Organic Functionalized Graphene Oxide Behavior in Water

Surface modified graphene oxide (GO) has received broad interest as a potential platform material for sensors, membranes, and sorbents, among other environmental applications. However, compared to parent (unmodified) GO, there is a dearth of information regarding the behavior of subsequently (secondary) modified GO, other than bulk natural organic matter (NOM) coating(s). Here, we systematically explore the critical role of organic functionalization with respect to GO stability in water. Specifically, we synthesized a matrix of GO-based materials considering a carefully chosen range of bound organic molecules (hydrophobic coatings: propylamine, tert-octylamine, and 1-adamantylamine; hydrophilic coatings: 3-amino-1-propanol and 3-amino-1-adamantanol), so that chemical structures and functional groups could be directly compared. GO (without organic functionalization) with varying oxidation extent(s) was also included for comparison. The material matrix was evaluated for aqueous stability by comparing critical coagulation concentration (CCC) as a function of varied ionic strength and type (NaCl, CaCl2, MgCl2, and MgSO4) at pH 7.0. Without surface derivatization (i.e., pristine GO), increased stability was observed with an increase in the GO oxidation state, which is supported by plate–plate Derjaguin, Landau, Verwey and Overbeek (DLVO) energy interaction analyses. For derivatized GO, we observed that hydrophilic additions (phi-GO) are relatively more stable than hydrophobic organic coated GO (pho-GO). We further explored this by altering a single OH group in the adamantane-x structure (3-amino-1-adamantanol vs. 1-adamantylamine). As expected, Ca2+ and monovalent co-ions play an important role in the aggregation of highly oxidized GO (HGO) and phi-GO, while the effects of divalent cations and co-ions were less significant for pho-GO. Taken together, this work provides new insight into the intricate dynamics of GO-based material stability in water as it relates to surface functionalization (surface energies) and ionic conditions including type of co- and counter-ion, valence, and concentration.


Synthesis of Graphene Oxide (GO)
GO was synthesized by the modified Hummers method [26,27] using graphite (1 g), KMnO 4 (3 g), and H 2 SO 4 (25 mL). To control the oxidation extent of GO, the mixtures were separately heated at 30 • C and 50 • C for 2 h [23]. After heat reaction, the mixture was cooled in an ice bath. The resulting mixture was diluted by 175 mL of deionized (DI) water (>18.2 MΩ cm resistivity, Milli-Q, Millipore Corp., Burlington, MA, USA). Then, the mixture was reheated at 60 • C for 30 min. After heating, we added H 2 O 2 to remove the residual permanganate. The resulting suspension was purified with DI water until the pH of washed solution became neutral. The purified suspension (graphite oxide power) was filtrated and dried at 60 • C for 24 h. The GO suspension was obtained by the probe-sonication of graphite oxide power (150 mg) in DI water (200 mL) for 2 h. After sonication, the suspension was centrifuged at 12,000 rpm for 2 h. The resulting supernatant (GO) was kept in a glass bottle covered with aluminum foil.

Functionalization of GO
The surface of GO was organically functionalized using propylamine, tert-octylamine, 1-adamantylamine, 3-amino-1-propanol, and 3-amino-1-adamantanol, respectively. We added 300 mL of GO suspension (500 mg/L) and 0.1 g of coating agent into a three neck flask and heated the solution at 60 • C for 4 h. After the reaction, the resulting solution was purified with DI water over six times using stirred cell (100K Dalton, Millipore, Burlington, MA, USA) at 1 bar (N 2 gas).

Critical Coagulation Concentration (CCC)
The CCC of the GO was measured by dynamic light scattering (DLS, NanoBrook Omni, Brookhaven Instruments Co., Holtsville, NY, USA) at 22 • C using NaCl, CaCl 2 , MgCl 2 , and MgSO 4 . Detailed information for measuring CCC has been described in our previous research and by others [28][29][30].

Hydrodynamic Diameter and Zeta Potential
The hydrodynamic diameter and zeta potential of GO materials were measured by DLS (NanoBrook Omni, Brookhaven Instruments Co., Holtsville, NY, USA) at 22 • C in 1 mM NaCl as background condition.

X-Ray Photoelectron Spectroscopy (XPS)
An XPS (PHI VersaProbe II, Physical Electronics Inc., Chanhassen, MN, USA) was used with a monochromatic Al Kα X-ray source with a 0.47 eV system resolution.

X-Ray Diffraction (XRD)
XRD patterns were obtained using a Rigaku SmartLab X-ray Diffractometer (Rigaku, Tokyo, JP) with a scan step of 10 • min −1 operating at 40 kV and 30 mA using Cu-Kα radiation.

The Plate-Plate Geometry DLVO Interactions
Classical Derjaguin, Landau, Verwey and Overbeek (DLVO) theory was used to calculate the interaction energy between the GO. Interaction energy (per area) profiles between GO and GO were interpreted upon the plate-plate geometry. Equations (1) and (2) for calculating electrostatic (V el ) and van der Waals (V vdw ) potential energy are as follows, respectively [31,32]: where σ is the surface charge density, ε is the dielectric constant of water, ε 0 is the permittivity of vacuum, κ is the inverse of the Debye length, d is the interaction distance between GO, A H is the Hamaker constants (GO−water−GO system), which were 4.9 × 10 −20 and 13.5 × 10 −20 J for GO and highly oxidized GO (HGO), respectively [4], and t is the thickness of GO, which was 6.7 Å [33]. The Debye length is given by where k B is the Boltzmann constant, T is the absolute temperature, N A is the Avogadro number, e is the elementary charge, and I is the ionic strength. The surface charge density is given by where ψ is the surface potential, which is calculated by the following equation: where d Stern is the thickness of the stern layer (0.7 Å) [4], and ξ is the zeta potential.

Schulze-Hardy Rule (CCC Relationship for Sphere-Sphere Interaction)
Critical coagulation concentration (CCC) relationships in terms of counter valence ion concentration are described by the Schulze-Hardy rule, which is derived from linear superposition of Gouy-Chapman and unretarded Hamaker expressions (Equations (6) and (7)) [14,16]: where a is diameters of particles, n ∞ is the bulk number density of ions, and γ is the reduced surface potential, which is given by where z is the valance. At the CCC, the total energy interaction (V T = V el + V vdw ) and the gradient dV T /dd are zero. In this point, κd should be 1. Therefore, the bulk number density of ions can be expressed by Simply, CCC is proportional to the following equation:

CCC Relationship for Plate-Plate Interaction
We derived the CCC relationship with respect to the valance of counter ions by classical plate-plate geometry DLVO interaction (Equations (1) and (2)). The total energy interaction and its gradient dV T /dd are zero at the CCC (2κ = d). Here, the surface potential is calculated by the following equation, which is valid at a low surface potential (zeψ < 2k B T): Calculated bulk number density of ions can be expressed by CCC is proportional to the following equation:

Synthesis and Characterization of GO Materials
GO and highly oxidized GO (HGO) were synthesized by the modified Hummers method [26,27], whereby the extent of oxidation was controlled by the reaction temperature of the second stage (detailed information is described in Materials and Methods) [23]. The synthesized GO was surface functionalized with a series of organic molecules, including propylamine, tert-octylamine, 1-adamantylamine, 3-amino-1-propanol, and 3-amino-1-adamantanol (termed here as propylamine GO, tert-octylamine GO, 1-adamantylamine GO, 3-amino-1-propanol GO, and 3-amino-1-adamantanol GO, respectively). All the synthesized GO materials were well dispersed in water. Synthesized GO materials were characterized using XPS, FTIR, XRD, and DLS. As shown in the XPS measurement ( Figure 1 and Table 1), the C1S spectra of the GO materials is deconvoluted into five types of carbon groups, represented as C-C, C-O, C-O-C, C=O, and O-C=O with binding energies centered at 284.6, 285.6, 286.7, 288.2, and 289.4 eV, respectively, as described by others [34]. As presented in Figure 1a, HGO has a higher percentage of oxidized carbon when compared with GO. The relative portion of oxidized carbon was 58.25% and 54.58% for HGO and GO, respectively. XPS survey indicated that the C/O atomic ratio of HGO and GO was 1.99 and 2.37, respectively. Upon organic functionalization, we observed an increase in C-O/C-N bonds as the amine groups were used as a linker for functionalization (in all cases). FTIR spectra in Figure 2a identifies the successful organic functionalization of GO, in which a amine peak (1600 cm −1 ) appears upon the functionalization for all materials [35]. We also observed decrease in % transmittance of OH stretch as the amine groups formed covalent amide bonds with the carboxyl or hydroxyl group of GO, which is in good agreement with the XPS results [36]. As shown in Figure 2b, XRD peak shifts were observed with organic functionalization. Diffraction peak at 10.1 • (0.87 nm of d-spacing) of a pristine GO was shift to 9.1 • (0.97 nm), 8.8 • (1.00 nm), and 8.7 • (1.01 nm) for propylamine GO, 3-amino-1-propanol GO, and tert-octylamine GO, respectively. As reported by others, functionalization increases interstitial spacing, thus the thickness of GO materials [37]. With respect to the adamantane groups, 2θ response of 1-adamantylamine GO and 3-amino-1-adamantanol GO disappeared, suggesting that bulk d-spacing(s) were irregular and/or uneven, which is likely due to the steric effect(s) of adamantane groups.       Lateral nanosheet size (L) and the surface zeta potential of the GO materials were measured using DLS at pH 7.0. Though the GO materials are not spherical particles, lateral nanosheet size (L) of GO materials are empirically estimated by DLS; L = (0.07 ± 0.03) × DH 1.5 ± 0.15 , where DH is the hydrodynamic diameter [38]. L of GO and HGO was 85.0 ± 4.0 and 83.4 ± 8.8 nm, respectively. L of organic functionalized GO was slightly increased; 98.3 ± 8.1, 89.3 ± 2.9, 89.4 ± 1.6, 88.0 ± 6.4, and 87.6 ± 5.2 nm for propylamine GO, tert-octylamine GO, 1-adamantylamine GO, 3-amino-1-propanol GO, and 3-amino-1-adamantanol GO, respectively. Similar material sizes indicate that GO functionalization occurs without aggregation and the resulting products are relatively stable. Zeta potential of GO and HGO was −33.1 ± 1.9 and −43.3 ± 1.7 mV, respectively. HGO showed large negative zeta potential compared with GO because HGO contained dense oxygen-containing functional groups. Hydrophobic organic coated GO (pho-GO) including propylamine GO, tertoctylamine GO, and 1-adamantylamine GO, have relatively small(er) negative zeta potential compared to GO (ca. 10 mV). Zeta potential of propylamine GO, tert-octylamine GO, and 1adamantylamine GO was −22.9 ± 1.1, −26.3 ± 2.1, and −22.2 ± 0.9 mV, respectively. Again, this is attributed to the fact that GO effectively loses carboxyl groups during functionalization. In case of hydrophilic organic coated GO (phi-GO), zeta potential of 3-amino-1-propanol GO and 3-amino-1adamantanol GO are similar to that of GO as the carboxyl groups of GO are replaced with hydroxyl groups of 3-amino-1-propanol and 3-amino-1-adamantanol. Zeta potential of 3-amino-1-propanol GO and 3-amino-1-adamantanol GO was −31.6 ± 0.8 and −31.4 ± 2.8 mV, respectively.

Effect of Oxidation Extent on Colloidal Stability of GO
The effect of the oxidation extent of GO on colloidal stability was evaluated using GO and highly oxidized GO (HGO). As shown in Figure 3, the CCC values were measured using NaCl, CaCl2, MgCl2, and MgSO4 at pH 7.0 ± 0.2, respectively (details are also summarized in Table 2). Under all the ionic conditions tested, HGO shows the higher colloidal stability compared to GO; the CCC values in NaCl, CaCl2, MgCl2, and MgSO4 were 246.38, 1.40, 3.97, and 4.75 mM for GO, and 387.18, 1.79, 4.75, and 5.47 mM for HGO, respectively. As reported by others, the C/O ratios were negatively correlated with colloidal stability for the crumpled GO and carbon nanotubes [39,40]. In addition, the EDL repulsion is increased as a function of surface zeta potential through the formation of effective primary energy barrier [31,32].  is the hydrodynamic diameter [38]. L of GO and HGO was 85.0 ± 4.0 and 83.4 ± 8.8 nm, respectively. L of organic functionalized GO was slightly increased; 98.3 ± 8.1, 89.3 ± 2.9, 89.4 ± 1.6, 88.0 ± 6.4, and 87.6 ± 5.2 nm for propylamine GO, tert-octylamine GO, 1-adamantylamine GO, 3-amino-1-propanol GO, and 3-amino-1-adamantanol GO, respectively. Similar material sizes indicate that GO functionalization occurs without aggregation and the resulting products are relatively stable. Zeta potential of GO and HGO was −33.1 ± 1.9 and −43.3 ± 1.7 mV, respectively. HGO showed large negative zeta potential compared with GO because HGO contained dense oxygen-containing functional groups. Hydrophobic organic coated GO (pho-GO) including propylamine GO, tert-octylamine GO, and 1-adamantylamine GO, have relatively small(er) negative zeta potential compared to GO (ca. 10 mV). Zeta potential of propylamine GO, tert-octylamine GO, and 1-adamantylamine GO was −22.9 ± 1.1, −26.3 ± 2.1, and −22.2 ± 0.9 mV, respectively. Again, this is attributed to the fact that GO effectively loses carboxyl groups during functionalization. In case of hydrophilic organic coated GO (phi-GO), zeta potential of 3-amino-1-propanol GO and 3-amino-1-adamantanol GO are similar to that of GO as the carboxyl groups of GO are replaced with hydroxyl groups of 3-amino-1-propanol and 3-amino-1-adamantanol. Zeta potential of 3-amino-1-propanol GO and 3-amino-1-adamantanol GO was −31.6 ± 0.8 and −31.4 ± 2.8 mV, respectively.

Effect of Oxidation Extent on Colloidal Stability of GO
The effect of the oxidation extent of GO on colloidal stability was evaluated using GO and highly oxidized GO (HGO). As shown in Figure 3, the CCC values were measured using NaCl, CaCl 2 , MgCl 2 , and MgSO 4 at pH 7.0 ± 0.2, respectively (details are also summarized in Table 2). Under all the ionic conditions tested, HGO shows the higher colloidal stability compared to GO; the CCC values in NaCl, CaCl 2 , MgCl 2 , and MgSO 4 were 246.38, 1.40, 3.97, and 4.75 mM for GO, and 387.18, 1.79, 4.75, and 5.47 mM for HGO, respectively. As reported by others, the C/O ratios were negatively correlated with colloidal stability for the crumpled GO and carbon nanotubes [39,40]. In addition, the EDL repulsion is increased as a function of surface zeta potential through the formation of effective primary energy barrier [31,32]. The Schulze-Hardy rule, which relates the valence of counter ion(s) to the colloidal stability, can be used to predict the CCC values under ideal conditions. When considering plate-plate geometries, the ratio of monovalent counter ions to divalent counter ions is 2 2 (as derived above). In the case of sphere-sphere geometry (Schulze-Hardy rule), the ratio is in the range between 2 2 and 2 6 , which depends on the zeta potential. For larger zeta potentials, the ratio of CCC is proportional to 2 6 due to the high EDL repulsion, and it moves towards 2 2 as the zeta potential gets smaller (more hydrophobic). The results obtained in this work show that the ratio of NaCl to CaCl2, MgCl2, and MgSO4 is 2 7.46 , 2 5.96 , and 2 5.70 for GO, and 2 7.76 , 2 6.45 , and 2 6.15 for HGO, respectively. These obtained ratios are close to the expectation of the Schulze-Hardy rule for sphere-sphere geometry. It is likely that 2D shaped GO sheet dynamics appears to be more like 3D sphere-like aggregates during the aggregation processes (hydration dynamics may also play a role) [20]. Among the divalent counter ions (Ca 2+ and Mg 2+ ) evaluated, Ca 2+ is stronger in influence for both GO and HGO, as Ca 2+ has a higher propensity to form cross-linking interactions than Mg 2+ [41]. We also observed that MgCl2 is more effective than MgSO4, which means that co-ions also, albeit to a lower extent, influence colloidal stability. With the same valance of counter ions, CCC values are inversely proportional to the valance of co-ions, which is termed the inverse Schulze-Hardy rule [42]. Compared with divalent co-ions, monovalent co-ions are more readily located near the GO surface, effectively neutralizing the surface charge [43].
To compare the experimental measurements with theoretical energy interactions, we calculated the classical plate-plate DLVO interaction where the attachment efficiency (α) is ca. 0.3, 0.8, and 1.0. Such plate-plate DLVO expressions are restricted to symmetric electrolytes because of the simplification of the Taylor series expansion in the EDL equation [44][45][46]. Thus, we selected NaCl as background ions, which is a 1:1 symmetric electrolyte. Figure 4 shows the DLVO interaction energy   The Schulze-Hardy rule, which relates the valence of counter ion(s) to the colloidal stability, can be used to predict the CCC values under ideal conditions. When considering plate-plate geometries, the ratio of monovalent counter ions to divalent counter ions is 2 2 (as derived above). In the case of sphere-sphere geometry (Schulze-Hardy rule), the ratio is in the range between 2 2 and 2 6 , which depends on the zeta potential. For larger zeta potentials, the ratio of CCC is proportional to 2 6 due to the high EDL repulsion, and it moves towards 2 2 as the zeta potential gets smaller (more hydrophobic). The results obtained in this work show that the ratio of NaCl to CaCl 2 , MgCl 2 , and MgSO 4 is 2 7.46 , 2 5.96 , and 2 5.70 for GO, and 2 7.76 , 2 6.45 , and 2 6.15 for HGO, respectively. These obtained ratios are close to the expectation of the Schulze-Hardy rule for sphere-sphere geometry. It is likely that 2D shaped GO sheet dynamics appears to be more like 3D sphere-like aggregates during the aggregation processes (hydration dynamics may also play a role) [20]. Among the divalent counter ions (Ca 2+ and Mg 2+ ) evaluated, Ca 2+ is stronger in influence for both GO and HGO, as Ca 2+ has a higher propensity to form cross-linking interactions than Mg 2+ [41]. We also observed that MgCl 2 is more effective than MgSO 4 , which means that co-ions also, albeit to a lower extent, influence colloidal stability. With the same valance of counter ions, CCC values are inversely proportional to the valance of co-ions, which is termed the inverse Schulze-Hardy rule [42]. Compared with divalent co-ions, monovalent co-ions are more readily located near the GO surface, effectively neutralizing the surface charge [43].
To compare the experimental measurements with theoretical energy interactions, we calculated the classical plate-plate DLVO interaction where the attachment efficiency (α) is ca. 0.3, 0.8, and 1.0. Such plate-plate DLVO expressions are restricted to symmetric electrolytes because of the simplification of the Taylor series expansion in the EDL equation [44][45][46]. Thus, we selected NaCl as background ions, which is a 1:1 symmetric electrolyte. Figure 4 shows the DLVO interaction energy profiles of GO and HGO with respect to attachment efficiency. Under favorable conditions (α = 1), there are no primary maximums for both GO and HGO interaction energies. For a high α condition (α ≈ 0.8), 1.66 × 10 −4 and 4.48 × 10 −4 mJ/m 2 of primary maximum were obtained at 1.3 and 0.9 nm (distance) for GO and HGO, respectively. For a low α condition (α ≈ 0.3), the highest primary maximum was observed; 1.01 × 10 −3 and 6.72 × 10 −3 mJ/m 2 at 1.0 and 0.6 nm for GO and HGO, respectively. While the GO formed 3D sphere-like aggregates during the aggregation, plate-plate DLVO energy interactions were in good agreement with the CCC experimental measurements.
profiles of GO and HGO with respect to attachment efficiency. Under favorable conditions (α = 1), there are no primary maximums for both GO and HGO interaction energies. For a high α condition (α ≈ 0.8), 1.66 × 10 −4 and 4.48 × 10 −4 mJ/m 2 of primary maximum were obtained at 1.3 and 0.9 nm (distance) for GO and HGO, respectively. For a low α condition (α ≈ 0.3), the highest primary maximum was observed; 1.01 × 10 −3 and 6.72 × 10 −3 mJ/m 2 at 1.0 and 0.6 nm for GO and HGO, respectively. While the GO formed 3D sphere-like aggregates during the aggregation, plate-plate DLVO energy interactions were in good agreement with the CCC experimental measurements.

Effect of Organic Coating on Colloidal Stability of GO
The effects of surface coating on colloidal stability were evaluated using GO with five types of surface coatings; hydrophobic coatings (propylamine, tert-octylamine, and 1-adamantylamine) and hydrophilic coatings (3-amino-1-propanol and 3-amino-1-adamantanol). To evaluate their colloidal stability, we measured the CCC of both hydrophobic organic coated GO (pho-GO) and hydrophilic organic coated GO (phi-GO) under varied salt conditions (NaCl, CaCl2, MgCl2, and MgSO4) at pH 7 ± 0.2 (summarized in Table 2). As presented in Figure 5, the colloidal stability of pho-GO was decreased, compared to the pristine GO. The CCC values in NaCl were 101.4, 104.6, and 118.3 mM for propylamine GO, tert-octylamine GO, and 1-adamantylamine GO, respectively. Typically for GO, carboxyl groups play a significant role in EDL and steric repulsion [5]. As observed in the characterization of GO materials (Figures 1 and 2), pho-GO effectively lost that carboxyl functionality. We also observed that the divalent cation type (Ca 2+ vs. Mg 2+ ) did not significantly influence the stability of pho-GO. CCC values in CaCl2, MgCl2, and MgSO4 were 1.1, 1.2, and 1.4 mM; 1.3, 1.6, and 1.7 mM; 2.0, 2.2, and 2.5 mM for propylamine GO, tert-octylamine GO, and 1-adamantylamine GO, respectively. In the presence of divalent cations, the colloidal stability of GO decreases through surface charge screening, bridging, intercalating, and cross-linking processes with oxygen-containing functional groups, including carboxyl, carbonyl, and hydroxyl groups [47]. Due to the deficit of the oxygen-containing (surface exposed) functionalities of pho-GO, the influence of type of divalent counter ions on colloidal stability is less significant.

Effect of Organic Coating on Colloidal Stability of GO
The effects of surface coating on colloidal stability were evaluated using GO with five types of surface coatings; hydrophobic coatings (propylamine, tert-octylamine, and 1-adamantylamine) and hydrophilic coatings (3-amino-1-propanol and 3-amino-1-adamantanol). To evaluate their colloidal stability, we measured the CCC of both hydrophobic organic coated GO (pho-GO) and hydrophilic organic coated GO (phi-GO) under varied salt conditions (NaCl, CaCl 2 , MgCl 2 , and MgSO 4 ) at pH 7 ± 0.2 (summarized in Table 2). As presented in Figure 5, the colloidal stability of pho-GO was decreased, compared to the pristine GO. The CCC values in NaCl were 101.4, 104.6, and 118.3 mM for propylamine GO, tert-octylamine GO, and 1-adamantylamine GO, respectively. Typically for GO, carboxyl groups play a significant role in EDL and steric repulsion [5]. As observed in the characterization of GO materials (Figures 1 and 2), pho-GO effectively lost that carboxyl functionality. We also observed that the divalent cation type (Ca 2+ vs. Mg 2+ ) did not significantly influence the stability of pho-GO. CCC values in CaCl 2 , MgCl 2 , and MgSO 4 were 1.1, 1.2, and 1.4 mM; 1.3, 1.6, and 1.7 mM; 2.0, 2.2, and 2.5 mM for propylamine GO, tert-octylamine GO, and 1-adamantylamine GO, respectively. In the presence of divalent cations, the colloidal stability of GO decreases through surface charge screening, bridging, intercalating, and cross-linking processes with oxygen-containing functional groups, including carboxyl, carbonyl, and hydroxyl groups [47]. Due to the deficit of the oxygen-containing (surface exposed) functionalities of pho-GO, the influence of type of divalent counter ions on colloidal stability is less significant.  Figure 6 presents the colloidal stability of phi-GO (3-amino-1-propanol GO and 3-amino-1adamantanol GO). Compared with the pho-GO, phi-GO was more colloidally stable as expected. CCC values in NaCl were 153.1 (51% increase) and 303.3 mM (156% increase) for 3-amino-1-propanol GO and 3-amino-1-adamantanol GO, respectively. The coatings of both pho-Go and phi-GO have an identical structure and composition, except for a single hydroxyl functional group of phi-GO. We speculate that hydroxyl groups in phi-GO not only provide strong EDL repulsion (larger zeta potential), but also effectively hinder the hydrophobic interactions. Interestingly, we observed that Ca 2+ has a stronger propensity to decrease the stability of phi-GO, compared with the pho-GO. CCC values in CaCl2, MgCl2, and MgSO4 were 1.2, 2.3, and 2.9 mM; 1.7, 2.8, and 3.5 mM for 3-amino-1propanol GO and 3-amino-1-adamantanol GO, respectively. In the presence of divalent counter ions, Ca 2+ ions play an important role as they uniquely coordinate between oxygen functionalities [5,47]. Additionally, we found that monovalent co-ions are more effective than divalent co-ions when considering the aggregation of phi-GO, compared to pho-GO. The role of co-ions in aggregation is based on surface charge screening [42,43]. It is likely that phi-GO has larger charged surface area (due to additional hydroxyl groups) than the pho-GO, which offers co-ion sensitivity for phi-GO.  Attachment efficiency of (a) propylamine GO, (b) tert-octylamine GO, and (c) 1-adamantylamine GO as a function of salt concentration; NaCl (blue), CaCl 2 (red), MgCl 2 (dark yellow), and MgSO 4 (dark green). Figure 6 presents the colloidal stability of phi-GO (3-amino-1-propanol GO and 3-amino-1adamantanol GO). Compared with the pho-GO, phi-GO was more colloidally stable as expected. CCC values in NaCl were 153.1 (51% increase) and 303.3 mM (156% increase) for 3-amino-1-propanol GO and 3-amino-1-adamantanol GO, respectively. The coatings of both pho-Go and phi-GO have an identical structure and composition, except for a single hydroxyl functional group of phi-GO. We speculate that hydroxyl groups in phi-GO not only provide strong EDL repulsion (larger zeta potential), but also effectively hinder the hydrophobic interactions. Interestingly, we observed that Ca 2+ has a stronger propensity to decrease the stability of phi-GO, compared with the pho-GO. CCC values in CaCl 2 , MgCl 2 , and MgSO 4 were 1.2, 2.3, and 2.9 mM; 1.7, 2.8, and 3.5 mM for 3-amino-1-propanol GO and 3-amino-1-adamantanol GO, respectively. In the presence of divalent counter ions, Ca 2+ ions play an important role as they uniquely coordinate between oxygen functionalities [5,47]. Additionally, we found that monovalent co-ions are more effective than divalent co-ions when considering the aggregation of phi-GO, compared to pho-GO. The role of co-ions in aggregation is based on surface charge screening [42,43]. It is likely that phi-GO has larger charged surface area (due to additional hydroxyl groups) than the pho-GO, which offers co-ion sensitivity for phi-GO.  Figure 6 presents the colloidal stability of phi-GO (3-amino-1-propanol GO and 3-amino-1adamantanol GO). Compared with the pho-GO, phi-GO was more colloidally stable as expected. CCC values in NaCl were 153.1 (51% increase) and 303.3 mM (156% increase) for 3-amino-1-propanol GO and 3-amino-1-adamantanol GO, respectively. The coatings of both pho-Go and phi-GO have an identical structure and composition, except for a single hydroxyl functional group of phi-GO. We speculate that hydroxyl groups in phi-GO not only provide strong EDL repulsion (larger zeta potential), but also effectively hinder the hydrophobic interactions. Interestingly, we observed that Ca 2+ has a stronger propensity to decrease the stability of phi-GO, compared with the pho-GO. CCC values in CaCl2, MgCl2, and MgSO4 were 1.2, 2.3, and 2.9 mM; 1.7, 2.8, and 3.5 mM for 3-amino-1propanol GO and 3-amino-1-adamantanol GO, respectively. In the presence of divalent counter ions, Ca 2+ ions play an important role as they uniquely coordinate between oxygen functionalities [5,47]. Additionally, we found that monovalent co-ions are more effective than divalent co-ions when considering the aggregation of phi-GO, compared to pho-GO. The role of co-ions in aggregation is based on surface charge screening [42,43]. It is likely that phi-GO has larger charged surface area (due to additional hydroxyl groups) than the pho-GO, which offers co-ion sensitivity for phi-GO. Figure 6. Attachment efficiency of (a) 3-amino-1-propanol GO, and (b) 3-amino-a-adamantanol GO as a function of salt concentration; NaCl (blue), CaCl2 (red), MgCl2 (dark yellow), and MgSO4 (dark green).

Conclusions
In this work, we systematically describe the critical role of organic functionality as it relates to the colloidal stability of derivatized GO in water. Matrix variables including the oxidation extent of surface, coating structure, and hydrophilicity, have significant influence on the aqueous stability of GO. Generally, a higher oxidation extent provides higher effective colloidal stability. Furthermore, GO materials with hydrophobic organic coatings (pho-GO) are much less sensitive to divalent counter ion type and concentration(s). The presence of hydrophilic coatings improves the colloidal stability of GO by providing (or reconstituting) steric and EDL repulsions. As co-ions are directly related to the surface charge neutralization, monovalent co-ions have a higher impact on for those GO materials having a larger number of oxygen-containing functional groups (HGO and phi-GO). Taken together, this work highlights the role of organic functionality as it governs the stability of GO in water. Future studies should explore the transport of similarly derivatized GO in an environmental system.