(45) Huangfu, P.; Atkinson, R. Long-term exposure to NO2 and O3 and all-cause and res-
piratory mortality: A systematic review and meta-analysis. Environ. Int. 2020, 144,
105998.

(46) Partoens, B.; Peeters, F. M. From graphene to graphite: Electronic structure around
the K point. Phys. Rev. B 2006, 74, 075404.

23
(47) Saroka, V. A.; Abdelsalam, H.; Demin, V. A.; Grassano, D.; Kuten, S. A.;
Pushkarchuk, A.; Pulci, O. Absorption in finite-length chevron-type graphene nanorib-
bons. Semiconductors 2018, 52, 1890–1893.

(48) Payod, R. B.; Grassano, D.; Santos, G. N. C.; Levshov, D. I.; Pulci, O.; Saroka, V. A.
2N+4-rule and an atlas of bulk optical resonances of zigzag graphene nanoribbons. Nat.
Commun. 2020, 11, 82.

(49) Zak, J. Berry’s phase for energy bands in solids. Phys. Rev. Lett. 1989, 62, 2747–2750.

(50) Bena, C.; Montambaux, G. Remarks on the tight-binding model of graphene. New J.
Phys. 2009, 11, 095003.

(51) Saroka, V. A. TBpack (version 0.4.0 and higher). 2021; https://github.com/

vasilsaroka/TBpack.

(52) Cai, J.; Ruffieux, P.; Jaafar, R.; Bieri, M.; Braun, T.; Blankenburg, S.; Muoth, M.;
Seitsonen, A. P.; Saleh, M.; Feng, X.; Müllen, K.; Fasel, R. Atomically precise bottom-
up fabrication of graphene nanoribbons. Nature 2010, 466, 470–473.

(53) Tepliakov, N. V.; Lischner, J.; Kaxiras, E.; Mostofi, A. A.; Pizzochero, M. Unveiling
and manipulating hidden symmetries in graphene nanoribbons. Phys. Rev. Lett. 2023,
130, 9045–9050.

(54) Son, Y.-W.; Cohen, M. L.; Louie, S. G. Energy gaps in graphene nanoribbons. Phys.
Rev. Lett. 2006, 97, 216803.

(55) Serra-Garcia, M.; Peri, V.; Süsstrunk, R.; Bilal, O. R.; Larsen, T.; Villanueva, L. G.;
Huber, S. D. Observation of a phononic quadrupole topological insulator. Nature 2018,
555, 342–345.

(56) Peri, V.; Serra-Garcia, M.; Ilan, R.; Huber, S. D. Axial-field-induced chiral channels in
an acoustic Weyl system. Nat. Phys. 2019, 15, 357–361.

24
(57) Xia, S.; Liang, Y.; Tang, L.; Song, D.; Xu, J.; Chen, Z. Photonic realization of a generic
type of graphene edge states exhibiting topological flat band. Phys. Rev. Lett. 2023,
131, 013804.

(58) Frisch, M. J. et al. Gaussian˜16 Revision C.01. 2016; Gaussian Inc. Wallingford CT.

(59) Lu, T.; Chen, F. Multiwfn: A multifunctional wavefunction analyzer. J. Comput. Chem.
2012, 33, 580–592.

(60) Becke, A. D. Density-functional thermochemistry. III. The role of exact exchange. J.

Chem. Phys. 1993, 98, 5648–5652.

(61) Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti correlation-energy
formula into a functional of the electron density. Phys. Rev. B 1988, 37, 785–789.

(62) Ditchfield, R.; Hehre, W. J.; Pople, J. A. Self-consistent molecular-orbital methods. IX.
An extended Gaussian-type basis for molecular-orbital studies of organic molecules. J.
Chem. Phys. 1971, 54, 724–728.

(63) Hehre, W. J.; Ditchfield, R.; Pople, J. A. Self—consistent molecular orbital methods.
XII. Further extensions of Gaussian—type basis sets for use in molecular orbital studies
of organic molecules. J. Chem. Phys. 1972, 56, 2257–2261.

(64) Povie, G.; Segawa, Y.; Nishihara, T.; Miyauchi, Y.; Itami, K. Synthesis and size-
dependent properties of [12], [16], and [24] carbon nanobelts. J. Am. Chem. Soc. 2018,
140, 10054–10059.

(65) Abdelsalam, H.; Sakr, M. A.; Saroka, V. A.; Abd-Elkader, O. H.; Zhang, Q. Nanoporous
graphene quantum dots constructed from nanoribbon superlattices with controllable
pore morphology and size for wastewater treatment. Surf. Interfaces 2023, 40, 103109.

(66) Abdelsalam, H.; Abd-Elkader, O. H.; Sakr, M. A. S.; Saroka, V. A.; Zhang, Q.
Nanoporous triangulene-based frameworks for the separation of petroleum hydrocar-

25
 bons: Electronic, magnetic, optical, and adsorption properties. ACS Appl. Nano Mater.
2023, 6, 15128–15137.

(67) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and accurate ab initio
parametrization of density functional dispersion correction (DFT-D) for the 94 elements
H-Pu. J. Chem. Phys. 2010, 132, 154104.

(68) Collier, T. P.; Saroka, V. A.; Portnoi, M. E. Tuning terahertz transitions in a double-
gated quantum ring. Phys. Rev. B 2017, 96, 235430.

(69) Cai, J.; Pignedoli, C. A.; Talirz, L.; Ruffieux, P.; Söde, H.; Liang, L.; Meunier, V.;
Berger, R.; Li, R.; Feng, X.; Müllen, K.; Fasel, R. Graphene nanoribbon heterojunc-
tions. Nat. Nanotechnol. 2014, 9, 896–900.

(70) Chen, L. et al. Oriented graphene nanoribbons embedded in hexagonal boron nitride
trenches. Nat. Commun. 2017, 8, 14703.

(71) Taylor, J.; Guo, H.; Wang, J. Ab initio modeling of quantum transport properties of
molecular electronic devices. Phys. Rev. B 2001, 63, 245407.

(72) Taylor, J.; Guo, H.; Wang, J. Ab initio modeling of open systems: Charge transfer,
electron conduction, and molecular switching of a C60 device. Phys. Rev. B 2001, 63,
121104.

(73) Datta, S. In Electronic transport in mesoscopic systems; Haroon, A., Pepper, M.,
Broers, A., Eds.; Cambrige University Press: Cambridge, United Kingdom, 1995.

(74) Tamersit, K. An ultra-sensitive gas nanosensor based on asymmetric dual-gate graphene

nanoribbon field-effect transistor: proposal and investigation. J. Comput. Electron.
2019, 18, 846–855.

(75) Klaassen, D. J.; Eek, L. A. G.; Rudenko, A. N.; Westende, E. D. v.; Castenmiller, C.;
Zhang, Z.; Boeij, P. d.; Houselt, A. v.; Ezawa, M.; Zandvliet, H. J. W.; Smith, C. M.;

26
Bampoulis, P. Realization of a one-dimensional topological insulator in ultrathin ger-
manene nanoribbons. http://arxiv.org/abs/2411.18156.

27
TOC Graphic

NO2 Z2=1 Z2=0

28