Graphitic carbon nitride nano sheets functionalized with selected transition metal dopants: an efficient way to store CO₂

A constant increase in energy demands over the years has accounted for the enhanced consumption of fossil fuels, causing drastic changes in the atmosphere due to the emission of greenhouse gases like CO₂. Although there is a strong surge for the use of clean energy, fossil fuels might continue to be the main contributor for at least the near future. Apart from fossil fuels, many industrial processes also contribute a large amount of CO₂ to the atmosphere. The capture of atmospheric CO₂ is also significant for alleviating the mesosphere as almost half of the emitted CO₂ cannot escape and remains in the atmosphere, which is yet another problem to be solved [1]. Thus for the protection of environment and restricting a sharp rise in the global temperature, an efficient capture of CO₂ is of immense importance [2, 3]. A modern technology of capturing and storing CO₂, called CCS, is believed to be effective in reducing CO₂ drastically and it has been implemented recently to a great effect [4]. However for a viable implementation of CCS ,an efficient capture of CO₂ is of great significance. Another way to adsorb CO₂ has been presented recently by Shi et al through chemical reaction by means of varying the water quantities in nanopores that would promote the hydrolysis process of CO₃²⁻ [5].

Owing to a possible interaction between CO₂ and various carbon-based nanostructures, material-based CO₂ storage seems to be a viable option and has been studied extensively in recent times [6–9]. However the choice of a proficient material capable of anchoring multiple CO₂ molecules with suitable binding energies is yet another hurdles to cross for efficient CO₂ storage. Moreover a deep understanding of the intrinsic mechanism of CO₂ interaction with the host material is of great importance and plays an important role in designing an efficient CO₂ sorbent [10, 11].

Two-dimensional (2D) nanostructures have lately attracted interest due to their promise in various basic and applied scientific endeavors post the successful synthesis of graphene, a truly single sheet of carbon atoms [12–15]. The maximized surface-to-volume ratio of graphene is particularly appealing to chemical gas sensors to achieve the ultimate level of gas binding. However the constraints on the large-scale production of graphene, and the negligibly weak binding of CO₂, compels researchers to go for not only different strategies in enhancing CO₂–graphene binding but also explore other post graphene 2D materials.

Tawfik et al recently employed first principles calculations to study the CO₂ storage properties of transition metal (TM) doped graphene [16]. They have considered as many as 16 different graphene systems with TMs bonded strongly to defected graphene, capable of anchoring multiple CO₂ molecules with reasonable binding energies. Another computational study by Fischer et al explained the interaction of CO₂ with TiO₂ supported graphene and graphene-oxide sheets by means of density funcional theory (DFT) [17]. It was concluded that the graphene/graphene-oxide supported TiO₂ clusters largely influence the binding of CO₂. In contrast to TMs and TiO₂, Koo et al used calcium (Ca) to functionalize graphene nanoribbons (GNRs) and studied the doped systems as efficient and selective CO₂ storage media though first principles calculations based on DFT [18]. Each Ca on GNRs was found to adsorb three CO₂ molecules selectively among H₂, N₂ and CH₄ with an average binding energy of 0.80 eV/CO₂. Another computational study by Sophia et al investigated the interaction of CO₂ with that SrTiO₃ (001) surface in connection to the band gap modulation of the latter [19].

The CO₂–SrTiO₃ (001) binding falls into the chemisorption range with the opening of the band gap of the surface, which differ with the CO₂–TiO₂ interaction. A relatively different approach was used by Tan et al to capture CO₂ by borophene nano sheets with the help of a charge-modulated mechanism [20]. By means of first principles calculations it was found that the borophene nano sheet could anchor multiple CO₂ molecules with reasonable binding energy upon getting extra electrons injected into the conductive sheet. Controlled, fast and reversible CO₂ adsorption/desorption could be achieved by using negatively charged borophene nano sheets, which makes it one of the excellent materials for CO₂ capture. There are few other reports discussing CO₂ adsorption by means of MoS₂ through electric field control [21] and TM functionalized benzene-complexes [22].

Graphitic carbon nitride (g-C₆N₈) is an important member of graphene-like 2D materials with uniform pores. It belongs to a family of porous 2D membranes, most of which are experimentally synthesized and have found their applications in various scientific and technological fields [23–28]. Different configurations of carbon nitride have already been investigated as efficient hydrogen storage materials both experimentally and theoretically [29–31]. These studies report that metal-doped carbon nitride sheets turn out to be excellent medium for ambient conditions hydrogen storage. Motivated by this property, we have investigated the promise of g-C₆N₈ as a high capacity CO₂ storage material. Our van der Waals corrected DFT calculations revealed that several TM dopants like Sc to Zn, bind strongly with g-C₆N₈, and anchor multiple CO₂ molecules with a binding energy that is much higher than its value on a pristine sheet.

The first principles calculations carried out throughout this project by using VASP code were based on DFT [32–34]. The electronic exchange-correlations interactions were described by means of the generalized gradient approximation (GGA) of Perdew-Burke-Ernzerhof whereas projector augmented-wave pseudopotentials were employed to consider the core electronic effects [35, 36]. Underestimation in the CO₂-g-C₆N₈@TMs bindings, which was caused by the GGA were taken care by including van der Waals correction of Grimme's DFT-D2 as implemented in VASP [37]. An energy cut-off of 500 eV was used for the structural optimization under the plane wave basis set. The Monkhorst-Pack scheme was used for the Brillouin zone sampling with 3 × 3 × 1 k-points for geometry optimization and a 7 × 7 × 1 k-point mesh for plotting the density of states (DOS). Convergence benchmarks of 10⁻⁵ eV were used for the relaxation process of the atomic positions and lattice constants, whereas a force criteria of 0.001 eV Å⁻¹ was used. We used a reasonably large 2 × 2 supercell of g-C₆N₈ with 56 atoms in this study having a vacuum space of 15 Å along the z-direction to avoid the possible vertical interactions. A single dopant of selected TMs was considered on a g-C₆N₈ supercell that constitutes a relatively small doping concentration of 1.785%.

The binding energies of TMs on g-C₆N₈ can be calculated by the following relation:

$$\begin{eqnarray}&&{{E}}_{{\rm{b}}}({X})={E}({\rm{g}} \mbox{-} {{\rm{C}}}_{6}{{\rm{N}}}_{8}@{X})-{E}({\rm{g}} \mbox{-} {{\rm{C}}}_{6}{{\rm{N}}}_{8})-{E}({X}).\end{eqnarray} \tag{ 1 }$$

Here, E_b(X) is the binding energy of the dopant. E(g-C₆N₈@X), E(g-C₆N₈) and E(X) are the total energies of g-C₆N₈ doped with TMs, pristine g-C₆N₈ and isolated dopant, respectively.

$$\begin{eqnarray*}&&{X}={\rm{Sc}},{\rm{Ti}},{\rm{V}},{\rm{Cr}},{\rm{Mn}},{\rm{Fe}},{\rm{Co}},{\rm{Ni}},{\rm{Cu}},{\rm{Zn}}.\end{eqnarray*}$$

The following relation can calculate binding energies of CO₂ molecules on g-C6N8@TMs

$$\begin{eqnarray}{{E}}_{{\rm{b}}}({{\rm{CO}}}_{2}) & = & [{E}({\rm{g}} \mbox{-} {{\rm{C}}}_{6}{{\rm{N}}}_{8}@{X}+{n}\,{{\rm{CO}}}_{2})\\ & & -{E}({\rm{g}} \mbox{-} {{\rm{C}}}_{6}{{\rm{N}}}_{8}@{X}+({n}-1){{\rm{CO}}}_{2})\\ & & -{n}\,{E}({{\rm{CO}}}_{2})]/{n}.\end{eqnarray} \tag{ 2 }$$

In equation (2), the first and second terms represent the total energies of TM functionalized g-C₆N₈ loaded with n and (n − 1) CO₂ molecules respectively, where n is the number and E (CO₂) is the total energy of CO₂ molecule.

Structural properties of g-C₆N₈ and g-C₆N₈@TMs

Before studying the CO₂ storage properties of g-C₆N₈@TMs, we will briefly describe the structural properties of the monolayers in their pristine and functionalized form. The optimized C–N bond lengths of the atoms that constitute the six member rings and the atoms that joined the rings are found to be 1.476 Å and 1.33 Å, respectively. The lattice parameter after relaxation is 7.132 Å. These values agree well with the literature [30]. The lowest energy configuration of a 2 × 2 supercell of g-C₆N₈ is given in figure 1.

Zoom In Zoom Out Reset image size

In its pure form, g-C₆N₈ is inert towards CO₂, thus to enhance g-C₆N₈–CO₂ interaction to an acceptable value for ambient condition CO₂ storage, we introduced selected TMs on different binding cites available in g-C₆N₈. A set of selected TMs from Sc–Zn has been considered for the doping process. While considering the suitability of the doping mechanism of foreign adatoms on 2D g-C₆N₈ monolayers, the most important parameters are stability and reversibility of the doped systems. Both of these parameters could be established through the determination of the E_b/E_c ratio [38–42]. For E_b/E_c > 1, a uniform scattering of dopants over the g-C₆N₈ is preferred that ensures stability and reversibility, whereas E_b/E_c < 1 would lead towards cluster formation instead of binding. Considering this fact, we have carefully calculated the E_b/E_c ratio for the selected TMs over g-C₆N₈.

For convenience we will explain here the case of Sc on g-C₆N₈. In order to find the most stable binding site, we introduced Sc on all the possible positions, which are C/N top, bridge and hollow sites formed by a small ring, a bridge of atoms linking the small rings and a big pore. The Sc dopant, similar to all the other considered TMs dopants, prefers to sit in plane with g-C₆N₈ in the big pore with significantly high binding energies (E_b) as shown in figure 2.

Zoom In Zoom Out Reset image size

In the case of Sc, the calculated E_b value of −12.21 eV is much higher than its experimental E_c and Sc sits exactly in the center of the big pore. An average distance of Sc with the six neighboring N atoms is 2.39 Å. After introducing Sc, the g-C₆N₈ monolayer does not remain exactly planar but crumbled a little. However this slight crumbling is reversible, not permanent, and the monolayer comes back to its planar form once Sc dopant is removed and the systems is optimized again. With E_b/E_c ∼ 3.13, Sc would stay on g-C₆N₈ comfortably even at significantly high temperature and nullifies the likelihood of cluster formation. Other TM dopants also ensured E_b/E_c > 1, for example Ti ∼ 2.6, V ∼ 2.6, Cr ∼ 3.9 and Mn ∼ 5.3, thus guaranteeing the stability and reversibility of the functionalized systems. It is worth mentioning here that the binding mechanism and E_b values of the considered TMs on g-C₆N₈ follow a similar trend to that of g-CN [43]. A comparative analysis of E_b to E_c of the studied TMs is given in figure 3.

Zoom In Zoom Out Reset image size

A strong E_b is evidence of the fact that a significant amount of charge is being transferred from the TMs (with smaller electronegativities) to the N atoms (with higher electronegativity) of g-C₆N₈, with which dopants are bonded. Let us consider the case of Sc–g-C₆N₈; Sc loses an electronic charge of around 1.701 e to its closest N atoms of g-C₆N₈. This depletion and accumulation of charges can be seen in the isosurface charge density plots in figure 4. Other TM dopants also acquire partial positive state upon losing a substantial amount of their electronic charges to their nearby N atoms of g-C₆N₈. This transfer of charge would also result into a strong hybridization between the TMs and g-C₆N₈, which is discussed in the electronic properties section.

Zoom In Zoom Out Reset image size

Adsorption mechanism of CO₂ on g-C₆N₈@TMs

After establishing a strong TMs-g-C₆N₈ binding, we introduced multiple CO₂ molecules on these functionalized systems one by one. It is worth mentioning here that after the Tawfik et al study of CO₂ storage by defected graphene doped TMs [16], the present study is the only attempt to report the storage of several CO₂ on TM functionalized g-C₆N₈. All the doped systems anchor CO₂ under a similar mechanism; however, as a representative system, we will explain the case of Sc@g-C₆N₈. For CO₂ adsorption, we have tried several adsorption configurations of CO₂ on Sc@g-C₆N₈ by considering different binding distances of CO₂ from cationic Sc⁺ as well as by changing the orientations of the incident CO₂ molecules like C-directed, O-directed and tilted molecule. Energetics analysis all the CO₂-Sc@g-C₆N₈ configurations reveal the most stable structure, as shown in figure 5.

Zoom In Zoom Out Reset image size

The lowest energy configuration of single CO₂ on Sc-g-C₆N₈ yields an E_b value of −0.77 eV at a binding distance of 2.07 Å, which agrees very well with the single CO₂ adsorption on Ca and selected TM doped carbon-based nanostructures [16, 18, 44]. This reasonably high E_b and appropriate binding distance suggest that Sc-g-C₆N₈ is a good option for reversible CO₂ capture. Upon CO₂ adsorption, Sc-g-C₆N₈ crumbles a little further and a CO₂ molecule also rearranges its structure with the elongation of C–O bond lengths to 1.21 Å and 1.29 Å, from an equilibrium value of 1.16 Å. The molecule also loses its planar geometry with the shrinking of the O–C–O angle to 136.4°. Now we study the coverage effect by introducing more CO₂ to Sc-g-C₆N₈, which is pre-adsorbed with single CO₂. While studying the multiple CO₂ storage mechanism, we adopted the similar principle regarding binding energies and binding distances as that of [16]. We kept on increasing the number of CO₂ on Sc-g-C₆N₈ as long as the resulting E_b values exceeded a threshold of −0.25 eV/CO₂ and the binding distance remained lower than 4.0 Å, which are the usual physisorption parameters [45]. The second CO₂ molecule was introduced by considering several configurations and maintaining a suitable distance not only with Sc-g-C₆N₈ but also with the pre-adsorbed CO₂, to avoid any possible repulsion between CO₂ molecules. A resultant ground state configuration yields average E_b and binding distance of −0.50 eV/CO₂ and 3.0 Å respectively. Upon structural relaxation, there is hardly any change in the geometry of the second CO₂, which has C–O bond lengths of 1.18 Å each. Now for the adsorption of the third CO₂, we considered the ground state configuration of Sc-g-C₆N₈ with 2CO₂ already adsorbed. Most stable configurations yield E_b values of −0.42 eV and −0.37 eV per CO₂ in the case of 3CO₂ and 4CO₂ on Sc-g-C₆N₈, respectively. Adsorption distances in all the cases up to 4CO₂ fall within the prescribed criteria mentioned above. However the adsorption of the fifth CO₂ on Sc-g-C₆N₈ pre-adsorbed with 4CO₂ would not only result into an E_b that is smaller than −0.25 eV/CO₂ but also at a larger distance, falling short to the laid down criteria. It is therefore concluded that Sc-g-C₆N₈ could take 4CO₂ with reasonable E_b and binding distances. Optimized structures of Sc-g-C₆N₈ with multiple CO₂ are shown in figure 5. Similar to Sc-g-C₆N₈, we have studied the mechanism of multiple CO₂ storage on g-C₆N₈ functionalized with other TMs and results of the respective E_b are given in figure 6. It is clear from figure 6 that with the increasing number of CO₂, the E_b decreases due the increase of distances from CO₂ and Sc-g-C₆N₈ and possible repulsion among the partially polarized pre-adsorbed CO₂.

Zoom In Zoom Out Reset image size

The analysis of DOS, shown in figure 7(a), indicates that g-C₆N₈ is semiconducting where its valence band and conduction band are predominantly formed by N-2p and the overlap of N-2p and C-2p, respectively. In particular, the remarkable overlap of N-2p and C-2p accounts for a sigma C–N bond (i.e., 0.5–2.5 eV as anti-bonding and −6.0 to −4.0 eV as bonding). This resembles the character of photocatalytic material g-C₃N₄. The integration of a Sc atom into the hole of g-C₆N₈ results in the significant shift in the Fermi level to the conduction band, manifesting the n-type impurity that originates from the donation of two outermost 4 s electrons of Sc to the π* state of the sheet [43]. As a consequence, the donating atom has become a positively charged Sc²⁺ ion of which there is a localized electron in the 3d orbital (d¹) as evidenced by the major spin-up close to the Fermi level (figure 7(b)). According to the optimized structure of Sc-doped C₆N₈, six N atoms surround the Sc dopant. (i.e., the hexagonal crystal field) This field leads to splitting of the d orbital where the d¹ electron is filled in the lowest degenerate states either d_xz or d_yz [43]. Therefore, it is prone to interact with adsorbates (here CO₂ molecules) via orbital overlapping.

Zoom In Zoom Out Reset image size

Figure 8(a) shows the DOS Sc-doped C₆N₈ adsorbed with various numbers of CO₂ molecules. There is the apparent overlap of Sc 3d with the LUMO (mainly derived from C and O 2p orbitals at around −0.5 to 0.5 eV) of the first CO₂ molecule. This overlap is responsible for the formation of strong covalent Sc–C and Sc–O bonds. As a result, they notably induce two major modifications (1) the elongation of two C–O double bonds in CO₂ from the initial value of 1.16 Å to 1.29 Å and 1.21 Å and (2) the bending of C–O–C from 180.0° to 136.4°, as schematically depicted in figure 8(b). The subsequent adsorption of the second CO₂ leads to the gradual disappearance of magnetic moment due to delocalization. Interestingly, further inclusion of the third and fourth CO₂ molecules yield negligible changes in the DOS, indicating that they are weakly bound to the functionalized sheet by van der Waals forces with the probable adsorption energy as described previously. The presence of the van der Waals interaction of the second to fourth molecules is also supported by the slight changes in their C–O bond length and C–O–C bond angle. Therefore, it is conclusive that a Sc-doped C₆N₈ sheet is a suitable material for CO₂ storage with exceptional storing capacity.

Zoom In Zoom Out Reset image size

In conclusion, we have performed rigorous ab initio calculations based on DFT, to design and investigate metallized g-C₆N₈ monolayers as potential CO₂ adsorbents for environment protection. A set of selected TMs that includes Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu and Zn have been functionalized on a supercell of g-C₆N₈ at a doping concentration of 1.79%. Our van der Waals induced calculations have revealed that each dopant sticks to a g-C₆N₈ nano sheet without forming metal clusters, which is extremely important for the reversibility of the doped system. The doping mechanism involves a transfer of electronic charge from TMs to the g-C₆N₈, altering the electronic properties of the latter without compromising its stability. Each TM's dopant in cationic form is capable of anchoring a maximum of four CO₂ molecules with binding energies, which are much higher than that of pristine g-C₆N₈. We strongly believe that these findings will provide the basis for the synthesis of a promising class of functional nanostructures with application in feasible CO₂ storage.

T H is indebted to the resources at NCI National Facility systems at the Australian National University through National Computational Merit Allocation Scheme supported by the Australian Government and the University of Queensland Research Computing Centre. R A acknowledges the Swedish Research Council (VR), Carl Tryggers Stiftelse för Vetenskaplig Forskning and StandUp for financial support. The SNIC and UPPMAX are also acknowledged for provided computing time. T K would like to acknowledge the Development and Promotion of Science and Technology Talent Project (DPST) for the financial support of this project. The Nanotechnology Centre (NANOTEC), NSTDA Ministry of Science and Technology (Thailand) also supports T K through its program of Centre of Excellence Network, Integrated Nanotechnology Research Centre Khon Kaen University (Thailand).