Download as pdf or txt
Download as pdf or txt
You are on page 1of 38

Accepted Manuscript

Title: Ab-initio investigation of adsorption of CO and CO2


molecules on graphene: Role of intrinsic defects on gas
sensing

Author: Nacir Tit Khadija Said Nadin M. Mahmoud


Summayya Kouser Zain H. Yamani

PII: S0169-4332(16)32168-7
DOI: http://dx.doi.org/doi:10.1016/j.apsusc.2016.10.052
Reference: APSUSC 34146

To appear in: APSUSC

Received date: 22-7-2016


Revised date: 7-10-2016
Accepted date: 8-10-2016

Please cite this article as: Nacir Tit, Khadija Said, Nadin M.Mahmoud, Summayya
Kouser, Zain H.Yamani, Ab-initio investigation of adsorption of CO and CO2 molecules
on graphene: Role of intrinsic defects on gas sensing, Applied Surface Science
http://dx.doi.org/10.1016/j.apsusc.2016.10.052

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Ab-initio Investigation of Adsorption of CO and CO2 Molecules on
Graphene: Role of Intrinsic Defects on Gas Sensing

Nacir Tit1,2,*, Khadija Said1, Nadin M. Mahmoud1,3, Summayya Kouser4 and Zain H. Yamani2

1Physics Department, UAE University, P.O. Box 15551, Al-Ain, United Arab Emirates
2Center for Research Excellence in Nanotechnology, King Fahd University of Petroleum and
Minerals, P.O. Box 5040, 31261 Dhahran, Kingdom of Saudi Arabia
3Physics Department, American University at Cairo, New Cairo 11835, Egypt
4TheoreticalSciences Unit, Jawaharlal Nehru Center for Advanced Scientific Research, Bangalore
560 064, India

(*) Corresponding author: ntit@uaeu.ac.ae

1
Highlights
• Absorptions of CO and CO2 molecules on graphene are studied using DFTB and QE.
• CO and CO2 molecules can alter chemisorption on vacancy of graphene.
• Vacancy-defected graphene is highly sensitive and selective toward detection of CO.
• Normalized Conductance varies linearly with CO or CO2 gas doses in consistency with
experimental data.

Abstract

We determine the chemical activity of (a) carbon site of pristine graphene (pG), (b) Stone-Wales (SW)
defect site, and (c) Single-vacancy of graphene (vG) site towards the adsorption of CO and CO2
molecules, through comparative analysis based on first-principles density-functional calculations
incorporating van der Waals (vdW) interactions, but excluding the heat effects (i.e., at T = 0 oK). The
results show that the chemisorption of both latter molecules to possibly occur only on vG. The response
(sensitivity) of vG towards detecting CO molecule was confirmed by the rise of conductance with the
increasing CO gas dose. The selectivity was investigated by testing the response of vG towards detecting
eight different gases (i.e., CO, CO2, N2, O2, H2O, H2S, H2, and NH3). Three gases are found to exhibit
physisorption (namely: N2, H2O, and H2S) and the other five gases alter chemisorption (namely: CO, CO2,
O2, H2, and NH3). The chemisorption of CO molecule is distinct by being direct and not involving
dissociation. This fact made defected graphene have the highest sensitivity and selectivity towards the
detection of CO molecules.

PACS Numbers: 81.05.U; 81.05.ue; 68.43.Fg; 68.43.Mn

Keywords: Carbon systems, Graphene, Ab-initio calculations, Adsorption kinetics, Gas sensing

2
1. Introduction
The continuous rise of concentration of carbon dioxide (CO2) in the atmosphere causes a real
threat to the lifelong security of mankind and naturally becomes amongst the prime issues of
global concerns. The reduction and capture of green-house gases have been among the most
challenging issues in environmental protection [1]. Solutions to this problem require
simultaneous efforts in many fronts: (i) Ultimate reduction of emissions of green-house gases
by switching to novel clean-energy resources; (ii) Increase of vegetation because plants have
the ability to transform CO2 into O2 via the photosynthesis process, (iii) Search for suitable
materials (with optimum cost and efficiency) capable of detecting CO2 in order to incorporate
them in the fabrication of devices such as sensors and filters; and (iv) Promote the awareness
through the media and further help poor countries to use filters; especially if reasonably low
costs become attainable.

In the last decade, research has been focused on materials to control CO2 emissions by
capturing and separating technologies such as absorption, adsorption, membranes and so forth
[2]. Among them, adsorption technologies using semiconducting metal oxides [3], zeolites [4],
metal-organic framework (MOFs) [5], or activated carbons [6] have been endeavored. In recent
years, various candidates as adsorber beds have been experimentally and theoretically
investigated for the purpose of adsorption of various atmospheric gases, including CO2, on
functionalized graphene, carbon nanotubes (CNTs), graphene nanoribbons (GNRs) and
graphene oxide (GO) [7-19]. In the theoretical side, density functional theory (DFT) has been
predominantly used as is appropriate to study the ground-state properties, such as the problem
of adsorption of molecules on surfaces.

The breakthrough success in fabricating single-layer graphene sheet earned Geim and
Novoselov [20] a Noble prize in physics in the year 2010 and has opened up a new horizon for
fundamental research, and a wide range of advanced technological applications [21]. Graphene
possesses unique characteristics such as being zero-gap semiconductor, where two bands
linearly crossing each other at K high-symmetry points of the Brillouin zone and thus producing
the so-called Dirac cones. This fact made graphene possess massless Dirac (relativistic) fermions

3
with high mobility. These characteristics are advantageous in the development of high-speed
next generation devices with properties exceeding those of silicon and conventional
semiconductors. Graphene is categorized among materials possessing multi-functionality
character with diversity of applications in various fields such as: (1) nanoelectronics (e.g.,
synthesis of the smallest transistor [22]); (2) photonics (e.g., band-gap engineering using
quantum confinement effects in nano-ribbons [23] and BN doping [24]); (3) spintronics (e.g.,
induction of ferromagnetism by the adsorbed hydrogen atoms in nano-ribbons [25]); and (4)
gas sensing (e.g., to detect gases like CO, CO2, NO, NO2 and NH3 [8,26,27]). A very nice coverage
of physical challenges involved in making graphene as the material of the 21st century for these
applications can be found in a recent review [28].

Furthermore, the reduced dimensionality of graphene itself decreases the number of possible
types of intrinsic defects [29]. Graphene lattice has the ability to reconstruct its structure by the
formation of non-hexagonal rings. Among the experimentally observed intrinsic defects, and
perhaps the most common ones, are: Stone Wales, single vacancy, inverse Stone Wales and
adatom defects. From the perspective of abundance and thermodynamic stability, the first
formal two defects should be the most interesting ones. (i) Stone-Wales defect is formed when
four hexagons are transformed into two pentagons and two heptagons (i.e., SW(55-77) defect)
[30], by rotating one of the C-C bonds by 90o. The SW(55-77) defect has a formation energy of
Ef ≈ 5 eV [31,32] with a barrier of approximately 9 eV. Once the defect is formed, the height of
the barrier would warrant its thermodynamic stability. (ii) Single vacancy is the simplest defect
in any material, and corresponds to a missing lattice atom. The atomic relaxation of this defect
leads to the formation of a pentagon and nine-membered ring (i.e., V1(5-9) defect). Ab-initio
calculations yield a formation energy of Ef ≈ 7.5 eV [33,34], which is much higher than that of
SW defect and also higher than the formation energy of vacancies in many materials (e.g., 4.0
eV in silicon [35]).

Native defects in materials usually play an important role in the modification of their
mechanical, thermal, and electronic characteristics. Among their positive effects in graphene,
defects can lead to chemisorption of some gases. For instance, using ab-initio method, Sanyal

4
and coworkers [36] have reported that N2 molecule can be dissolved in the vicinity of a
divacancy in a graphene layer. Similar dissociation and adsorption but by a single vacancy in
graphene were reported for other small gas molecules, such as H2 [37], O2 [38], and H2S [39,40].
Among the recent works on the adsorption of CO2 on graphene with defects, Yoon and
coworkers [8] reported experimental gas sensing results that support the capturing CO2. Under
ambient conditions and at room temperature, they reported a linear increase of conductance
when the concentration of CO2 is increased from 10 to 100 ppm when temperature ranges
within 22-60 oC. From computational point of view and using density functional theory (DFT),
Allouche and Ferro [41] have shown that CO2 and N2 have no interaction with defect of
graphene whereas carbon monoxide (CO) is incorporated into the vacancy with activation
energy of 1.5 eV. Huang and coworkers [42] have studied the adsorption of gas molecules (CO,
CO2, NO, NO2, O2, N2 and NH3) on graphene nano-ribbons (GNRs) with armchair edges using DFT
and demonstrated that only NH3 can be chemisorbed. Furthermore, their quantum transport
calculations corroborated that NH3 molecules can be detected out of the other gas molecules
by the GNR-based sensor. Cabrera-Sanfelix’s DFT calculations [43] confirmed a strong
physisorption of CO2 on defective graphene with a weak binding energy of ~ 136 meV. A bit
stronger physisorption was further reported by Liu and Wilcox [44] to be about ~ 210 meV.

In the present work, we use DFT to investigate the adsorption of gas molecules (CO, CO 2, N2, O2,
H2, H2O, H2S, and NH3) on defective graphene. Throughout this work, special attention will be
given to the adsorption of CO and CO2 molecules because their influences on the global
warming and other environmental and health issues. Our findings will show that only vacancy-
defected graphene can have chemisorption with five molecules (CO, CO2, O2, H2, and NH3). We
will show also that the interaction with CO molecule is distinct by being direct and does not
involve dissociation. With increasing the dose of CO gas, our results show an increase of density
of states at Fermi level, which in turn should enhance the conductance. The focus of our study
will be on the role of vacancies on the sensitivity and selectivity of detecting CO and CO2
molecules. Next section will give a brief description on the computational model and methods.
Section 3 describes and discusses the results. Section 4 summarizes our main findings.

5
2. Computational Method and Model
In the present work, we use two computational methods. These two methods yield the same
physical trends. The first one is the self-consistent-charge density-functional tight-binding (SCC-
DFTB) method [45,46], as implemented in the DFTB+ package. This method is used to perform
atomic relaxations of structures, including super-cell size in order to release pressure and
further minimize the total energy. In this method, we use Slater-Koster (SK) parameter files [47]
from the ‘mio-0-1’ [48,49] set to parameterize the inter-atomic interactions. Van der Waals
(vdW) interaction was accounted for by using the Lenard-Jones dispersion model as in DFTB+
with parameters taken from the universal force field (UFF) [45,46]. On one hand, approximating
and parameterizing Fock-Matrix elements, an effective one-electron Kohn-Sham (KS)
Hamiltonian is derived from density functional theory (DFT) calculations. On the other hand,
DFTB is in close connection to tight-binding method. So, it can be seen as tight-binding method,
parameterized from DFT, and this overcomes the problem of parameters’ transferability and
makes the method more accurate. So, its basis set does not rely on plane-waves nor Gaussian
functions, but rather is a minimal basis set based on pseudo-atomic orbitals (Slater orbitals and
spherical harmonics). Based on this basis set, DFTB gains its speed and ability to deal with large
systems. So, in contrast to “full” DFT methods such as quantum Espresso, DFTB can easily
handle calculations of large systems with reasonably large Monkhorst-Pack (MP) grid and
perform atomic relaxations. Furthermore, DFTB was augmented by a self-consistency treatment
based on atomic charges in the so-called self-consistent charge density-functional tight-binding
(SCC-DFTB) method; charge density is expressed in terms of Milliken charges [50]. Because the
wave functions in DFTB are well defined as KS-like orbitals, one can easily derive expressions for
any property in the same way as within a “full” DFT scheme. This has indeed paved the way for
DFTB to extend its domain applications to even comprise biological systems [51]. Its strength
stems from the transparent derivation, the inclusion of electron correlation on the DFT-GGA
level and the updating parameterization process. This led to a robust method that predicts
molecular geometries quite reliably. Nevertheless, among the limitations in DFTB is the
incomplete availability of Slater-Koster files for all elements in the periodic table and this
remains among the main challenges in the next years.

6
For the purpose of benchmarking the binding energy results, we explore another ab-initio
code, which is quantum espresso (QE) package [52]. QE is based on DFT, plane-wave basis sets
and pseudopotentials (both norm-conserving and ultrasoft). In our particular study of
adsorption of molecules on surfaces, as the charge density is expected to vary rapidly in space,
because of defects and adsorbed molecules, we use a generalized gradient approximation
(GGA) of Perdew Burke-Ernzerhof (PBE) parameterized form [53] and the interaction between
ionic core and valence electrons is represented using ultra-soft pseudo-potentials [54].
Furthermore, we use Grimme scheme [55] to capture the long-range interactions namely van
der Waals (vdW) like. We use plane-wave-basis set with energy cutoffs of 30 Ry and 180 Ry in
representing orbital wave-functions and charge density, respectively. For the Brillouin-zone
sampling, for instance in the case of a super-cell (SCell) of graphene composed of 5x5 primitive
cells, a uniform mesh of 5x5x1 k-points is used, and the occupation numbers of electronic
states is smoothened with a smearing width of order kBT (i.e., about 0.04 eV) using Fermi-Dirac
distribution function. Regarding the accuracy of the two methods, we carried out a small test of
total energy calculations on a pristine graphene sample of size 5x5 primitive cells (i.e.,
containing 50 carbon atoms) and both DFTB and QE agree within a discrepancy of 10 meV when
a relatively large MP mesh of about 12x12x1 is used in QE code.

Relaxed structures were determined through minimization of the total energy until
Hellmann-Feynman forces on each atom became smaller than 0.03 eV/Å in magnitude. The 2D
sheet of graphene was simulated using a super-cell geometry, with vacuum layer of 20 Å
separating adjacent periodic images of the sheet. Except when studying the dependence of
adsorption capacity on the concentration of gas adsorbed on graphene, we used a supercell
size of 5x5x1 unit cells (i.e., containing 50 carbon atoms). The Brillouin zone (BZ) was sampled
using the Monkhorst-Pack technique [56], with a mesh of 26x26x1 in case of application of
DFTB on the latter supercell.

The binding energy of a single gas molecule (i.e., the adsorption energy, Ebind) is calculated using
the following convention:

𝐸𝑏𝑖𝑛𝑑 = 𝐸(𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒+𝑔𝑟𝑎𝑝ℎ𝑒𝑛𝑒) − 𝐸(𝑔𝑟𝑎𝑝ℎ𝑒𝑛𝑒) − 𝐸(𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒) (1);

7
where E(molecule+graphene), E(graphene), and E(molecule) stand for the total energies of the relaxed
molecule on graphene system, isolated graphene, and isolated molecule, respectively. In case
of adsorption of several molecules (for instance N molecules), the binding energy per molecule
would be defined as

𝐸(𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠+𝑔𝑟𝑎𝑝ℎ𝑒𝑛𝑒) −𝐸(𝑔𝑟𝑎𝑝ℎ𝑒𝑛𝑒) −𝑁𝐸(𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒)


𝐸𝑏𝑖𝑛𝑑 = (2);
𝑁

As a matter of fact, the binding energy of molecule on a surface can be taken as a parameter
which reveals the type of adsorption taking place between molecule and surface. Specifically,
physisorption (i.e., weak van der Walls-like interaction) usually occurs with Ebind < 1.0 eV,
whereas chemisorption often takes place with Ebind > 1.0 eV [57,58]. The physisorption is
always reversible; whereas in the chemisorption the molecule may alter into a chemical
reaction of dissociation, that is irreversible or after whuch the molecule cannot be recovered by
desorption.

The sensitivity of a gas sensor is studied by looking at the variation of the electrical
conductance versus gas dose (i.e., variation of conductance versus number of molecules landed
on the adsorbent). The conductivity is evaluated based on Drude formula [59,60], which should
be valid for transport in metals:

𝑛𝑒 2 𝜏
𝜎= = 𝑛𝑒𝜇 (3);
𝑚∗

where n is the number density of free electrons, e is the free electron charge, τ is the average
collision time, m* is electron’s effective mass, and μ is the electron mobility. This formula
remains valid in the case of graphene, which is a 0-gap semiconductor with large mobility (i.e.
conductor). Neglecting the heat effects (at 0 K) and assuming that the mobility of the adsorbent
is independent of gas dose, the sensitivity [61-63] then would be proportional to the following
quantity:

|𝐼𝑔 −𝐼𝑜 | |𝐺𝑔 −𝐺𝑜 |


𝑆= × 100% = × 100% (4);
𝐼𝑜 𝐺𝑜

where the pairs of Ig, Io and Gg, Go stand for the electric current intensity and the conductance in
presence and absence of gas adsorption, respectively. Here, we take by assumption the validity
of Ohm’s law (i.e., I=GV, where V is the applied voltage). Under the same assumption, the
1 𝐴
conductance is related to conductivity by the relation: 𝐺 = 𝑅 = 𝜎 𝐿 and with consideration
that the cross-sectional area A and length L of sample to be constant, the sensitivity becomes:
|𝜎𝑔 −𝜎𝑎 | |𝑛𝑔 −𝑛𝑎 |
𝑆= × 100% = × 100% (5);
𝜎𝑎 𝑛𝑎

8
Furthermore, it should be emphasized that the conductivity depends only on the properties of
the electrons at the Fermi surface, not on the total number of electrons in the metal. The high
conductivity of metals is to be ascribed to the high current, jF=nevF, carried by the few electrons
at the top of the Fermi distribution, rather than to the total density of free electrons (i.e, vF is
Fermi velocity, and electrons of lower energy than Fermi energy are slowly drifting). Thus, the
electrical conductivity gains its main contribution from states near Fermi surface [58]. Assuming
that Fermi energy remains the same after the adsorption of a molecule, then the sensitivity
might be written as:
(𝑔) (𝑎)
|𝑁𝐹 −𝑁𝐹 |
𝑆= (𝑎) × 100% (6);
𝑁𝐹
(𝑔) (𝑎)
where 𝑁𝐹 and 𝑁𝐹 are the density of states (DOS) at Fermi level with and without gas
molecule, respectively.

The selectivity is studied by keeping the gas dose constant but varying the gas type, and
all sensitivities are compared on a unified scale. In the present work, the sensitivity is studied
versus the following gases: CO, CO2, N2, O2, H2O, H2S, H2 and NH3. The results of structural
relaxations, electronic structure calculations, and both sensitivity and selectivity will be
discussed in the next section.

3. Results and Discussions


3.1. Relaxation and atomic structures
Figure 1 displays the atomic structures of adsorbent with/without adsorbate systems after
being relaxed by the DFTB+ code. In all our calculations a super-cell containing 5x5 unit cells
(i.e., 50 atoms of carbon atoms) of graphene is used. This supercell is chosen as being the
optimum size to host diverse defects of our interest like Stone-Wales and multi-vacancy
defects. We emphasize that in case of pristine graphene, relaxation tests of either CO or CO2
molecules were carried out starting from initial positions within approximately a bond-length
distance from three different places: (i) on-site, (ii) on C-C bridge site, and (iii) on hollow site. All
these configurations yield physisorption situations where the molecule relaxes far above the
surface at a distance of about 3.3 Å. Furthermore, we found that the algebraic values of the
𝑜𝑛−𝑠𝑖𝑡𝑒 𝑜𝑛−𝑏𝑟𝑖𝑑𝑔𝑒 𝑜𝑛−ℎ𝑜𝑙𝑙𝑜𝑤
binding energies are ranked in the following order: 𝐸𝑏𝑖𝑛𝑑 > 𝐸𝑏𝑖𝑛𝑑 > 𝐸𝑏𝑖𝑛𝑑 , which
reveals that on-hollow site is the most stable one as it corresponds to the least total and

9
binding energies. Nevertheless, in the present section, we focus on discussing the results of
adsorptions of CO and CO2 molecules only on vacancy-defected graphene, as this latter
adsorbent is found chemically active and can interestingly yield both chemisorption and
physisorption with CO and CO2 molecules, respectively. Figure 1a shows the geometry structure
of pristine graphene, whose equilibrium bond length is about 1.42 Å, consistent with
experimental data existing in literature. Figure 1b shows the relaxed structure of graphene with
a single vacancy (i.e., it contains 49 carbon atoms). Figure 1c shows both top-view and side-
view of CO2 molecule being relaxed on vacancy defected graphene. The CO2 molecule is
released from a place just above the vacancy vicinity and it moves to stabilize away from
graphene sheet at a distance of about 3.30 Å and having a binding energy of – 0.685 eV. While
it is clear that this is a physisorption state, chemisorption is also discovered possible and
scenarios for it to occur will be discussed in the last sub-section below. We emphasize that in
the ab-initio relaxation process, all the atoms in the supercell were given the freedom to move.
Figure 1d shows the top-view and side-view of the relaxed CO molecule on vacancy-defected
graphene. It is found that the carbon atom of CO molecule inserts itself into the vacancy site to
accomplish the formation of a pristine graphene structure then the oxygen atom adjusts itself
along the C-C bond and makes single bond with each of the carbon atoms along the “old” C-C
bond neighboring the “old” vacancy. The binding energy of CO inside vG is large -5.063 eV.
Whereas, the binding energy of an oxygen atom on pristine graphene is about -3.215 eV. The
picture of side-view makes the evidence that this is a clear case of chemisorption.

Table 1 shows the adsorption (binding) energies of CO and CO2 molecules on various
adsorbents, namely: (i) pristine graphene, (ii) Stone-Wales defected graphene and (iii) Single-
vacancy-defected graphene. The shortest distance from molecule to graphene is shown
between brackets in Å units. Scenarios yielding physisorption of CO2 molecule are displayed in
Figure 1. In the physisorption states, the strongest van der Waals interaction happens to occur
with the SW-defect. Whereas, CO molecule does have a strong physisorption with SW-defect,
nonetheless, it can further reach a chemisorption state on a vacancy.

10
3.2. Band structures

Figure 2 displays the band structures corresponding to the respective systems of panels of
Figure 1. Namely, Figure 2a corresponds to a super-cell of pristine graphene containing 50
atoms (i.e., 200 states). Of course, while the bulk band structure requires 2 atoms, we chose to
use a SCell of 5x5 unit cells in consistency with the ones used for studying adsorption for sake of
comparison. So, in Figure 2a many bands are just folded and we show there 20 highest valence
bands (VBs) and 20 lowest conduction bands (CBs), and Fermi level is taken as an energy
reference. Figure 2a clearly shows that graphene has a zero gap, as two bands are crossing each
other at K corner of the hexagonal Brillouin zone. The dispersions of these latter two bands are
also linear near the K-point to produce the so-called “Dirac cone”. This fact makes graphene
metallic having high motilities of charge carriers.

Figure 2b shows the bands of graphene with a single vacancy (i.e., SCell containing 49 carbon
atoms). As expected, the removed carbon will result in leaving the three neighboring carbon
atoms having three dangling bonds, which will produce localized states at Fermi level. Actually
the three flat bands (one just below EF level and two almost degenerate just above EF level) are
attributed to the three dangling bonds directed towards the vacancy vicinity. Besides this, one
may notice an opening of an energy gap of about 0.149 eV between the dangling bonds’ eigen-
states. The rest of bands are dispersive as to correspond to extended states of Bloch-like
wavefunctions.

Figure 2c shows the bands of the system of vacancy-defected graphene with CO2 molecule (in
case of physisorption). The initial state in the relaxation has led to physisorption where CO2
molecule weakly interacts with vG via van der Waals like forces and the molecule stabilizes at a
distance of about 3.3 Å away from vG. Actually, within the energy range shown, the molecule
possesses two discrete eigen-states of energies E1 = -3.47 eV and E2 = 5.71 eV. These two states
are shown by green dashes as guide to the eye. The almost flat bands above Fermi level remain
to correspond to the dangling bonds pointing toward the vacancy’s vicinity. The molecular
eigenstates have remained discrete as an indication of physisorption to take place. The opening
of band-gap energy is of order 0.354 eV.

11
Figure 2d displays the band structure of the system of vacancy-defected graphene with CO
molecule after achieving full relaxation. As consequence of chemisorption, all the bands are
dispersive and very similar to the case of pristine graphene (shown in Fig.2a) except that an
oxygen atom is added to cause the opening of an energy gap of value of about 0.331 meV. Even
the gap remains located at K corner of the Brillouin zone and is direct.

3.3. Density of states

Figure 3 displays the DOSs corresponding to the same systems of the preceding figure. In each
panel, we use a super-cell of size 5x5 unit cells (i.e., containing roughly about 50 atoms) and use
DFTB code with 290 k-vectors in sampling the Brillouin zone to integrate the Fourier transform
of charge density in each self-consistent calculation. Using this number of k-vectors, together
with a Lorentzian broadening/smoothening of width 0.01 eV, seems to be sufficient to yield
appropriate DOSs. Figure 3a shows the DOS for pristine graphene. We focus more on ground
states as DFT overlooks the excited states and Fermi level is taken as an energy reference.
Figure 3a shows the closure of bandgap at Fermi level in consistency with the zero-gap
graphene’s bands of Figure 2a. However, in Figure 3b, the removal of one carbon atom in
creating a vacancy, the dangling bonds of the three carbon atoms around the vacancy vicinity
introduce localized states at Fermi level. The localized peak at Fermi level (EF=0) should be
attributed to those dangling bonds (i.e., corresponding to one flat band just below EF and two
flat bands just above EF as shown in Figure 2b).

The next two panels Figure 3(c-d) show the partial density of states (PDOS) of carbon atoms in
black curve and of oxygen atoms in green curve, while the total density of states (TDOS) in red
solid curve. Figure 3c corresponds to the case of CO2 molecule relaxed on vacancy-defected
graphene (case of physisorption). This Figure shows oxygen atoms’ PDOS being discrete, which
in turn reveals that CO2 molecule is completely isolated from graphene. This is consistent with
physisorption state, as discussed in the previous figures. On the other hand, Figure 3d
corresponds to the case of CO molecule relaxed on vacancy-defected graphene in
chemisorption state. The PDOS of oxygen is shown to form a whole band contributing to the
TDOS of the whole system. We emphasize that the number of eigen-states in isolated CO

12
molecule is less than that of isolated CO2 molecule. Furthermore, the PDOS of oxygen atom in
Figure 3d does corroborate the fact of existence of chemisorption where the whole molecule is
chemically bonded to the graphene; as carbon atom of CO merges inside the vacancy site and
the oxygen atom makes a bridge with 2 covalent bonds with the two carbon atoms along the C-
C bond of graphene.

3.4. Simulation of gas sensing

From the experimental side, Yoon and coworkers [8] reported an efficient way to detect CO2
molecules using graphene-based sensor. The graphene-based sensor was fabricated by
mechanical cleavage. The normalized change of conductance is shown to increase linearly with
CO2 gas dose in the range 10-100 ppm. Their sensor was reported to have high sensitivity, fast
response time, short recovery time and low power consumption.

In the present work, we have performed calculations of adsorption of CO and CO2 molecules on
the vacancy-defected graphene’s sample of size 5x5 unit cells. The results show that CO
molecule always exhibits a chemisorption, in which the carbon atom merges inside the vicinity
of vacancy and occupies the empty site while the oxygen atom makes a bridge bonding with its
original molecular carbon atom and the nearest neighboring carbon atom along the “old” sigma
bond of vG (see Figure 4a). The binding energy of CO molecule on vG is -5.063 eV. Whereas,
CO2 molecule can exhibit both chemisorption and physisorption depending on the initial state
of relaxation. Furthermore, its chemisorption involves in its process the dissociation of the
molecule and that costs energy. If it happens, the chemisorption would yield two atoms of
oxygen doping pG and thus enhancing the conductivity even more than the chemisorption of
CO molecule on vG. Nonetheless, we will discuss in more details this case in an independent
sub-section below. In case of physisorption, the binding energy is about -0.685 eV and a
nominal distance between CO2 molecule and vG is of about 3.30 Å. Whereas, in case of
chemisorption, the binding energy is -1.023 and the process involve the dissociation of
molecule whose cost is about 5.52 eV (at room temperature).

13
We emphasize that the chemisorption processes of CO on vG merge to yield similar lattice,
having oxygen atoms bind on the surface. To investigate the effect of gas dose on the
conductivity (i.e, the response function), we have considered the detection of multiple of CO
molecules, up to 6 molecules, on multiple-vacancy defected graphene (up to 6 vacancies,
selected in a uniformly scattered fashion on the sample to take account of a random
occurrence of single vacancies and avoid the cases of extensive defects such as divacancy) and
we assessed whether they will boost or hamper the conductivity.

Panels 4b-d show the converged geometries of chemisorption of 1 CO up to 6 CO molecules on


1-6 vacancies (i.e., computationally speaking, the process of full oxidation of multiple-vacancies
by CO molecules is similar to adding 1 oxygen atom up to 6 oxygen atoms on bridge sites of
pristine graphene, then let the systems relax). Taking this similarity into account, Table 2 shows
the binding energies corresponding to the respective oxygen ad-atoms on pristine graphene as
well as the binding energies of CO molecules on vG. More importantly, each relaxation yields a
configuration with density of states clearly showing an enhancement at Fermi level versus the
increasing number of oxygen ad-atoms (this will be further discussed in next Figure). It is worth
emphasizing that the relaxation process involves all the atoms in the system. While doing this,
we notice that whenever the oxygen ad-atoms get close within an average distance of about 2a
= 2.84 Å, the oxygen atoms tend to couple into one molecule and gets liberated from the
graphene and stays away from it within a distance of 3.0 Å via weak vdW interaction. This
scenario is displayed Fig. 4, panels (e-f). Figure 4e shows the initial state where 4 oxygen atoms
are put on top-sites of 4 carbon atoms within a bond length of d CO = 1.118 Å. Then, the
relaxation is carried out using DFTB code to yield both chemisorption of two oxygen atoms and
physisorption of molecule O2 at distance of about 3.081 Å. In terms of adsorption of 4 CO
molecules, one may rewrite about the preceding observation saying that chemisorption of CO
molecule occurs on vacancy if the separation between vacancies is larger than 2a but the
chemisorption of CO molecules yields dissociation of CO molecule into carbon atoms filling the
vacancies and production of O2 in case of extended vacancies (i.e., whose inter-distance is at
the order of 2a, where a is the C-C bond-length).

14
Figure 5a shows the partial density of states (PDOS) contributed by oxygen atoms for a
number of O atoms ranging from 1 to 6. Fermi levels of various structures are taken as a
common energy reference. The integrations of PDOSs up to Fermi level yield values of average
of electric charges on the outer shells of anonymous oxygen atom to be in the range 6.358 ≥ q ≥
6.323 revealing excess of electric charge of oxygen atoms as an effect of their being superior in
electronegativity than carbon atoms. It is clear that oxygen atoms participate in sigma bonding
and predominantly to the composition of valence band. Consequently, the increase in number
of oxygen ad-atoms would enhance the density of states at Fermi level as is displayed by the
inset in Figure 5a and also in Figure 5b and 5c. Figure 5b shows the DOS at Fermi level (NF)
versus number of oxygen ad-atoms NO in linear scale. Whereas, Figure 5c shows the same
profile but on semi-log scale. It shows that the increase is logarithmic. The increase of DOS at
Fermi level should yield an enhancement of conductivity. One should expect an increase of
conductivity versus the dose of CO molecules because electronic DOS at Fermi level increases.
The slope of the line in Figure 5c is proportional to the response function (which is proportional
to the sensitivity) of the detector.

Figure 6 shows the response function versus number of CO molecules as being adsorbed
on vacancies. Figure 6 corroborates the experimental data of Yoon and coworkers [8], as to
show that sensitivity increases linearly with the gas dose. The slop in Figure 6 is 0.23 whereas
the experimental sensitivity has a slop of about 0.18 [8]. Last remark is that the experimental
sensitivity is shown to increase a little bit with the rise of temperature (from 22 oC to 60 oC). In
our present work, we did not explore the effect of temperature.

3.5. Adsorption of other gases (selectivity)

So far, it has been shown that the most chemically active adsorbent to be the vacancy-defected
graphene. Furthermore, since the predominant gases altering the composition of air are O2 and
N2; then, it is interesting to investigate the selectivity of the vacancy-defected graphene
towards the adsorption of CO gas by inspecting the adsorptions of air molecules and others,
such as: O2, N2, H2, H2O, H2S, and NH3.

15
Figure 7 displays a comparison of the adsorptions of O2 molecule on both pristine graphene and
vacancy-defected graphene. The results clearly show that the O2 molecule exhibits a weak
physisorption with pristine graphene with a binding energy of about -0.137 eV. Figure 7a shows
the configuration after relaxation, where O2 molecule stabilizes at a distance of about 3.08 Å
away from the vG sheet. Figure 7c shows the PDOS and TDOS corresponding to O2 molecule on
pristine graphene, respectively, where the molecular states remain discrete. This corroborates
the fact that the molecule is not chemically bonded to the surface as it is in a physisorption
state. On the other hand, Figure 7b shows the configuration of O2 after its dissociation by the
vacancy (after relaxation). The dangling bonds of the vacancy were able to split the oxygen
atoms of molecule and get them involved in covalent bonds with the carbon atoms neighboring
the vacancy. Figure 7d shows the PDOS contribution of oxygen atoms to be spread out in
energy along the valence band and thus revealing the strong binding occurred with graphene.
The binding energy of the O2 molecule is -5.148 eV, with inclusion of cost of the dissociation
energy. This is at similar magnitude as the binding energy of CO molecule chemisorbed in
vacancy-defected graphene, which is about -5.063 eV but this latter does not involve
dissociation. As far as the large potential barrier involving the O2 dissociation is concerned, it is
likely that the adsorption of CO should be more favorable event to occur on a vacancy than the
adsorption of O2 molecule.

Table 3 shows the binding energies of a selection of gas molecules and compare them to
the one of adsorption of CO molecule. All adsorptions are on vacancy-defected graphene. It
seems that in some cases, physisorption is the destiny of some molecules like: N2, H2O and H2S.
The chemisorption cases (such as: CO2, O2, H2, NH3) do involve the dissociation of the
molecules, whose dissociation energies are listed in Table 4. It is noticeable that CO molecule
remains distinct to exhibit chemisorption on vG without dissociation. Hence, the defected
graphene would be more selective towards the detection of CO gas. For example, using Table 4,
one can note that O2 molecule requires about 5.11 eV (at 0 oK) to get dissociated then bind to
the vacancy. Whereas, in binding to the vacancy, CO molecule avoids to pay that huge
dissociation energy. This fact makes CO molecule easier to couple with a vacancy with a chance
and a mechanism faster than the chemisorption of O2 molecule with a vacancy. Hence, one may

16
conclude from this analysis that while defected graphene is sensitive towards detecting CO and
O2 gases, it is rather more selective to detect CO gas. Furthermore, as can be noticed from
Table 3 and is mentioned in section 2, the binding energy of a physisorption case is shown to be
Ebind < 1.0 eV; whereas Ebind > 1.0 eV in a case of chemisorption. Last but not least,
benchmarking to other ab-initio results is crucially important. Table 3 shows a comparison of
the binding energies for all molecules on vG obtained by both DFTB and Quantum Espresso
methods. There are good agreements between the results of both methods in respect to the
energetics (namely, the binding energy of molecule to vG). The small discrepancy in binding
energy, in almost all cases except adsorption of O2 where huge dissociation energy gets into
account, basically stems from the difference in the technicality of the two methods and more
specifically the difference in the MP mesh.

3.6. Chemisorption of CO2

To illustrate a possible scenario, in which chemisorption of CO2 could take place, Figure 8 shows
the initial and final states of relaxation carried out by DFTB code. Somehow, if the CO2 molecule
nominally penetrates into the vicinity of the vacancy, then a chemisorption process could start
leading to the complete dissociation of the molecule. Figure 8a shows both top-view and side-
view of the possible initial state configuration, where CO2 molecule is nominally merged inside
the vacancy. The atomic relaxation would yield the dissociation of the molecule (CO 2  CO +
O), where one oxygen atom makes two sigma bonds (one with the old C atom while filling the
vacancy and the second oxygen atom couples with the neighboring C atom of vG). This second
dissociated oxygen atom stabilizes a bit far from the place of “old” vacancy on the other surface
of the vG. Figure 8b shows both top-view and side-view of the relaxed configuration. The
binding energy of such chemisorption is -1.023 eV but this also includes the energy cost of the
dissociation process, which is about 5.46 eV (at 0 oK as shown in Table 4). Furthermore,
chemisorption of CO can be initiated by electrostatic interaction while the dangling bonds of
the molecule and on the vacancy site would help processing it. In contrast, the chemisorption
of CO2 should be initiated by kinetic energy and occurs if the molecule succeeds to reach the

17
heart of vacancy vicinity in nominal incidence. For the above reasons, the chemisorption of CO
molecule has more chances to occur than CO2 molecule.

4. Conclusions

Using ab-initio methods, the adsorptions of CO and CO2 have been investigated on graphene
sheet, in three forms: (i) pristine graphene, (ii) Stone-Wales defected graphene, and (iii)
vacancy-defected graphene. Our results show the occurrence of a chemisorption of CO and CO2
molecules to possibly take place on only vacancy-defected graphene, which is considered the
most active adsorbent and thus has been extensively studied in the present investigation.

To investigate the response (i.e., sensitivity), we have considered the adsorption of several CO
molecules on multiple-vacancy graphene (i.e., which is similar to chemisorption of CO2 on vG)
and found that the total density of states (TDOS) at Fermi level rises with the increase of gas
dose. This enhancement causes rise of conductance. Consequently, the response increases
linearly with the gas dose in similar way to the experimental findings reported by Yoon and
coworkers [8].

To inspect the problem of selectivity, we have studied the adsorption of eight gases (CO, CO2,
N2, O2, H2, H2O, H2S, and NH3) on vacancy-defected graphene. The results show the occurrence
of physisorption with 3 gases (N2, H2O, and H2S) and chemisorption with 5 gases (CO, CO2, O2,
H2, and NH3). In exception of CO molecule, the chemisorption of gas molecules on vG involves
the dissociation of the molecule (which is very costly at order of 5 eV), whereas the one of CO is
direct (i.e., process does not require dissociation and should be easier and faster). Thus, the
results suggest that vacancy-defected graphene (vG)-based sensors should be highly sensitive
and selective towards the detection of CO gas. The obtained trends using DFTB technique are
actually benchmarked and corroborated by the Quantum Espresso results.

18
Acknowledgements

The authors are indebted to thank Drs. Ahmad Ayesh and Bashar Issa for their critical readings
of the manuscript. The students K.S. and N.M. were supported by the UAEU summer
undergraduate research experience program (SURE-2015, grant No. 31S178). We also thank the
UAEU-UPAR (grant No. 31S057), the UAEU Research-Center based (grant No. 31R068) and the
KFUPM for their partial financial supports.

19
References
[1] R.F. Service, Science 305 (2004) 962.

[2] A.B. Rao and A.R. Rubin, Environ Sci. Technol. 36 (2002) 4467.

[3] G.F. Fine, L.M. Cavanagh, A. Afonja, and R. Binions, Sensors 10 (2010) 5469.

[4] P. Nugent, Y. Belmabkhout, S.D. Burd, A.J. Cairns, R. Luebke, Nature 495 (2013) 80.

[5] O. Shekhah, Y. Belmabkhout, Z. Chen, V. Guillerm, A. Cairns, K. Adil, and M. Eddaoudi,


Nature Commun. 5 (2014) 4228.

[6] M.S. Shafeeyan, W.M.A. Wan Daud, A. Houshmand, A. Shamiri, J. Anal. Appl. Pyrolysis 89
(2010) 143.

[7] A.K. Mishra and S. Ramaprabhu, Energy Environ. Sci. 4 (2011) 889.

[8] H.J. Yoon, D.H. Jun, J.H. Yang, Z. Zhou, S.S. Yang, M.M. Cheng, Sensor & Actuators B 157
(2011) 310.

[9] K.K. Paulla and A.A. Farajian, J. Phys. Chem. C 117 (2013) 12815.

[10] Q-Y. Wu, J.H. Lan, C.Z. Wang, C.L. Xiao, Y.L. Zhao, Y.Z. Wei, Z.F. Chai and W.Q. Shi, J. Phys.
Chem. A 118 (2014) 2149.

[11] Q-Y. Wu, J.H. Lan, C.Z. Wang, Y.L. Zhao, Z.F. Chai and W.Q. Shi, J. Phys. Chem. A 118 (2014)
10273.

[12] G. Zhao, L. Jiang, Y. He, J. Li, H. Dong, X. Wang and W. Hu, Adv. Mat. 23 (2011) 3959.

[13] Z.Q. Bai, Z.J. Li, C.Z. Wang, L.Y. Yuan, Z.R. Liu, J. Zhang, L.R. Zheng,Y.L. Zhao, Z.F. Chai and
W.Q. Shi, RSC Adv. 4 (2014) 3340.

[14] Y. Sun, S. Yang, Y. Chen, C. Ding, W. Cheng and X. Wang, Environ. Sci. Technol. 49 (2015)
4255.

[15] Z. Jin, X. Wang, Y. Sun, Y. Ai and X. Wang, Environ. Sci. Technol. 49 (2015) 9168.

[16] Y.J. Zhang, J.H. Lan, C.Z. Wang, Q.Y. Wu, T. Bo, Z.F. Chai and W.Q. Shi, J. Phys. Chem C. 119
(2015) 5783.

[17] T. Bo, J.H. Lan, C.Z. Wang, Y.L. Zhao, C.H. He, Y.J. Zhang, Z.F. Chai and W.Q. Shi, J. Phys.
Chem. C 118 (2014) 21935.

20
[18] B. Huang, Z. Li, Z. Liu, G. Zhou, S. Hao, J. Wu, B.L. Gu and W. Duan, J. Phys. Chem. C 112
(2008) 13442.

[19] O. Leenaerts, B. Partoens and F.M. Peeters, Phys. Rev. B 77 (2008) 125416.

[20] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, Y. Zhang, S.V. Dubonos, I.V. Grigorieva,
A.A. Firsov, Science 306 (2004) 666.

[21] K.S. Novoselov, V.I. Fal’ko, L. Colombo, P.R. Gellet, M.G. Schwab, K. Kim, Nature 490 (2012)
192.

[22] F. Schwierz, Nat. Nanotechnol. 5 (2010) 487.

[23] Y.W. Son, M.L. Cohen, S.G. Louie, Phys. Rev. Lett. 97 (2006) 216803.

[24] I. Ci, I. Song, C. Jin, D. Jariwala, D. Wu, Y. Li, A. Srivastava, Z.F. Wang, K. Storr, I. Balicas, F.
Liu, P.M. Ajayan, Nat. Mater. 9 (2010) 430.

[25] D. Soriano, F. Munoz-Rojas, J. Fernandez-Rossier, J.J. Palacios, Phys. Rev. B 81 (2010)


165409.

[26] R.K. Joshi, H. Gomez, F. Alvi, A. Kumar, J. Phys. Chem. C 114 (2010) 6610.

[27] Y.H. Zhang, Y.B. Chen, K.G. Zhou, C.H. Liu, J. Zneg, H.L. Zhang, Y. Peng, Nanotechnol. 20
(2009) 185504.

[28] S.S. Varghese, S. Lonkar, K.K. Singh, S. Swaminanthan, A. Abdala, Sensors & Actuators B:
Chemical 218 (2015) 160.

[29] F. Banhart, J. Kotakoski and A.V. Krasheninnikov, ACS Nano 5 (2011) 26.

[30] A.J. Stone, D.J. Wales, Chem. Phys. Lett. 128 (1986) 501.

[31] L. Li, S. Reich, J. Robertson, Phys. Rev. B 72 (2005) 184109.

[32] J. Ma, D. Alfe, A. Michaelides, E. Wang, Phys. Rev. B 80 (2009) 033407.

[33] A.V. Krasheninnikov, P.O. Lehtinen, A.S. Foster, R.M. Nieminen, Chem. Phys. Lett. 418
(2006) 132.

[34] A.A. El-Barbary, R.H. Telling, C.P. Ewels, M.I. Heggie, P.R. Briddon, Phys. Rev. B 68 (2003)
144107.

[35] N. Fukata, A. Kasuya, M. Suezawa, Physica B 308-310 (2001) 1125.

21
[36] B. Sanyal, O. Eriksson, U. Jansson, H. Grennberg, Phys. Rev. B 79 (2009) 113409.

[37] M. Ricco, D. Pontiroli, M. Mazzani, M. Choucair, J.A. Stride, O.V. Yazyev, Nano Lett. 11
(2011) 4919.

[38] T.P. Kaloni, Y.C. Cheng, R. Faccio, U. Schwingerschlogl, J. Mater. Chem. 21 (2011) 18284.

[39] V.I. Hegde, S.N. Shirodkar, N. Tit, U.V. Waghmare, Z.H. Yamani, Surf. Sci. 621 (2014) 168.

[40] S. Kouser, U.V. Waghmare, and N. Tit, Phys. Chem. Chem. Phys. 16 (2014) 10719.

[41] A. Allouche, Y. Ferro, Carbon 44 (2006) 3320.

[42] B. Huang, Z. Li, Z. Liu, G. Zhou, S. Hao, J. Wu, B.L. Gu, W. Duan, J. Phys. Chem. C 112 (2008)
13442.

[43] P. Cabrera-Sanfelix, J. Phys. Chem. A 113 (2009) 493.

[44] Y. Liu, J. Wilcox, Environ. Sci. Technol. 45 (2011) 809.

[45] T. Frauenheim et al., J. Phys. Condens. Matter 14 (2002) 3015; www.dftb.org.

[46] P. Koskinen, V. Mäkinen, Comp. Mat. Sci. 47 (2009) 237.

[47] J.C. Slater and G.F. Koster, Phys. Rev. 94 (1954) 1498.

[48] M. Elstner, D. Porezag, G. Jungnickel, J. Elsner, M Haugk, Th. Frauenheim, Phys. Rev. B 58
(1998) 7260.

[49] T.A. Niehaus, M. Elstner, Th. Frauenheim, S. Suhai, J. Mol. Struct. (Theochem) 541 (2001)
185.

[50] C.M. Goringe, D.R. Bowler and E. Hernandez, Rep. Prog. Phys. 60 (1997) 1447.

[51] M. Elstner, Theor. Chem. Acc. 116 (2006) 316.

[52] P. Giannozzi, S. Baroni, et al., J. Phys: Condens. Matter 21 (2009) 395502

[53] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892-7895.

[54] J.P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865-3868.

[55] S. Grimme, J. Comput. Chem. 27 (2006) 1787-1799.

[56] H.J. Monkhorst and J.D. Pack, Phys. Rev. B 13 (1976) 5188.

22
[57] F. Rouquerol, J. Rouquerol, K.S.W. Sing, P. Llewellyn, and G. Maurin, “Adsorption by
Powders and Porous Solid: Principles, Methodology and Applications”, 2nd Ed. (Elsevier, Oxford,
UK, 2014).

[58] M. Batzill and U. Diebold, Phys. Chem. Chem. Phys. 9 (2007) 2307.

[59] N.W. Ashcroft and N.W. Mermin, “Solid State Physics”, Ed. D.G. Crane (Harcourt College

Publishers, New York, 1976), Chapter 2, p.29.

[60] J.M. Ziman, “Principles of the theory of Solids”, 2nd Edition (Cambridge University Press,

1979), Chapter 7, p. 211.

[61] A.I. Ayesh, J. Alloys Compd. 689 (2016) 1-5.

[62] A.I. Ayesh, A.F.S. Abu-Hani, S.T. Mahmoud and Y. Haik, Sens. Actuators B 231 (2016) 593.

[63] M. Abu-Haija, A.I. Ayesh, S. Ahmed and M.S. Katsiotis, Appl. Surf. Sci. 369 (2016) 443.

[64] V.I. Vedeneyev, L.V. Gurvich, V.N. Kondrat’yev, V.A. Medvedev, and Y.L. Frankevich, “Bond
Energies, Ionization Potentials, and Electron Affinities” (St. Martin’s Press, New York, 1962).

[65] R.C. Shiell, X.K. Hu, Q.J. Hu, J.W. Hepburn, J. Phys. Chem. A 104 (2000) 4339.

[66] D.D. Wagman, W.H. Evans, V.B. Parker, I. Halow, S.M. Bailey, and R.H. Schumm, Nat. Bur.
Stand (U.S.), Tech. Note 270-4 (1969).

23
Figure Captions
Figure 1: Relaxed atomic structures of systems: (a) Pristine graphene; (b) Graphene with a
single vacancy; (c) CO2 molecule in physisorption on vacancy-defected graphene; (d) CO
molecule in chemisorption on vacancy-defected graphene. EF is Fermi energy of the system as
referred to vacuum level; and Ebind is the binding energy of the molecule, both given in eV units.
“D” is the molecule-to-surface distance given in Å units.

Figure 2: Band structures corresponding to the four systems of Figure 1 are shown with Fermi
energy taken as an energy reference (i.e., EF = 0). Ebind and NF stand for the binding energy (in
eV units) and DOS at Fermi level (in units of 1/eV per Hexagon).

Figure 3: Densities of states for the same systems of Figure 1 are shown with Fermi energy
taken as an energy reference. Panel (c) shows discrete oxygen states revealing the
physisorption of CO2 on vG whereas panel (d) shows formation of oxygen band as part of TDOS
revealing the chemisorption of CO molecule on vG. EF, Ebind and D stand for Fermi energy,
binding energy and molecule-surface distance, respectively.

Figure 4: Relaxed atomic structures of multiple-vacancy-defective graphene versus CO


molecules (dose) (i.e., equivalent to adding O atoms on pristine graphene, where number of
oxygen atoms is varied 1, 2, 4, and 6 in respective panels 4 a-d). Panels 4e-f show the critical
dose of 4 oxygen atoms added on pristine graphene of size 5x5 unit cells, where the relaxation
might yield coexistence of chemisorption and physisorption.

Figure 5: (a) Partial densities of states (PDOSs) of cases of 1, 2, 4, 6 ad oxygen atoms on


pristine graphene sample of size 5x5 unit cells. Fermi energy EF is taken as an energy reference.
The inset shows DOS at Fermi level. (b) DOS at Fermi level versus number of O ad-atoms (i.e.,
like versus CO gas dose) is shown in linear scale. (c) Same as Panel 5b but is shown in semi-log
scale.

Figure 6: Estimated response versus number of CO molecules chemisorbed on vacancy-


defected graphene. Sample taken is of size 5x5 unit cells.

Figure 7: Comparison of adsorption of one O2 molecule on: (a,c) Pristine graphene and (b,d)
single-vacancy defected graphene, to show physisorption and chemisorption, respectively. The
chemisorption of O2 molecule is associated with dissociation.

Figure 8: Scenario of possible occurrence of chemisorption of CO2 molecule on a single-


vacancy-defected graphene: (a) Before relaxation, and (b) After relaxation.

24
(a) Pristine Graphene (b) Graphene with Vacancy

(c) Physisorption of CO2 on Vacancy (d) Chemisorption of CO on Vacancy

Top View Top View

Side View Side View

FIGURE-1

25
(a) (b)

(c) (d)
d)

FIGURE-2

26
(a) (b)

(c) (d)

FIGURE-3

FIGURE-3

27
(a) 1 O atom on pG (relaxed) (b) 2 O atoms on pG (relaxed)

Top View Top View

Side View Side View

(c) 4 O atoms on pG (relaxed) (d) 6 O atoms on pG (relaxed)

Top View Top View

Side View Side View

(e) 4 O atoms on pG (f) 4 O atoms on pG


(before relaxation) (After relaxation)

Top View Top View

28
Side View Side View

FIGURE-4

29
FIGURE-5

30
FIGURE-6

31
(a) 1 O2 Molecule on pG (b) 1 O2 Molecule on Vacancy

Top View Top View

Side View Side View

(c) DOS of system above (d) DOS of system above

FIGURE-7

32
(a) 1 CO2 Molecule on vG (b) 1 CO2 Molecule on vG
(Before Relaxation) (After Relaxation)

Top View Top View

Side View Side View

FIGURE-8

33
Table Captions

Table 1: Binding energies of CO and CO2 on various adsorbents, namely: (i) Pristine graphene;
(ii) Vacancy-defected graphene; and (iii) Stone-Wales-defected graphene.

Table 1
Adsorbent CO molecule CO2 molecule
EF (eV) Ebind (eV) EF (eV) Ebind (eV)
{D(Å)} {D(Å)}
Pristine Graphene -4.650 -0.125 -4.644 -0.184
{3.467} {3.345}
SW-defected -4.661 -1.301 -4.655 -1.371
Graphene {3.00} {2.96}
Vacancy-defected -4.652 -5.063 -4.695 -0.685
Graphene {Chemisorption} {3.30}

34
Table 2: Chemisorption energy on vacancy-defected graphene versus CO gas dose is given by
Ebind in eV-units. Fermi energy is given with respect to vacuum level. Average charge per oxygen
atom and DOS at Fermi level are also shown.

Table 2

System EF (eV) Ebind (eV) of Ebind (eV) of O Charge of NF (1/eV-


CO atom on atom on pG Oxygen (e) Hexagon)
vG
1 CO /1 vG -4.652 -5.063 -3.215 6.358 0.0065
2 CO /2 vG -4.810 -5.162 -3.280 6.350 0.0095
4 CO /4 vG -5.003 -3.510 -3.201 6.331 0.0131
6 CO /6 vG -4.913 -3.838 -3.175 6.323 0.0155

35
Table 3: The binding energies for 8 different gas molecules on single-vacancy in graphene
using DFTB code are shown and compared to those obtained by Quantum Espresso package.

Table 3

System EF (eV) Ebind (eV) per D (Å) Adsorption State


molecule
vG-N2 -4.682 -0.604, -0.42b 3.38 Physisorption
vG-H2O -4.874 -0.658, -0.65 b DC-H= 2.22 Physisorption
DC-O= 3.53
vG-H2S -4.879 -0.760, -0.63 b DC-H= 2.07 Physisorption
DC-O= 3.44
vG-CO2 -4.695 -0.686, -0.52 b 3.30 Case of Physisorption
vG-CO2 -4.836 -1.023, -1.11b < 2.0 Case of Chemisorption
(with dissociation)
vG-H2 -4.6537 -4.327, -4.01b < 2.0 Case of Chemisorption
(with dissociation)
vG-NH3 -4.313 -2.616, -2.47b < 2.0 Chemisorption
(with dissociation)
vG-CO -4.652 -5.063, -5.15b < 2.0 Chemisorption
(without dissociation)
vG-O2 -5.170 -5.148, -6.83b < 2.0 Chemisorption
(with dissociation)
(b) Results obtained using Quantum Espresso package.

36
Table 4: Dissociation energies of various molecules at 0 oK and 298 oK (room temperature)
expressed in both kJ/mole and eV/molecule units.

Table 4

Molecule Reaction Edissoc(kJ/mole) Edissoc (kJ/mole)


{eV/molecule} {eV/molecule}
at 0 oK at 298 oK
O2 O2  2 O 493 {5.11}a 498 {5.16}a
H2S H2S  HS + H 376 {3.9}b 381 {3.95}b
HS  S + H 348 {3.61}b 353 {3.66}b
N2 N2  2 N 942 {9.77}a 945 {3.81}a
H2 H2  2 H 432 {4.48}c 436 {4.52}c
NH3 NH3  NH2 + H 431 {4.47}c 435 {4.51}c
H2O H2O  OH + H 493 {5.11}c 498 {5.16}c
CO2 CO2  CO + O 526 {5.46}c 532 {5.52}c
CO CO  C + O 1072 {11.12}c 1076 .16}c
(a) Reference [64], (b) Reference [65], (c) Reference [66]

37

You might also like