The phase transition between two solvated phases was studied by DSC for graphite oxide (GO) powders immersed in water–methanol mixtures of various compositions. GO forms solid solvates with two different compositions when immersed in methanol. Reversible phase transition between two solvate states due to insertion/desertion of methanol monolayer occurs upon temperature variations. The temperature point and the enthalpy (ΔH) of the phase transition are maximal for pure methanol and decrease linearly with increase of water fraction up to 30%.
Graphite oxide (GO) is a non-stoichiometric material with layered structure obtained by strong oxidation of graphite 1–3. The attachment of oxygen and OH groups to the graphene layers results in some buckling of the planes and the interlayer spacing increases up to ∼6 Å 4–6. GO is hydrophilic and easily absorbs water and other polar solvents when exposed to vapor or liquid with an increase of interlayer distance 7–9.
A surprising application of solvated GO was recently found by Nair et al. 10 who reported synthesis and unusual properties of submicrometer-thick membranes made from Hummers graphene oxide. The membranes were found to be permeable to water but not ethanol or other polar solvents and not even by helium gas.
An explanation of these unusual permeation properties requires better understanding of the structural properties of GO when hydrated or solvated by various solvents. Our recent experiments showed that the incorporation of water into the GO structure is strikingly different compared to any other polar solvents 8, 11.
Graphite oxide produced by Brodie's method forms solvate phases when immersed in excess of polar solvents at ambient conditions. Insertion of additional solvent occurs upon either cooling at ambient pressure 8, 12 or pressure increase at ambient temperature 12–14. For all solvents except water (methanol, ethanol, acetone, and dimethylformamide (DMF)) we observed phase transitions into low temperature (or high pressure) phases with a stepwise increase of the interlayer distance which correlates to intercalation of an additional monolayer of solvent molecules 11. The low temperature solvate phases are stable even below the freezing point of solvents.
In contrast, GO in excess of water shows quite different hydration/dehydration behavior upon temperature and pressure variations. Cooling of GO/water samples results first in a gradual change of the interlayer spacing with a stepwise contraction at the point of water media freezing 8. These results suggest that part of the water in fully hydrated solvate remains in the liquid-like state and keeps the translational motion ability when confined in slit pores of GO. Molecules of other polar solvents (so far tested) are incorporated into the GO structure in more fixed positions forming disordered crystalline solvates and in contrast to water cannot be considered as liquid in nanosized pores.
However, so far the phase transitions between different solvate phases were studied only for pure water and methanol with only a couple of high pressure experiments performed for water/methanol mixtures of certain compositions 12, 13.
Here we present a differential scanning calorimetry (DSC) study of phase transitions in GO powder samples immersed in excess amount of water/methanol solvent mixed in various proportions. Reversible transitions between the ambient and low temperature phases were observed upon temperature variations only for a specific interval of water–methanol compositions.
The GO sample was prepared using Brodie's method and showed a composition of CO0.38H0.12 according to elemental analysis 1, 12. DSC was used to monitor the heat effect of the phase transformations using a Perkin Elmer PYRIS Diamond™ DSC. GO powder and excess amount of solvent (e.g., 2.0 mg GO in 12.7 mg methanol) were loaded into sealed aluminum sample capsules of ca. 50 µL volume. The solvent was prepared by mixing water and methanol in various proportions. The temperature was programmed to change by a rate of 10 K min−1 both on cooling and heating. The samples were first cooled from room temperature to ∼213 K and then heated up.
3 Results and discussion
According to our previously published XRD study 8, cooling down GO/water sample below the point of water solidification results in stepwise contraction of the GO interlayer distance by ∼25% due to partial withdrawal of water from the hydrated structure. It is not possible to detect the heat effect from this transition using DSC because it is negligibly small compared to the heat effect from water solidification which occurs at exactly the same temperature point.
In contrast, GO immersed in pure methanol shows a very clear exothermic peak upon cooling due to a phase transition from ambient temperature Solvate A phase to low temperature Solvate L phase and an endothermic peak for the reverse transition when samples are heated back after cooling 11. The reversible phase transition for GO immersed in excess of water-free methanol was correlated with insertion of one additional methanol monolayer into the GO structure upon cooling and desertion of this monolayer upon heating back to ambient temperatures, as revealed in our previous study. It could be anticipated that adding water into methanol, thus forming binary mixtures, will result in strong modifications of the phase transition. Therefore, we performed a set of experiments with progressively increased proportion of water (volume fraction, X%) in water–methanol mixtures while monitoring the changes in the temperature and heat effects of the phase transition.
The samples were cooled to below the point of Solvate L formation and then heated back to observe the reverse transformation. The evolution of the endothermic peak due to phase transition from Solvate L to Solvate A for water–methanol liquid media with variation of water fraction X% is shown in Fig. 1. A sharp endothermal peak is observed between 275 and 295 K for a GO sample immersed in water-free methanol liquid medium. Adding 5% of water into the methanol medium gives a similar DSC curve for the reversible phase transition as shown in Fig. 2, with a phase transition temperature drop by ∼9 K compared to the GO/pure methanol sample. When more water is added into methanol the endothermic peak shifts to even lower temperature, and becomes weaker and broader. The trend could be followed for compositions of binary mixtures with water content 5–30%. No clear peaks on heat flow traces could be observed for amounts of water above 30% in the water–methanol mixture.
The phase transition temperature, Tc, (the temperature point for maximum value of the heat flow peak), decreases almost linearly by ∼1.56 K per percent of water in the solvent mixture range from 0 to 25% (Fig. 3). A similar trend is also observed for the enthalpy of the phase transition (ΔH; Fig. 4). The largest ΔH value (11.45 J g−1(GO)) is observed for the GO sample immersed in pure methanol. The value decreases linearly with increased amount of water in the water–methanol mixture. Extrapolation of the linear trend shows that the ΔH would go to zero at a content of water X about 40–45%.
The linear trend observed for the temperature points and the ΔH values of the phase transition also correlates with trends in the temperature dependence of the melting points for water–methanol binary solutions (Fig. 3). The freezing point of water–methanol binary mixtures increases as the content of water increases (when X > 10%) 15 while the temperature of phase transition in GO/water–methanol mixtures goes down.
The cross-over point for the two dependences shown in Fig. 3 can be obtained by extrapolation of Tc versus X (volume fraction of water in mixture with methanol) and gives a temperature of about 217 K for a water content X = 43%. Remarkably, almost the same value is obtained by an extrapolation of the ΔH versus X dependence shown in Fig. 4.
It can be concluded that a phase transition into the low temperature solvate phase should not be observed when the content of water is increased above 40–45%; the extrapolated phase transition temperature would be below the freezing point of the liquid medium. Finding the exact structural changes which occur in the GO system with water–methanol mixtures upon temperature variations require further experiments using X-ray diffraction. However, using DSC data we can predict that GO in water–methanol mixtures with content of water over 45% most likely will not exhibit step-like structural changes as observed for a pure methanol solvent medium previously 11 and the structural parameters of GO will be changing gradually with variation of temperature as for pure water 8. Gradual variations of interlayer distance between graphene oxide sheets was observed in our previously published study under pressure increase when a water/methanol 2:1 mixture was used as liquid medium 13.
This effect can possibly be explained assuming that water and methanol are capable to intercalate GO structure only by monolayers of pure solvent even when immersed in water–methanol mixtures of various compositions. It is also possible to anticipate that the permeation of the graphene oxide membranes described in Ref. 10 will be sharply distinct for mixtures of water with methanol of various compositions, reflecting the different mechanisms of solvent incorporation into the GO structure revealed in our study. It is also clear that the temperature dependence for permeation of such membranes could be quite complex, reflecting changes in the phase composition of GO solvates which occur upon cooling and heating.
Phase transitions between the ambient Solvate A and the expanded Solvate L phases of GO in water–methanol mixture liquid media were studied by DSC as a function of water content in the liquid medium. The position of the endothermic anomaly which corresponds to a transition from Solvate L to Solvate A phase (desertion of a solvent monolayer from the interlayer space of GO) is shifted to low temperatures when water is added to the methanol. The enthalpy difference (ΔH) measured for the phase transition decreases linearly with an increase of the water fraction and goes to zero when the content of water in the water–methanol mixture reaches 40–45%. Our results are of general importance for solution-based chemistry of GO and applications like selective filters/membranes.
We thank Prof. Åke Fransson in Umeå University for providing the access to DSC and the help with setting up the measurement. Part of this work was financially supported by the Swedish Research Council, grant 621-2010-3732.