Time-dependent density-functional study of hydrogen adsorption and scattering on

graphene surfaces
Samuel S. Taylor,1 Nicholas Skoufis,1 Hongbo Du,2 Cody Covington,3 and Kálmán Varga1, ∗
1
Department of Physics and Astronomy, Vanderbilt University, Nashville, Tennessee, 37235, USA
2
Department of Physics, School of Science, Xi’an Technological University, Xi’an, Shaanxi 710032, China
3
Department of Chemistry, Austin Peay State University, Clarksville, USA
Time-dependent density-functional theory simulations are performed to examine the effects of
varying incident points and kinetic energies of hydrogen atom projectiles on a graphene-like struc-
ture. The simulations reveal that the incident point significantly influences the hydrogen atom’s
kinetic energy post-interaction, the vibrational dynamics of the graphene lattice, and the scattering
arXiv:2412.06939v1 [cond-mat.mtrl-sci] 9 Dec 2024

angles. Incident points that do not directly collide with carbon atoms result in prolonged interaction
times and reduced energy transfer, increasing the likelihood of overcoming the graphene’s potential
energy barrier and hydrogen atom adsorption. The study also explores the role of initial kinetic
energy in determining adsorption, scattering, or transmission outcomes. These results emphasize
the critical influence of initial parameters on the hydrogenation process and provide a foundation
for future experimental validation and further exploration of hydrogen-graphene interactions.

I. INTRODUCTION theoretical and computational methods have been

employed—including classical molecular dynamics (MD)
Graphene, a material consisting of C atoms arranged in simulations and ab initio density functional theory
a honeycomb lattice, is well known for its exceptional me- (DFT) simulations.
chanical strength [1, 2] and remarkable electrical conduc- Classical MD simulations are widely used due to their
tivity [3, 4]. Graphene’s distinctive chemical, electronic, computational efficiency and ability to sample a large
and magnetic properties have made it a focal point of number of conditions [27, 35–43]. For example, classi-
recent research due to its potential applications in elec- cal MD with a reactive empirical bond-order (REBO)
tronics [5–7], solar cells [8, 9], and energy storage [10, 11]. potential has been used to explore how varying incident
The unique properties of graphene have also driven angles and points affect outcomes [36]. However, these
studies into its chemical modification to induce new func- methods inherently neglect quantum effects, which have
tionalities [12–16]. One of the most promising modifica- been shown to significantly influence hydrogen–graphene
tions is hydrogenation, where hydrogen (H) atoms bond interactions, including altering sticking probabilities [34].
with graphene’s carbon (C) atoms to induce a bandgap, DFT is often used as a more precise computational
granting graphene the qualities required for semiconduc- technique for studying H atom and graphene interac-
tor applications [17]. Understanding the interaction of tions, having been proven accurate in describing adsorp-
hydrogen with graphene provides critical insights into tion [23–26, 30] and vibrational relaxation [27]. Despite
aspects of graphene’s behavior and material properties, these successes, previous DFT studies have not system-
such as hydrogen adsorption mechanisms and the for- atically investigated the role of varying the incident point
mation of C–H bonds [18–26], vibrational redistribution within graphene’s benzene rings or its impact on scatter-
reactions [19, 27], hydrogen storage [28–30], and mag- ing angular distributions and energy transfer dynamics.
netism [31, 32]. Additionally, it provides insights into This study aims to address this gap by employing time-
molecular dynamics [33, 34] and helps identify the con- dependent density-functional theory (TDDFT) to inves-
ditions that maximize hydrogen adsorption probabilities tigate the scattering distributions and energy transfer
while minimizing scattering [18, 35, 36] dynamics of hydrogen interacting with graphene. The
The outcomes of hydrogen–graphene interac- use of such simulations allows for detailed snapshots and
tions—adsorption, scattering, or transmission—depend kinetic energy profiles to elucidate the mechanisms un-
on several parameters. Key factors include the kinetic derlying hydrogen scattering and C–H bond formation
energy of the incoming H atom, its incident angle processes. Furthermore, additional simulations explore
relative to the graphene surface, and the specific inci- the influence of varying initial hydrogen velocities while
dent point on the graphene lattice [18, 35, 36]. These keeping the incident point and angle fixed, providing in-
parameters critically influence the scattering angles and sight into how kinetic energy affects H atom adsorption,
energy transfer during the interaction, determining the scattering, and transmission reactions.
overall dynamics and reaction pathways. To identify
the optimal combination of parameters that maximize
hydrogen adsorption on the graphene surface, various II. COMPUTATIONAL METHOD

The simulations were performed using TDDFT for

∗ [email protected] modeling the electron dynamics on a real-space grid with
 2

real-time propagation [44], with the Kohn-Sham (KS) the simulations is determined by the grid spacing, which
Hamiltonian of the following form is the key parameter that can be adjusted. In our simu-
lations, we used a grid spacing of 0.25 Å and placed 81
ℏ2 2 points along each of the x- and y-axes, and 65 points are
ĤKS (t) = − ∇ + Vion (r, t) + VH [ρ](r, t)
2m (1) placed along the z-axis.
+VXC [ρ](r, t). Motion of the ions in the simulations were treated clas-
sically. Using the Ehrenfest theorem , the quantum forces
Here, ρ is the electron density, defined as the sum of the on the ions due to the electrons are given by the deriva-
densities of all occupied orbitals: tives of the expectation value of the total electronic en-
∞ ergy with respect to the ionic positions. These forces are
then fed into Newton’s Second Law, giving
X
ρ(r, t) = fk |ψk (r, t)|2 , (2)
k=1
N
d 2 Ri X ions
Zi Zj (Ri − Rj )
where fk is the occupation number of the orbital ψk , Mi = Z i E laser (t) +
dt2 |Ri − Rj |3
which can take values 0, 1, or 2 (fk = 2 is allowed due Z
j̸=i (7)
to spin degeneracy). Additionally, fk must satisfy the
P∞
constraint k=1 fk = N , where N is the total number of −∇Ri Vion (r, Ri )ρ(r, t) dr,
valence electrons in the system.
Vion in eq. 1 is the external potential due to the ions, where Mi , Zi , and Ri are the mass, pseudocharge (va-
represented by employing norm-conserving pseudopoten- lence), and position of the i-th ion, respectively, and Nions
tials centered at each ion as given by Troullier and Mar- is the total number of ions. This differential equation was
tins [45]. VH is the Hartree potential, defined as time propagated using the Verlet algorithm at every time
step δt.
ρ(r′ , t) ′
Z
VH (r, t) = dr , (3) To model a graphene sheet, coronene (C24 H12 ) was
|r − r′ | selected due to its structural similarity to graphene.
Coronene, depicted in Fig. 1, consists of 7 benzene rings
and accounts for the electrostatic Coulomb interactions at its core, resembling the honeycomb lattice of graphene.
between electrons. The last term in eq. 1, VXC , is the Its relatively small size allows it to fit comfortably within
exchange-correlation potential, which is approximated by the simulation grid. A larger molecule would significantly
the adiabatic local-density approximation (ALDA), ob- increase the grid dimensions and highly increase compu-
tained from a parameterization to a homogeneous elec- tational costs. Graphene super cell with periodic bound-
tron gas by Perdew and Zunger [46]. ary conditions also computationally expensive because a
At the beginning of the TDDFT calculations, the large supercell has to be used to avoid the interaction
ground state of the system is prepared by performing between neighboring cells and allow scattering with an
a Density-Functional Theory (DFT) calculation. With angle with respect to the surface.
these initial conditions in place, we then proceed to prop- The coronene molecule was positioned at z = -2 Å
agate the Kohn–Sham orbitals, ψk (r, t) over time by us- in the xy-plane, and H atom was positioned 7 Å above
ing the time-dependent KS equation, given as the coronene at z = 5 Å. These positions fit well within
∂ψk (r, t) the 20 Å × 20 Å × 16 Å grid centered at the origin,
i = Ĥψk (r, t). (4) with sufficient space along the z-axis to accommodate
∂t
the trajectory of the hydrogen projectile.
Eq. 4 was solved using the following time propagator In the following section, we present the results of 28
! simulations in which a hydrogen projectile was directed
iĤKS (t)δt at a 4 × 7 grid of incident points on the graphene-like
ψk (r, t + δt) = exp − ψk (r, t). (5) target surface. To further investigate the interaction dy-
ℏ
namics and explore conditions favorable for hydrogen ad-
This operator is approximated using a fourth-degree Tay- sorption, an additional set of 7 simulations was performed
lor expansion, given as with varying projectile kinetic energies at a distinct in-
cident point and angle separate from the first set of 28
4  n simulations.
X 1 −iδt
ψk (r, t + δt) ≈ ĤKS (t) ψk (r, t). (6)
n=0
n! ℏ
III. RESULTS
The operator is applied for N time steps until the final
time, tf inal = N ·δt, is obtained. A time step of δt = 1 as
was used in the simulations. A. Incident Points and Scattering
In real-space TDDFT, the Kohn-Sham orbitals are rep-
resented at discrete points in real space. These points are Fig. 1 illustrates the incident points where the hy-
organized on a uniform rectangular grid. The accuracy of drogen was aimed in each of the 28 simulations. The
 3

FIG. 1. Diagram illustrating the incident points of H atom projectiles on the xy-plane of the coronene molecule.

coronene sheet that the hydrogen was aimed at lies in the x and z directions.
the xy-plane. A 4 × 7 point grid of incident points was The initial positions of the H atoms were selected
used for the x and y values, respectively, with a grid spac- to align with the incident points illustrated in Fig. 1.
ing of 0.3 Å. In all simulations, the hydrogen projectile The height above the coronene molecule was consistently
was launched at an angle of 27.4◦ from the z-axis (or- maintained at 7 Å for all simulations, ensuring unifor-
thogonal to the coronene plane) with a kinetic energy of mity in the initial z-coordinate. Additionally, the initial
1.89 eV (velocity of 0.19 Å/fs). These parameters for velocity and its angle were kept constant across all sim-
the initial kinetic energy and angle were motivated by ulations. As a result, the only variable between simula-
a recent study [18]. The initial kinetic energy remained tions was the H atom’s initial position in the xy-plane at
identical across all simulations, with components only in z = 5 Å, enabling the H atom to precisely target each of
 4

FIG. 2. Kinetic energy (in eV) of the H atom (red solid line) and the net kinetic energy of all atoms in the coronene sheet
(green dashed line) as a function of time throughout the simulation. Each graph corresponds to a different simulation from the
column of x1 incident points, ranging from y1 to y7 .

FIG. 3. Kinetic energy (in eV) of the H atom (red solid line) and the net kinetic energy of all atoms in the coronene sheet
(green dashed line) as a function of time throughout the simulation. Each graph corresponds to a different simulation from the
column of x2 incident points, ranging from y1 to y7 .

the incident points shown in Fig. 1. barrier. In all simulations, the H atom failed to penetrate
Fig. 2 displays the kinetic energy (in eV) of the H atom this barrier or form a bond with a C atom. Instead, the H
throughout the simulations for all incident points along atom experienced strong repulsion, leading to a sudden
the x1 column. In most cases, the H atom reaches the change in direction (e.g., from 35 fs to 40 fs at x1 y4 ).
coronene surface at approximately 35 fs. As the H atom After being repelled away from the coronene (beyond
approaches the coronene (e.g., during 30 fs to 35 fs at 40 fs), the H atom’s velocity gradually decreases, sug-
incident point x1 y4 ), its kinetic energy sharply decrease gesting the presence of an attractive force exerted by the
due to interactions with the coronene’s potential energy coronene. This deceleration stabilizes into a constant ve-
 5

FIG. 4. Kinetic energy (in eV) of the H atom (red solid line) and the net kinetic energy of all atoms in the coronene sheet
(green dashed line) as a function of time throughout the simulation. Each graph corresponds to a different simulation from the
column of x3 incident points, ranging from y1 to y7 .

FIG. 5. Kinetic energy (in eV) of the H atom (red solid line) and the net kinetic energy of all atoms in the coronene sheet
(green dashed line) as a function of time throughout the simulation. Each graph corresponds to a different simulation from the
column of x4 incident points, ranging from y1 to y7 .

locity once the H atom reaches a sufficient distance from energy transfer is largely converted into vibrational en-
the coronene, where it is no longer influenced by signifi- ergy within the coronene molecule, as indicated by the
cant forces. initial rise in net kinetic energy followed by oscillatory
The scattered H atom transferred a portion of its ki- behavior observed after 40 fs.
netic energy to the coronene structure, as evidenced in Comparing the kinetic energy profiles of the H atom
each kinetic energy graph where the total kinetic energy across all simulations in the x1 column (Fig. 2) provides
of all atoms in the coronene begins to increase upon in- important insights into the interaction dynamics. No-
teraction with the H atom at around 35 fs (Fig. 2). This tably, the interaction time between the H atom and the
 6

coronene sheet is significantly longer for incident points x1 x2 x3 x4

that are farther from the C atoms. For instance, at x1 y1 , y1 1.24 0.76 0.91 1.00
the incident point furthest from the C atoms in the x1 y2 1.28 1.28 1.18 1.21
column, the interaction lasts approximately 20 fs. Dur- y3 1.19 1.27 1.40 1.24
ing this period, the H atom’s kinetic energy decreases y4 1.32 1.31 1.46 1.23
before reaching the coronene (from 30 fs to 40 fs), fol- y5 1.13 1.41 1.34 1.47
lowed by a prolonged escape phase after scattering (from y6 1.18 1.15 1.54 1.38
40 fs to 50 fs). In contrast, at points closer to the C y7 1.35 1.30 1.54 1.56
atoms, such as x1 y7 , the interaction is much shorter. At
x1 y7 , the velocity decreases around 35 fs, reaches a min- TABLE I. Energy differences (in eV) of the H atom between
imum at 38 fs, and then increases again until 40 fs. This its initial kinetic energy (1.89 eV) and the kinetic energy after
comparison highlights that the interaction time at x1 y1 the collision, measured at each xi yj incident point.
is roughly 20 fs—four times longer than the 5 fs interac-
tion observed at x1 y7 , where the H atom interacts more
x1 x2 x3 x4
briefly with the coronene.
y1 0.25 0.41 0.39 0.42
H atoms aimed at incident points farther from C atoms y2 0.52 0.48 0.47 0.48
experience longer interaction times because they are not y3 0.45 0.50 0.50 0.53
immediately reflected or scattered upon approaching the y4 0.48 0.44 0.52 0.54
graphene surface. This suggests that it is easier for the y5 0.53 0.47 0.49 0.56
H atom to penetrate coronene’s potential energy barrier y6 0.57 0.51 0.44 0.54
at these incident points. Conversely, when the H atom y7 0.54 0.51 0.47 0.53
is aimed directly at a C atom, the repulsion between
the positively charged nuclei and their associated elec- TABLE II. Vibrational energy differences (in eV) of the
tron densities causes immediate reflection. This limits corenene between the ground state and after hydrogen col-
the time the H atom spends near the graphene plane, re- lision at each xi yj incident point.
ducing its ability to overcome the potential energy bar-
rier, rehybridize the C atom from sp2 to sp3 , and form a
C-H bond. initial kinetic energy at 0 fs to its final kinetic energy at
This trend—incident points further from C atoms hav- 80 fs. The exact values of kinetic energy loss for the H
ing a longer interaction period as opposed to immediate atom across all incident points are provided in Table I.
reflection—persists across all kinetic energy graphs, as This lost energy is absorbed by the coronene and con-
shown in Fig. 2, Fig. 3, Fig. 4, and Fig. 5. In each of verted into vibrational motion of its atoms, as shown in
these figures, the y1 incident points, which are furthest Fig. 2–5 as well. The energy transfer is further illus-
from any C atoms, exhibit significantly longer interaction trated in Table II, which displays the kinetic energy sum
periods, with changes in kinetic energy beginning around of all atoms (vibrational energy) in the coronene at 80 fs,
30 fs and continuing until approximately 50 fs. In con- highlighting the increase due to this interaction.
trast, incident points near C atoms, such as x2 y6 and Based on Table I, the calculated mean H atom kinetic
x3 y6 , demonstrate rapid, immediate repulsion and sharp energy loss is 1.27 eV. The kinetic energy loss exceeds
kinetic energy changes. 1 eV in all cases except for x2 y1 , x3 y1 , and x4 y1 , where
The increased probability of C-H bond formation at the losses are 0.76 eV, 0.91 eV, and 1.00 eV, respec-
these incident points is particularly evident at the x4 y1 tively. This lower kinetic energy loss at incident points
incident point (Fig. 5). In this graph, the H atom kinetic further from C atoms is consistent with the previously
energy reaches a relative maximum at 40 fs. The kinetic discussed findings, where these points exhibit longer in-
energy decrease following the peak suggests the presence teraction times and are more likely to form a C-H bond.
of additional attractive forces within the coronene that The smaller energy loss at these incident points suggests
act on the H atom after it reaches the incident point. that the H atom transfers less energy to the coronene
Although the H atom was ultimately scattered (as ob- compared to points where the H atom is immediately re-
served in all simulations across various incident points), flected from the C atom it strikes, such as x3 y6 , where
the extended interaction period and the attractive forces the H atom loses 1.54 eV (see Table I). This observation
from the electron densities of the C atoms allowed the aligns with the energy differences in coronene, as shown
H atom to remain near the coronene plane for a longer in Table II, where the y1 row of incident points (x1 y1 ,
duration. This increased interaction time and attractive x2 y1 , x3 y1 , and x4 y1 ) exhibit the smallest energy dif-
forces, shown in kinetic energy graphs for the x1 y1 , x2 y1 , ferences. The average energy difference for all incident
x3 y1 , and x4 y1 (Fig. 2-5), enhances the probability of points, as detailed in Table II, is 0.48 eV.
C-H bond formation. Notably, the energy differences for the H atom (Ta-
During the scattering interaction between the H atom ble I) are much larger than those for coronene (Table II).
and coronene, energy transfer occurs. The H atom loses This discrepancy arises because not all of the kinetic en-
kinetic energy, as shown in Fig. 2–5, by comparing its ergy of the H atom is transferred to coronene as vibra-
 7

tional energy—some of the energy is dissipated within the Furthermore, the data reveal that the hydrogen generally
electron density. As shown in all the kinetic energy fig- reflects at an angle closer to the coronene plane than its
ures (Fig. 2–5), the vibrational energy of the coronene initial trajectory. In all simulations, the hydrogen ap-
molecule increases upon interaction with the H atom. proaches the graphene sheet at an angle of 27.4◦ and, on
However, after the initial interaction, this energy oscil- average, reflects at an angle of 33.5◦ .
lates as the molecules move, stretch, and repel each other. Interestingly, the scattering angles for the first row of
The increase in vibrational energy following the hydro- incident points—previously identified as having a higher
gen scattering event results from the excitation of the probability of adsorption—do not show a significant de-
coronene structure due to the energy transferred by the viation from the overall average reflection angle. The
H atom. reflection angles for incident points x1 y1 , x2 y1 , x3 y1 , and
x4 y1 are 63.7◦ , 43.7◦ , 15.2◦ , and 14.2◦ , respectively. The
average reflection angle for these points is 34.2◦ , closely
aligning with the overall average reflection angle of 33.5◦
across all incident points.

B. Variable Kinetic Energy and Adsorption

Fig. 7 presents snapshots from a simulation where the

H atom was successfully adsorbed onto the graphene sur-
face. In this simulation, the initial kinetic energy of the
H atom was identical to that used in Fig. 2-5 (1.89 eV),
but the incident angle was adjusted to 35◦ relative to the
z-axis, making it more parallel to the xy-plane compared
to the 27.4◦ angle in the previous simulations. The H
atom was targeted at an incident point of -0.82 Å on the
x-axis, chosen based on earlier findings that the y1 row
of incident points maximizes the probability of hydrogen
adsorption. Additionally, the angle was altered to make
the hydrogen approach closer to the C atoms, facilitat-
ing its ability to form a C-H bond after overcoming the
FIG. 6. Angular distribution of scattered H atoms from the coronene potential energy barrier.
28 simulations. The plot shows the final kinetic energy (at As illustrated in Fig. 7, the H atom initially breaks
80 fs) and the angle θ, which is the angle between the velocity through the potential energy barrier of the coronene and
vector and the z-axis. The colors red, blue, yellow, and green forms a transient bond at 45 fs. At 68 fs, the hydrogen is
correspond with simulations aimed at the x1, x2, x3, and x4 displaced and temporarily bonds with another C atom.
columns of incident points, respectively.
By the end of the simulation at 85 fs, the hydrogen forms
a stable bond with a C atom at the edge of the coronene
The scattering angles and kinetic energies of the H molecule. Although this final bonding is influenced by
atoms as they are deflected from the coronene surface edge effects inherent to the coronene model, the H atom
are presented in Fig. 6. This figure demonstrates the sig- double-bounce trajectory and kinetic energy loss transfer
nificant influence of the incident point on the scattering following 45 fs suggests that, in a larger graphene system,
angle and final kinetic energy of the H atoms. Depend- a bond would likely form with a C atom in an interior
ing on the specific impact location, the hydrogen can be benzene ring.
scattered nearly orthogonal to the coronene surface (e.g., Throughout this process, the H atom’s kinetic energy
x2 y4 , which reflects with an angle of 3.4◦ ) or at a less or- is transferred to the coronene structure, as shown in the
thogonal angle (e.g., x4 y7 , which scatters with an angle kinetic energy graph in Fig. 7, by the increasing net ki-
of 72.9◦ ). The incident point x2 y1 results in the high- netic energy of the coronene from 35 fs onward. This
est final kinetic energy (1.14 eV), while x4 y7 produces rise in energy corresponds to the vibrational excitation
the lowest (0.33 eV). This wide variability in scattering of the coronene atoms, which absorb the incoming hy-
angles and velocities underscores the critical role of the drogen’s kinetic energy. Unlike scattering interactions,
incident point in determining the H atom’s trajectory where some of the kinetic energy remains with the hydro-
and kinetic energy after interaction. gen, in this adsorption scenario, the hydrogen’s kinetic
From the data, the mean and median final kinetic en- energy is entirely transferred to the coronene, initiating
ergy of the H atoms are 0.62 eV, and the mean deflection vibrational motion across the molecule.
angle is 33.5◦ . These results indicate that, on average, Seven simulations exploring the effect of varying the
the H atom loses 1.42 eV of kinetic energy after scatter- initial kinetic energy of the H atom are summarized in
ing, as corroborated by the mean of the values in Table I. Table III. In each simulation, the H atom was launched
 8

Simulation Initial KE Result Angle of KE loss IV. SUMMARY

(eV) (A/S/T) reflection (◦ ) (eV)
1 1.89 A ... ... Time-dependent density functional theory (TDDFT)
2 2.07 A ... ... simulations were conducted to investigate the effects of
3 2.51 A ... ... varying incident points and velocities of H atom projec-
4 3.50 A ... ...
tiles on a graphene-like structure. The findings reveal
5 4.66 S 79.87 3.91
6 6.35 S 57.34 3.45 that the location where the H atom strikes the graphene
7 9.14 T -38.33 4.29 significantly influences its kinetic energy, the vibrational
motion of the graphene lattice, and the angles at which
the H atoms are scattered.
TABLE III. Outcomes of 7 simulations with varying initial Clear trends emerged in the data based on the inci-
kinetic energy of the H atom. The collision results in one of dent points. Incident points farther from C atoms result
three events: adsorption (A), scattering (S), or transmission in longer interaction times between the H atom and the
(T). For cases where the H atom does not form a bond with
graphene, leading to smaller kinetic energy differences for
any C atom in the coronene (S and T), the angle of reflection
and the kinetic energy loss are reported. both the H atom and the coronene structure after scatter-
ing. This extended interaction time and reduced energy
transfer increases the likelihood of the H atom overcom-
from the same initial position, with identical orientation ing the graphene’s potential energy barrier, rehybridizing
and incident point on the coronene. Consistent with the a C atom from sp2 to sp3 , and forming a covalent C-H
setup depicted in Fig. 7, the H atom was positioned 7 Å bond.
above the coronene, aimed at x = -0.82 Å, and incident Building on these observations, a specific incident
at an angle of 35◦ relative to the z-axis. Consequently, point and a velocity more parallel to the graphene plane
the first row in Table III corresponds to the simulation were selected to achieve a successful simulation of H atom
presented in Fig. 7. adsorption. Using this configuration, the effects of vary-
Each simulation in Table III produced a distinct out- ing initial kinetic energies were tested. The results in-
come depending on the initial kinetic energy of the H dicate that kinetic energies within the range of 1.89 eV
atom. The four simulations with the initial H atom ki- to 3.50 eV enable adsorption. Higher kinetic energies
netic energy ranging from 1.89 eV to 3.50 eV resulted (4.66 eV to 6.35 eV) allow the H atom to penetrate the
in adsorption (A), indicating that kinetic energies within potential energy barrier but result in strong repulsion
this range at the given incident angle were sufficient to from the positively charged carbon nuclear core, caus-
overcome the potential energy barrier of the graphene ing scattering. At much higher kinetic energies (9.14 eV
sheet and form a bond with one of the C atoms. In and above), the H atom transmits through the graphene
contrast, the two simulations with higher initial kinetic sheet, losing energy to the structure while refracting at
energies of 4.66 eV and 6.35 eV also overcame the po- a specific angle.
tential barrier due to their increased energy but did so These results highlight the critical role of initial pa-
with excessive kinetic energy. This caused the H atom to rameters, such as incident points and kinetic energies, in
approach too closely to the nuclear core of the C atom, determining the outcomes of the hydrogenation process
resulting in scattering (S). The scattering behavior varied of graphene. The findings underscore their impact on ad-
between these two cases, with the angles of reflection rel- sorption probability, scattering distributions, and energy
ative to the xy-plane provided in Table III. Additionally, transfer dynamics. Future work could focus on exper-
the amount of kinetic energy lost by the H atom during imentally validating these results and further exploring
these simulations differed, with the higher-energy sim- the interplay of incident points and velocities.
ulation (Simulation 6) losing less kinetic energy despite
starting with a greater initial value.
Finally, in the simulation with the highest initial ki- ACKNOWLEDGMENTS
netic energy of 9.14 eV, the H atom had sufficient veloc-
ity to not only surpass the graphene’s potential energy This work has been supported by the National Science
barrier but also transmit (T) through the sheet. In this Foundation (NSF) under Grant No. DMR-2217759.
case, 4.29 eV of the H atom’s energy was absorbed by This work used ACES at TAMU through alloca-
the graphene, and the H atom exited the sheet with a re- tion PHYS240167 from the Advanced Cyberinfrastruc-
fracted angle of −38.33◦ relative to the xy-plane (where ture Coordination Ecosystem: Services & Support (AC-
the negative angle indicates a velocity component below CESS) program, which is supported by National Sci-
the xy-plane, corresponding to a negative z-component ence Foundation grants #2138259, #2138286, #2138307,
of velocity). #2137603, and #2138296 [47].

[1] D. G. Papageorgiou, I. A. Kinloch, and R. J. Young, nanocomposites, Progress in Materials Science 90, 75
Mechanical properties of graphene and graphene-based
 9

FIG. 7. Snapshots of a hydrogen projectile with an initial kinetic energy of 1.89 eV (velocity of 0.19 A/fs) being absorbed
by a coronene molecule. The corresponding H atom kinetic energy (red solid line) is plotted over time, with green markers
indicating the specific simulation times at which each snapshot was captured. The net kinetic energy of all the coronene atoms
throughout the simulation is plotted as well (green dashed line)

(2017). synthesis and application for solar cells, Journal of Ma-

[2] C. Lee, X. Wei, J. W. Kysar, and J. Hone, Measure- terials Research 29, 299–319 (2014).
ment of the elastic properties and intrinsic strength [10] A. Olabi, M. A. Abdelkareem, T. Wilberforce, and E. T.
of monolayer graphene, Science 321, 385 (2008), Sayed, Application of graphene in energy storage device
https://www.science.org/doi/pdf/10.1126/science.1157996. – a review, Renewable and Sustainable Energy Reviews
[3] L. Fritz, J. Schmalian, M. Müller, and S. Sachdev, Quan- 135, 110026 (2021).
tum critical transport in clean graphene, Phys. Rev. B [11] J. Zhu, D. Yang, Z. Yin, Q. Yan, and H. Zhang,
78, 085416 (2008). Graphene and graphene-based materials for en-
[4] A. B. Kashuba, Conductivity of defectless graphene, ergy storage applications, Small 10, 3480 (2014),
Phys. Rev. B 78, 085415 (2008). https://onlinelibrary.wiley.com/doi/pdf/10.1002/smll.201303202.
[5] K. S. Novoselov, V. I. Falko, L. Colombo, P. R. Gellert, [12] J. Liu, J. Tang, and J. J. Gooding, Strategies for chemical
M. G. Schwab, and K. Kim, A roadmap for graphene, modification of graphene and applications of chemically
Nature 490, 192 (2012). modified graphene, J. Mater. Chem. 22, 12435 (2012).
[6] P. Avouris and F. Xia, Graphene applications in electron- [13] H. W. Ha, A. Choudhury, T. Kamal, D.-H. Kim, and
ics and photonics, MRS Bulletin 37, 1225 (2012). S.-Y. Park, Effect of chemical modification of graphene
[7] M. Sang, J. Shin, K. Kim, and K. J. Yu, Electronic and on mechanical, electrical, and thermal properties of poly-
thermal properties of graphene and recent advances in imide/graphene nanocomposites, ACS Applied Materials
graphene based electronics applications, Nanomaterials & Interfaces 4, 4623 (2012).
9, 10.3390/nano9030374 (2019). [14] T. Kuila, S. Bose, A. K. Mishra, P. Khanra, N. H. Kim,
[8] T. Mahmoudi, Y. Wang, and Y.-B. Hahn, Graphene and and J. H. Lee, Chemical functionalization of graphene
its derivatives for solar cells application, Nano Energy and its applications, Progress in Materials Science 57,
47, 51 (2018). 1061 (2012).
[9] S. Das, P. Sudhagar, Y. S. Kang, and W. Choi, Graphene [15] K. C. Kemp, H. Seema, M. Saleh, N. H. Le, K. Mahesh,
 10

V. Chandra, and K. S. Kim, Environmental applications huega, Atomic-scale control of graphene magnetism
using graphene composites: water remediation and gas by using hydrogen atoms, Science 352, 437 (2016),
adsorption, Nanoscale 5, 3149 (2013). https://www.science.org/doi/pdf/10.1126/science.aad8038.
[16] X. Li and L. Zhi, Graphene hybridization for energy stor- [32] D. W. Boukhvalov, M. I. Katsnelson, and A. I. Lichten-
age applications, Chem. Soc. Rev. 47, 3189 (2018). stein, Hydrogen on graphene: Electronic structure, total
[17] R. Balog, B. Jørgensen, L. Nilsson, M. Andersen, energy, structural distortions and magnetism from first-
E. Rienks, M. Bianchi, M. Fanetti, E. Lægsgaard, principles calculations, Phys. Rev. B 77, 035427 (2008).
A. Baraldi, S. Lizzit, Z. Sljivancanin, F. Besen- [33] L. Shi, M. Schröder, H.-D. Meyer, D. Pelaez, A. M.
bacher, B. Hammer, T. G. Pedersen, P. Hofmann, and Wodtke, K. Golibrzuch, A.-M. Schönemann, A. Kandrat-
L. Hornekær, Bandgap opening in graphene induced by senka, and F. Gatti, Full quantum dynamics study for h
patterned hydrogen adsorption, Nature Materials 9, 315 atom scattering from graphene (2024), arXiv:2410.07246
(2010). [physics.chem-ph].
[18] H. Jiang, M. Kammler, F. Ding, Y. Dorenkamp, [34] L. Shi, M. Schröder, H.-D. Meyer, D. Peláez, A. M.
F. R. Manby, A. M. Wodtke, T. F. Miller, Wodtke, K. Golibrzuch, A.-M. Schönemann, A. Kan-
A. Kandratsenka, and O. Bünermann, Imag- dratsenka, and F. Gatti, Quantum and classical
ing covalent bond formation by h atom scat- molecular dynamics for h atom scattering from
tering from graphene, Science 364, 379 (2019), graphene, The Journal of Chemical Physics 159,
https://www.science.org/doi/pdf/10.1126/science.aaw6378. 194102 (2023), https://pubs.aip.org/aip/jcp/article-
[19] L. Hornekær, Stabilizing a c–h bond on pdf/doi/10.1063/5.0176655/19665109/194102 1 5.0176655.pdf.
graphene with sound, Science 364, 331 (2019), [35] A. Ito, A. Takayama, and H. Nakamura, Hydrogen ad-
https://www.science.org/doi/pdf/10.1126/science.aax1980. sorption of back side of graphene, (2007).
[20] M. Bonfanti, S. Achilli, and R. Martinazzo, Sticking of [36] S. Saito, A. M. Ito, and H. Nakamura, Molecular dy-
atomic hydrogen on graphene, Journal of Physics: Con- namics simulation of the incident angle dependence of
densed Matter 30, 283002 (2018). reactions between graphene and hydrogen atom, Plasma
[21] S. Casolo, O. M. Løvvik, R. Martinazzo, and G. F. Tan- and Fusion Research 5, S2076 (2010).
tardini, Understanding adsorption of hydrogen atoms [37] H. Nakamura and A. Ito, Molecular dynamics
on graphene, The Journal of Chemical Physics 130, simulation of sputtering process of hydrogen and
054704 (2009), https://pubs.aip.org/aip/jcp/article- graphene sheets, Molecular Simulation 33, 121 (2007),
pdf/doi/10.1063/1.3072333/15425402/054704 1 online.pdf. https://doi.org/10.1080/08927020601078471.
[22] V. V. Ivanovskaya, A. Zobelli, D. Teillet-Billy, [38] A. ITO and H. NAKAMURA, Molecular dynamics simu-
N. Rougeau, V. Sidis, and P. R. Briddon, Hydrogen ad- lation of collisions between hydrogen and graphite, Jour-
sorption on graphene: a first principles study, The Euro- nal of Plasma Physics 72, 805–808 (2006).
pean Physical Journal B 76, 481 (2010). [39] A. Ito, H. Nakamura, and A. Takayama, Molecu-
[23] L. Jeloaica and V. Sidis, Dft investigation of the adsorp- lar dynamics simulation of the chemical interaction
tion of atomic hydrogen on a cluster-model graphite sur- between hydrogen atom and graphene, Journal of
face, Chemical Physics Letters 300, 157 (1999). the Physical Society of Japan 77, 114602 (2008),
[24] X. Sha and B. Jackson, First-principles study of the https://doi.org/10.1143/JPSJ.77.114602.
structural and energetic properties of h atoms on a [40] L. Delfour, A. Davydova, E. Despiau-Pujo, G. Cunge,
graphite (0001) surface, Surface Science 496, 318 (2002). D. B. Graves, and L. Magaud, Cleaning graphene:
[25] Y. Ferro, F. Marinelli, and A. Allouche, Density A first quantum/classical molecular dynam-
functional theory investigation of h adsorption and ics approach, Journal of Applied Physics 119,
h2 recombination on the basal plane and in the 125309 (2016), https://pubs.aip.org/aip/jap/article-
bulk of graphite: Connection between slab and pdf/doi/10.1063/1.4945034/15181049/125309 1 online.pdf.
cluster model, The Journal of Chemical Physics [41] M. Bonfanti, B. Jackson, K. H. Hughes,
116, 8124 (2002), https://pubs.aip.org/aip/jcp/article- I. Burghardt, and R. Martinazzo, Quantum dy-
pdf/116/18/8124/19152973/8124 1 online.pdf. namics of hydrogen atoms on graphene. ii. stick-
[26] H. Tachikawa, T. Iyama, and H. Kawabata, Density func- ing, The Journal of Chemical Physics 143,
tional theory and direct molecular dynamics study of the 124704 (2015), https://pubs.aip.org/aip/jcp/article-
hydrogen atom on a finite-sized graphene, Japanese Jour- pdf/doi/10.1063/1.4931117/14800085/124704 1 online.pdf.
nal of Applied Physics 49, 01AH06 (2010). [42] H. Nakamura, A. Takayama, and A. Ito,
[27] C. P. Herrero and R. Ramı́rez, Vibrational properties Molecular dynamics simulation of hydro-
and diffusion of hydrogen on graphene, Phys. Rev. B 79, gen isotope injection into graphene, Contri-
115429 (2009). butions to Plasma Physics 48, 265 (2008),
[28] V. Jain and B. Kandasubramanian, Functionalized https://onlinelibrary.wiley.com/doi/pdf/10.1002/ctpp.200810046.
graphene materials for hydrogen storage, Journal of Ma- [43] J. Petucci, C. LeBlond, M. Karimi, and G. Vi-
terials Science 55, 1865 (2020). dali, Diffusion, adsorption, and desorption of
[29] V. Tozzini and V. Pellegrini, Prospects for hydrogen stor- molecular hydrogen on graphene and in graphite,
age in graphene, Phys. Chem. Chem. Phys. 15, 80 (2013). The Journal of Chemical Physics 139, 044706
[30] H. G. Shiraz and O. Tavakoli, Investigation of graphene- (2013), https://pubs.aip.org/aip/jcp/article-
based systems for hydrogen storage, Renewable and Sus- pdf/doi/10.1063/1.4813919/14138673/044706 1 online.pdf.
tainable Energy Reviews 74, 104 (2017). [44] K. Varga and J. A. Driscoll, Monte carlo calculations, in
[31] H. González-Herrero, J. M. Gómez-Rodrı́guez, P. Mal- Computational Nanoscience: Applications for Molecules,
let, M. Moaied, J. J. Palacios, C. Salgado, M. M. Clusters, and Solids (Cambridge University Press, 2011).
Ugeda, J.-Y. Veuillen, F. Yndurain, and I. Bri- [45] N. Troullier and J. L. Martins, Efficient pseudopoten-
 11

tials for plane-wave calculations, Phys. Rev. B 43, 1993 and J. Towns, Access: Advancing innovation: Nsf’s ad-
(1991). vanced cyberinfrastructure coordination ecosystem: Ser-
[46] J. P. Perdew and A. Zunger, Self-interaction correction vices & support, in Practice and Experience in Advanced
to density-functional approximations for many-electron Research Computing 2023: Computing for the Common
systems, Phys. Rev. B 23, 5048 (1981). Good , PEARC ’23 (Association for Computing Machin-
[47] T. J. Boerner, S. Deems, T. R. Furlani, S. L. Knuth, ery, New York, NY, USA, 2023) pp. 173–176.