Conductive Graphitic Carbon Nitride as an Ideal Material for Electrocatalytically Switchable CO₂ Capture

Abstract

Good electrical conductivity and high electron mobility of the sorbent materials are prerequisite for electrocatalytically switchable CO₂ capture. However, no conductive and easily synthetic sorbent materials are available until now. Here, we examined the possibility of conductive graphitic carbon nitride (g-C₄N₃) nanosheets as sorbent materials for electrocatalytically switchable CO₂ capture. Using first-principle calculations, we found that the adsorption energy of CO₂ molecules on g-C₄N₃ nanosheets can be dramatically enhanced by injecting extra electrons into the adsorbent. At saturation CO₂ capture coverage, the negatively charged g-C₄N₃ nanosheets achieve CO₂ capture capacities up to 73.9 × 10¹³ cm⁻² or 42.3 wt%. In contrast to other CO₂ capture approaches, the process of CO₂ capture/release occurs spontaneously without any energy barriers once extra electrons are introduced or removed and these processes can be simply controlled and reversed by switching on/off the charging voltage. In addition, these negatively charged g-C₄N₃ nanosheets are highly selective for separating CO₂ from mixtures with CH₄, H₂ and/or N₂. These predictions may prove to be instrumental in searching for a new class of experimentally feasible high-capacity CO₂ capture materials with ideal thermodynamics and reversibility.

Introduction

At the current rate of emissions of greenhouse gases, for which carbon dioxide (CO₂) is the main component, global warming and climate change will continue to rise^1,2,3. One crucial issue facing efficiently separating, capturing, storing and/or converting CO₂ is the development of a practical sorbent material^4,5,6. Liquid-amine, which is the most common adsorbent for current industrial process for CO₂ capture, suffers from relatively low efficiency, equipment corrosion, solvent loss and toxicity^7,8,9,10. Alternatively, various solid materials have been proposed as attractive adsorbents for CO₂ capture, including metal-organic frameworks (MOFs)^{11,12,13,14,15}, aluminum nitride (AlN)¹⁶, carbon^17,18,19, hexagonal boron nitride (h-BN)²⁰ and silicon carbide (SiC)^21,22 nanostructures. However, the difficult regeneration processes due to the large adsorption energy, which generally demands high temperatures to release captured CO₂, significantly hinders their practical applications.

Recently, electrocatalytically switchable CO₂ capture scheme has been proposed as a controllable, high selective and reversible CO₂ capture strategy for bare h-BN nanomaterials²³. Specifically, CO₂ molecules are weakly adsorbed (i.e. physisorbed) on neutral h-BN. By injecting extra electrons into h-BN adsorbent, density functional theory (DFT) calculations reveal that CO₂ adsorption can be dramatically enhanced via a charge-induced chemisorption interaction. The chemically adsorbed CO₂ can in principle be released when the extra electrons are removed. In contrast to previous methods, the CO₂ capture/release occurs spontaneously once extra electrons are introduced or removed and the process of CO₂ capture/release can be simply controlled and reversed by switching on/off the charges carried by h-BN nanomaterials. However, h-BN is wide-gap semiconductor with band gap around 5.8 eV^24,25 and it is not clear how to charge up bare h-BN due to its insulating character.

To overcome the above disadvantage, Jiao et al.²⁶ have investigated carbon nanotubes with pyridinic-nitrogen as an alternative absorbent to electrocatalytically switchable CO₂ capture because of their good electron conductivity. On the other hand, we have proposed layered h-BN and graphene (hybrid BN/G) nanosheets, consisting of a single or double-layer h-BN and a substrate graphene layer, as an experimentally feasible approach to induce the requisite charge on h-BN for electrocatalytically switchable CO₂ capture²⁷. However, the synthesis of carbon nanomaterials with pyridinic nitrogen doping and hybrid BN/G are difficult to control in experiment. One natural question arise: can we find a conductive sorbent material for electrocatalytically switchable CO₂ capture, which avoids complicated synthesis route?

Very recently, intense attention has been attracted by a new class of two-dimensional conjugated polymer, graphitic carbon nitride, due to the anisotropic two-dimensional geometric morphology and the aromatic π-conjugated framework. This endows carbon nitride nanosheets with attractive bandgap- and surface-engineered applications in both energy- and environment-related topics, such as photocatalysis for water splitting^28,29, hydrogen evolution³⁰, CO2 reduction³¹, organosynthesis³², amongst others³³.g-C₃N₄ and g-C₄N₃ are two kinds of two-dimensional conjugated nanosheets, which have been recently synthesized by using cross-linking nitride-containing anions in ionic liquid^34,35. Different from each other, g-C₃N₄ is semiconductor³⁴, while g-C₄N₃ shows half-metallic property³⁶.

Here we show that electrocatalytically switchable CO₂ capture is indeed possible by considering conductive g-C₄N₃ nanosheets, of which the charge states can be easily modified experimentally because of the good electrical conductivity and high electron mobility. Using first-principle calculations, we found that the adsorption energy of CO₂ molecules on g-C₄N₃ nanosheets can be dramatically enhanced from 0.24 to 2.52 eV by injecting extra electrons into the adsorbent. At saturation CO₂ capture coverage, the negatively charged g-C₄N₃ nanosheets achieve CO₂ capture capacities up to 73.9 × 10¹³ cm⁻² or 42.3 wt%. Once the extra electrons are removed, the captured CO₂ molecules can easily desorb from the adsorbent. In contrast to other CO₂ capture approaches, the process of CO₂ capture/release occurs spontaneously without any energy barriers once extra electrons are introduced or removed and these processes can be simply controlled and reversed by switching on/off the charging voltage. In addition, these negatively charged g-C₄N₃ nanosheets are highly selective for separating CO₂ from mixtures with CH₄, H₂ and/or N₂. These predictions might pave the way in searching for a new class of experimentally feasible high-capacity CO₂ capture materials with ideal thermodynamics and reversibility.

Results

Since good electrical conductivity and high electron mobility are prerequisite for injecting extra electrons into electrocatalytically switchable CO₂ capture materials, we first studied the electronic structures of isolated g-C₄N₃. The lowest-energy configurations and the calculated band structures of g-C₄N₃ are shown in Fig. 1. Consistent with previous studies³⁶, g-C₄N₃ is a (2 × 2) reconstructed structure with half-metallic state. This indicates that g-C₄N₃ has good electrical conductivity and high electron mobility, which should readily facilitate electron injection/release for electrocatalytically switchable CO₂ capture.

Figure 1

Top (upper) and side (lower) views of (a) a (2 × 2) reconstructed g-C4N3 supercell. The blue and grey balls represent N and C atoms, respectively and the unit cell of g-C₄N₃ is indicated by red dot lines. C₁ and C₂ denote different C atoms in g-C₄N₃ unit cell. The calculated band structures of (b) a (2 × 2) reconstructed g-C₄N₃. The blue dashed line denotes the Fermi level. The red and black lines in (b) denote the spin-up and spin-down states, respectively.

Single CO₂ Adsorption on Neutral and 2 e⁻ Negatively Charged g-C₄N₃ Nanosheets

We next shift our attention to a single CO₂ adsorption on neutral and negatively charged g-C₄N₃. Since g-C₄N₃ is a (2 × 2) reconstructed structure, there are many different adsorption sites for a CO₂ molecule. Here, we considered all the adsorption sites: directly on top of a C or N atom, above the midpoint of a bond linking the C and N atoms and above the center of a honeycomb-like hexagon. Figure 2 shows the lowest-energy configurations of a CO₂ absorbed on neutral and 2 e⁻ negatively charged g-C₄N₃. On neutral g-C₄N₃ (Fig. 2(a)), the linear CO₂ molecule is parallel to g-C₄N₃ and locates on top of three nitrogen atoms. The distance between the C atom of CO₂ and closest N atom is 2.966 Å and the linear CO₂ molecule shows little structural change compared to a free CO₂ molecule with the O-C-O angle and two double C=O bonds being 178.2° and 1.176 Å, respectively. Mulliken population analysis suggests that the amount of transferred electron from the absorbed CO₂ molecule to g-C₄N₃ is negligible (about 0.004 e⁻). For the neutral case, the CO₂ molecule is weakly adsorbed (i.e. physisorbed) onto neutral g-C₄N₃ with small adsorption energy of 0.24 eV.

Figure 2

Top and side views of the lowest-energy configurations of a single CO₂ molecule absorbed on the (a) neutral and (b) 2 e⁻ negatively charged g-C₄N₃. The blue, grey and red balls represent N, C and O atoms, respectively and the adsorption energies of the CO₂ molecule on neutral and 2 e⁻ negatively charged g-C₄N₃ are listed.

After injecting two extra electrons into the g-C₄N₃ supercell (Fig. 2(b)), the CO₂ is strongly adsorbed at surface N atom and changes from physisorption into chemisorption on 2 e⁻ negatively charged g-C₄N₃. The distance between the C atom of CO₂ and surface N atom of g-C₄N₃ is shortened to 1.569 Å, the O-C-O angle is bent from 178.2° to 131.8°, the two double C=O bonds are elongated from 1.176 to 1.246 Å and the charge transfer from g-C₄N₃ to CO₂ increase to 0.56e⁻. In this case, the adsorption energy of a CO₂ remarkably increases to 1.20 eV, which is much larger than the adsorption energies of CO₂ on other high-performance adsorbents (0.4–0.8 eV)⁶, indicating that the negatively charged g-C₄N₃ is an excellent adsorbent for CO₂ capture.

To understand the underlying mechanism of CO₂ capture on negatively charged g-C₄N₃, we plotted the deformation electronic density of neutral and 2 e⁻ negatively charged g-C₄N₃ by subtracting the electronic density of isolated N and C atoms from the sheet in Fig. 3. Obviously, for the neutral case (Fig. 3(a)), some electrons are extracted from the C atoms and delocalized over the N atoms, as implied by the green regions. Mulliken population analysis indicated that the electrons distribute at N, C₁ and C₂ are −0.302, 0.294 and −0.036 |e|, respectively. When two extra electrons are introduced (Fig. 3(b)), the extra electrons are almost evenly distributed on N and C atoms. Mulliken population analysis suggest that each atom gains −0.07 ~ −0.09 |e| and the electrons distribute at N, C₁ and C₂ are −0.383, 0.222 and −0.122 |e|, respectively. Compared with the neutral case, more electrons are distributed and delocalized at N atoms, as implied by the green regions in Fig. 3(b). As CO₂ is a Lewis acid and it prefers to accept, rather than donate, electrons during reaction, the N atom of negatively charged g-C₄N₃ can donate electrons to CO₂ and form a new bond between the C atom of CO₂ and surface N atom of g-C₄N₃ (Fig. 3(d)), which is significantly different from the case that CO₂ on neutral g-C₄N₃ (Fig. 3(c). This is the reason why the CO₂ molecule has a strong interaction with negatively charged g-C₄N_3.

Figure 3

The deformation electronic density of (a) neutral and (b) 2 e⁻ negatively charged g-C₄N₃. Green and yellow refer to electron-rich and -deficient area, respectively. The isosurface value is 0.02 e/au. (c) The total charge density distribution of a single CO₂ molecule on (c) neutral and (d) 2 e⁻ negatively charged g-C₄N₃. The isosurface value is 0.8 e/au. The overlap of the electron densities of the C atom of CO₂ and surface N atom of g-C₄N₃ in (d) indicates the formation of a new bond.

In order to investigate the kinetic process of CO₂ capture/release on 2 e⁻ negatively charged g-C₄N₃, we next studied the energy change of a CO₂ molecule adsorbed on g-C₄N₃ after the introduction or removal of the two extra electrons. In Fig. 4(a), we started with the lowest-energy configuration of neutral g-C₄N₃ with a physisorbed CO₂ molecule. Two electrons are then added to the neutral g-C₄N₃ and we examined the energy changes as the system relaxes to the 2 e⁻ negatively charged optimized state. In Fig. 4(b), we started with the lowest-energy configuration of the 2 e⁻ negatively charged g-C₄N₃ with a chemisorbed CO₂ molecule. Two electrons are removed and then the system is allowed to relax, forming a physisorbed CO₂ molecule. When two extra electrons are introduced into g-C₄N₃, the interactions between the CO₂ molecule and the 2 e⁻ negatively charged g-C₄N₃ are significantly larger than that with neutral g-C₄N₃ and the CO₂ molecule spontaneously relaxes to chemisorption configuration. This process is exothermic by 1.08 eV without any energy barrier. On the other hand, when two extra electrons are removed from the 2 e⁻ negatively charged g-C₄N₃, the CO₂ molecule spontaneously returns to the weakly bound state and desorbs from g-C₄N₃. This process is also exothermic by 1.39 eV without any energy barrier. Therefore, the CO₂ storage/release processes on negatively charged g-C₄N₃ are reversible with fast kinetics and can be easily controlled via adding/removing the extra electrons.

Figure 4

The energy change of (a) the relaxation (capture) of a CO₂ molecule on g-C₄N₃ after two extra electrons are introduced and (b) the reverse relaxation (release) process of a captured CO₂ molecule from g-C₄N₃ after two extra electrons are removed from the adsorbent.

The Effects of Charge Density on Single CO₂ Capture on Negatively Charged g-C₄N₃ Nanosheets

To investigate the effects of charge density on CO₂ capture on negatively charged g-C₄N₃, we investigated a CO₂ adsorption on negatively charged g-C₄N₃ with different charge densities. Here, we defined the charge densities of g-C₄N₃ as follows

where ρ, Q and S are the charge densities of g-C₄N₃, the total charge and the surface area in 2 × 2 supercell, respectively. For g-C₄N₃, the surface area in 2 × 2 supercell can be calculated as , where a is the lattice constant of 2 × 2 supercell.

Figure 5 shows the adsorption energies of a CO₂ on negatively charged g-C₄N₃ and the charge transfer between CO₂ and g-C₄N₃ as functions of charge densities. For small charge density case (<13.9 × 10¹³ cm⁻²), the adsorption energy of CO₂ is small (0.24 ~ 0.35 eV) and charge transfer between CO₂ and g-C₄N₃ is less than 0.06 e⁻. When charge density is larger than 13.9 × 10¹³ cm⁻², the adsorption energy of CO₂ and the charge transfer from g-C₄N₃ to CO₂ increase dramatically with increasing charge density, indicating the CO₂ molecule can only adsorb on negatively charged g-C₄N₃ with large charge density. Considering the adsorption energies of CO₂ on other high-performance adsorbents is 0.4−0.8 eV ⁶, we define the minimum charging density for CO₂ capture on negatively charged g-C₄N₃ is about 17.0 × 10¹³ cm⁻².

Figure 5

The adsorption energies of a CO₂ on negatively charged g-C₄N₃ and the charge transfer between CO₂ and g-C₄N₃ as functions of charge densities.

The gray area indicates the adsorption region.

CO₂ Capture Capacity on Negatively Charged g-C₄N₃ Nanosheets

To estimate CO₂ capture capacity on negatively charged g-C₄N₃, we studied the maximum number and the average adsorption energy of captured CO₂ molecules on negatively charged g-C₄N₃ with different charge densities (Fig. 6(a)). Here, we determinate the maximum number of captured CO₂ for each negatively charged g-C₄N₃ with different charge density by gradually increasing the number of CO₂ molecules on negatively charged g-C₄N₃ until no more CO₂ can be absorbed. The average adsorption energy of captured CO₂ is calculated as the total adsorption energy divided by the maximum number of captured CO₂. The results show that no CO₂ molecules can be captured by negatively charged g-C₄N₃ with small charge density (≤12.3 × 10¹³ cm⁻²). As the charge density increase from 18.5 × 10¹³ to 61.6 × 10¹³ cm⁻², the negatively charged g-C₄N₃ can capture two, four and six CO₂ molecules with the average adsorption energy of captured CO₂ molecules ranging from 0.72 to 3.58 eV. We note that a further increase in the number of CO₂ molecules leads to some CO₂ molecules moving far away from the adsorbent during the geometry optimization even if we further increase the charge density of g-C₄N₃. Therefore, we define six CO₂ molecules in each 2 × 2 supercell (i.e. CO₂ capture capacity 73.9 × 10¹³ cm⁻² or 42.3 wt%) as the likely saturation CO₂ capture coverage (Fig. 6(b,c)). It should be noted that surface defective sites such as N vacancies or un-condensed amino group could lower CO₂ capture capacity. However, considering the high CO₂ capture capacity of negatively charged g-C₄N₃, we believe this may nevertheless represent a feasible high-capacity CO₂ capture material.

Figure 6

(a) The maximum number and the average adsorption energies of captured CO₂ molecules on negatively charged g-C₄N₃ with different charge densities. (b) Top and (c) side views of the lowest-energy configuration of six CO₂ molecules adsorbed on negatively charged g-C₄N₃ with charge density 61.7 × 10¹³ cm⁻².

Interestingly, we note that the CO₂ molecules do not all bind equally but the capture process occurs discretely two at a time. To further confirm this phenomenon, we put four CO₂ molecules on neutral g-C₄N₃ and gradually increase the charge density of negatively charged g-C₄N₃ until four CO₂ are all captured (corresponding lowest-energy configurations are shown in Figure S1 of the Supporting Information). Clearly, four CO₂ are physically adsorbed on neural and 1 e⁻ negatively charged g-C₄N₃ (Figure S1(a,b), Supporting Information). On 2 e⁻ negatively charged g-C₄N₃, two CO₂ are chemisorbed while other two CO₂ are physisorbed on adsorbent (Figure S1(c), Supporting Information). When three electrons are introduced, all the CO₂ are captured on 3 e⁻ negatively charged g-C₄N₃ (Figure S1(d), Supporting Information).

CH₄, H₂ and N₂ Adsorption on g-C₄N₃ Nanosheets

CH₄, H₂, N₂ are three types of gas mixtures that are currently most interesting for CO₂ capture technologies, namely, postcombustion (predominantly CO₂/N₂ separation), natural gas sweetening (CO₂/CH₄) and precombustion (CO₂/H₂) capture³⁷. In order to demonstrate the high selectivity of negatively charged g-C₄N₃ nanosheets for CO₂ capture, we calculated the adsorption energies of CH₄, H₂ and N₂ on neutral and negatively charged g-C₄N₃ and compared with those of CO₂. In Fig. 7 we list the comparative adsorption energies of CO₂, CH₄, H₂ and N₂ on neutral, 1 e⁻ and 2 e⁻ negatively charged g-C₄N₃ (corresponding lowest-energy configurations are shown in Figure S2 of the Supporting Information). Clearly, the adsorptions of CH₄, H₂ and N₂ on neutral, 1 e⁻ and 2 e⁻ g-C₄N₃ are all physical rather than chemical. The distance between the carbon atom of CH₄ (the hydrogen atom of H₂, the nitrogen atom of N₂) and g-C₄N₃ is 3.157–3.159 (2.111–2.539, 2.865–3.236) Å, respectively. The adsorption energies of CH₄, H₂ and N₂ on neutral, 1 e⁻ and 2 e⁻ g-C₄N₃ range from 0.06 to 0.39 eV. In contrast, although CO₂ is physically adsorbed at neutral and 1 e⁻ g-C₄N₃ with small adsorption energy in the range from 0.24 to 0.32 eV, CO₂ is tightly chemisorbed on 2 e⁻ g-C₄N₃ with large adsorption energy of 1.20 eV. The above comparisons demonstrate that negatively charged g-C₄N₃ has very high selectivity for capturing CO₂ from CH₄, H₂ and/or N₂ mixtures.

Figure 7

The adsorption energies of CO₂, CH₄, H₂, N₂ and H₂O on neutral, 1 e^– and 2 e^– negatively charged g-C₄N₃.

Water Adsorption on g-C₄N₃ Nanosheets

Since water saturates most industrial gas streams, including flue gas, we also studied the adsorption energies of H₂O on neutral and negatively charged g-C₄N₃ and compared with those of CO₂, as shown in Fig. 7 (corresponding lowest-energy configurations are shown in Figure S2 of the Supporting Information). On neutral g-C₄N₃, both CO₂ and H₂O are physically adsorbed on adsorbents with small adsorption energies of 0.24 and 0.38 eV, respectively. On 1 e⁻ g-C₄N₃, the adsorption energy of CO₂ slightly increases to 0.32 eV, while the adsorption energy of H₂O significantly increases to 0.60 eV, which is twice as much as that of CO₂. On 2 e⁻ g-C₄N₃, both CO₂ and H₂O are chemically adsorbed on adsorbents with large adsorption energies of 1.20 and 1.18 eV, respectively. These results indicate that the negatively charged g-C₄N₃ cannot selectively capture CO₂ from a gas mixture with H₂O present and we should utilize some absorbent to eliminate water prior to CO₂ adsorption. In fact, since the adsorption energy of H₂O is twice as much as that of CO₂ on 1 e⁻ g-C₄N₃, utilization of 1 e⁻ g-C₄N₃ to eliminate water prior to CO₂ adsorption is one potentially viable approach. In this scanario, we could utilize g-C₄N₃ at lower voltage to eliminate water prior to a second stage of CO₂ adsorption at slightly higher voltage.

Discussion

In summary, we have shown that modification of the charge state of conductive g-C₄N₃ nanosheets provides an experimentally feasible approach for electrocatalytically switchable CO₂ capture. Compared with other CO₂ capture approaches, the process of CO₂ capture/release occurs spontaneously without any energy barriers once extra electrons are introduced or removed and these processes can be simply controlled and reversed by switching on/off the charging voltage. In addition, these negatively charged g-C₄N₃ nanosheets are highly selective for separating CO₂ from mixtures with CH₄, H₂ and/or N₂.

Good electrical conductivity and high electron mobility of the sorbent materials are prerequisite for electrocatalytically switchable CO₂ capture. The aim of the present paper is to explore conductive and easily synthetic sorbent material as an experimentally feasible adsorbent for electrocatalytically switchable CO₂ capture. These predictions may prove to be instrumental in searching for a new class of high-capacity CO₂ capture materials with ideal thermodynamics and reversibility and we hope that this work will stimulate further theoretical and experimental research in this direction.

Methods

Our DFT calculations employed the linear combination of atomic orbital and spin-unrestricted method implemented in Dmol³ package³⁸. The generalized gradient approximation (GGA) in Perdew-Burke-Ernzerhof (PBE) functional form³⁹ together with an all-electron double numerical basis set with polarization function (DNP) were adopted. Since the standard PBE functional is incapable of giving an accurate description of weak interactions, we adopted a DFT+D (D stands for dispersion) approach with the Grimme’s vdW correction in our computations⁴⁰. The real-space global cutoff radius was set to be 4.1 Å.

To study CO₂ capture/release on g-C₄N₃ nanosheets, we employed 2 × 2 supercell with periodic boundary conditions in the x-y plane (Fig. 1(a)). The vacuum space was set to larger than 20 Å in the z direction to avoid interactions between periodic images. In geometry optimizations, all the atomic coordinates were fully relaxed up to the residual atomic forces smaller than 0.001 Ha/Å and the total energy was converged to 10⁻⁵ Ha. The Brillouin zone integration was performed on a (6 × 6 × 1) Monkhorst-Pack k-point mesh⁴¹.

In order to investigate the gas adsorption on adsorbent, we defined the adsorption energy E_ads of CO₂, CH₄, H₂ and N₂ molecules on g-C₄N₃ as follows

where , , and are the total energy of isolated g-C₄N₃ nanosheets, isolated gas molecule, g-C₄N₃ with the adsorbed gas and number of gas molecules adsorbed on g-C₄N₃. According to this definition, a more positive adsorption energy indicates a stronger binding of the gas molecule to g-C₄N₃. The electron distribution and transfer mechanism are determined using the Mulliken method⁴².