3.2. Characterisation of Hydrogels
The polymer hydrogels were characterised with FTIR (Figure S3 in Supplementary Materials
). All spectra exhibit a characteristic peak at 1711 cm
assigned to the stretching vibration of C=O bonds of carboxylate groups of PAAc, and two absorbance bands at 1624 cm
and 1540 cm
attributed to the secondary amide stretching of PNIPAAm (aka amide I and II bonds, respectively). The broad bands in the range of 3100–3600 cm
are due to the N–H stretching, while a C–N stretching band appears at ∼1455 cm
, confirming the presence of amide groups.
The morphological features of the swollen freeze-dried hydrogels were evaluated via SEM studies. The micrographs shown in Figure 2
reveals that all gels exhibit a heterogeneous interconnected porous-like structure. During the freeze-drying process, the formation of ice crystals within the network upon immersing the samples into liquid nitrogen acts as a template for pore generation resulting in macropores structure [31
]. Therefore, this observed morphology is compact aggregates of the polymer chains rather than the mesh of hydrogels at their molecular level. Nevertheless, the macrostructure of freeze-dried hydrogels depends remarkably on the PAAc-TTC concentration. With higher PAAc content, the grafted hydrogel samples show larger internal pores due to the lower polymer fraction in the hydrogel that corresponds to higher water content. In contrast, the chain length of the PAAc appears to have a minor influence on the porous structure.
3.2.1. Swelling Properties
The swelling behaviour of the comb-type grafted PAAc gels was investigated in both brine and deionised (DI) water. It is well known that the swelling and osmotic pressure within the hydrogel increases with increasing the ionic component and thus the equilibrium-swelling ratio of the hydrogel will be increased. Figure 3
shows that indeed the equilibrium-swelling ratio of all synthesised hydrogels increases with increasing the PAAc content. The length of grafted chains also affects the swelling behaviour of the hydrogel. Therefore, the gels with longer PAAc chains show higher equilibrium-swelling ratios as shown in Figure 3
The strong hydration of the gels is related to the fact that longer chains are structurally separated from the backbone of the cross-linked network. The hydrogels with different PAAc-grafted chain lengths may have different cross-linking density. The effect of a change in crosslink density on the equilibrium-swelling ratios was not studied. No effect of the chain length on the equilibrium-swelling ratio in brine was found indicating that the contribution of charges to the osmotic pressure is the same in all systems. We therefore expect samples containing the same total amount of PAAc to show a similar salt depletion capacity.
The dynamic of the swelling process of the dry materials was also quantified via the half-swelling time (
), at which 50% water was taken up (Figure S4 in Supplementary Materials
). Generally, the swelling process can be described by three transition steps [33
]: the diffusion of water into the glassy polymer matrix, the transition of polymer phase from glassy state to rubbery state (due to polymer hydration), and the collective diffusion of polymer network toward the surrounding water. Figure 4
shows that the effect of both the graft chain amount and the length on the swelling kinetics was evident. The lower the content of PAAc in the hydrogel was, the faster was the water uptake. Gels with higher grafted PAAc content fill more voids in the hydrogel network and this leads to more compact structure. Hence, the transition of polymeric materials from glassy to rubbery state will be slower. The other factor, which may hinder the diffusion of water into the hydrogels, is the hydrogen bonding interactions between the grafted PAAc chains and the PNIPAAm backbone. In contrast, at a fixed amount of grafted PAAc chains, gels with a longer chain length undergoes faster hydration rate. This contradictory behaviour can be attributed to the faster collective diffusion of water in systems with longer polymer chain length upon the transition of the polymer phase from glass to rubbery state.
3.2.2. Rheological Investigation and Network Structure
The mechanical properties of the fully swollen hydrogels in deionised water were characterised by measuring the influence of the angular frequency on elastic modulus (
) and loss modulus (
) (Figure S5 in Supplementary Materials
). All gels exhibit a plateau, i.e., frequency independent response of the storage modulus in the range from 0.1 to 100 rad s
, which reveals that those gels indeed exhibit a soft rubbery-like behaviour. At higher frequencies, gels with a PAAc content of 40 wt % show a slight increase in
. This behaviour is most likely attributed to the more or less flexible grafted PAAc chains that exhibit longer relaxation times, i.e., they require lower applied frequencies to rearrange themselves on the time scale of the imposed mechanical motion resulting in a gradual rise of
However, the change in modulus arise from entropy changes in the rubbery range. Therefore, the change in toughness or stiffness of the hydrogel networks can be correlated with their composition. Figure 5
shows that there is indeed an increase of
with both the content and the length of the grafted PAAc chains. Gel networks with more flexible grafted PAAc chains exert a lower elastic force and thus a higher loss tangent due to the strong dependence on the water content of hydrogels at fully swollen state. As we stated regarding the swelling properties presented above, the graft chains’ mobility affects the gel swelling behaviour, namely the equilibrium swelling ratio. Therefore, a correlation between storage modulus and swelling ratio at equilibrium could be confirmed (Figure 6
). Gels with more water per unit volume exhibit a reduction of the elastic modulus.
The mesh size of hydrogels, as a measure of their microstructure, was estimated from the rheological data using Equation (9
). As expected, the mesh size increases with increasing both, the PAAc content and the length of the grafted PAAc chains (Figure 7
). This result is in agreement with the SEM observations and the swelling behaviour, where high PAAc content shows larger internal pores in SEM and more water in equilibrium-swelling compared with those with less PAAc content.
3.3. Dewatering Behaviour of Hydrogels and Salt Rejection
In order to get an insight into the dewatering of gel matrices, the type of water binding inside the fully swollen hydrogels was determined by means of DSC analysis. Generally, water inside hydrogel networks can be thermodynamically classified as one of three different states [34
]: (1) free water, which does not take part in hydrogen bonds with the polymer chains; this free interstitial water should be easily removed under high temperature conditions as it is physically entrapped within the polymer networks, and thus it behaves similarly to pure water as far as freezing and melting is concerned; (2) bound water, which directly binds to polymer chains via hydrogen bounding; this type of water is an integral part of the hydrogel structure and cannot easily be separated from it; thus it does not show any endothermic peak within the normal temperature range associated with pure water; and (3) semi-bound water, that exhibits weak interactions with polymeric chains within the network of hydrogels and freezes/melts at a temperature shifted with respect to that of free water. DSC thermograms of the fully swollen hydrogels in DI water (Figure 8
) display that the melting of water in all samples starts at temperatures lower than that of pure water (dashed line
). Generally, for all studied gels, the endothermic peaks are broad and structured.
The different fraction of freezable and non-freezable water for the studied hydrogels can be correlated with their swelling ratio at equilibrium, which is also associated with both grafted PAAc content and chain length. The grafted PAAc with freely mobile ends can hydrate sufficiently and contain a large amount of freezable water. Figure 9
shows that the fraction of freezable water increases gradually with increasing PAAc fraction in the hydrogel. A slight increase in the freezable water content was also observed for increasing grafted PAAc chain length. Accordingly, the different swelling ratios at equilibrium between the gels lead to a different freezable water content.
As we stated in the Introduction, the solution taken up by gels should not only be depleted from salt, but also can swiftly be released under thermal stimulus. Therefore, the phase transition behaviour of hydrogels was explored by measuring the changes of the swelling ratios at equilibrium as a function of external temperature in deionised water as shown in Figure 10
. All hydrogels displayed a sigmoid curve of swelling ratio versus temperature, where the gels swell at lower temperature and shrink above the LCST of PNIPAAm that occurred at around 32
C. This is because the hydrophobic interactions between the hydrophobic groups of the hydrogels become dominant and thus the gel matrixes shrinks. Optically, gels with 20 and 30 wt % PAAc changed their appearances from transparent at 15
C to opaque at 50
C while gels with 40 wt % PAAc kept their transparency. This may be attributed to the loose structure of the latter gels. Hence, the macropores of certain size still retain water even in a shrunken state, thus keeping the samples transparent. This would qualitatively also correlate with the SEM data for the frozen gels (cf. Figure 2
The LCST at which the volume phase transition of the gels start is obtained by a fit of a sigmoidal function to the data (Figure 10
). Gels with PAAc content of 20 wt % show a LCST at 32
C, while the phase transition temperature for gels with higher PAAc content increased up to 43
C. Moreover, the relative change of the swelling ratio or the percentage of released water in the temperature range of 30 to 35
C decreases and the transition itself becomes broader with increasing PAAc content and chain length. The highly hydrophilic nature of PAAc enhances the hydration of gels via a repulsive force between the strongly hydrated and charged carboxylate anions. However, the increased repulsion also restricts the hydrophobic network aggregation resulting in a distribution of LCST over a broader temperature range as the fraction of ionic moiety increases.
The water release from the polymer hydrogels (Figure 11
) was achieved via external heating and the percentage of water recovery was determined after 60 min of dewatering at a relatively low temperature (50
shows the percentage of water recovery from swollen gels with various grafted PAAc contents and chain length starting with powder after equilibrium loading in both deionised water (DI) and brine solution of 2 g L
NaCl. For all hydrogels, the water recovery decays dramatically as PAAc content increases. Hydrogels with 20 wt % of PAAc show the most powerful dewatering ability. The recovered liquid water fraction from the equilibrium swelling state was above 50%, while the water recovery for gels with 30 and 40 wt % of PAAc exhibit a reduction by a factor of 1.6 and 5, respectively. This is because the hydrophobic aggregation of the thermoresponsive moieties at the utilised temperature and on the time scale of the imposed dewatering process is restricted by the highly hydrated carboxylate groups that minimises the possibility of depletion processes.
In addition, gels with longer PAAc chain length have a slightly better dewatering performance, resulting probably from the increase in void volume within the polymer network upon the dehydration of PNIPAAm chains (in qualitative agreement with SEM data: cf. Figure 2
). This is also supported by the slight dependence of the freezable water content with increasing chain length (cf. Figure 9
), where the relative change is roughly the same. It is worth noting that faster dewatering responses were also observed for gels with longer grafted chains.
The salt rejection was measured via the ion concentrations of the salt solutions before swelling and after 60 min dewatering for each hydrogel. The rejection of salinity increases with increasing ionic moiety fraction as shown in Figure 13
. As we expected, no effect on the salt rejection has been observed with altering the PAAc-grafted chain length, since those hydrogels contain the same total fraction of ionic component. However, the salt rejection triples from 20 to 40 wt % PAAc content.
The opposing behaviour of gels to absorb water with less salinity and easily release it in the dewatering process is nicely illustrated when the water recovery is plotted versus the salt rejection (Figure 14
). Increasing the hydrophilic volume fraction, i.e., the ability of salt rejection counteracts the water recovery that is triggered via the hydrophobic volume fraction of the hydrogel. Therefore, to predict the optimum performance of such materials as a function of the hydrophilic or the hydrophobic content, a relative percentage of both the salt rejection and the desalination capacity (water recovery) was separately calculated according to the best working materials. The obtained data were evaluated in terms of the intersection of the two linear functions as shown in Figure 15
or as a sum function of the two relative performances (Figure S6 in Supplementary Materials
). Clearly find that the gel with 30 wt % of PAAc and longer chain length shows the best trade-off for such desalination experiments. It worth noting that, in the same experimental conditions, the water recovery of PNIPAAm gel does not exceed 70%. This is supported by the Donnan membrane theory according to which highly charged gels are expected to show the largest effect of salt depletion [35
]. However, even pure polyacrylic acid gels could not achieve a high salt rejection, and their performance decreases as the initial brine concentration increases [10
In this sense, we performed a consecutive swelling/deswelling experiment with the best trade-off sample (GG10k–30) to determine the number of consecutive steps required to achieve a complete salt rejection (Figure 16
). An initial salt concentration of 2000 ppm can be totally depleted only after four desalination steps. Every measurement step was carried out using new dry hydrogel material and the feed saline water was prepared with respect to the salt rejection measured in the previous step. After the optimisation of the working material, which is the only requirement for our proposed desalination to be controlled, the efficiency of the desalination process was also estimated. The efficiency was calculated from the fractions of water recovery and the salt concentration reduction in each step of the 4-stage process, for a total salt depletion from the initially 2 g L
brine solution. The value is found to be ∼0.4 kg hydrogel per square metre freshwater using the best trade-off hydrogel material. Since the salt rejection is a function of the respective hydrogel structure and composition, we measured it for the same type of hydrogel. In this context, salt concentration of the feed solution as a function of salt rejection was also determined using the same hydrogel material. As shown in Figure 17
, the influence of the salt concentration on the desalination process using a GG10k–30 hydrogel can be separated into two regimes. In the first regime (from 200 up to 750 ppm), the salt is almost totally rejected and the salt rejection is independent of the input concentration because the ionic content (the charge concentration of the polyelectrolyte) in the gel phase is higher than outside. In contrast, the salt rejection decreases exponentially in the second regime as the salt concentration in the feed solution exceeds a critical value of about 750 ppm of NaCl. However, this critical value is dependent on the effective charge density of the polyelectrolyte, which is correlated with the network structure on a given polymer volume fraction in the respective swollen state.
From an application point of view, it is also important to study the reversibility of the desalination process. We evaluated the reversibility of the swelling ratio and the salt rejection on temperature cycles between 20 and 50
C over 2 h for a GG10k–20 gel (Figure 18
). After each cycle, the salt depleted water was taken out from the mixture and the completely dried gel at ambient conditions was reused for the next cycle. It can be observed that both the dewatering level and the salt rejection start to decrease after the first two cycles, the total water content decreases slightly, whereas more water stays in upon heating.albeit the relative swelling ratio is nearly constant, i.e., dewatering is slower than swelling over the time scale. This may be attributed to the formation of denser or more stable hydrated polymer layers within the hydrogel network resulting from the reorientation of the grafted PAAc chains upon each cycle. The reduced percentage of the recovered water could be also caused by the delayed formation of hydrophobic nuclei during gel shrinking.