Graphene Oxide as Foam Stabilizing Agent for CO2 EOR

Graphene oxide (GO), nanographene oxide (nGO) and partially reduced graphene oxide (rGO) have been studied as possible foam stabilizing agents for CO2 based enhanced oil recovery (EOR). GO was able to stabilize CO2/synthetic sea water foams. rGO was not able to stabilize foams likely due to the high reduction degree of the material. Particle size had a strong influence on foamability and stability. GO hydrophilicity increased as the particle size decreased and no foams were created when particle size was below 1 µm (nGO). GO brine dispersions showed immediate gel formation, which improved foam stability. Particle growth due to layer stacking was also observed. This mechanism was detrimental for foam formation and stabilization. nGO dispersed in synthetic sea water rapidly formed hydrogels and was not filterable. This work indicates that the particles studied are not suitable for CO2 EOR purposes.


Introduction 21
CO2-based methods for enhanced oil recovery (EOR) in water flooded reservoirs faces various 22 technical challenges. The low density of CO2 compared to water causes CO2 migration towards 23 upper zones impairs oil recovery in the lower zones of the reservoir. The low CO2 viscosity leads to 24 viscous fingering and excessive flow in high permeability layers. The net effect can be early CO2 25 breakthrough, reduced sweep efficiency and low oil recovery. Counteracting these effects can be 26 achieved by decreasing the CO2 mobility, either by adding thickeners into CO2 or dispersing it into 27 brine (CO2 foam) [1][2][3][4]. 28 Direct thickeners CO2 require molecules that are CO2 soluble and have groups that interact 29 giving the increased viscosity. Efforts to develop thickeners have been ongoing through the last 30 decades. Up to now the best results have been obtained with a fluoroacrylate-styrene copolymeric 31 thickener that at typical reservoir conditions is able to increase the CO2 viscosity by the order of ten 32 using low concentrations (< 1 wt. %) [1]. However, due to costs and environmental concerns these 33 types of additives are unlikely to have practical application. 34 The apparent viscosity of CO2 dispersed into foams may be very high, depending on the type 35 of surfactant used. One potential problem is that many foams are sensitive to the presence of oil 36 giving destabilisation through several mechanisms including spreading and entering phenomena 37 [5][6][7]. During miscible CO2 flooding oil sensitivity may be an advantage as foam formation is desired 38 where the oil already has been displaced, diverting the CO2 into oil containing parts of the reservoir. 39 However, also minute amounts of oil remaining after miscible flooding may have detrimental effects 40 on foam propagation [8]. 41 CO2 foams can also be stabilized by nanoparticles. Nanoparticles adsorb strongly at interfaces 42 which contribute to a higher stability. However, the mixing energy required to adsorb at interfaces 43 2 is larger than traditional surfactant stabilized systems. This is an important disadvantage for oil 44 recovery [9][10][11][12]. Typically, oil field flow velocities do not exceed a few feet/day which results in low 45 mixing energy. It may therefore be difficult to utilise nanoparticle stabilized foams under normal 46 process conditions. 47 The use of binary mixtures of surfactant and nanoparticles may improve foamability at low 48 flooding rates. Singh et al. [13] showed that a mixture of anionic surfactant and fly ash could reduce 49 CO2 mobility more than anionic surfactant alone. Cationic surfactant had the opposite effect, 50 however. Manan et al. [14] demonstrated that mixtures of surfactant and different types of 51 nanoparticles improved oil recovery during CO2 flooding compared to only surfactant, the 52 improvement depended on type of particles used. 53 Patel et al. [15] studied oil-in-brine emulsion stability of silica nanoparticles in presence of 54 sodium dodecyl sulfate. The presence of surfactant allowed to increase suspension stability by 55 diminishing particle flocculation. Even though they observed that nanoparticles were more effective 56 for stabilizing oil emulsions, binary mixtures of particles and surfactant were detrimental for 57 emulsion stability compared to only nanoparticle-stabilized emulsions. These observations are 58 consistent with the competitive adsorption conclusion of Pichot et al. [16]. 59 A binary system with surfactant and nanoparticles will be vulnerable to separation of the 60 constituents during transport in porous media and beneficial system properties may thus be lost. 61 Ideally the foam stabilising agent should be uniform to avoid loss of performance. to the strong interactions between the GO nanosheets and the molecules of asphaltenes and resins. 76 GO is a candidate for application within the oil industry. However, studies for its applications 77 in EOR has up to now not been published. In this work we have studied the possible use of GO/rGO 78 particles as foam stabilizing agents for CO2 EOR. 79 In this article dispersions of aqueous solutions and CO2 is referred as foams whereas dispersions 80 of aqueous solutions and organic solvents are named emulsions. 81

Materials 83
GO and rGO materials were supplied by Graphenea S.A., San Sebastián, Spain. GO was 84 delivered in a water suspension with a total concentration of 4 mg/ml. The particle size was 85 polydisperse, ranging from 4 to 30 µm. GO suspensions with smaller particle size (nGO) were also 86 provided by Graphenea S.A. The particle sizes of the latter suspension were determined using a 87 Zetasizer Nano ZS (Malvern Instruments Ltd). rGO was obtained in solid powder form. The particle 88 size of rGO ranged from 260 to 295 nm as specified by the provider. The elemental composition of 89 both materials was also obtained from the provider and is described in The bottle tests were carried out with a volume fraction of 50 % of each phase (6 ml of total 101 volume), SSW and organic solvents. The particle concentrations used were 1 mg/ml and 0.5 mg/ml 102 of GO in the SSW. The mixture was shaken by hand vigorously for 10 seconds and placed in a 103 graduated glass test tube where the volumes of each phase were read. 104

Phase equilibria studies 105
An internally stirred windowed high-pressure, high-temperature pVT-cell allowing visual 106 observation along the whole volume of the cell was used for the measurements. The cell was placed 107 inside a temperature-controlled heating cabinet. Equal volumes of dense CO2 and aqueous solutions 108 were injected into the cell by using high-pressure pumps and the pressure was adjusted to the initial 109 test-pressure. The total test volume was approximately 60 ml. The pressure varied typpically less 110 than 2% during the tests.

114
The foams were created using a magnetic stirrer located inside the cell. The system was stirred 115 for one minute. During stirring, the cell was tilted 180 degrees vertically one time and returned to its 116 original position for favouring phase contact. After the stirring stopped the phase heights (CO2, 117 foam, SSW) were determined over time using a cathetometer. 118

Emulsion stability 120
Initial tests with organic solvents were carried out just to observe the ability of GO to stabilize 121 organic solvent/water emulsions.

128
All systems showed good stability and, after the separation observed during the first minutes, 129 the emulsions remained stable for weeks. Toluene and n-decane had a similar emulsified volume, 130 while hexane showed a lower emulsion volume (~ 41 vol. %). 131 The samles were immersed in a water bath to study the temperature effect on emulsion stability 132 at 30 °C, 50 °C, and 80 °C. No differences were observed between room temperature, 30 °C, and 50 133  °C. However, at 80 °C larger droplets of solvent were observed trapped inside the emulsion phase 134 and there was a slight solvent phase volume increase. 135 The effect of the particle concentration was also studied using concentrations ranging from 0.1 136 to 1.0 mg/ml. Figure 3 shows results for 0.5 mg/ml and 0.3 mg/ml. The stability for 0.7 mg/ml and 137 0.5 mg/ml was as for 1.0 mg/ml. However, there was a drop in the stability when the concentration 138 was reduced from 0.5 to 0.3 mg/ml. The results obtained with 0.1 mg/ml were the same as the results 139 obtained with 0.3 mg/ml.  GO-stabilized CO2 foams were studied using the pVT cell. 30 ml of CO2 were first charged into 146 the cell. The pressure and temperature were stabilized. Then, 30 ml of GO suspension in SSW (1  147 mg/ml) were injected into the cell, giving a total volume of 60 ml at the desired conditions. The 148 system was dispersed by stirring for one minute. After the stirring was stopped, the brine and CO2 149 heights were measured. Then, 24 h after charging the cell, the same system was re-stirred using the 150 same procedure and the interphase heights were measured again. This process was repeated also 48 151 h after the cell charge. Hereinafter, the time specified will always refer to the cell charging time.

164
After 72 hours, the sample was again re-stirred and a significant stability reduction was 165 observed. It was not possible to disperse the particles as effectively as observed during the previous 166 days. Just a few minutes after stirring stopped, aqueous and CO2 phases were separated with a 167 considerable foam phase reduction. Thus, the system stability showed a time-depending effect. 168 Phase volumes were not recorded, however. 169 The cell was next charged with new fluids at the same conditions as before. After two days 170 ageing (when the previous experiment had shown maximum stability), the temperature was 171 increased from room temperature to 50 °C at constant pressure, 78 bar. Due to the heating, the CO2 172 phase expanded exceeding the 50 vol. % as can be seen in Figure 5. Shortly after, the pressure was 173 increased to 152 bar at constant temperature. 174 Increased temperature was detrimental to foam stability as observed from comparing Error! 175 Reference source not found. at 48 h and

184
The time-dependent foam stability was studied further using a lower concentration of particles. 185 The cell was filled with 50 vol. % of each phase at a particle concentration of 0.5 mg/ml at 153 bar. 186 The foam stability was measured after stirring once the sample was injected, after 5 days, after 11 187 days and 11 days at 226 bar (the system was re-stirred each time). 188 The results are shown in Figure 6. When the new system was charged into the cell the formation 189 of free water appeared after 5 minutes under a large foam phase which immediately started to 190 coalesce. After one hour, the system consisted of 40 vol. % of coarse foam and 40 vol. %of stable foam 191 (Figure 6a). The phase observed under the free water phase was foam that adhered to the piston 192 corners and the magnetic stirrer (Figure 6a, Figure 6b, Figure 7a, and Figure 7b). After 5 minutes the 193 system remained almost unchanged. 194 After 5 days, CO2 phase appeared immediately after the stirring was stopped (ca. 31 vol. %). 195 The total foam volume decreased to 54 vol. %after approximately 30 minutes. 10 vol. % was coarse 196 foam and 15 vol. % was free aqueous phase (Figure 6b). After 30 minutes, only small changes in 197 phase volumes were observed. 198 After 11 days, the appearance of the foam changed. It adopted a self-folding structure 199 appearance which acted as a plug when re-stirring attempts were made. The system was composed 200 of 37 vol. % CO2, 10 vol. % aqueous phase, and 53 vol. % of foam (Figure 6c). Then the system was 201 pressurized to 226 bar and a new attempt to stir was made, but the foam phase continued acting as 202 a plug.

Foam morphology 208
The pVT cell was filled with 1 mg/ml GO in SSW and CO2 (50 vol. % of each phase) at 226 bar 209 and room temperature and stirred. The system was re-stirred after 1 hour, after 3 hours and after 11 210 days. Pictures of the cell were taken approximately 5 minutes after each stirring and are shown in 211 Figure 7.

212
There was an evident change in the foam morphology between the freshly made system and 213 after 3 hours. After 3 hours the foam phase occupied the whole volume and large trapped CO2 214 bubbles were observed (Figure 7c). Then the system was left in static conditions for 11 days. 215 Figure 7d shows the system after 11 days for the previous experiments carried out with a 216 concentration of 0.5 mg/ml. However, the systems with 0.5 and 1 mg/ml after 11 days had the same 217 appearance. The 11 days aged system showed a noticeable appearance difference with the systems 218 during the first day (Figure 7a, Figure 7b, and Figure 7c). The foam phase exhibited a different 219 morphology, suggesting the formation of film-like structure. 220 Once the system was re-stirred, the foam phase acted as a rigid plug not allowing mixing of the 221 phases. Thus, the foam phase had solidified supporting the assumption that self-folding structures 222 were formed.

228
The increase of stability and its consecutive stability reduction might be the effect of two 229 competitive effects: gelation and precipitation. The presence of divalent ions in the brine could 230 promote system gelation, increasing aqueous phase viscosity and thus, improve stability. Bai et al. 231 [24] already reported the ability of divalent ions (Ca 2+ and Mg 2+ ) to promote hydrogel formation with 232 GO in brine. This hydrogel formation was already observed when the GO was dispersing into SSW. 233 The solution became heterogeneous, and the apparition of gel-like aggregates were noticed. 234 On the other hand, stacking of GO sheets would decrease their interfacial activity and their 235 ability to stabilize dispersed systems. In Figure 7d, particle associations in a film-like structure in the 236 foam phase can be intuited. The stacking of parallel layers is energetically more stable than gel 237 structures due to its larger contact area between GO sheets. This can also be strengthened by the 238 presence of the divalent ions like Ca 2+ and Mg 2+ [25]. However, this last stacking mechanism seemed 239 to happen at a lower rate than gelation did. 240

Partially reduced Graphene Oxide (rGO) 241
A CO2/SSW system using a concentration of rGO of 1 mg/ml in SSW was studied at 78 bar and 242 21 C. However, it was not possible to disperse the CO2 into the aqueous phase. Moreover, all 243 particles flocculated and gathered at the interfaces, as observed in Figure 8a. 244 An attempt to disperse n-hexane into water using 1 mg/ml rGO failed. (Figure 8b)

249
According to Liu et al. [22] reducing the GO increases the CO2-philicity of the particles which 250 leads to a better foam stability. However, the present observations are not necessarily opposed to 251 Liu et al. conclusions. In their research, partially reduced GO was used, but the reduction degree was 252 not specified nor was an elemental analysis of the particles provided. In this study the oxygen 253 content was reduced from 41 -50 % (GO) to 13 -22 % (rGO). It is possible that these particles were 254 more reduced, decreasing the sites for interaction between particles and CO2 (epoxy groups) and 255 thus, increasing both their CO2-phobicity and the hydrophobicity. 256 3.2.4. Nanographene Oxide (nGO) 257 The first experiments using GO particles for stabilisation of CO2/SSW foam were promising with 258 respect to stabilisation. However, the particles used were large (4 to 30 µm), and the size should be 259 reduced to avoid pore blockage during flow through porous media. Graphenea supplied a reduced 260 particle size nGO suspension in pure water. A diluted sample of the new batch was ultrasonicated 261 and filtrated through a 0.45 µm syringe filter and afterwards, the particle size distribution was 262 measured. Figure 9 depicts the nGO particle size distribution. The sample showed a peak around 263 333 nm and a polydispersity ranging from 70 nm to 1.5 µm. The presence of large particles indicate 264 that particle aggregation occurred shortly after the filtration. 265 266 267 Figure 9. Particle size distribution for nGO particles after ultrasonication and filtration. The nGO was found to be too hydrophilic for forming and stabilizing foams, remaining in the 278 aqueous phase and not adsorbing at the CO2/SSW interface. A size-reduced GO particle would have 279 a higher density of -COOH and epoxy groups, increasing hydrophilicity [21,22]. However, it may 280 be possible to reduce the particles to an optimal degree. At this hypothetic point, the density of -281 COOH and epoxy groups for a given particle size may reduce hydrophilicity keeping the particles 282 CO2-philic enough for interfacial adsorption. 283 In addition, a filtration test using the initial 1 mg/ml suspension in SSW (heterogeneous 284 suspension as described before) was done using a 1.2 µm cellulose nitrate filter. The filter was 285 blocked immediately. This showed that the particles aggregated in the presence of divalent ions most 286 likely forming hydrogels. Thus, hydrogel formation can stabilise foam, as observed with larger 287 particles, but makes the system useless for injection into porous media. 288

Conclusions 289
Large GO sheets can be used effectively to disperse CO2 in brine, but the particles tested were 290 too large (4 to 30 µm) for flow through porous media. Smaller particles were considered. However, 291 reducing particle size had a determinant effect on foamability. nGO with particle size below 1 µm 292 was not able to form foams. This can possibly be adjusted by partially reducing nGO to make the 293 particles less hydrophilic, but the reduction degree must be carefully considered. 294 CO2 in SSW foams stabilized by GO showed a time-dependent stability. This was likely the 295 result of the competitive effect of two mechanisms, hydrogel formation and GO layer staking. Both 296 mechanisms apperared to be triggered by the presence of divalent ions. Hydrogel formation was the 297 faster mechanism and played a beneficial role for foam stability which initially increased reaching a 298 maximum after two days. Contrarily, GO layer stacking was a slower process and cancelled out the 299 stabilizing hydrogel formation mechanism giving a rigid dispersion of CO2 and brine. 300

Particle size [nm]
Reduced graphene oxide (rGO) with 13 -22 % oxygen content was not able to form foams. This 301 contradicts the results obtained by Liu et al.[22] and may be due to a the high degree of reduction of 302 the particles used. Thus, the degree of reduction must be taken carefully into account as low contents 303 of oxygen would increase both hydrophobicity and CO2-phobicity. 304 Hydrogel formation in presence of divalent ions can make graphene oxide particles not suitable 305 for EOR purposes. A dispersion of nGO in SSW could not flow through 1.2 µm cellulose nitrate 306 filters, most likely due to hydrogel formation. Passing the filtration test was set as a requrement for 307 carrying on with core flooding experiments. Even though GO can stabilize CO2/SSW foams, the 308 results indicate that they are not suitable for CO2 EOR.