Synergetic hydrogen-bond network of functionalized graphene and cations for enhanced atmospheric water capture

Significance

This study uncovers a remarkable synergy of hydrogen-bond networks at the solid–liquid interface, showing how calcium ions on graphene oxide significantly increase water capture. We experimentally observe that the water uptake per atom of a combined cation-oxygen functional groups system is significantly larger than the sum of the individual parts. Via the experimental approach of Van’t Hoff estimation and density functional theory (DFT), we demonstrate the strengthening of hydrogen bonds due to interfacial polarization. We present a mechanism for understanding the water hydrogen-bond network, giving rise to the atmospheric water capture process using GO-based materials.

Abstract

Water molecules at the solid–liquid interface display intricate behaviors sensitive to small changes. The presence of different interfacial components, such as cations or functional groups, shapes the physical and chemical properties of the hydrogen-bond network. Understanding such interfacial hydrogen-bond networks is essential for a large range of applications and scientific questions. To probe the interfacial hydrogen-bond network, atmospheric water capture is a powerful tool. Here, we experimentally observe that a calcium ion on a calcium-intercalated graphene oxide aerogel (Ca-GOA) surface captures 3.2 times more water molecules than in its freestanding state. From experimental Van’t Hoff estimation and density functional theory (DFT) calculations, we uncover the synergistically enhanced hydrogen-bond network of the calcium ion–epoxide complex due to significantly larger polarizations and hydrogen bond enthalpies. This study reveals valuable insights into the interfacial water hydrogen-bond network on functionalized carbon–cation complexed surfaces and potential pathways for future atmospheric water generation technologies.

Under humid or aqueous conditions, interfacial water molecules are ubiquitous in nature and technology. The solid–liquid interface is vital in countless physical, chemical, or biological processes and applications (1–4). One important factor to understand the structural properties of interfacial water is the hydrogen-bond network at the interface, which is notoriously difficult to probe (1, 3). Recent studies revealed an exciting way to study the interfacial hydrogen network within MOFs (5, 6) and graphene capillaries (7) via atmospheric water capture (AWC) (8–10). It is now appealing to further broaden the use of this methodology. First, extend AWC beyond the specific case of MOFs or the perfect graphene plane as the solid adsorbent. These materials consider special cases which have few technological and natural analogues. In contrast, functionalized carbon in humid or aqueous environments is ubiquitous and thus relevant in many systems. For example, the hydrophilic groups in lipid bilayers or DNA strands (11, 12) as well as in hydrophilic polymer membranes (13, 14), graphene oxide (GO)-based membranes and aerogels (15–22), or single-atom catalysts (23, 24) all obtain carbon–oxygen interfaces with water.

All these examples exist in aqueous environments that contain cations. However, there is lack of studies looking at the carbon–oxygen interface with water in the presence of cations despite its enormous fundamental and technological relevance. Hence, conceiving an AWC study that can investigate the interfacial hydrogen network in functionalized carbon under the presence of cations is highly desirable. Here, we solve this issue by utilizing functionalized graphene (graphene oxide) as a representative of functionalized carbon. Graphene oxide (GO) has shown great potential for fast vapor transport, and it can be readily intercalated with cations, which allows probing the interfacial hydrogen network of functionalized carbon in the presence of cations via AWC (25).

Graphene oxide is composed of a graphene plane along with various oxygen functionalities (26–31). The presence of functionalized areas gives rise to stable dipole sites and the development of a hydrogen-bond network with water molecules (15, 16). Water transport via GO-based materials has been widely studied in experiments and simulations to better understand its surface interaction with water molecules (32–39). Yet, direct experimental evidence is still limited for progressive understanding of the hydrogen-bond network on the GO surface. The most recent water transport study highlighted that the intercalated cation on the GO surface causes controllable friction with water molecules via hydrogen bonding (16). The observations suggest that intercalated cations may enrich the hydrogen-bond network of water molecules on GO surface. With that, cation intercalated GO offers a unique platform to investigate the hydrogen network of functionalized carbon in a cationic aqueous solution, which is highly relevant in many fields.

In this study, we reveal that the interaction of cation and oxygen functionalized carbon induces synergistic enhancement of the interfacial hydrogen-bond network in a moist environment. Here, we choose calcium ions due to their outstanding atmospheric water capture ability (40). This allows us to sensitively detect changes in the hydrogen-bond network in GO due to the presence of the cations. We synthesized calcium-intercalated GO (Ca-GO) in aerogel form and measured the atmospheric water capture capability of the as-prepared material. Surprisingly, we experimentally observed that calcium-intercalated GO aerogel (Ca-GOA) presents significantly different water uptake ability than original GO and CaCl₂. Via further in-depth experimental analysis and computational simulations, this study revealed an enhanced water hydrogen-bond network, which is governed by a synergistic enhancement between oxygen functionalities of GO and hydrated cations. We found that this strong hydrogen-bond network gives rise to AWC technology using GO-based materials.

Results and Discussion

Based on our aim to investigate the functionalized carbon–cation interfacial water hydrogen-bond network, we prepared GO/CaCl₂ aerogel via the solution intercalation method (41–43), which is schematically shown in SI Appendix, Fig. S1A (SI Appendix, Supplementary Note 1) and described in detail in the Methods section. The solution intercalation method allows the metal ions to sufficiently interact with GO nanosheets, thus achieving a more uniform distribution of intercalation. The as-synthesized samples are marked as GOA for graphene oxide aerogel and Ca-GOA for Ca²⁺ intercalated GO aerogel, respectively. Exemplary images of GOA and Ca-GOA are displayed in SI Appendix, Fig. S1 B–D.

We characterized the structure and chemical properties of the samples before and after intercalation. Scanning electron microscopy (SEM) with energy dispersive spectroscopy (EDS) images are shown in Fig. 1 A–C and SI Appendix, Fig. S2 B and C (SI Appendix, Supplementary Note 2), illustrating the sample morphology.

Fig. 1.

Characterization of Ca-GOA. (A) Scanning electron microscopy (SEM) image of the overall structure of Ca-GOA. (B) SEM image of Ca-GOA showing typical wrinkles. (C) SEM image of Ca-GOA with energy dispersive spectroscopy (EDS) showing elemental images of carbon, oxygen, and calcium. (D) X-ray photoelectron spectrometry (XPS) survey of GO and Ca-GOA showing C1s, O1s, and Ca2p3 peak with carbon/oxygen (C/O) ratio. (E) XPS spectra (black curve) showing C1s peak of GOA and Ca-GOA sample with peak fitting of C=C/C–C bond at ~284.5 eV (blue and brown curve), C–O bond (red curve) at ~286 eV, C=O bond (purple curve) at ~288 eV and O–C=O (pink curve) at ~289 eV. (F) Atomic percentage of elemental composition including calcium, oxygen, and carbon of GO and Ca-GOA samples. (G) X-ray diffraction (XRD) pattern of Ca-GOA. The inset diagram in (G) illustrates the XRD pattern of CaCl₂. (H) XRD pattern of (001) peak of GOA and Ca-GOA.

The EDS analysis confirms the presence of carbon, oxygen, and calcium as the main elemental composition of Ca-GOA and verifies the homogeneous distribution of calcium ions intercalation on the GO flakes. The synthesized Ca-GOA samples display a porous structure similar to the GOA sample with a typical wrinkled large surface area. We conducted the Brunauer–Emmett–Teller (BET) nitrogen adsorption analysis for porosity distribution and surface area analysis of aerogel samples. GOA and Ca-GOA present similar trends for nitrogen adsorption and pore width distribution, which indicates that the intercalation of calcium ions has limited impact on the porous structure (SI Appendix, Supplementary Note 7). X-ray photoelectron spectrometry (XPS) reveals that the carbon to oxygen ratio (C/O) of GO remains constant at ~2 before and after intercalation (Fig. 1D). C1s XPS spectra further suggest that Ca-GOA (Fig. 1E) have similar carbon–carbon and carbon–oxygen bond composition compared to GO samples (SI Appendix, Fig. S3 A–D in Supplementary Note 3). After intercalation, the atomic percentage of C-O bonds decreased from 33.7 to 20.6%, C–C/C=C bonds increased from 27.6 to 31.8% and C=O bonds vary from 2.9 to 4.3%. The atomic percentage of each element from the XPS survey for both GO and Ca-GOA are shown in Fig. 1F. Around 5.1% of calcium atoms are detected on the Ca-GOA surface, while in the original GO, no calcium was detected. With XPS, we further confirm the homogeneous intercalation of Ca ions into the Ca-GOA samples (SI Appendix, Supplementary Note 5), which agrees with the EDS analysis. It is noted that a small number of chlorine and sulfur are also detected on Ca-GOA, however, are not the main concern in this study. X-ray diffraction (XRD) patterns shown in Fig. 1 G and H illustrate the d-spacing change of Ca-GOA after intercalation (16) (for GOA see SI Appendix, Supplementary Note 6), and the absence of CaCl₂ peaks in the Ca-GOA samples suggest that the intercalated Ca is in the form of ions rather than CaCl₂.

To investigate the hydrogen-bond network in the Ca-GOA, we first performed AWC measurements including cycling performance test (SI Appendix, Supplementary Note 4) as described in the Methods section and compared the results to literature values from our own studies and others (40, 44). Fig. 2 A–C shows the recorded and reported adsorption isotherms. The water uptake (m/m₀) of Ca-GOA was recorded with varying relative humidity (RH%) at a constant room temperature (298 K). The relative humidity is then converted into the partial pressure of water vapor to visualize the isotherm. The results show that AWC ability of Ca-GOA samples is significantly improved to around 2.1 g/g from pure GO samples (GOM and GOA) with around 0.6 g/g, which indicates the enriched hydrogen network on the Ca-GOA surface. Ca-GOA shows the most significant improvement in water uptake compared to other metal ions intercalated GOA (SI Appendix, Supplementary Note 16). It should also be noted that the aerogel samples maintained a solid structure during the whole AWC measurement, confirming that the adsorption process mainly occurs on the surface of the material. This is different from the CaCl₂ which follows the solution adsorption mechanism(45). Moreover, pure GOM and GOA are observed to have similar AWC abilities (SI Appendix, Supplementary Note 2), indicating that the studied hydrogen-bond network is independent from the materials morphology.

Fig. 2.

Adsorption behaviors of CaCl_2, GO, and Ca-GOA. (A–C) Water uptake of CaCl₂ (A), GO (B), and Ca-GOA (C) under different ambient water partial pressure (kPa). The error bar indicates the SD. Black dashed lines illustrate the fractal BET fitting of CaCl₂, GO, and Ca-GOA. (D–F) Freundlich linear fitting (black dashed line) of CaCl₂ (D), GO (E), and Ca-GOA (F). The inset diagram in (F) indicates the heterogeneous adsorption behaviors of Freundlich model. The red balls represent the adsorption sites (calcium ions). The blue balls represent the adsorbed water molecules, which are closely attached to the adsorption sites. R²represents the coefficient of determination of the fitting.

Further analysis of the adsorption behaviors allows us to get a deeper understanding of interfacial water molecules during the AWC. All three different materials, CaCl₂ (40), GO (44), and Ca-GOA, fit well under the BET model (46, 47). However, the fitted BET equations are notably different from each other (see SI Appendix, Supplementary Note 8 for fitting details). The degree of the polynomial function of the BET equation represents the number of adsorbed water molecule layers (n) on the surface of the material (40, 47, 48). Here, n = 7 for CaCl₂, n = 4 for GO, and n = 6 for Ca-GOA. Hence, the intercalation of Ca into Ca-GOA results in a stronger hydrogen network on the surface compared to the pure GO. In other words, the intercalation of cation enhances the interfacial water hydrogen network at the oxidized carbon surface; however, the mechanism of such enhancement is still unclear.

Based on the BET model fitting analysis, we established that the Ca-intercalation offers strong adsorption sites for water molecules on the Ca-GOA surface. If so, the adsorption isotherm should follow the Freundlich model, as it represents a heterogeneous surface adsorption process (49). In particular, the Freundlich model describes a material with special adsorbing sites which are heterogeneously distributed on the adsorbent surface (49–51). In this case, the Ca ions are the strong water-attracting adsorbing sites, well-dispersed on the GO plane, allowing coverage of the whole surface with several layers of water molecules (52). As shown in Fig. 2 D–F, the water adsorption isotherm of Ca-GOA can be fitted in high agreement with the typical Freundlich isotherm model, while being less in agreement with GO and CaCl₂. The detailed information for Freundlich model fitting is shown in SI Appendix, Supplementary Note 9.

To understand the heterogeneous adsorption phenomenon, we observed the water adsorption of Ca-GOA which contains varying amounts of calcium ions. As shown in Fig. 3A, the water adsorption exhibits a nonlinear increasing trend, correlating with the number of intercalated calcium ions. The results indicate that the strong attraction to water molecules not solely depends on the quantities of calcium ions which were intercalated to the oxidized carbon surface. Moreover, there could be a synergetic interaction between the two components, which is vital to result in the enhanced hydrogen-bond network. To understand the mechanisms of this phenomenon, we further studied the samples with the highest intercalation ratio as it provides the best uniformity of ion interaction and the most significance of water uptake. Keeping in mind that water molecules follow the multilayered adsorption model. We analyzed the thermodynamic equilibrium of water adsorption from an energy perspective to probe this synergetic interaction in the hydrogen-bond network. To achieve this, we measured the water adsorption and desorption isotherm for GOA and Ca-GOA in a small quantity of samples (in milligrams, represented as Ca-GOA-S) under various temperatures. The results are shown in Fig. 3B and SI Appendix, Fig. S9. For clarity, Ca-GOA-S represents the small-scale Ca-GOA samples measured using precise adsorption analyzer (SI Appendix, Supplementary Note 10). Via the Van’t Hoff equation, these measurements allow us to link the pressure and temperature, further, to estimate the water adsorption enthalpy (ΔH _{ads, m}), entropy (ΔS _{ads, m}), and Gibbs free energy (ΔG _{ads, m}) for GOA and Ca-GOA (6, 53, 54). The results are as shown in Fig. 3 C–E and SI Appendix, Fig. S11 C–F.

Fig. 3.

Adsorption isotherm of GOA and Ca-GOA. (A) Water adsorption performance of Ca-GOA with varying concentrations of intercalated calcium ions. (B) Comparison of water adsorption and desorption isotherm between GOA and Ca-GOA-S under different ambient water partial pressure (kPa) at 298 K. Ca-GOA-S represents the Ca-GOA samples measured in milligrams for precise analysis (SI Appendix, Supplementary Note 10). (C and D) Estimated adsorption enthalpy (ΔH _{ads, m}) of (C) GOA and (D) Ca-GOA using the Van’t Hoff method in correlation to water uptake. The error bar represents the SE of the linear regression at each loading increment. The ΔH_{GOA/Ca-GOA, sat} denotes the adsorption enthalpy at the saturated water uptake capacity of aerogel samples. (E) Adsorption Gibbs free energy (ΔG_{ads, m}) of Ca-GOA correlate to water uptake, estimated using the Van’t Hoff method in 298 K. ΔG_{Ca-GOA, sat} corresponds to the ΔG_{ads, m} of Ca-GOA at its saturated water uptake capacity. The green line indicates the estimated hydration Gibbs free energy of per mole calcium ions (ΔG_Ca²⁺) (55, 56), refer to SI Appendix, Supplementary Note 11. (F and G) Comparison of estimated water molecules per oxygen (F) and calcium atom (G) between GOM, GOA_, Ca-GOA, and Ca-GOA-S under different ambient water partial pressure (kPa). (ref.) and (exp.) denotes the data from referenced literature and our experimental observations, respectively. (H) Estimation of the highest number of water molecules per oxygen/calcium atom in CaCl₂, Ca-GOA-S, GOA, and GOM based on our experimental observation of water uptake.

We first examined the adsorption enthalpy differences between GOA and Ca-GOA. Adsorption enthalpy is a state function to reflect the average exothermic process of hydrogen bond formation on the entire surface (57). For GOA, we observe consistent enthalpy at around -45 kJ/mol (Fig. 3C). This indicates the formation of a stable hydrogen-bond network across various levels of water uptake. In contrast, Ca-GOA exhibits a different behavior: The initial adsorption enthalpy was approximately –43.9 kJ/mol however it decreased significantly to -61.7 kJ/mol upon reaching saturation (Fig. 3D). The notable change of enthalpy reflects the heterogeneous adsorption process, while the lower enthalpy suggests the formation of stronger hydrogen bonds on the Ca-GOA surface. This aligns closely to the heterogeneous adsorption behaviors of Ca-GOA, which follows the Freundlich isotherm model for water adsorption. More importantly, we noticed that the numeric difference of enthalpy between GOA and Ca-GOA is higher than the hydration enthalpy of intercalated calcium ions (56, 58, 59). This observation suggests the synergetic interaction between calcium ions and oxidized carbon surface. The methods and discussion of this estimation are illustrated in detail in SI Appendix, Supplementary Note 11.

We also analyzed the adsorption Gibbs free energy for GOA and Ca-GOA at the ambient temperature of 298 K. Similarly, ΔG_{ads, m}of GOA remains relatively stable for GOA, indicating the consistent thermodynamic equilibrium for the formation of a homogeneous hydrogen-bond network. In comparison, ΔG_{ads, m}of Ca-GOA decreases significantly with increasing water uptake, suggesting a more stable thermodynamic state of a stronger hydrogen-bond network on the Ca-GOA surface. Notably, upon saturation, each unit mole of Ca-GOA has a lower Gibbs free energy compared to the hydration of the same amount of pure calcium ions (SI Appendix, Supplementary Note 11). Furthermore, we observed the trend of water molecules per oxygen and calcium atom of GO, GOA, Ca-GOA, and CaCl₂ versus the water partial pressure, as shown in Fig. 3 E–G. The uptake of water molecules per oxygen atom on Ca-GOA is up to 8.7 times higher than that of GOM and GOA under all ambient environments. Surprisingly, the water uptake per calcium ion is up to 3.2 times higher in Ca-GOA compared to CaCl₂ at the highest water partial pressure (Fig. 3G).

Here, we can conclude that the water uptake per calcium and oxygen atom in Ca-GOA is much higher than in the individual materials, CaCl₂ and GO. Keeping in mind that the Ca-sites were identified as the main contributor to the water uptake in Ca-GOA, it is surprising to see that the water uptake per calcium is 3.2 times higher in the Ca-GOA compared to CaCl₂. Moreover, the results of Van’t Hoff estimated adsorption enthalpy strongly suggest that this enhancement is linked to an interplay between the functional groups of GO and the intercalated cations to form a synergistically enhanced hydrogen-bond network. To investigate this hypothesis, we further performed density functional theory (DFT) calculations.

We examine the hydrogen bond properties between oxygen functionalities and water molecules with and without the existence of calcium ions using DFT calculations. We select a graphene plane with a bare epoxide group as a simplified model of a GO nanoflake surface (Fig. 4A). This represents the dominant oxygen functionalities (epoxides and hydroxyls) found on the GO basal plane (28, 60). Since the basal plane constitutes the majority of the surface area available for interaction within the porous Ca-GOA aerogel structure used experimentally, we anticipate that the adsorption of calcium ions and subsequent water interactions will predominantly occur via these basal plane groups. Epoxide groups have been experimentally confirmed to have a strong hydrogen bonding interaction with water molecules (37, 61). The DFT calculations were performed at the PBE0-D3BJ/def2-QZVPP level of theory (62–64).

Fig. 4.

Graphene plane–based models for DFT calculations. (A) Single epoxide group. (B and C) Epoxide group with 1 (B) or 2 (C) water molecules. (D and E) GO oxygen coordinated to hydrated calcium ion with 1 (D) or 2 (E) water molecules. ΔH_{298, bind} represents hydrogen bond enthalpies. Dashed lines indicate hydrogen bond lengths (selected H-bond distances given in Å).

Fig. 4 B and C indicates the process of the epoxide hydrogen bonded to one and two water molecules. We obtain hydrogen bond distances and energies for typical moderate hydrogen bonds (65–67). Namely, both water molecules hydrogen bonded to the epoxide group with a distance range of 1.96 to 1.99 Å and enthalpy of –14.8 kJ/mol and –13.4 kJ/mol for the first and second water molecule, respectively. However, when the hydrated calcium cation is bound to the epoxide group, the hydrogen bonding network presents significantly different properties. We note that when the epoxy oxygen is coordinated to the Ca cation, the oxygen is bound to the GO surface via one C–O bond, as illustrated in Fig. 4 D and E. We note that the bonding situation in Fig. 4 D and E also represents the bonding of a hydrated Ca cation to a C–O functional group on the GO surface (68). Hereinafter, the oxygen connected to the GO surface with a single C–O bond will be referred to as the GO oxygen, rather than the epoxy oxygen.

With the presence of calcium cation, the GO oxygen is coordinated to the hydrated calcium cation. This dramatically enhances the hydrogen-bond network surrounding GO oxygen. Fig. 4D presents a scenario when one water molecule is hydrogen bonded to both the GO oxygen and the hydrated calcium ion. Compared with the system in Fig. 4C, this water molecule is now hydrogen bonded via three hydrogen bonds, one with the GO oxygen and two with the hydrating water molecules around the calcium ion. As shown in Fig. 4D, the hydrogen bond formed with the GO oxygen (1.546 Å) is significantly shorter than the hydrogen bonds formed with the water molecules hydrating the calcium cation (1.699 and 1.725 Å). Furthermore, this hydrogen bond is also significantly shorter than that formed between the water molecule and bare epoxide group presented in Fig. 4B. Similarly, in Fig. 4E, we further show the scenario when the GO oxygen is hydrogen bonded to a second water molecule. The calculation shows that the length of the hydrogen bond between the GO oxygen and the second water molecule is also largely shortened, from 1.982 Å (Fig. 4C) to 1.689 Å (Fig. 4E).

From the hydrogen bond enthalpy point of view, our calculations show that the hydrogen bonding between the GO oxygen and water molecules are highly reinforced by the hydrated calcium ion. In the scenario of the bare epoxide group on the GO surface, the hydrogen bond enthalpy between the oxygen and water molecules is around –14 kJ/mol, as mentioned above. However, in the presence of the hydrated calcium ion, the hydrogen bond enthalpy between the GO and water molecules in the system shown in Fig. 4D increases to as much –66.7 kJ/mol and in Fig. 4E comes to –52.9 kJ/mol. Thus, the H-bond enthalpy in the presence of the hydrated calcium ion is about 3 to 4 times higher than that of a bare system. Both the increased hydrogen bond enthalpy and shorter hydrogen bond lengths are attributed to both the larger hydrogen-bond network and stronger hydrogen bonding acceptor strength of the calcium ion decorated GO oxygen.

We further investigate the hydrogen bonding acceptor properties of the GO oxygen based on the atomic polar tensor (APT) charges (69, 70). The results are shown in Table 1.

Table 1.

Atomic polar tensor (APT) charge (q) in a.u. involved in the hydrogen bonds and on the Ca atom for the systems in Fig. 4

Model	q (Owater)	q (Hwater)	q (Oepoxy)	q (Ca)
4A	N/A*	N/A	–0.40	N/A
4B	–0.49	+0.27	–0.47	N/A
4C	–0.48, –0.45	+0.26, +0.23	–0.51	N/A
4D	–0.63	+0.37	–0.84	+1.34
4E	–0.61, –0.60	+0.40, +0.35	–0.84	+1.25

^*N/A corresponds to the absence of atomic polar tensor charge.

We obtain the following APT charges of the epoxide oxygen, –0.40 e, –0.47 e, and –0.51 e in the system shown in Fig. 4 A–C, respectively. As expected, the negative charge on oxygen increases with the number of hydrogen bonds involved. The charge on the hydrogen and oxygen atoms of the water molecules remains relatively constant for systems as shown in Fig. 4 B and C. Namely, they range between +0.23 e and +0.27 e for the hydrogen and between –0.45 e and –0.49 e for the oxygen. Coordination of the hydrated calcium cation to the GO oxygen results in a dramatic change in the oxygen charge. In particular, the coordination of the hydrated calcium cation increases the negative charge on the GO oxygen from –0.40 e to as much as –0.86 e. This significant increase in negative charge on the GO oxygen makes it a much stronger hydrogen bond acceptor. Accordingly, the H-bond distance with the water molecule is shortened from 1.965 Å (Fig. 4B) to 1.546 Å (Fig. 4D). We note that the later hydrogen bond distance represents an exceptionally short hydrogen bond for a HOH•••O system (i.e., a water molecule coordinated to an oxygen atom) (65, 66). It also reveals that coordination of calcium ion to the GO oxygen alters the atomic charges on the hydrogen and oxygen of the hydrogen-bonded water molecules. For example, for the systems depicted in Fig. 4 B and D, the positive charge on the hydrogen increases from +0.27 e to +0.37 e, and the negative charge on the oxygen increases from –0.49 e to –0.63 e (Table 1). Remarkably, this demonstrates significant medium-range effects of the hydrated calcium cation on the charge of an oxygen center to which it is bound via a hydrogen-bond network (i.e., not covalently bound). The above results indicate that the hydrogen bonding ability of one epoxide group on the GO surface is enhanced by the coordination of the hydrated calcium ion. Such enhancement was shown via the increasing hydrogen bond acceptor strength of GO oxygen and additional hydrogen bonding interactions with the first hydration sphere of calcium cation. It is evident, both theoretically and experimentally, that epoxide groups on a GO surface tend to cluster in islands rather than be uniformly distributed across the surface (30, 71). This leads to a natural question, whether this single hydrated calcium cation can interact with more than one epoxide groups on the GO surface.

We further investigate the system with two epoxides on a graphene plane coordinated to a single hydrated calcium ion (SI Appendix, Supplementary Note 13). Each of the two oxygens on the GO surface is able to hydrogen bond to two water molecules. For the first three water molecules, we obtain binding enthalpies that are similar to those obtained from the functionalized systems in Fig. 4, namely 70.2, 58.5, and 63.6 kJ/mol, respectively. For the fourth water molecule, we obtain a lower binding enthalpy of 29.9 kJ/mol; this reduction is partly attributed to two (rather than three) hydrogen bonds in which this water molecule is involved in (SI Appendix, Supplementary Note 13). Importantly, all these binding enthalpies are significantly larger than those obtained from the systems with the absence of hydrated calcium cation (Fig. 4 B and C).

With the results above, we can now further optimize the hydrogen bond enthalpies from the perspective of the experimental conditions. This is because the calcium intercalated GO surface in the experimental settings is expected to be intermediate between the solid state and a bulk aqueous solution. It is well established that hydrogen bond strengths are influenced by the effect of the solvent. In particular, the H-bond strength with the GO surface decreases with the polarity of the medium in the order of solid-state > nonpolar solvents > polar solvents. Thus, the calculated hydrogen bond enthalpies above, which do not include solvent corrections, are expected to represent the upper limits for the experimental setting. Therefore, it is instructive to calculate the hydrogen bond enthalpies in bulk aqueous solution to obtain lower limits for the hydrogen-bond enthalpies. For this purpose, we use the conductor-like polarizable continuum model (CPCM) (72), which has been found to provide good performance for aqueous solution (73, 74). The solvation corrections reduce the hydrogen bond enthalpies for the unfunctionalized GO models by a factor of ~2.5, whereas they reduce the bond enthalpies for the Ca-functionalized GO models by a factor of ~1.5. As shown in Table 2, the inclusion of the solvation corrections widens the gap between the hydrogen bond enthalpies of the GO models with and without coordination of the calcium ion. Considering that the GO surface in the experimental settings is expected to be intermediate between the solid state and a bulk aqueous solution, this is strong evidence to explain our experimental observation.

Table 2.

Comparison between the hydrogen binding enthalpies at 298 K (∆H_{298, bind}, in kJ/mol) for the systems in Fig. 4 obtained in the solid state and in bulk aqueous solution

	∆H298,bind (kJ/mol)
Model	Solid state	Aqueous solution
4A	N/A*	N/A
4B	–14.8	–6.8
4C	–13.4	–5.3
4D	–66.7	–42.7
4E	–52.9	–32.9

^*N/A corresponds to the absence of hydrogen binding enthalpy.

Conclusion

Here, we uncover the synergistic hydrogen-bond network of functionalized carbon in the presence of a hydrated cation. Our experimental and computational results both show a strong increase in hydrogen bond strength in the system of an epoxy functional group in close range to a hydrated calcium ion on graphene plane. Via AWC experiments, we observed that the water uptake per calcium is dramatically increased, by up to a factor of 3.2 times higher in Ca-GOA compared to pure CaCl₂. Similarly, the water uptake per oxygen of GOA is increased by a factor of 8.7 after intercalation of Ca ions. Further in-depth thermodynamic analysis reveals that Ca-GOA has a significantly lower adsorption enthalpy and Gibbs free energy compared to GOA, indicating a strongly enhanced hydrogen-bond network on the surface.

Via extensive DFT calculations, we uncover that the system of hydrated calcium ion and epoxy functional group on graphene planes synergistically enhances the overall binding strength of the hydrogen-bond network. Particularly, the calcium ion increases the charge polarization of the oxygen and hydrogen atoms of the C-O bond and water molecules. This leads to higher enthalpies and shorter hydrogen bond lengths, explaining the experimentally observed enhancement of water uptake per calcium ion.

This study holds significance for numerous systems where hydrated ions are in proximity to carbon-based functional groups. As these are omnipresent in nature and technology, our study may help to bring a unique perspective to a wide range of natural phenomena and technology applications that involve our described model system. Remarkably, we show that the AWC ability of GO can be enhanced to similar levels as pure CaCl₂ in terms of water uptake per gram. This may open exciting opportunities to utilize Ca-GOA as a powerful desiccant in atmospheric water generation and energy-efficient dehumidification.

Materials and Methods

Materials.

The GO solution (15 mg/ml) was prepared via Hummer’s method and was supplied by NiSiNa Materials Japan. Calcium chloride (CaCl₂) powder was purchased from Sigma-Aldrich.

Synthesis of Aerogels.

The Ca-GO solution was prepared by mixing predetermined volume of CaCl₂ salt solution and GO solution, followed by magnetic stirring for 30 mins at 1,000 rpm. The as-prepared solution was then freeze-dried using a vacuum freeze drier at –60 °C to synthesize Ca-GOAs. The GOA samples were prepared by freeze-drying the GO solution without further modification. All prepared aerogel samples were stored in vacuum condition until experimental measurements. Synthesized samples are shown in SI Appendix, Supplementary Note 1 (SI Appendix, Fig. S1 B–H).

Characterization of Aerogels.

The X-ray photoelectron spectrometer (XPS, Thermo Scientific, UK ESCALAB250i) with monochromatic Al K alpha (energy 1486.68 eV) X-ray source was conducted to investigate the chemical composition of the synthesized aerogels. The C1s peak (284.5 eV) of graphite was used as a reference. The interlayer space of synthesized aerogels was analyzed using the XRD (PANalytical Empyrean IV) at 45 kV 40 mA with Cu Kα radiation (λ = 0.154 nm). The surface morphologies of the aerogels were visualized using a field-emission scanning electron microscope (FEI Nova NanoSEM 230 FE-SEM) with an energy dispersive spectroscopy (EDS) detector (Bruker SDD-EDS) for mapping the qualitative elemental composition. The material porosity distribution was analyzed using a BET-based porosity analyzer (TriStar II Plus, Micromeritics).

Atmospheric Water Capture Tests.

The large-scale atmospheric water capture (AWC) tests of GOAs and Ca-GOAs were carried out in a custom-designed humidity-controlling system (SI Appendix, Supplementary Note 14). The weight changes of the aerogels were continuously recorded every 10 s using a computer-controlled mass balance during the AWC process. The humid environment was maintained stable overnight ($\mathrm{R}\mathrm{H}\pm 3\%$) before AWC tests. Each experiment was repeated more than 2 times to study the SD. The adsorption isotherm studies were carried out using an adsorption analyzer (BELSORP-MAX II, Microtrac) using small-scale aerogel samples. The cycle performance of aerogels was conducted in a humidity chamber maintaining constant humidity of RH 90 ± 2%). After each cycle, the aerogels were vacuum-dried to the initial weight. The weight of aerogels was recorded before and after AWC for 20 cycles.

Computational Details.

DFT calculations were performed to gain further insights into the experimental findings using the Gaussian 16 software (75). All geometries were fully optimized using the PBE0 exchange-correlation functional (62) in conjunction with def2-SVP basis set (63) (SI Appendix, Supplementary Note 15). Empirical dispersion corrections (64) are included using the Becke–Johnson potential (denoted by the suffix D3BJ) (76). Zero-point vibrational energy and enthalpic temperature corrections have been obtained from these calculations. The equilibrium structures were verified to have all real harmonic frequencies, confirming they are local minima on the potential-energy surface. The final electronic energies were refined using the PBE0-D3BJ functional in conjunction with the much larger quadruple-z def2-QZVPP basis set (63). Bulk solvent effects in a queous solution were also considered using the conductor-like polarizable continuum model (CPCM) continuum solvation model (72).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.