The peak positions of graphene plasmon resonance can be controlled to overlap with those of the infrared absorption spectra of gas molecules, allowing highly sensitive detection and identification by graphene nanoribbons. In this study, we investigate the adsorption of gas molecules, including SO2, SO3, H2S, and NH3, on graphene and characterize its effects on the relative positions of the two peaks using density functional theory and the finite difference time domain method. It is demonstrated that the binding energies are stronger, and the amounts of charge transfer are greater in the case of SO2 and SO3 adsorbed on n-doped graphene than in other cases. Electron acceptance by SO2 and SO3 adsorbates on n-doped graphene redshifts the graphene plasmon resonance peaks and their stretching and wagging infrared absorption peaks. However, the former is significantly further redshifted, leading to narrower peak-position-matching ribbon widths in n-doped graphene than in p-doped graphene. The amounts of charge transfer are relatively small regardless of the doping type in the case of NH3 and H2S, mitigating the doping-type dependence compared to SO2 and SO3. The wagging peaks of NH3 on n-doped graphene are shown to be further blueshifted than on p-doped graphene, rendering their peak-position-matching ribbon widths further closer to each other. These results suggest that the effects of doping and adsorption on the two types of peaks should be considered to optimize the performance of graphene plasmon-based gas sensing and identification.

Graphene has been suggested and demonstrated to be useful as an active material in various applications owing to its unique physical and chemical properties.1,2 One of the unique physical properties of graphene is the ease of control over its Fermi level. The Fermi level of graphene is significantly varied by charge transfer from a small number of molecular adsorbates or a small gate voltage applied in field effect devices due to the low density of states near the Dirac point. The Fermi level of graphene strongly affects its physical properties such as the Dirac voltage, infrared absorbance, resonance Raman scattering cross section, and plasmon resonance.3–7 This allows the use of graphene to create novel molecule-sensing devices other than those based on the modulation of electrical conductivity or refractive index. The submicrometer and nanopatterning of graphene leads to infrared plasmon resonances whose peaks strongly depend on the Fermi level. Recent studies have shown that the resonance peaks can be tuned by gating to approach the infrared absorption peaks of adsorbed gas molecules, substantially enhancing the detection limit.8–10 The positions of infrared absorption bands strongly depend on the masses of the constituent atoms and molecular structures, allowing the identification of gas molecule analytes. They can also depend on the Fermi level of graphene because charge transfer between the gas adsorbates and graphene can induce structural changes in the adsorbates. The binding energies (BEs) of adsorbates, which affect the sensitivity of sensors, have been shown to depend on the Fermi level of graphene.11 Hence, it is important to predict the effects of the Fermi level of graphene on the infrared absorption bands and binding energies of adsorbates to generally apply the novel principle for sensing various gas molecules. Density functional theory (DFT) has been widely used to investigate the adsorption of gas molecules on graphene.11–16 DFT calculations have shown that the physisorption of electron-accepting molecules (p-type dopant) significantly decreases the Fermi level, whereas that of electron donating molecules (n-type dopant) has an insignificant effect. The n-type pre-doping of graphene has been shown to strengthen the binding energies of electron-accepting molecules. The effect of the Fermi level on the adsorbate binding energy was explained in terms of the positions of molecular orbitals relative to the Fermi level of graphene.16 In this study, DFT is used to calculate the binding energies and infrared absorption peak positions of various gas molecules adsorbed on graphene with varied doping concentration. The effects of the Fermi level of graphene on the calculation results are characterized and explained in terms of adsorption-induced charge transfer and structural change. The finite difference time domain (FDTD) method is used to calculate the peak positions of graphene plasmon resonance. We analyze the effects of the doping and charge transfer on the two peak positions and discuss graphene doping and patterning conditions for matching them.

Spin-polarized DFT calculations with DFT-D4 dispersion correction17,18 were performed to obtain the binding energies and the most stable configurations of gas molecules on graphene. The real-space project-augmented wave method and the Perdew Burke Ernzerhof (PBE) exchange correlation functional19 were used under general gradient approximation as implemented in the GPAW package.20 The grid spacing was set to 0.18 Å. A graphene monolayer was modeled using a 4 × 4 × 1 orthogonal supercell with a 20 Å-thick vacuum layer. Dipole correction21 was applied in the z-direction to prevent interaction between the periodic images of graphene. The Brillouin zone was sampled to a 2 × 2 × 1 Monkhorst–Pack mesh.22 The Fermi level of graphene was varied by introducing a 1 Å-thick Jellium slab at a distance of 5 Å from the back side of the graphene surface.23 Atomic structures used in the calculation in this study were built using the Atomic Simulation Environment (ASE) package.24 The geometry optimization of the structures was performed using the quasi-Newton algorithm as implemented in ASE until the maximum force was less than 0.03 eV/Å. The BEs of the adsorbates were obtained from the calculation results of the total electronic energies as follows:
B E = E g r a p h e n e + a d s E g r a p h e n e E g a s ,
where E g a s, E g r a p h e n e, and E g r a p h e n e + a d s are the total electronic energies of the gas molecules, graphene, and graphene with an adsorbate, respectively. The infrared spectra of adsorbates were calculated for the geometry-optimized structures using the infrared module in ASE.

The transmission spectra of graphene nanoribbon arrays on a dielectric substrate were calculated using the FDTD method implemented in the MEEP package.25 The relative permittivity of graphene was expressed as ε = ε r + i ( σ / ε 0 ω Δ ), where ε r is the background relative permittivity, ε 0 is the vacuum permittivity, σ is the optical conductivity of graphene, ω is the angular frequency, and Δ is the graphene thickness. The optical conductivity of graphene was approximated using the expression σ = i e 2 E F / π 2 ( ω + i Γ ), where E F is the Fermi level of graphene and Γ is the scattering rate, assuming the dominance of intraband conductivity.22 In the FDTD simulations in this study, Δ, ε r, and Γ were set to the grid size (1 nm), 2.5, and 3.33 × 1012 s−1, respectively.6,26,27

Figure 1 shows the most stable adsorption configuration of four different gas molecules on undoped graphene (see Supplementary Figs. S1–S4 for binding energies in various adsorption configurations of each molecule).38 The H2S binding energy on undoped graphene is the second strongest in the configuration in Fig. 1(d), as shown in Supplementary Fig. S4,38 but it becomes the strongest on n-doped and p-doped graphene. In all cases, graphene maintains its planar structure with no buckling during adsorption. To study the effects of the graphene doping concentration on the adsorbate binding energies, electrons were added or removed from a graphene supercell with jellium charge compensation.23 The doping concentrations were varied between −3.58 × 1013 and 3.58 × 1013 cm−2; the plus/minus signs indicate p-doping and n-doping, respectively. The adsorption configurations shown in Fig. 1 do not significantly change with the doping condition for all the four molecules with a few minor exceptions. For example, SO2 molecules are adsorbed to be inclined such that oxygen atoms are closer to the surface of n-doped graphene than a sulfur atom, as shown in Supplementary Fig. S5 (see Supplementary Figs. S5–S7 for other exceptions).38 The effects of doping on the binding energies differ from one another, as shown in Fig. 2(a). In the case of NH3 and H2S, the changes in the binding energies by doping are insignificant and smaller than or at most similar to ∼10 meV within the abovementioned doping range. The binding energies of SO2 and SO3 are strengthened by 0.143 and 0.098 eV, respectively, as the n-doping concentration is increased to 3.58 × 1013 cm−2, whereas the effect of p-doping is as insignificant as in NH3 and H2S. Considering the calculation results of binding energies, n-doping might be beneficial in terms of increasing the sensitivity of sensing SO2 and SO3.

FIG. 1.

Most stable adsorption configurations of (a) SO2, (b) SO3, (c) NH3, and (d) H2S molecules on pristine graphene. As mentioned in the main text, the binding energy of H2S is calculated to be the second strongest in the configuration shown in the figure, but it becomes the strongest on doped graphene. The stable position and orientation do not significantly change with the graphene doping concentration (see Supplementary Figs. S5–S7) (Ref. 38).

FIG. 1.

Most stable adsorption configurations of (a) SO2, (b) SO3, (c) NH3, and (d) H2S molecules on pristine graphene. As mentioned in the main text, the binding energy of H2S is calculated to be the second strongest in the configuration shown in the figure, but it becomes the strongest on doped graphene. The stable position and orientation do not significantly change with the graphene doping concentration (see Supplementary Figs. S5–S7) (Ref. 38).

Close modal
FIG. 2.

(a) Binding energies of adsorbates and (b) amounts of adsorption-induced charge transfer as a function of the doping concentration of graphene.

FIG. 2.

(a) Binding energies of adsorbates and (b) amounts of adsorption-induced charge transfer as a function of the doping concentration of graphene.

Close modal

The adsorption of a molecule leads to electron transfer from graphene to the molecule, except for NH3, as shown in the Bader charge analysis28 results of Fig. 2(b) (see Supplementary Table S1] for the quantitative data).38 The amounts of charge donated to graphene by NH3 are calculated to be nearly constant at approximately 0.012 e regardless of the graphene doping concentration. H2S accepts only 0.006–0.007 e from graphene in the aforementioned range of doping concentration. The numbers of electrons accepted by a SO2 or SO3 molecule are calculated to be relatively large compared to NH3 and H2S; 0.063 and 0.084 e are transferred from undoped graphene to an SO2 and an SO3 molecule, respectively. The amounts of charge transfer are nearly unaffected by p-doping, but they increase with the n-doping concentration to 0.219 e and 0.236 e, respectively, at 3.58 × 1013 cm−2.

The results of Bader analysis are consistent with the effects of graphene doping concentration on the binding energies of the gas molecules. As shown in Fig. 2, the strength and doping-concentration susceptibility of the binding energy are positively correlated with the amounts of charge transfer that occur during adsorption. The binding energies of SO2 and SO3 are stronger than those of the other molecules and change more sharply under n-doping conditions in which the amount of charge transfer is greater. The lowest unoccupied molecular orbital levels of the two molecules are lower than the Fermi level of n-doped graphene, allowing them to accept relatively large amounts of electron from graphene until the Fermi level is equal to the ionized energy states of the molecules. As the n-doping of graphene raises its Fermi level, the amount of electron transfer increases with the graphene doping concentration.16 

Charge transfer often also leads to changes in the characteristics of an adsorbate structure such as the bond length and angle.16 Hence, the normal mode frequencies of an adsorbate can depend on the amounts of charge transfer during adsorption.29–31  Figures 3(a) and 3(b) show the symmetric and asymmetric stretching-vibration peaks, respectively, in the calculated infrared absorption spectra of SO2 adsorbed on graphene under different initial doping conditions. They are the two strongest absorption peaks of SO2 adsorbates on undoped graphene and appear at 1122.2 and 1313.3 cm−1. The peak positions are clearly redshifted when graphene is initially n-doped. They appear at 1053.9 and 1254.6 cm−1, respectively, at an initial graphene doping concentration of 3.58 × 1013 cm−2. These redshifts can be understood in terms of the increased length of S–O bonds. The bond length is increased by n-doping from 1.458 Å to 1.463, 1.469, and 1.471 Å as the initial doping concentration is set as 1.19 × 1013, 2.39 × 1013, and 3.58 × 1013 cm−2, respectively, as shown in Fig. 3(c). The increases in the bond lengths are consistent with the amounts of electron transfer from graphene to SO2, which increases with the initial doping concentration. As mentioned earlier, p-doping does not significantly affect the charge transfer. As a result, the bond length barely changes, leading to smaller shifts in the infrared absorption peaks compared to the n-doping cases.

FIG. 3.

Infrared absorption spectra and geometries of [(a)–(c)] SO2 and [(d)–(f)] SO3 adsorbates adsorbed on graphene with different doping concentrations. (a) and (b) show the symmetric and antisymmetric stretching absorption peaks of SO2 as indicated in the inset diagram. (d) and (e) show the antisymmetric stretching and wagging absorption peaks of SO3 as indicated in the inset diagram. The numbers in the legends of (a), (b), (d), and (e) indicate the initial doping concentration in the unit of 1012 cm−2 . The sign implies the doping type as mentioned in the main text. (c) and (f) show the equilibrium bond lengths of SO2 and SO3 at different graphene doping concentrations. In (f), the difference in the equilibrium Z coordinates of S and O atoms in SO3 at different graphene doping concentrations is shown.

FIG. 3.

Infrared absorption spectra and geometries of [(a)–(c)] SO2 and [(d)–(f)] SO3 adsorbates adsorbed on graphene with different doping concentrations. (a) and (b) show the symmetric and antisymmetric stretching absorption peaks of SO2 as indicated in the inset diagram. (d) and (e) show the antisymmetric stretching and wagging absorption peaks of SO3 as indicated in the inset diagram. The numbers in the legends of (a), (b), (d), and (e) indicate the initial doping concentration in the unit of 1012 cm−2 . The sign implies the doping type as mentioned in the main text. (c) and (f) show the equilibrium bond lengths of SO2 and SO3 at different graphene doping concentrations. In (f), the difference in the equilibrium Z coordinates of S and O atoms in SO3 at different graphene doping concentrations is shown.

Close modal

Similar trends are observed in the asymmetric stretching absorption peaks of SO3, as shown in Fig. 3(d). The n-doping of graphene leads to clear redshifts of the peak, whereas the effect of p-doping is negligible. The S–O bond length increases from 1.445 to 1.451, 1.457, and 1.463 Å with the aforementioned change in n-doping concentration, which is also consistent with the amounts of electron transfer shown in Fig. 2(b) and Table S1 in the supplementary material.38 The wagging mode peak at 459.8 cm−1 is also redshifted to 444.7 and 398.1 cm−1 at 2.39 × 1013 and 3.58 × 1013 cm−2, respectively, as shown in Fig. 3(e). In the two doping conditions, the S atoms are notably lowered relative to the O atoms, which might also be associated with the redshifts of the wagging vibration frequencies [see Fig. 3(f) for the aforementioned structural changes of an SO3 molecule]. Some shifts are not explained in terms of the structural changes mentioned above. For example, the asymmetric stretching peak position at 3.58 × 1013 cm−2 is redshifted by a smaller amount compared to that at 2.39 × 1013 cm−2 in spite of the longer S–O bond length. The S atom is lowered at an initial n-doping concentration of 1.19 × 1013 cm−2, but the wagging mode frequency is slightly blueshifted.

In the case of an NH3 adsorbate, the strongest peak corresponds to the wagging mode appearing at 1009.9 cm−1. The peak is calculated to be blueshifted regardless of the doping type and concentration, as shown in Fig. 4(a). The N–H bond lengths do not change with doping and are approximately 1.021 Å in all cases. Instead, doping leads to the rise of the N atom relative to the H atoms and the rise distances are greater at 2.39 × 1013 and 1.19 × 1013 cm−2 than at other doping concentrations, as shown in Fig. 4(b). The two doping conditions correspond to the ones in which the blueshifts are the greatest, suggesting that the rise might be associated with the blueshift. The strongest peak of H2S adsorbed on undoped graphene appears at 2648.8 cm−1, which corresponds to the symmetric stretching mode. The peak positions change between ∼2630 and ∼2665 cm−1 with the doping conditions mentioned above, as shown in Fig. 4(c), but no correlation clearly appears between the amount of the peak shift and the doping concentration. The H–S bond length barely changes with the doping concentration, which is consistent with the small change in the peak positions of the symmetric stretching vibration. It should also be noted that the amount of charge transfer is smaller compared to other molecules as mentioned earlier.

FIG. 4.

Infrared absorption spectra and geometries of [(a) and (b)] NH3 and (c) H2S adsorbates adsorbed on graphene with different doping concentrations. (a) and (c) show the wagging and symmetric stretching absorption peaks as indicated in the inset diagrams. In (b), the difference in the equilibrium Z coordinates of N and H atoms in NH3 at different graphene doping concentrations is shown.

FIG. 4.

Infrared absorption spectra and geometries of [(a) and (b)] NH3 and (c) H2S adsorbates adsorbed on graphene with different doping concentrations. (a) and (c) show the wagging and symmetric stretching absorption peaks as indicated in the inset diagrams. In (b), the difference in the equilibrium Z coordinates of N and H atoms in NH3 at different graphene doping concentrations is shown.

Close modal

The relative permittivity of graphene depends on its Fermi level as in the equations mentioned in Sec. II, leading to a doping-induced change of the plasmon resonance peak position in graphene. Figure 5(a) shows the transmission spectra of various graphene nanoribbon arrays on a dielectric substrate whose Fermi level is set to be deviated from the Dirac point by the indicated values ranging from 0.3 to 0.75 eV. The refractive index of the substrate is set constant as 1.46 for simplification. The gaps between the ribbons are set to be equal to the ribbon widths. As shown in the figure, the plasmon resonance peaks are blueshifted as the ribbon width is decreased and the Fermi level is deviated further from the Dirac point. The Fermi level of graphene relative to the Dirac point can be estimated by entering the carrier concentration in the following equation: ± v F π C, where vF is the Fermi velocity and C is the carrier concentration.32 The plus/minus sign is applied to n-doped and p-doped graphene, respectively. The carrier concentration after adsorption can be obtained from the results of Bader analyses. The adsorption of electron-accepting and donating molecules changes the carrier concentration such that the Fermi level of graphene is down-shifted and up-shifted, respectively. For example, in the calculation for SO2, the hole concentration of graphene doped with an initial concentration of −3.58 × 1013 cm−2 is increased to −4.34 × 1013 cm−2 after adsorption. This corresponds to a Fermi level change from −0.768 to −0.846 eV relative to the Dirac point. In the case of n-doped graphene with the same initial doping concentration, SO2 adsorption changes the Fermi level from 0.768 to 0.402 eV. As shown earlier, an NH3 adsorbate donates approximately 0.012 e to graphene regardless of the initial doping concentration. The charge transfer slightly increases the Fermi level of the abovementioned p-doped and n-doped graphene from −0.768 and 0.768 eV to −0.754 and 0.785 eV, respectively.

FIG. 5.

Calculation results of the peak positions of graphene plasmon resonance and adsorbate infrared absorption spectra. (a) Transmission spectra of graphene nanoribbon arrays with different Fermi levels (the indicated Fermi level is the absolute value of the deviation from the Dirac point). The widths of nanoribbons are indicated on the upper-right of each set of plots. (b) Overlay of infrared absorption peak position data onto graphene-plasmon peak position data. The inset indicates the ribbon width. In the legend, “asym”, “symm,” and “wag” represent the asymmetric stretch, symmetric stretch, and wagging mode data, respectively; “n” and “p” indicate the doping type.

FIG. 5.

Calculation results of the peak positions of graphene plasmon resonance and adsorbate infrared absorption spectra. (a) Transmission spectra of graphene nanoribbon arrays with different Fermi levels (the indicated Fermi level is the absolute value of the deviation from the Dirac point). The widths of nanoribbons are indicated on the upper-right of each set of plots. (b) Overlay of infrared absorption peak position data onto graphene-plasmon peak position data. The inset indicates the ribbon width. In the legend, “asym”, “symm,” and “wag” represent the asymmetric stretch, symmetric stretch, and wagging mode data, respectively; “n” and “p” indicate the doping type.

Close modal

The effects of doping and adsorption on the Fermi level-dependent plasmon resonance peak should be compared to those on the infrared absorption peak of a molecule to specify a graphene doping condition in which the positions of the two peaks match each other. The plasmon resonance peak positions are linearly proportional to | E F | and 1 / W, where W is the ribbon width, as shown in Fig. 5(b) and Supplementary Fig. S8,38 respectively, which is consistent with the analytical expression obtained by quasi-static analysis.6,27,33 The positions of strong SO2, SO3, and NH3 peaks in Figs. 3 and 4 are overlaid in Fig. 5(b). As shown in Fig. 5(b), for SO2 and SO3, the peaks obtained in n-doped graphene are significantly shifted to the left side compared to those obtained in p-doped graphene, implying that the n-doped ribbon should be patterned to be narrower than the p-doped ribbon for peak position matching. This difference in the two doping types arises from the relatively large amounts of electron transfer from n-doped graphene to SO2 and SO3, which shifts the Fermi level toward the Dirac point. As shown in Fig. 3, the increased initial n-doping concentrations of 2.39 × 1013 and 3.58 × 1013 cm−2 lead to significant redshifts of the stretching and wagging peaks. However, the shifts are insignificant compared to the peak shifts of graphene plasmon resonance, implying that the latter plays a dominant role in determining the width of a peak-position-matching ribbon. In the case of NH3 and H2S, the amounts of adsorption-induced charge transfer are not sufficiently large to result in significant doping-type dependence observed in the case of SO2 and SO3. As shown in Fig. 5(b), both doping types render the wagging mode adsorption peaks of NH3 adjustable to be matched to the plasmon resonance peaks of nanoribbons whose widths range between ∼100 and ∼200 nm. NH3 donates electrons during adsorption and the amounts of electron transfer do not significantly depend on the doping type and concentration, as shown in Fig. 2(b) and Table S1 in the supplementary material.38 The charge transfers slightly blue- and redshift the plasmon resonance of n-doped and p-doped graphene ribbons, respectively. The blueshifts of the wagging absorption peaks are greater in n-doped graphene than in p-doped graphene at a particular initial doping concentration [see Fig. 4(a)], causing the widths of the peak-position-matching nanoribbons to be more similar, as shown in Fig. 5(b) (see the three pairs of open circle data).

These results show that the effects of adsorption on the peak positions of graphene plasmon resonance and adsorbate infrared absorption significantly depend on the electron affinity and molecular structure of an adsorbate. The calculation methods used in this study can be applied to other molecules, which is expected to allow the establishment of more general principles on adsorption-induced peak position shifts. It should be noted that the calculations in this study were performed at a specific number density of adsorbates. We suggest extensive calculations for different adsorption densities will be an important future research focus, as the adsorption number density might affect the amount of charge transfer, the adsorption geometry,9 and electromagnetic coupling between graphene plasmon and molecular vibration.34,35

It is noteworthy that the sensitivity and recovery time also need to be considered under certain conditions. The sensitivities for SO2 and SO3 will be higher than those for NH3 and H2S, regarding the relatively stronger binding energies of SO2 and SO3 adsorbates, which will lead to larger surface coverages at a given partial pressure.11 Both the infrared absorbances and binding energies are stronger under n-doping for SO2 and SO3 adsorbates, suggesting that n-doping is beneficial for improving the sensitivity. However, the stronger binding energies also increase the response time (see Supplementary Fig. S9).36–38 Hence, a p-doping can be favored in applications where a fast recovery is critical. In the case of NH3 and H2S, recovery times are estimated to be substantially shorter and there will be little dependence of the sensitivity on the doping conditions.

In this study, we used DFT and FDTD calculations to characterize the effects of graphene doping and adsorption on the peak positions of graphene plasmon resonance and adsorbate infrared absorption. We analyzed the conditions under which graphene-plasmon-resonance peak positions are matched to the strong infrared absorption peaks of various gas adsorbates. The calculation results showed that the charge transfer and binding strength are relatively greater and increase with the initial doping concentration in the case of SO2 and SO3 adsorbed on n-doped graphene. Electron acceptance by SO2 and SO3 adsorbed on n-doped graphene leads to significant redshifts of the graphene plasmon resonance peaks and stretching-vibration peaks, but the former is significantly larger. This renders the ribbon width to match the two peaks significantly narrower compared to the case of p-doped graphene. In the case of NH3 and H2S for which the amounts of charge transfer are relatively smaller, the doping-type dependence is significantly less. NH3 donates electrons during adsorption regardless of the initial doping condition of graphene. This leads to redshift and blueshift of the plasmon resonance peaks of n-doped and p-doped graphene. Their wagging vibration absorption peaks were calculated to be more blueshifted on n-doped graphene than on p-doped graphene, resulting in further similarity in the peak-matching ribbon width for the two doping types. These results suggest that the effects of doping and adsorption-induced charge transfer on the Fermi level of graphene and the vibration frequencies of adsorbates should be considered to optimize the performance of graphene plasmon-based gas sensing and identification.

The author gratefully acknowledges the support from KISTI (No. KSC-2021-CRE-0111).

The authors have no conflicts to disclose.

Jongpil Ye: Conceptualization (lead); Formal analysis (lead); Funding acquisition (lead); Investigation (lead); Project administration (lead); Resources (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
Y. W.
Zhu
,
S.
Murali
,
W. W.
Cai
,
X. S.
Li
,
J. W.
Suk
,
J. R.
Potts
, and
R. S.
Ruoff
,
Adv. Mater.
22
,
3906
(
2010
).
2.
Y.
Zhang
,
L. Y.
Zhang
, and
C. W.
Zhou
,
Acc. Chem. Res.
46
,
2329
(
2013
).
3.
H. G.
Yan
,
F. N.
Xia
,
W. J.
Zhu
,
M.
Freitag
,
C.
Dimitrakopoulos
,
A. A.
Bol
,
G.
Tulevski
, and
P.
Avouris
,
ACS Nano
5
,
9854
(
2011
).
4.
J.
Christensen
,
A.
Manjavacas
,
S.
Thongrattanasiri
,
F. H. L.
Koppens
, and
F. J. G.
de Abajo
,
ACS Nano
6
,
431
(
2012
).
5.
X.
Ling
,
L. G.
Moura
,
M. A.
Pimenta
, and
J.
Zhang
,
J. Phys. Chem. C
116
,
25112
(
2012
).
6.
H. S.
Chu
and
C. H.
Gan
,
Appl. Phys. Lett.
102
,
231107
(
2013
).
7.
S. X.
Huang
,
X.
Ling
,
L. B.
Liang
,
Y.
Song
,
W. J.
Fang
,
J.
Zhang
,
J.
Kong
,
V.
Meunier
, and
M. S.
Dresselhaus
,
Nano Lett.
15
,
2892
(
2015
).
8.
H.
Hu
,
X. X.
Yang
,
X. D.
Guo
,
K.
Khaliji
,
S. R.
Biswas
,
F. J. G.
de Abajo
,
T.
Low
,
Z. P.
Sun
, and
Q.
Dai
,
Nat. Commun.
10
,
1131
(
2019
).
9.
A.
Marini
,
I.
Silveiro
, and
F. J. G.
de Abajo
,
ACS Photonics
2
,
876
(
2015
).
10.
D.
Rodrigo
,
O.
Limaj
,
D.
Janner
,
D.
Etezadi
,
F. J. G.
de Abajo
,
V.
Pruneri
, and
H.
Altug
,
Science
349
,
165
(
2015
).
12.
D.
Tristant
,
P.
Puech
, and
I. C.
Gerber
,
J. Phys. Chem. C
119
,
12071
(
2015
).
13.
I. C.
Gerber
and
R.
Poteau
,
Theor. Chem. Acc.
137
,
156
(
2018
).
14.
L. M.
Kong
,
A.
Enders
,
T. S.
Rahman
, and
P. A.
Dowben
,
J. Phys.: Condens. Mater.
26
,
443001
(
2014
).
15.
Y.
You
,
J.
Deng
,
X.
Tan
,
N.
Gorjizadeh
,
M.
Yoshimura
,
S. C.
Smith
,
V.
Sahajwalla
, and
R. K.
Joshi
,
Phys. Chem. Chem. Phys.
19
,
6051
(
2017
).
16.
T.
Hu
and
I. C.
Gerber
,
J. Phys. Chem. C
117
,
2411
(
2013
).
17.
E.
Caldeweyher
,
S.
Ehlert
,
A.
Hansen
,
H.
Neugebauer
,
S.
Spicher
,
C.
Bannwarth
, and
S.
Grimme
,
J. Chem. Phys.
150
,
154122
(
2019
).
18.
E.
Caldeweyher
,
J. M.
Mewes
,
S.
Ehlert
, and
S.
Grimme
,
Phys. Chem. Chem. Phys.
22
,
8499
(
2020
).
19.
J. P.
Perdew
,
K.
Burke
, and
M.
Ernzerhof
,
Phys. Rev. Lett.
77
,
3865
(
1996
).
20.
J.
Enkovaara
et al,
J. Phys.: Condens. Mater.
22
,
253202
(
2010
).
21.
22.
H. J.
Monkhorst
and
J. D.
Pack
,
Phys. Rev. B
13
,
5188
(
1976
).
23.
C. J.
Jin
,
F. A.
Rasmussen
, and
K. S.
Thygesen
,
J. Phys. Chem. C
120
,
1352
(
2016
).
24.
A. H.
Larsen
et al,
J. Phys.: Condens. Mater.
29
,
273002
(
2017
).
25.
A. F.
Oskooi
,
D.
Roundy
,
M.
Ibanescu
,
P.
Bermel
,
J. D.
Joannopoulos
, and
S. G.
Johnson
,
Comput. Phys. Commun.
181
,
687
(
2010
).
26.
L. A.
Falkovsky
,
J. Phys.: Conf. Ser.
129
,
012004
(
2008
).
27.
W. L.
Gao
,
J.
Shu
,
C. Y.
Qiu
, and
Q. F.
Xu
,
ACS Nano
6
,
7806
(
2012
).
28.
W.
Tang
,
E.
Sanville
, and
G.
Henkelman
,
J. Phys.: Condens. Mater.
21
,
084204
(
2009
).
29.
S. A. C.
McDowell
and
A. D.
Buckingham
,
J. Am. Chem. Soc.
127
,
15515
(
2005
).
30.
W. Z.
Wang
,
Y.
Zhang
,
B. M.
Ji
, and
A. M.
Tian
,
J. Chem. Phys.
134
,
224303
(
2011
).
31.
E.
Kraka
,
W. L.
Zou
, and
Y. W.
Tao
,
WIREs Comput. Mol. Sci.
10
,
e1480
(
2020
).
32.
S.
Das Sarma
,
S.
Adam
,
E. H.
Hwang
, and
E.
Rossi
,
Rev. Mod. Phys.
83
,
407
(
2011
).
34.
F.
Liu
and
E.
Cubukcu
,
Phys. Rev. B
88
,
115439
(
2013
).
35.
K.
Khaliji
,
S. R.
Biswas
,
H.
Hu
,
X. X.
Yang
,
Q.
Dai
,
S. H.
Oh
,
P.
Avouris
, and
T.
Low
,
Phys. Rev. Appl.
13
,
011002
(
2020
).
36.
S.
Peng
,
K.
Cho
,
P.
Qi
, and
H.
Dai
,
Chem. Phys. Lett.
387
,
271
(
2004
).
37.
M. M.
Husain
,
M. T.
Ansari
, and
A.
Almohammedi
,
Mater. Today Commun.
39
,
108725
(
2024
).
38.
See supplementary material online for the binding energies of adsorbates on pristine graphene, the most stable adsorption configurations on graphene with different doping concentrations, the Bader analyses of charge transfer, FDTD results showing the effect of graphene nanoribbon width, and the calculated recovery time as a function of graphene doping concentration.

Supplementary Material