3.1. Chemical-Physical Characterization
3.1.1. FTIR-ATR Characterization
Lupamin 9095 and purified PVAm films (PVAm-LG and PVAm-HG) were analyzed by IR spectroscopy to assess the efficiency of the purification process; for completeness, the intermediate purification step post Soxhlet treatment (PVAm-LG+) was analyzed as well (Figure 5
). The latter step did not cause massive variation of impurities concentration but was shown to greatly improve the efficiency of final ion exchange and was therefore maintained in the final purification procedure.
The peaks at 770, 1350, 1570, 2725 and 2840 cm−1
refer to the presence of the sodium formate salt (by comparison with the NIST’s IR spectra [64
]. They are well visible in the commercial Lupamin 9095 spectrum but lose intensity in the purified PVAm spectra. In particular, PVAm-LG and PVAm-LG+ show depressed absorbance at the mentioned wavelengths with respect to Lupamin 9095, but the peaks are still present, suggesting a significant but not complete purification. In PVAm-HG films, on the other hand, the characteristic signals of the salt are almost completely absent (except for the one at 770 cm−1
), indicating an excellent purification of the polymer. It is worth noticing that an intensification of the characteristic polymer signals (880 cm−1
N–H wagging, 1380 and 1440 cm−1
C–H bending/rocking, 1660 cm−1
N–H bending, 2920 cm−1
C–H stretching, and 3150–3400 cm−1
broad peak N–H stretching) is also observed in these materials, which corresponds to a better degree of purification.
Concerning the polymer peaks, then, an interesting change in the different spectra is the shift of the peak related to N–H stretching before and after the resin treatment (PVAm-HG). This one moves from 3100 to 3300 cm−1
, which is quite reasonably associated to the neutralization of amine group from the R–NH3+
to the R–NH2
form, as also confirmed by the results obtained by Annenkov et al. [66
]. This neutralizing effect is caused by the hydroxylated resin, which exchanges hydroxyl anions in solution, while retaining salt anions.
3.1.2. Thermal Characterization
Thermal and thermo-oxidative stability of Lupamin 9095 and purified PVAm was assessed through TGA analysis (Figure 6
). The overall behavior of all tested samples is quite similar and the most important differences concern the entity of the weight loss (visible in Table 1
) and final residues.
Until 180 °C, TGA curves display weight loss due to absorbed water. The polymer degradation occurs in two main steps with onset temperature at 190–220 and 305–320 °C with only very minor differences among different materials, as visible in Table 1
. During degradation, ammonia and hydrazine are released by the polymer, as stated by IR spectra of the evolved gas.
Sample weight is stable in the range 500–600 °C, indicating the end of the thermal degradation. Organic residues were oxidized by switching in air atmosphere at 600 °C and keeping the sample at that temperature for 30 min. After this treatment, purified films showed a significantly lower residue with respect to untreated Lupamin 9095. In particular, PVAm-LG and PVAm-HG showed residues in the order of 15% and 8% of the initial weight, respectively, confirming the better purification obtained when the ion exchange step was used after ethanol washing.
No major modifications of the TGA curves of the polymer were observed upon graphene or graphene oxide addition; these nanofiller, therefore, did not seem to affect the thermo-oxydative stability of the matrix.
Interestingly, the degree of purification and nanofiller loading affected the glass transition temperature (Tg
) of the films, as shown by DSC analyses reported in Figure 7
In general, PVAm Tg
decreases when purification is improved, a trend that is clearly related with the salt content, which seems to make the overall system more rigid. For similar reasons, the presence of nanofillers tends to increase the Tg
in a different way, depending on the material considered. The determined glass transition temperatures are shown in Figure 7
and summarized in Table 1
More specifically, in the case of Lupamin9095, the very high salt content creates, upon film deposition, macroscopic heterogeneous agglomerates, which are clearly visible to the naked eye. This material therefore presents a two-phase system, in which the Tg is associated to the polymeric phase.
Lightly purified PVAm, on the other hand, showed very small crystals with micrometric dimensions (see SEM analysis below), which seem to act as a nanometric filler able to efficiently rigidify and reinforce the matrix through polar interactions with the polymeric chains. This hypothesis can explain the higher Tg with respect to the PVAm-HG and the very small effect of GO addition, which indeed causes only a 3 °C increase on the Tg in PVAm-LG + 3% GO with respect to PVAm-LG. The presence of micrometric sodium formate inclusions in low purified Lupamin indeed can interact and attract the polar GO inside to the compatible ionic structure of such salt, thus keeping unmodified the macromolecular rigidity.
On the other hand, 3% in weight addition of both G and GO to PVAm-HG determined a sharp increase of Tg in High Grade films (+20 °C), as in this case no rigid inclusions are present in the highly purified polymer, thus the nanofiller can properly act as reinforcement, hindering macromolecular motion.
3.1.3. SEM Results
Through SEM analysis, the fractured surfaces of produced films were examined to investigate the dispersion and the interaction of G and GO with the polymeric matrix.
As shown in Figure 8
, the fracture surface of the PVAm-HG film (Figure 8
c) is much smoother than the PVAm-LG (Figure 8
a), most likely due to the presence of a considerable amount of sodium formate in the low-grade film, still visible also in PVAm-LG + 3%GO film (Figure 8
b). The addition of G and GO in the high-grade film (Figure 8
d,e), on the other hand, does not appears to modify the structure of the materials, even though the presence of a slightly higher heterogeneity could be argued, possibly due to the nano-reinforcement presence.
Differences in smoothness between the fracture surfaces may be considered as further proof of the achieved purification: the most purified materials presented fewer salt impurities on the surface.
3.1.4. Water Sorption Results
shows the water sorption isotherms obtained for the various materials at a temperature of 35 °C.
In the chart, it can be seen how Lupamin 9095, the commercial form, can absorb the largest amount of water, reaching 90 wt% of water mass gained at 60% water activity, as also shown in a previous work [40
]. It is followed by PVAm-LG (60–70 wt% increase at 75–80% water activity), and PVAm-HG (10–40% mass gained at 80–90% water activity). As expected, the removal of a hygroscopic phase such as the saline one, largely reduced the water uptake and the matrix swelling.
The addition of a graphene-based phase presents different effects on the water sorption of the different materials. Indeed, the reduction of water uptake upon filler loading is negligible for PVAm-LG composites, while it results significantly higher for PVAm-HG based materials.
In the latter case, even at relatively small amounts, the graphene-based nanofillers significantly decreases the water uptake of PVAm-HG based composites with respect to that of the non-loaded corresponding polymer. For example, pure PVAm-HG presents an uptake of 21 wt% at around 60% activity, while the addition of 3 wt% graphene and 3 wt% graphene oxide lower that value to, respectively, 15.1 and 7.8 wt%. This decrement can be interpreted as a positive interaction between the matrix and the filler, which is capable of stiffening the overall material structure, preventing an excessive swelling at high relative humidity. This result is more evident in the case of GO rather than G, likely due to the larger aspect ratio of the sheets and to the possible presence of electrostatic interaction among the carboxylic groups of GO and the primary amine groups of PVAm.
Concerning the PVAm-LG + 3% GO behavior, it can be noted instead that it is in accordance with that discussed above regarding the materials glass transition temperature variation: the limited effect of the filler seems indeed to be related to the presence of finer dispersion of sodium formate crystals, which somewhat interact with the GO decreasing its ability to impact the properties of the polymeric matrix, and also to the cationic nature of low-grade samples, which impairs the possible interactions among the amine groups of the polymer and the GO’s carboxylic acid groups.
3.2. Permeation Results
The single gas permeation results for carbon dioxide for the different membranes tested are presented in Figure 10
as a function of relative humidity and for further clarity in Table 2
, where CO2
selectivity are also reported, when available. As quite common for hydrophilic membranes, permeability increased exponentially with the degree of humidification [67
] due to the high water sorption and consecutive membrane swelling.
Apart from the general behavior and considering the different materials, as a first observation, it can be noted that for PVAm-HG a single point at intermediate humidity was obtained. This is due to the fact that PVAm in its pure form appeared to be quite unstable as a self-standing film, especially when purified. This led to the impossibility of running tests at higher humidity, due to the rupture of the film, verified on several specimens. For the case of PVAm-LG, the presence of residual sodium formate appeared to slightly increase the mechanical stability, allowing a full permeation curve to be acquired, at least for CO2; in the case of nitrogen tests, no reliable data could be obtained, likely due to the longer experimental time required, which resulted in an excessive stress on the film. On the other hand, all films prepared using a graphene-based filler resulted mechanically stable and allowed extensive tests to be performed without any loss in permselective properties for several days.
From the experimental data, it can be seen how PVAm-HG presents a permeability of 4.2 Barrer at 56% RH, but no particular trend can be obviously inferred. For PVAm-LG, permeability of CO2 varies from 16.5 Barrer at 63% RH, up to 73.8 Barrer at 93% RH. For the same polymer, filled with 3 wt% of graphene oxide, values range from 1.7 Barrer at 53% RH to 71.0 Barrer at high humidity. PVAm-HG, loaded with the same quantity of graphene oxide, presented instead a permeability for carbon dioxide of 1.6 and 25.1 Barrer at 75% and 93% RH, respectively. When loaded with few-layer graphene, the same matrix presented values from 2.0 Barrer at 77% RH to 23.1 Barrer at 92% RH.
The non-loaded materials therefore present a higher permeability at the same relative humidity with respect to their counterparts containing a graphene-based nanophase. This is a somewhat expected result, since the platelets of graphene, due to their high aspect ratio, can easily increase the tortuosity of the diffusion pathway of the gases in the matrix. This effect appeared to be less pronounced in the case of PVAm-LG, possibly due to the polar interaction between graphene oxide and salt nanocrystal in these materials as already discussed by considering DSC and water sorption experimental results.
Regarding PVAm-HG, it is quite interesting to notice how no particular differences in gas permeability can be discerned between the materials obtained using two different fillers. This could be contradictory considering that GO usually presents a higher aspect ratio than few-layers graphene and that PVAm-HG + GO showed a lower water uptake than PVAm-HG + G. Usually, indeed, when no other factors are in play, a lower concentration of humidity in the matrix resulted in a lower CO2
permeability, as also shown by previous works [26
]. The fact that the two composites show comparable permeabilities therefore suggests the existence of a different structure and possibly of higher interaction of carbon dioxide with the polarly charged GO, which increase the intrinsic materials permeability thus compensating negative factor mentioned above.
As previously mentioned, nitrogen permeation tests were only possible on the nanocomposite membranes, due to their enhanced stability. For this reason, the ideal selectivity could be calculated only for these materials, using an exponential interpolation to report data from different gases at the same RH values. Figure 11
presents the results from this calculation in a Robeson plot. It can be seen how for all materials a monotonous trend of selectivity versus permeability is followed, with both quantities increasing with relative humidity, indicated next to each experimental point for reader convenience. These behaviors were observed in previous studies [32
] and were somewhat expected because water favors both CO2
solubilization and facilitated transport execution, as previously shown in Equations (1)–(5) and Figure 1
, thus increasing its permeability more than that of nitrogen.
In the case of PVAm-LG+GO in particular, selectivity was shown to vary from 3.1 to 59.2 when humidity was raised from 60% to 95%. For PVAm-HG, instead, this value ranges from 3.0 at 75% RH to 80.7 at 95% RH for the graphene oxide sample and from 1.1 at 82% RH to 45.2 at 92% RH for the graphene one.
Analyzing PVAm-LG and -HG reinforced with graphene oxide, therefore, it resulted that the low-grade polymer has higher maximum CO2
permeability (≈70 Barrer), with lower CO2
selectivity (≈60) compared to the high-grade one, which has higher selectivity (≈80) and lower permeability (≈35 Barrer). These differences are probably due to the swelling difference of the two composite materials, as already discussed, and shows a PVAm-LG + 3% GO more hydrophilic than the PVAm-HG-based composite. These results fit also with the chemical structure differences between the two materials: PVAm-HG has an almost totally neutral charge upon the polymeric chain, leading to a tighter packing of the chain itself, which cause the overall lower permeability, while the PVAm-LG material has much higher ionic repulsion [56
]. Nevertheless, the higher neutral secondary amine group concentration in PVAm-HG promotes the facilitated transport mechanism shown in Figure 1
and Equations (1)–(5), which results in higher CO2
Despite these differences, the two materials show a very similar behavior in terms of overall permselective performances as they lie very close, but are not able to overcome the CO2
Robeson’s upper bound [16
], even at the highest humidity inspected.
From this point of view, therefore, contrary to what expected, the further purification step seems not to give substantial advantages in terms of membrane performance.
Among the many papers found in the literature about PVAm based system, it is very difficult to find references for a consistent comparison of the present results in order to give a more general picture of the overall potential of such membranes. Available data on FTM are usually measured at higher temperatures and/or different upstream pressures, and often refer to very complex systems also including mobile carriers. Among others, Zhao and coworkers, studying PVAm-based mixed-matrix membranes, tested PVAm thin films at 22 °C and 100% RH and reported CO2
permeabilities in the order of 50 Barrer and CO2
selectivity lower than 10 [68
]. On the other hand, 260 Barrer and a selectivity of 24 were found by Hamouda et al. when testing at the same temperature and pressure (1 bar and 25 °C) a polyetherimine/polyvinyl alcohol/polyethyleneglycol polymer blend [31
], at 20% RH; in the former case, lower selectivity was achieved with similar permeability, while, in the latter, higher permeability but lower selectivity was obtained with respect to the here considered membranes. A more interesting comparison in this concern is given by the data for a supported purified PVAm membranes reported by Kim et al. [43
as also reported in Figure 11
. In this case, values are generally higher than those obtained in the present work and confirm that graphene addition mainly affects polymer permeability, while it has a limited effect on selectivity.
This reduction of the separation performance with respect to the supported pure polymer from Kim et al. [43
however, is accompanied by a remarkable improvement of mechanical strength of the material: pristine PVAm sample indeed cannot be tested as pure polymer self-standing membrane and also literature data [25
] were obtained from thin supported films; the present composites instead were tested routinely with no problem for more than a month without showing any rupture or degradation.