Stabilization of small organic molecules on V₂C and V₂CO₂ MXenes: first-principles insights into the performance of van der Waals functionals and the effect of oxygen vacancies

Abstract

The adsorption of small organic molecules on pristine V₂C MXene and its derivatives is investigated by first-principles density functional theory calculations. By employing state-of-the-art van der Waals (vdW) density functionals, the binding affinity of studied molecules, i.e., CH₄, CO₂, and H₂O on MXene adsorbents is well described by more recent vdW functionals, i.e., SCAN-rvv10. Although both CH₄ and CO₂ are nonpolar molecules, on pristine and oxygen-vacancy surfaces, they show a different range of adsorption energies, in which CH₄ is more inert and has weaker binding than CO₂. CO₂ stays intact in its molecular forms for most of the tested functionals, except for the case of the vdW-DF functional, where CO₂ exhibits a dissociation regardless of its initial adsorption geometry. For full surface terminations, the adsorption affinity of all involved species is comparable within the same range, varying from −0.10 to −0.20 eV, attributed to either weak dispersion interactions or hydrogen bonds. The binding of H₂O is much more pronounced compared to CO₂ and CH₄ in the presence of oxygen vacancies with the highest adsorption energy of −1.33 eV, vs. −0.67, and −0.20 eV obtained for H₂O, CO₂, and CH₄ respectively. H₂O can dissociate with a small activation energy barrier of 0.40 eV, much smaller than its molecular adsorption energy, to further saturate itself on the surface. At high oxygen-vacancy concentrations, stronger bindings of adsorbates are found due to a preferred attachment of adsorbates to induced undercoordinated metal sites. The findings propose a potential scheme for greenhouse gas separation based on the surface modification of novel two-dimensional structures.

---

1. Introduction

The interactions of organic molecules with metal surfaces are of great interest for both fundamental understanding and practical regulation, concerning many urgent adsorption and detection technology applications.^1–4 However, small organic molecules often encounter technical difficulties in adsorptive practices due to their inert properties. The binding of molecules onto the surface relies on both physical and chemical adsorption mechanisms.^5,6 Particularly, physisorption through van der Waals (vdW) interaction is known to be relevant for the attachment of nonpolar molecules,⁷ in reverse to the binding of actively polar species relying on the exposure of chemical bonds. The treatment of such interactions is well-established for fist-principles calculations via the introduction of vdW functionals with either pairwise-additive vdW corrections,^8,9 or truly nonlocal vdW functionals.¹⁰ Both approaches are successful in interpreting structures from biological molecules to layered compounds, as well as their physicochemical interactions.¹¹ Molecular binding affinities strongly depend on the activity of both the adsorbed molecules and the adsorbents, i.e., the binding sites. Searching for suitable adsorbents for specific adsorbates is an indispensable prerequisite for advanced applications.

MXenes are novel layered materials emerging for relevant technologies based on their diverse structural compositions and stoichiometries, along with intriguing physical-chemical properties.¹² Those graphene-like structures can be experimentally synthesized via the exfoliation of “A” layer metals from its parent MAX phase M_n+1AX_n by HCl or HF acids, followed by a sonification treatment to separate layers by layers.^13–15 The derived MXene surfaces are usually covered by terminal groups such as O, OH, or F in wet etching environments during the synthesis.^14,15 The chemical formula of a monolayer MXene is M_n+1X_nT_x, with X being the carbon C or nitrogen N sandwiched by early transition metals M (Sc, Ti, V, Mo, etc., n = 1, 2, 3), and T_x (T = F, O, OH, x = 0, 1, 2) is the terminal groups. Those functional groups could be removed from the surface via chemical washing or thermal methods to obtain clean, pristine structures.^16,17 Featuring as high chemical and thermal stabilities, electrical conductivities, high surface areas, and abundant functional groups, MXenes are extensively studied for a variety of applications, especially for energy storage,^18–22 spintronics,²³ gas separation,⁴ carbon capture,^24,25 gas sensors,^26–28 and catalysts.²⁹

V₂C, a family member of carbon-based MXenes, has been known to be stable at 150 °C.³⁰ V₂CT_x, a functionalized form of V₂C, shows much superior by its high thermal stability, up to 300 °C in both oxidative and reductive environments, or up to 600 °C in inert environments.¹⁶ Compared to the well-studied Ti₃C₂T_x counterparts,^4,14,31 V₂CT_x benefits from its high surface area per mass and chemical activity relying on multiple oxidation states of V against Ti metal atoms.³² V₂CT_x structures are widely studied as sensing materials to volatile organic compounds,^33,34 methane,³⁵ ammonia,³⁶ methylamine adsorption,³⁷ methylene blue removal,³⁸ carbon dioxide capture,^24,39 catalysts for hydrocarbon conversion,⁴⁰ or water dissociation.⁴¹ Lee et al. experimentally reported an ultrahigh sensitivity of methane on V₂CT_x at room temperature provided that the functional groups are active sites for physical or chemical interactions.³⁵ A large number of M₂C (M = Ti, Zr, Hf, V, Nb, Ta, Cr, Mo, W) MXenes show strong bindings with carbon dioxide molecules compared to zeolites or graphene oxides.³⁹ Gouveia et al. computationally demonstrated that water can be dissociated on almost all pristine M₂X (M = Ti, Zr, Hf, V, Nb, Ta, Cr, Mo, and W) with low activation energy barriers.⁴¹ Though many seminal works on the adsorption of organic molecules over V-based MXene structures can be found, bare MXene structures as adsorbents for small organic molecules are much more focused than the structures with the presence of explicit terminal groups or the existence of surface defects. The partial missing of terminating groups and the presence of other surface vacancies are inevitable in strongly acidic etching environments during the synthesis or in high thermal conditions for post-synthesis MXenes.^17,42,43 For example, terminal oxygens are most abundant, occupied about 84% terminations, among other functional groups on Ti₃C₂ MXene under synthesis conditions.⁴⁴ As a result, oxygen vacancies can be generated via multiple routes such as an incompletion termination during the synthesis, the chemical washing, or the thermal removal as post-synthesis modifications. Additionally, oxygen vacancies can also be produced and controlled by the oxidation of carbon monoxide on the MXene surface, under the presence of H₂, via the chemical reaction, CO + H₂ + MXene → HCOOH + MXene_O_v, proposed by Zhang and coworkers.⁴⁵ In this work, the oxygen vacancy on Ti₂CO₂ plays the role of a catalytic site for capturing and driving CO₂ reduction. To enhance the catalytic activity, single transition-metal atoms are deposited to the oxygen-vacant sites featuring the efficiency of oxygen evolution reactions⁴⁶ on Mo₂CO₂ and the methane-to-methanol conversion on V₃C₂O₂.⁴⁷ It has been known that bare MXenes without terminations are very active in capturing most of the molecules by chemical bonds relying on top-layer transition metals.^25,41 Besides, functionalized MXenes, as V₂CT_x, characterize rich surface chemistry exhibiting hydrophilic features due to their O- or OH-terminated groups, favoring hydrogen bondings or weak vdW interactions.^48,49 The presence of surface defects on functionalized structures proposes more active sites at realistic conditions. The hybrid of multiple phases by surface engineering is advantageous for the efficient treatment of mixed gases.

Systematic insights into the response of V₂CT_x MXenes against small organic molecules, i.e., greenhouse gases, regarding their adsorption mechanisms and affinities are fundamentally and technologically important. This work provides the interaction trends of both polar and nonpolar molecules with the V₂CT_x MXenes focusing on their adsorption performances on either the pristine, functionalized, or defective surfaces. Based on first-principles calculations, we attempt to benchmark the stability of adsorbed molecules at their binding sites concerning the choice of van der Waals functionals. The more advanced vdW functionals will be used for further calculations. The functionalizations and defect modifications allow us to better determine the selectivity of various target molecules. Following the adsorption, the calculated dissociation processes are the key signals showing the activity of the induced surface sites, in which the oxygen vacancies play a paramount importance role.

2. Computational method

Calculations based on density functional theory (DFT) are performed on the simulation suite Vien Ab initio Simulation Package (VASP 5.4.4).⁵⁰ The project augmented wave (PAW) pseudopotential⁵¹ with a cutoff kinetic energy of 500 eV is chosen for all atoms. The lattice parameters of the host structures and adsorption affinities of gas molecules are evaluated under the light of generalized gradient approximation (GGA) functionals of Perdew, Burke, and Ernzerhof (PBE),^52,53 and PBE with a D3 correction scheme.⁸ The nonlocal van der Waals functionals of optPBE-vdW,⁵⁴ optB88-vdW,⁵⁴ vdW-DF,⁵⁵ and rev-vdW-DF2 (ref. 56) are also included to improve long-range interactions of adsorbates. In addition, the meta-GGA functional of SCAN-rvv10 (ref. 57) is also used to benchmark either the intralayer lattice constants or the long-range binding of molecular adsorptions. For optimizations, the atoms are allowed to relax until the atomic force exerted on each atom balances at 0.01 eV Å⁻¹ and the convergence of the total energy of 10⁻⁵ eV.

Spin-polarization is applied for all calculated structures to account for the magnetic feature of the vanadium atoms. We first benchmark the primitive cells 1 × 1 of V₂C and V₂CO₂ using 9 × 9 × 1 k-point meshes within the Gamma sampling scheme. For the adsorption study, structures of 3 × 3 are directly generated from the previous pre-optimized primitive cells, shown in Fig. 1. For all calculations, a vacuum slab of at least 17 Å is chosen, which is sufficiently large to ignore spurious interactions. The oxygen-vacancy V₂CO₂_O_v configuration is also produced by removing one functionalized-oxygen atom. The k-point meshes of 3 × 3 × 1 are now used for all the adsorption calculations. The electron configurations of involving atoms are as follows V: [Ar] 4s² 3d³, C: [He] 2s² 2p², O: [He] 2s² 2p⁴, and H: 1s¹. The isolated gas molecules CH₄, H₂, CO₂, and H₂O are modeled in a cubic box of 20 Å length in each dimension. The adsorption energy E_ads of a molecule M is defined as eqn (1),

E_ads = E_M-host − E_M − E_host (1)

where E_M-host, E_M, and E_host stand for the total energies of the structures of the host with the adsorbed molecule, isolated molecule, and pristine structure, respectively. Excluding a surface oxygen atom costs an amount of energy as follows eqn (2),where E_host-Ov, E_O2, and E_host are the total energies of the oxygen-vacancy structure, isolated molecular oxygen, and pristine structure, respectively. The Mulliken charge analysis employed in LOBSTER code⁵⁸ (Local Orbital Basis Suite Towards Electronic-Structure Reconstruction) is used to evaluate atomic charges of the structures before and after adsorptions. The VESTA visualizing program, version 3.5.7,⁵⁹ is exploited to display all geometries. The transition state search is performed under the climbing-image nudged elastic band (CINEB) method.⁶⁰ Accordingly, nine intermediate images along the reaction path are generated and optimized. The obtained transition state is further confirmed by the fact that there is only one imaginary frequency corresponding to the transition mode of the interested reaction channel.

Fig. 1 Monolayers of 3 × 3 sizes are displayed in the side-view and top-view for the pure V₂C, in (a) and (b), and functionalized V₂CO₂, in (c) and (d), respectively. The spheres in blue, green, brown, and red correspond to vanadium V on the top layer, V on the bottom layer, carbon C, and oxygen O atoms. The adsorption sites on the surface are denoted as V-top, V-bridge, C-hollow, V-hollow, and O-vacancy (O_v). The C-hollow and V-hollow sites are located in the center of three V atoms with the underneath C and V atoms, respectively. The black lines in (a) and rhombus in (b) define the boundaries of the primitive cell in the side and top view, respectively.

3. Result and discussion

3.1 Geometry optimization

Monolayer pristine V₂C is obtained by manually eliminating atomic aluminum layer “Al” from the MAX phase V₂AlC unit cell, a layered hexagonal cell of the P6₃/mmc space group (space group number 194).^31,61 The V₂C primitive cell, shown in Fig. 1(a) and (b), is composed of three atoms arranged in ternary atomic layers, where carbon atoms are sandwiched between top and bottom vanadium atoms. The primitive cell is optimized in the way that lattice constants “a” and “b” are simultaneously increased or decreased, followed by the relaxation of atomic positions, until its optimal total energy is found. The lattice constants for the primitive cells of both pristine V₂C and V₂CO₂ are calculated within various functionals, including references to the previous work,⁶² as the results are shown in Table 1. For V₂C, it was found that the first four used functionals, i.e., PBE, PBE-D3, optPBE-vdW, and optB88-vdW, have similar lattice constants of about 2.89–2.90 Å and V–V distances of 2.16–2.17 Å, while vdW-DF predicts those values as high as 2.92 Å and 2.21 Å, respectively. Conversely, the functionals rev-vdW-DF2 and SCAN-rvv10 provide smaller values of only 2.87–2.88 Å and 2.18–2.19 for the lattice constants and V–V distances, respectively. For V₂CO₂, we have seen that “a” and “b” are almost unchanged or slightly increased for all functionals. The expansion of the intra-atomic layers in the “c” direction, or the functionalized orientation, is obvious by the increase of V–V distances of around 0.19–0.26 Å (∼10%) compared to ones for V₂C for all the functionals, see Table 1, V₂CO₂, d_V–V. Among benchmarked functionals, vdW-DF shows its highest lattice parameters and atomic-layer distances, while SCAN-rvv10 is on the reverse. This also implies that bond lengths described by vdW-DF are more elongated and debilitated than others. Accordingly, as shown in Table 1, the average V–O bond lengths, d_V–O, for vdW-DF is the longest, about 1.98 Å compared to other functionals. SCAN-rvv10 predicts this bond length is only 1.94 Å, which is the shortest one.

Table 1 Optimized parameters with different exchange–correlation functionals for the primitive cell, a tetragonal geometry. Lattice constants “a” and “b” are equal. The expansion of intra-atomic layers is determined by the distance between the top and bottom V atoms d_V–V. The bond length d_V–O is defined as the average distance of the V atom to the three connected surface oxygen atoms

Lattice constant (Å)		vdW functional
		PBE	PBE-D3	optPBE-vdW	optB88-vdW	vdW-DF	Rev-vdW-DF2	SCAN-rvv10	Ref. 62
a Calculated with a PBE method.^62
V2C	a = b	2.90	2.89	2.90	2.90	2.92	2.88	2.87	2.89^a
	dV–V	2.16	2.17	2.16	2.16	2.21	2.19	2.18	2.18^a
V2CO2	a = b	2.90	2.89	2.90	2.90	2.94	2.89	2.87	2.90^a
	dV–V	2.38	2.37	2.42	2.41	2.45	2.38	2.39	2.38^a
	dV–O	1.95	1.95	1.96	1.96	1.98	1.95	1.94	N/A

---

Obtaining the most compact V₂CT_X structures by SCAN-rvv10 is important to distinguish their energetic performances related to structural modifications.

3.2 Adsorption of CH₄, CO₂ and H₂O

Terminated by V¹⁺ atoms, V₂C adsorbate is very active in the binding of small gas molecules through different adsorption mechanisms. Functionalizing of V₂C with –O groups causes a reduction of V¹⁺ to V²⁺ due to the strong negativity of surface oxygen atoms. The surface chemistry of V₂CO₂ is subsequently dominated by the presence of terminated groups O¹⁻. Fig. 2 shows all the adsorption geometries of small molecules over both V₂C and V₂CO₂. The atomic positions for each structure are provided in the ESI, Section S1.†

Fig. 2 Most stable adsorption configurations are shown for all investigated molecules. The labels (a and b), and (c–e), and (f–i) denote the adsorption of target molecules on V₂C, V₂CO₂, and V₂CO₂_O_v, respectively. The displayed numbers (in Å), calculated via SCAN-rvv10, show either the bond lengths or nearest distances (dash lines) between the molecules and surface atoms.

3.2.1 Adsorption affinity.
3.2.1.1 CH₄. CH₄, a nonpolar molecule, is inactive for binding to many materials. CH₄ molecules can be captured by metal atoms via an electrostatic interaction.^63–65 Here, CH₄ molecules prefer binding to V-top atoms on the V₂C surface with a η² mode, see Fig. 2(a). This adsorption geometry agrees well with previous works, where CH₄ resides on top of single metal atoms.^64–66 The electrostatic attraction arises from the surface V¹⁺ atom and a partially negative charge of carbon C of the CH₄ molecule. The adsorption energies vary from a weak binding of less than 0.10 eV to a few eV depending on the performance of functionals, as shown in Table 2, entry 1, and Fig. 3(a). It can be seen that there is a large discrepancy in values derived from different functionals. Similar adsorption energies could be found in groups of PBE, PBE-D3, and SCAN-rvv10 functionals, while the other functionals show overbinding of CH₄ being larger than unity. On V₂CO₂, CH₄ molecules attach to the C-hollow site with three hydrogen atoms pointing towards surface oxygens, so-called a η³-like mode, see Fig. 2(d). There is another configuration, i.e., a η²-like mode, in which the energy is only about 2 meV smaller than the η³-like one accounted for SCAN-rvv10. The adsorption energies are higher for most of the functionals upon the functionalization compared to ones on the pristine, except for rev-vdW-DF2 and SCAN-rvv10. Considering equilibrium distances between CH₄ and the V₂CO₂ surface, it has been found that SCAN-rvv10 estimated a shortest distance of 2.90 Å, compared to the values of greater than 3.0 Å for the others. Expectedly, SCAN-rvv10 provides the upper limit with a small binding energy of 0.11 eV for CH₄ against various vdW-corrected functionals, cf. Table 2, entry 2. For the presence of O_v, the overall trend is inverse, as shown in Fig. 3(a), adsorption energies are enhanced for rev-vdW-DF2 and SCAN-rvv10 (see Table 2, entry 3), while it shows decreasing for other functionals.

Table 2 Adsorption energies (in eV) for the most stable adsorption configurations. The adsorbed species retain their molecular states on the V₂CT_x surface unless otherwise noted. On the pristine V₂C surface, CH₄ binds to the V atom via a η² mode, H₂ resides on the V-top site with a side-on geometry, CO₂ prefers bridging configuration, while H₂O prefers V-top configuration. Three entries in each molecule correspond to the adsorption of that molecule on the pristine V₂C, V₂CO₂, and V₂CO₂_O_v, respectively

Gas/V2CTx		vdW functional
		PBE	PBE-D3	optPBE-vdW	optB88-vdW	vdW-DF	rev-vdW-DF2	SCAN-rvv10
a Dissociated into CO and O fragments.
CH4	1	−0.01	−0.35	−1.99	−1.94	−2.63	−1.26	−0.25
	2	−0.35	−0.36	−2.46	−2.29	−3.24	−1.19	−0.11
	3	−0.02	−0.33	−2.11	−1.98	−2.73	−1.32	−0.20
CO2	4	−2.00	−3.52	−3.41	−3.62	−4.43^a	−2.55	−2.60
	5	−0.22	−0.41	−0.68	−0.61	−0.79	−0.17	−0.18
	6	−0.09	−0.49	−0.49	−0.53	−0.79	−0.53	−0.67
H2O	7	−0.72	−0.95	−0.84	−0.85	−0.68	−0.88	−0.97
	8	−0.40	−0.38	−0.61	−0.53	−0.73	−0.35	−0.16
	9	−0.79	−1.17	−1.06	−1.10	−0.92	−1.20	−1.33

---

Fig. 3 Adsorption energies E_ads of molecules (a) CH₄, (b) CO₂, and (c) H₂O on pristine V₂C and its functionalized structures are calculated with various van der Waals functionals. The solid lines connected points to points are for eye guidance.

3.2.1.2 CO₂. On the pristine V₂C, CO₂ molecules form very strong chemical bonds with surface V atoms with high adsorption energies of −2.0 eV to over −4.0 eV for all the investigated functionals. Here, the π orbitals of CO₂ and d-orbitals of metals are hybridized in a way of the so-called Dewar mechanism,⁶⁷ resulting in such pronounced binding affinity. The most stable adsorption is found, in which CO₂ is strongly bent from its linear structure, as shown in Fig. 2(b). C and two O atoms prefer residing on the V-hollow and C-hollow sites, respectively. This geometry corresponds to the most stable adsorption model, i.e., η³-CO₂-μ⁵-C_MO_CO_C, found by Á. Morales-García and co-workers in the earlier work.³⁹ Particularly, we have seen that CO₂ easily breaks into –CO and –O fragments on the pristine V₂C surface calculated with vdW-DF functional regardless of initial adsorption configurations, while CO₂ can keep its molecular structure for other functionals. The adsorption energy for CO₂ is −4.43 eV for vdW-DF, much higher than others due to the breaking of the O–CO bond, see value in Table 2, entry 4. On V₂CO₂ adsorbate, vdW binding is much smaller than chemical one for all functionals, see Fig. 2. On V₂CO₂_O_v, CO₂ molecules occupy the O_v sites with a single-bond geometry, see Fig. 2(h). This preferable orientation of nonpolar CO₂ significantly enhances its adsorption energies concerning PBE-D3, rev-vdW-DF2, and SCAN-rvv10, from −0.41, −0.17, and −0.18 eV to −0.49, −0.53, and −0.67 eV, respectively. Those functionals are appropriate for the treatment of such chemical bonds.
3.2.1.3 H₂O. H₂O molecules are most stable and stay intact at V-top sites, see Fig. 2(c), among possible adsorption sites on the V₂C surface. Calculated with SCAN-rvv10, the binding energy for H₂O located at the V-bridge is about 5 meV lower than the V-top one. All functionals well predict adsorption energies without much difference, in a range of −0.68 to −0.97 eV, cf. Table 2, entry 7. PBE-D3 and SCAN-rvv10 provide very similar values of −0.95 and −0.97 eV, respectively. On V₂CO₂, it is dominated by the hydrogen bonding between terminating group O and H of H₂O, see Fig. 2(f). The adsorption energy shifts towards lower binding of H₂O compared to the pristine one, as shown in Fig. 3(c). The functional vdW-DF exhibits an exception by showing overbinding of H₂O via hydrogen bond against the chemical bond, i.e., −0.73 vs. −0.68 eV, respectively, while SCAN-rvv10 pronouncedly softens this binding with only −0.16 eV. In the presence of an oxygen-vacant O_v site, all the functionals show their high adsorption affinity towards the occupation of H₂O compared to those on the pristine. Here, H₂O is bound to top-layer vanadium atoms via a triple-bond geometry, see Fig. 2(i), based on three dangling bonds remaining due to the missing functional groups –O. In this case, the energy trend derived for all functionals is identical to the one for V₂C concerning the chemical bonding mechanism. SCAN-rvv10 shows its highest adsorption energy among other functionals with −1.33 eV, while vdW-DF is still underestimated with only −0.92 eV, see Table 2, entry 9.

3.2.2 Insight into the electronic property. To determine the effect of adsorption on the electronic structure of host structures, the density of states (DOS) is plotted, as shown in Fig. 4–6, for V₂C, V₂CO₂, and V₂CO₂_O_v, respectively. Compared to the gas phase, the DOS of bonding states are shifted to lower energy regions. The shift is larger for stronger bindings and vice versa. For CH₄ adsorption, the pristine V₂C surface noticeably modifies σ_p rather than σ_s, as can be seen in Fig. 4(a). Accordingly, σ_p is broader in the adsorbed state compared to the one near the Fermi level in the gas phase. This electronic structure modification is because of a transferring of 0.16e from the surface to CH₄, as presented in Table 3. For CH₄ adsorbed on V₂CO₂ and V₂CO₂_O_v, weak binding of CH₄ is exhibited by a relative shifting of peaks without the influence of the hosts, as plotted in Fig. 5(a) and 6(a), respectively. For CO₂, high charge transferring from the V₂C, about 1.34e, see Table 3, causes a broadening of the 1π_u peak in the adsorbed state, at around −7.5 eV, along with a strong reduction of DOS of the host V₂C near the Fermi level, as plotted in Fig. 4(b). There is also a strong fluctuation of DOS near the Fermi level due to the preferable binding of CO₂ at the vacant site on V₂CO₂_O_v, see Fig. 6(b), compared to the unchanged state on V₂CO₂, shown in Fig. 5(b). For H₂O adsorption, we see that the DOS of the systems are almost similar on V₂C and V₂CO₂ before and after adsorption, while there is a clear lowering shift of the DOS on V₂CO₂_O_v in response to a saturation of adsorbed H₂O at the vacant site.

Fig. 4 Density of state for the adsorption of (a) CH₄, (b) CO₂, and (c) H₂O on V₂C. Dashes arrows indicate shifting of the peaks.

Fig. 5 Density of state for the adsorption of (a) CH₄, (b) CO₂, and (c) H₂O on V₂CO₂. Dashes arrows indicate shifting of the peaks.

Fig. 6 Density of state for the adsorption of (a) CH₄, (b) CO₂, and (c) H₂O on V₂CO₂_O_v. Dashes arrows indicate shifting of the peaks.

Table 3 The charge transfer between the V₂CT_X surfaces and adsorbed molecules. The positive and negative values indicate the charge donating from the surface to the adsorbate and vice versus, respectively

V2CTX	Charge transfer (e)
	CH4	CO2	H2O
V2C	0.16	1.34	−0.03
V2CO2	0.02	0.01	0.00
V2CO2_Ov	0.09	−0.06	−0.09

---

3.3 Oxygen-vacancy effects

3.3.1 Vacancy formation energy. We first determine whether chemical bond strengths depend on the choice of functionals or not by considering the energy cost, or formation energy of vacancy, of eliminating one functional oxygen atom. The calculated result is shown in Fig. 7. All the obtained formation energies are over 2 eV, indicating chemical bonds are formed between surface oxygen and the underlying vanadium atoms. It can be seen that PBE and vdW-DF provide similar formation energies of around 2.65 eV, well in agreement with the literature,¹⁷ of which are smaller than other tested functionals. Formation energy values for PBE-D3, optPBE, and optB88 together distribute over a higher range from 2.90 to 3.10 eV, while ones for rev-vdW-DF2 and SCAN-rvv10 are most pronounced at about 3.20–3.30 eV. Higher formation energies obtained by rev-vdW-DF2 and SCAN-rvv10 are consistent with the fact that V–O bond lengths of these structures are nearest, as mentioned in Section 3.1, leading to stronger binding of O atoms. The trend of formation energy found here also reflects an overestimation of lattice constants and atomic distances of vdW-DF functional, where the bond lengths and bond strengths between atoms are longer and weaker. The bond lengths are improved by other functionals in the tested group, particularly for SCAN-rvv10, and so for the binding of terminal oxygen atoms.

Fig. 7 Formation energies of oxygen vacancy E_Ov were calculated for different van der Waals functionals. The dashed line is for eye guidance.

3.3.2 High vacancy regime. The second and third oxygen atoms are further removed from the surface to generate double- and triple-oxygen vacancies. Here, only the nearest neighbors are considered resulting in two- and three-body cluster defects. The calculated formation energies of single-, double-, and triple-oxygen vacancies obtained by SCAN-rvv10 are 3.28, 3.59, and 4.05 eV, respectively. This implies the formation energies tend to increase for higher defect concentrations. Regarding the adsorption, Table 4 shows the influence of vacancy concentration, i.e., 1/9, 2/9, and 3/9 ML, on the adsorption affinity. As higher oxygen vacancy densities, the adsorption energies of all adsorbates significantly increase, about −0.20 to −0.40 eV, at the highest concentration of 3/9 ML, in response to binding preference to undercoordinated vanadium atoms induced by removing the functional groups. Accordingly, for 3/9 ML, CH₄ molecules retain their η² binding modes, similar to the one on the pristine V₂C, while CO₂ and H₂O molecules attach atop under-coordinated surface vanadiums through monodentate configurations. Those adsorption geometries are provided in the ESI, Section S2.† For 2/9 ML, a slight decrease in the binding of CO₂ and H₂O is found, which might be the competition between the structural change and the adsorption.

Table 4 Adsorption energy (eV) at high vacancy concentrations. The single, double, and triple oxygen vacancies are denoted as 1/9 ML, 2/9 ML, and 3/9 ML, respectively. ML stands for “monolayer”. 9 is the total number of identical adsorption sites on the investigation surface

Gas	Oxygen-vacancy concentration
	1/9 ML	2/9 ML	3/9 ML
CH4	−0.20	−0.22	−0.64
CO2	−0.67	−0.52	−0.87
H2O	−1.33	−1.27	−1.58

---

3.3.3 Adsorption and dissociation of adsorbed H₂O molecules. It has been introduced in the previous section that the absence of surface atomic oxygen creates dangling bonds as active sites for the capturing of target molecules. Among investigated species, H₂O, followed by CO₂, show their highest affinities towards the O_v sites to saturate undercoordinated vanadium atoms. Calculated by SCAN-rvv10 functional, there is a significant energy shift from −0.16 eV to −1.33 eV via the hydrogen-to-chemical bond concerning before and after the presence of O_v, respectively. This adsorption could be characterized by considering the vibrational modes of H₂O before and after binding. Accordingly, we have obtained a lowering of about 200 cm⁻¹ for asymmetry and symmetry stretching modes of adsorbed H₂O compared to its isolated gas-phase counterpart, i.e., 3698 and 3606 vs. 3963 and 3848 cm⁻¹, respectively, while the bending modes are unaffected. This happens in response to the small elongation of O–H bond lengths from 0.96 to 0.97 Å for adapting from the isolated-to-adsorbed states. Moreover, upon binding to undercoordinated vanadium atoms, adsorbed H₂O reveals a weak charge withdrawal of about 0.09 e from surface V atoms for its stabilization. Through first-principles molecular dynamics simulations, the early work on Ti₃C₂O₂, a family of V₂CT_x, showed that H₂O can recover the oxygen at the oxygen-vacancy site by its decomposition reaction, however, it did not address the dissociative barrier.⁶⁸ In this work for V₂CO₂, Fig. 8 shows the calculated energy path along the dissociation channel for H₂O with the intrinsic energy barrier. Adsorbed at O_v site, or IS, one of the hydrogen atoms of H₂O leaves its bonding state and hops to the available neighboring oxygen site, depicted as transition state (TS) and final state (FS) geometries in Fig. 8. For the TS, the activation barrier for this dissociation is small at 0.40 eV, approximately corresponding to the temperature activation of 140 K by heating or electron exposure,⁶⁹ calculated by SCAN-rvv10 functional, while it is found to be 0.30 eV by PBE-D3 as a reference. The imaginary frequency is approximately 1400 cm⁻¹ in magnitude for SCAN-rvv10, compared to 1126 cm⁻¹ obtained by PBE-D3, corresponding to a stretching vibration of the hydrogen atom. This relatively low energy is attributed to a pre-elongation of the H–O bond of adsorbed H₂O and the presence of the nearby functionalized active center –O for the stabilizing of the dissociated-unbound hydrogen atom.

Fig. 8 The energy paths, calculated with SCAN-rvv10, with a reference via PBE-D3, are shown for the dissociation of H₂O initiated at O_v sites. It also includes the geometries of adsorption/initial state (IS), transition state (TS), and final state (TS). The orange background highlights the active region of H₂O on the surface for a top view. The imaginary frequencies denoted as “f/i” are provided for transition states. The black arrow indicates the vibrational direction of the dissociated hydrogen atom at the TS.

Fig. 9 displays the density of state (DOS) for the structures before and after functionalization, the structure with an oxygen vacancy, and the adsorbed H₂O. Compared between Fig. 9(a) and (b), it can be seen that the oxygen functionalization results in the shift of the entire bonding region around the Fermi level, down to −5.0 eV, because more bondings are formed between V and surface oxygen at the O_v site. The DOS of V atoms becomes broadened for both bonding and antibonding states upon missing surface oxygen at that site, see Fig. 9(c). This is beneficial for the adsorption of water molecules, which its DOS is plotted in Fig. 9(d). The O_p orbital of the O atom in H₂O replaces the role of O_v, however, it shows a weaker binding to V than O_v with fewer bonding states around the Fermi level. We can see here that three more peaks, i.e., p1, p2, and s, appear concerning the hybridizations of O_s and O_p orbitals of H₂O with the s orbital of H. One of the O_p peaks, i.e. p1, is higher in energy, and closer to the Fermi level, than the others. At the TS state, Fig. 9(e), the peak p1 is broadened and smaller, showing the weakening of the O–H bond in water. For the final state, the DOS is shown in Fig. 9(f), a new p3 state appears in between p1 and p2 in energy corresponding to the formation of the surface O–H bonding states. The closer orbital energy of the p1 peak against the Fermi level, or in other words closer to DOS of nearby oxygens, enhances the dissociation of water with a small barrier aforementioned and further stabilizes surface oxygens. This is important in showing the potential vacancy-adsorption site for actively catalytic occurrences.

Fig. 9 The projected density of state (PDOS) represents the change in molecular states upon the adsorption and dissociation of H₂O molecules. The subfigure (a)–(c) are the PDOS for the pristine V₂C, O-functionalized V₂CO₂, and oxygen vacancy V₂CO₂_O_v, respectively. The PDOS of H₂O at the O_v site is plotted in (d)–(f) for the adsorption, transition, and final states, respectively. O_s, O_p, and H correspond to the s and p orbitals of O, and the hydrogen atom in H₂O. O_v represents the functionalized oxygen atom at the vacancy site. Here, we only consider the DOS of three V atoms connected to O_v, presented by the black line. The vertical dash line at 0 eV indicates the Fermi level.

4. Conclusions

We study the stability and energetic performances of small organic molecules over V₂C and V₂CT_x by employing first-principles DFT-based calculations. It was demonstrated that commonly used pairwise-additive and nonlocal vdW functionals can similarly describe the adsorption affinity of both polar and nonpolar molecules on V₂CT_x. Recently developed van der Waals functionals, i.e., rev-vdW-DF2 or SCAN-rvv10, are more rational for energetic performances by lessening or enhancing the binding of nonpolar and polar molecules such as CH₄ and H₂O, respectively. The early vdW functional, i.e., vdW-DF, obviously overestimates the atomic bond lengths of both host and target molecules resulting in the adsorption dissociations of extremely stable CO₂ molecules on the pristine surface. For full terminations, adsorption energies, i.e., less than −0.20 eV, are on the same scale for the studied gases showing the binding strengths of weak vdW or hydrogen bonding types. High selectivity is observed for oxygen-vacancy structures with the preferred binding of H₂O at the vacant sites via the chemical bonds. Upon the adsorption, the charges are found to be transferred from the surface to the adsorbed molecules for the pristine V₂C and in an opposite direction for the oxygen vacancy surfaces. The adsorption causes shifting of the DOS toward a lower energy range, resulting in more stable configurations and strong changes in the density of states near the Fermi levels. Higher oxygen-vacancy concentrations offer stronger bindings of all adsorbates relying on the coordination at the vacancy-induced vanadium sites. Preadsorbed single water molecule proceeds a dissociation channel with a small activation barrier, about 0.40 eV, in response to the close orbital overlaps of adsorbed and functionalized atoms. The oxygen-vacancy site serves as an active center for both capturing and activating such dissociation reactions. The different energetics for specific structural phases of adsorptions suggest that CH₄, CO₂, and H₂O respond similarly in full termination surfaces. However, a small number of single-atom defects presented, i.e., oxygen vacancies, enhance the adsorption of H₂O, followed by CO₂. Further removing functional groups, resulting in a pristine surface, is beneficial for the capture of CO₂ only. Our work provides a potential scheme for greenhouse gas separation applications provided that control of the surface chemistry is pivotal in realistic conditions. Certainly, the positions and concentrations of vacancies should be considered and manipulated completely on purpose. In such circumstances, experimental validation is highly valuable not only for greenhouse gas treatment as in this study but also for activating other adsorbates. The engineering of functionalized MXene surfaces paves the way for the enhancement of its chemical activities regarding its multiple roles as binding sites, catalysts, or catalyst beds. Consequently, it is essential to establish an intrinsic correlation between vacancy density and catalytic activity. The co-adsorption of target molecules is also another issue in practice. For example, the clustering of water molecules might happen to the hydrophilic MXene surfaces, influencing either the adsorptive affinity or catalytic activity, which has been disregarded in the current work. All these factors are greatly attributed to both fundamental understandings and practical performances that should be carefully taken into account.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors greatly acknowledge the Center for Innovative Materials and Architectures, Vietnam National University Ho Chi Minh City, Vietnam, Institute for Materials Research, Tohoku University, Japan, and the Institute of Atomic and Molecular Sciences, Academia Sinica, Taiwan for providing authorized computational packages, as well as high-performance computing resources. T. N. M. L. would like to thank Dr Jer-Lai Kuo at the Institute of Atomic and Molecular Sciences, Academia Sinica, Taiwan for the assistance.

References

1. W. Liu, A. Tkatchenko and M. Scheffler, Acc. Chem. Res., 2014, 47, 3369–3377.
2. W. Liu, F. Maass, M. Willenbockel, C. Bronner, M. Schulze, S. Soubatch, F. S. Tautz, P. Tegeder and A. Tkatchenko, Phys. Rev. Lett., 2015, 115, 036104.
3. S. Mehdi Aghaei, A. Aasi and B. Panchapakesan, ACS Omega, 2021, 6, 2450–2461.
4. S. Massoumilari and S. Velioglu, ACS Omega, 2023, 8, 29859–29909.
5. A. Tkatchenko, L. Romaner, O. T. Hofmann, E. Zojer, C. Ambrosch-Draxl and M. Scheffler, MRS Bull., 2011, 35, 435–442.
6. J. E. Lennard-Jones, Trans. Faraday Soc., 1932, 28, 333–359.
7. K. Berland, C. A. Arter, V. R. Cooper, K. Lee, B. I. Lundqvist, E. Schroder, T. Thonhauser and P. Hyldgaard, J. Chem. Phys., 2014, 140, 18A539.
8. S. Grimme, S. Ehrlich and L. Goerigk, J. Comput. Chem., 2011, 32, 1456–1465.
9. N. Marom, A. Tkatchenko, M. Rossi, V. V. Gobre, O. Hod, M. Scheffler and L. Kronik, J. Chem. Theory Comput., 2011, 7, 3944–3951.
10. D. Chakraborty, K. Berland and T. Thonhauser, J. Chem. Theory Comput., 2020, 16, 5893–5911.
11. K. Berland, V. R. Cooper, K. Lee, E. Schroder, T. Thonhauser, P. Hyldgaard and B. I. Lundqvist, Rep. Prog. Phys., 2015, 78, 066501.
12. Y. Gogotsi and B. Anasori, ACS Nano, 2019, 13, 8491–8494.
13. M. Naguib, V. N. Mochalin, M. W. Barsoum and Y. Gogotsi, Adv. Mater., 2014, 26, 992–1005.
14. A. Iqbal, J. Hong, T. Y. Ko and C. M. Koo, Nano Convergence, 2021, 8, 9.
15. V. Shukla, Mater. Adv., 2020, 1, 3104–3121.
16. R. Thakur, A. VahidMohammadi, J. Moncada, W. R. Adams, M. Chi, B. Tatarchuk, M. Beidaghi and C. A. Carrero, Nanoscale, 2019, 11, 10716–10726.
17. J. D. Gouveia and J. R. B. Gomes, Phys. Rev. Mater., 2022, 6, 024004.
18. Z. Xu, X. Lv, J. Chen, L. Jiang, Y. Lai and J. Li, Phys. Chem. Chem. Phys., 2017, 19, 7807–7819.
19. L. Wang, D. Liu, W. Lian, Q. Hu, X. Liu and A. Zhou, J. Mater. Res. Technol., 2020, 9, 984–993.
20. K. Han, Y. Ji, Q. Hu, Q. Wu, D. Li and A. Zhou, Phys. Chem. Chem. Phys., 2024, 26, 18030–18040.
21. Q. Wu, Z. Wang, Q. Hu, Y. Ji, D. Li, J. Wang, Q. Xia, L. Wang and A. Zhou, Phys. Chem. Chem. Phys., 2023, 25, 14406–14416.
22. Q. Hu, Y. Ji, D. Li, Z. Wang, K. Han, Q. Wu, J. Wang, Q. Xia, L. Wang and A. Zhou, Phys. Chem. Chem. Phys., 2023, 25, 9428–9436.
23. S. Li, Y. Wang, Y. Wang, S. Sanvito and S. Hou, J. Phys. Chem. C, 2021, 125, 6945–6953.
24. I. Persson, J. Halim, H. Lind, T. W. Hansen, J. B. Wagner, L. A. Naslund, V. Darakchieva, J. Palisaitis, J. Rosen and P. O. A. Persson, Adv. Mater., 2019, 31, e1805472.
25. R. Morales-Salvador, J. D. Gouveia, Á. Morales-García, F. Viñes, J. R. B. Gomes and F. Illas, ACS Catal., 2021, 11, 11248–11255.
26. D. H. Ho, Y. Y. Choi, S. B. Jo, J. M. Myoung and J. H. Cho, Adv. Mater., 2021, 33, e2005846.
27. J. T. Mazumder and R. K. Jha, Sens. Actuators Rep., 2023, 6, 100174.
28. Q. Hu, W. Liu, D. Li, Q. Wu, Y. Chang, J. Wang, Q. Xia, L. Wang and A. Zhou, Mater. Today Commun., 2024, 40, 109467.
29. M. Keyhanian, D. Farmanzadeh, Á. Morales-García and F. Illas, J. Mater. Chem. A, 2022, 10, 8846–8855.
30. M. Wu, B. Wang, Q. Hu, L. Wang and A. Zhou, Mater., 2018, 11, 2112.
31. M. Naguib, M. Kurtoglu, V. Presser, J. Lu, J. Niu, M. Heon, L. Hultman, Y. Gogotsi and M. W. Barsoum, Adv. Mater., 2011, 23, 4248–4253.
32. K. Matthews, T. Zhang, C. E. Shuck, A. VahidMohammadi and Y. Gogotsi, Chem. Mater., 2021, 34, 499–509.
33. M. Wu, Y. An, R. Yang, Z. Tao, Q. Xia, Q. Hu, M. Li, K. Chen, Z. Zhang, Q. Huang, S.-h. Ma and A. Zhou, ACS Appl. Nano Mater., 2021, 4, 6257–6268.
34. S. M. Majhi, A. Ali, Y. E. Greish, H. F. El-Maghraby and S. T. Mahmoud, Sci. Rep., 2023, 13, 3114.
35. E. Lee, A. VahidMohammadi, Y. S. Yoon, M. Beidaghi and D. J. Kim, ACS Sens., 2019, 4, 1603–1611.
36. N. Liu, J. Li, S. Ma and Z. Jiao, Phys. E, 2023, 148.
37. N. Salami, Appl. Surf. Sci., 2022, 581, 152105.
38. H. Lei, Z. Hao, K. Chen, Y. Chen, J. Zhang, Z. Hu, Y. Song, P. Rao and Q. Huang, J. Phys. Chem. Lett., 2020, 11, 4253–4260.
39. Á. Morales-García, A. Fernández-Fernández, F. Viñes and F. Illas, J. Mater. Chem. A, 2018, 6, 3381–3385.
40. R. Thakur, M. Hoffman, A. VahidMohammadi, J. Smith, M. Chi, B. Tatarchuk, M. Beidaghi and C. A. Carrero, ChemCatChem, 2020, 12, 3639–3643.
41. J. D. Gouveia, Á. Morales-García, F. Viñes, F. Illas and J. R. B. Gomes, Appl. Catal., B, 2020, 260, 118191.
42. A. Bandyopadhyay, D. Ghosh and S. K. Pati, Phys. Chem. Chem. Phys., 2018, 20, 4012–4019.
43. R. Ibragimova, P. Rinke and H.-P. Komsa, Chem. Mater., 2022, 34, 2896–2906.
44. M. A. Hope, A. C. Forse, K. J. Griffith, M. R. Lukatskaya, M. Ghidiu, Y. Gogotsi and C. P. Grey, Phys. Chem. Chem. Phys., 2016, 18, 5099–5102.
45. X. Zhang, Z. Zhang, J. Li, X. Zhao, D. Wu and Z. Zhou, J. Mater. Chem. A, 2017, 5, 12899–12903.
46. S. Ram, G. H. Choi, A. S. Lee, S.-C. Lee and S. Bhattacharjee, J. Phys. Chem. C, 2023, 127, 12576–12585.
47. J. Xu, T. Wang, C. Geng, Y. Zhen, M. Lin, Z. Huang, Y. Shi, L. Zhao and W. Xing, ACS Appl. Nano Mater., 2024, 7, 20974–20984.
48. S. Jung, U. Zafar, L. S. K. Achary and C. M. Koo, EcoMat, 2023, 5, e12395.
49. A. Junkaew and R. Arroyave, Phys. Chem. Chem. Phys., 2018, 20, 6073–6082.
50. G. Kresse and J. Hafner, Phys. Rev. B:Condens. Matter Mater. Phys., 1993, 47, 558–561.
51. P. E. Blochl, Phys. Rev. B:Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
52. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
53. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1997, 78, 1396.
54. J. Klimes, D. R. Bowler and A. Michaelides, J. Phys.:Condens. Matter, 2010, 22, 022201.
55. M. Dion, H. Rydberg, E. Schroder, D. C. Langreth and B. I. Lundqvist, Phys. Rev. Lett., 2004, 92, 246401.
56. I. Hamada, Phys. Rev. B:Condens. Matter Mater. Phys., 2014, 89, 121103(R).
57. H. Peng, Z.-H. Yang, J. P. Perdew and J. Sun, Phys. Rev. X, 2016, 6, 041005.
58. S. Maintz, V. L. Deringer, A. L. Tchougreeff and R. Dronskowski, J. Comput. Chem., 2016, 37, 1030–1035.
59. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.
60. G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904.
61. M. Naguib, O. Mashtalir, J. Carle, V. Presser, J. Lu, L. Hultman, Y. Gogotsi and M. W. Barsoum, ACS Nano, 2012, 6, 1322–1331.
62. A. Champagne, L. Shi, T. Ouisse, B. Hackens and J.-C. Charlier, Phys. Rev. B, 2018, 97, 115439.
63. G. W. Roffe and H. Cox, J. Phys. Chem. A, 2013, 117, 3017–3024.
64. O. V. Lushchikova, S. Reijmer, P. B. Armentrout and J. M. Bakker, J. Am. Soc. Mass Spectrom., 2022, 33, 1393–1400.
65. T. N.-M. Le, T. B. N. Le, P. T. Nguyen, T. T. Nguyen, Q. N. Tran, T. T. Nguyen, Y. Kawazoe, T. B. Phan and D. M. Nguyen, RSC Adv., 2023, 13, 15926–15933.
66. S. Impeng, P. Khongpracha, J. Sirijaraensre, B. Jansang, M. Ehara and J. Limtrakul, RSC Adv., 2015, 5, 97918–97927.
67. V. Parey, B. M. Abraham, S. H. Mir and J. K. Singh, ACS Appl. Mater. Interfaces, 2021, 13, 35585–35594.
68. H. Song and D. E. Jiang, Nanoscale, 2023, 15, 16010–16015.
69. S. Maier and M. Salmeron, Acc. Chem. Res., 2015, 48, 2783–2790.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ra06676f

---