3.1. Characterization of Neat and Functionalized PES Membranes
As the first step of PES membrane modification, amine functionalization was performed. The common technique to immobilize amine groups on PES is through HNO
3 and H
2SO
4 mixture acid washing followed by a reduction reaction in Na
2S
2O
4 solution [
72,
73]. While the technique is properly capable of immobilizing amine functional groups on PES membranes, long reaction times, drying periods, and neutralization procedures make the approach less favorable for fast production. Aminolysis could be used as a substitute method of amine functionalization in only one step of membrane soaking [
41,
42]. Common aminolysis reaction conditions are reported as 10 wt.% of amine-containing monomer (DETA or dipropyltriamine (DPTA) aqueous solution), 48 h of reaction time, and at 90 °C [
41,
42]. We have assessed the initial reported conditions. We also assessed if amine functionalization could be performed at room temperature and within a shorter reaction time.
Figure 3 reflects the spectra for PES in its neat and amine-functionalized state. The characteristic peaks for the PES membrane are the weak asymmetric stretching vibration peak around 1340 and the symmetric vibration peak at 1134 cm
−1 [
38]. The sulfone characteristic ring could also be located in a different location (stretching vibrations of 1242 cm
−1 and 1124 cm
−1 for the PSF membrane [
74]). Aminolysis resulted in the formation of a weak peak around 1294 cm
−1, which represents the stretching vibration band of C-N, confirming that nitrogen is attached to the carbon backbone of the PES [
38]. Common IR amine peaks are reported to be between the wave numbers of 3200 and 3500 cm
−1 [
75]. Previous reports of aminolysis performed on PES also suggest the same range for the desired amine [
76]. Accordingly, the wide peak in this region reflects the successful amine functionalization of PES. After the functionalization of PES by the aminolysis reaction, the stretching peak for NH appeared near 1636 (secondary amine) and between 3300 and 3500 cm
−1 The two bands of NH appeared at 3390.53 and 3416.13 cm
−1, and the deformation band of NH
2 was at 1618.92 cm
−1 [
77]. The obvious peak between 2300 and 2400 cm
−1 represents the C-N bond, which approves the chemical attachment of amine functional groups [
78]. In addition, the peak that appeared between 2300 and 2400 cm
−1 represents the C-N bond, which approves the chemical attachment of amine functional groups [
78].
The effect of DETA concentration was studied at a fixed reaction time and temperature. As depicted in
Figure 3, the aminolysis reaction resulted in more pronounced peaks in the range of 3200 to 3500, which means more NH structures are created on PES membranes. DETA-assisted amine functionalization of other membranes has also been reported by Al-Shaedi et al. [
79]. Another IR peak is around 660–670 cm
−1. As presented in
Figure 3, the peaks in the announced range are also intensified at the 50% concentration of DETA. The C-N peak around 2350 cm
−1 is also intensified for the PES membrane functionalized with 50% DETA solution.
Figure 4 reflects the XPS of the PES membrane functionalized with 50% DETA solution. The peak near 400 eV reflects the existence of the nitrogen-containing structure in the functionalized membrane. Accordingly, aminolysis with moderate temperature and reaction time and a higher concentration of DETA monomers could successfully result in the amine functionalization of PES membranes.
After proving the aminolysis step, we assessed the functionalization of SB and CB, as shown in
Figure 3. ATR-FTIR assessment of the PES-CB membrane sample reveals the formation of the zwitterionic structure by the simultaneous presence of two peaks. The first peak near 1640 cm
−1 was attributed to quaternary nitrogen [
80]. The COO
− was confirmed by the peak at 1380 cm
−1 [
81].
The chemical footprint of the PES-SB reflects the comparison between PES-NH2 and PES-SB. Since PES has sulfone structure, the peak related to the S=O group exists at 1160 cm
−1. However, the small peak between 1030 and 1050 cm
−1 reflects the sulfonate functional group as the end moiety of the SB ZW structure [
82]. The peak near 1640 cm
−1 is attributed to quaternary nitrogen [
80]. The obvious peak between 2300 and 2400 cm
−1 is also mentioned to represent the C-N bond which approves the chemical attachment of amine functional groups [
78].
To further assess the final chemical structure of PES-CB, we made a high-resolution C1s XPS scan.
Figure 5a reflects the XPS spectra of PES-CB. The peaks around about 284.8 and 285.4 eV are attributed to C-H (as well as C-C) and C=O structures. The broad peak at 289.5 could be attributed to the O=C-O of the CB moiety of the zwitterionic structure [
34].
The key steps taken with ATR_FTIR analysis for the surface modification of PES with CB are locating amine sites on the PES using aminolysis proven by the amine and hydrogen bonding peaks around 1650 cm
−1 and after 2300 cm
−1, respectively, and locating the final negative carboxyl moiety proven by the peak at 1380 cm
−1. Similar assessment on the implication of carboxybetaine on porous substrates reflects a similar chemical footprint of symmetric stretching bond for COO
− from 1380 cm
−1 to 1454 cm
−1, along with overlap of the peaks with symmetric and asymmetric peaks of -CH
3 [
83].
The path for identifying successful implementation of SB on the membrane surface starts with locating the amine at 1650 cm
−1, and adding a sulfonate functional group at 1030 cm
−1 and 1050 cm
−1. The wave number is also in agreement with the location of the sulfonate of sulfobetaine methacrylate at 1033 cm
−1 [
83].
Figure 5b reflects the oxygen assessment of the PES-CB membranes. The appearance of oxygen peaks proves the CB-negative moiety. Zwitterionization of the PES membrane resulted in a broadened oxygen peak, which could be split in binding energies and could be divided into two extra split peaks rather than the C-O-H peak. The two peaks, 528.5 eV for the C-O bond and 531.5 eV for C=O, represent [O-C=O] and [O-C=O
−] structures in the CB [
84].
XPS analysis of the PES-SB membrane is presented in
Figure 6. The wide-angle scan of the membrane is shown in
Figure 6a. The peak at 167 eV (S 2p) represents the sulfur in the prepared membranes (
Figure 6b). The deformed S 2p peak for the PES-SB membrane is due to the addition of the SO
3− moiety of the ZW structure, which could be divided into three peaks similar to previously reported characteristics of SBMA [
85]. O 1s analysis of PES-SB also shows the peak of SO
3− at 532 eV (
Figure 6c) [
86].
Table 6 reflects the elemental analysis of the neat and modified membranes through XPS spectroscopy. The nitrogen percentage was less than 1% in the PES sample, which could be due to the trace amount of nitrogen trapped in the microporous structure of the membrane surface or to probable pollution on the surface. Aminolysis resulted in a higher amount of nitrogen on the membrane surface, from 0.92% to 4.26%. The PES membrane had 5.90% of sulfur. Surface coating of the amine and carboxyl functional groups resulted in the addition of a top layer with a diluted amount of sulfur in the case of PES-NH2 and PES-CB (5.21% and 4.90%, respectively). However, the sulfonate group located on the PES-NH2 membrane to form PES-SB resulted in a higher content of sulfur on the top surface of the membrane (6.61%). The highest carbon and oxygen contents (75.39% and 16.54%) on the modified membrane surfaces belong to PES-CB, which has the highest content of the carboxyl functional group.
3.2. Hydrogen Bonding Assessment
There are different approaches to identifying the stability of water molecules in a polymeric matrix. We have previously reported spectroscopy-based techniques such as IR-spectroscopy and NMR for qualitative and DSC for the quantitative assessment of the three types of water (namely, free water, intermediate water, and non-freezing water [
3]). The advantage of IR spectroscopy over other techniques is its accessibility and low cost. Previous research efforts over the IR-assisted water behavior study reflected two regions for water molecules. The broad peak at around 3200 cm
−1 is related to the strong hydrogen bonding of water molecules, while the broad peak around 3500 cm
−1 is of the less strong hydrogen bonds. In other words, stronger hydrogen bonds tend to have lower wave numbers [
87,
88]. It is important to note that the adjustment of the humidity as well as the amount of water exposed to the membrane material affects the shape of the peaks [
89]. More importantly, the experiment could be time-sensitive, depending on the membrane material used [
90].
Figure 7 reflects the ATR-FTIR spectra of the PES control sample (
Figure 7a), PES-CB (
Figure 7b), and PES-SB (
Figure 7c). The black spectra are for the dry samples, and the red ones are hydrated with the measured amount of water. For the PES sample, most of the hydrogen bonding is leaning toward the weaker region (3500 cm
−1). The PES-CB has most of the hydrogen bonds between (3000 and 3200 cm
−1). The black PES-SB spectra in dry mode have the same peak as the hydrated PES-CB, in the strong hydrogen bond region. This means that even in the vacuum chamber, the hydration layer, which is strongly associated with the surface, could not be removed. After adding water, the peak in the 3000 and 3200 cm
−1 became broader, which reflects a higher quantity of strong hydrogen bonding. On the other hand, weaker hydrogen bonds at higher frequencies were also increased.
3.3. Surface Roughness
Figure 8 reflects the scanned surface of the membrane using the AFM microscope and the roughness parameters.
Figure 8 reflects the values in nm for Sa and Sq.
Figure 8b,c,d reflect the scans for neat PES, modified PES-CB and PES-SB membranes, respectively. Both roughness parameters have significantly dropped after the surface modification.
The mean roughness significantly decreased from 52.61 nm for neat PES to 6.3 nm for CB (88% decrease) and 7.7 nm(85% decrease) for SB. RMS roughness declined from 68.16 nm for neat PES to 8.49 nm (87% decrease) for CB and 9.9 nm(85% decrease) for SB.
Figure 8d reflects the PES-SB membrane. As it could be seen visually (as well as quantitative data in
Figure 8a), CB addition to the PES membrane resulted in lower roughness. Depending on the nature of the modification layer’s chemical structure and the geometry of the material, the ZW top layer could fill the pores of the membrane to an extent and reduce the roughness [
91]. Modifications that reduce the roughness of the surface could result in a higher degree of hemocompatibility due to a lower hemolysis ratio [
92,
93]. According to our previous investigations, we have proved that lowering the roughness could enhance hemocompatibility [
69,
70].
Commonly, smoother surfaces (less rough, lower values for S
a and S
q) are known to act in favor of hemocompatibility. This is due to the fact that rough surfaces would rapture the blood cells, and this would ultimately result in a higher hemolysis factor (lower hemocompatibility). Another point to consider is that the bloodstream is loaded with more than 3700 types of proteins, which makes the membrane prone to fouling intensively. Accordingly, a less tough surface for the modified surface could result in lower plasma protein fouling. Provocation of the plasma proteins is the initial stage for their activation, which consequently results in complement, thrombogenesis and coagulation, inflammation, and leukocyte activation [
1,
2,
3].
3.4. Surface Charge Analysis
Surface charge is a crucial characteristic for interfacial interaction assessment, fouling, biocompatibility, and hemocompatibility. A common understanding of dialysis membranes is used to infer that more hydrophilic groups, such as carboxyl and hydroxyl functional groups, would result in better hemocompatibility [
1]. However, negative surface charges were reported to trigger the kallikrein/kinin system and blood coagulation cascade toward an incompatible membrane-blood interaction [
94]. The emergence of the third generation of dialysis membrane modifiers reflected that not all negatively charged surfaces are blood-friendly, and a better hemocompatible surface would be obtained by electroneutral surfaces [
1,
9]. Accordingly, an ideal modification to our PES membranes would have been an electroneutral CB. As reflected in
Figure 9a, the surface charge of the neat PES membrane was −6.92 mV. The addition of amine as the positive block of the zwitterion increased the membrane’s surface charge to a value near 0 mV. The addition of the carboxyl final structure to form the zwitterion decreased the charge to a value of −12.9 mV. The last membrane sample, named PES-CB-L, showed a loss of surface charge to a value between neat PES and PES-NH2. The increase of the surface charge after aminolysis and its drop to a negative value, besides the FTIR and XPS characterizations, reflect that we have successfully immobilized the CB structure on the surface. However, since the number of amino and carboxyl functional groups was not optimized, more negative functional groups were located on the surface. Keeping the membranes in deionized water resulted in the loss of loosely attached negative charges, and the final product, i.e., PES-CB-L, had a less charged surface.
Figure 9b reflects the same trend for the SB-modified PES membrane. Compared to PES-CB, PES-SB was more negatively charged. Moreover, the PES-SB-L lost more negative charge in comparison with PES-CB-L. Accordingly, SB-modified membranes experienced a wider range. Since PES-CB-L experienced a less intense charge loss, we conclude that CB has a more stable chemical structure on the PES membrane. While a similar trend for the two zwitterionic structures was expected, the difference in charge range could be due to the different nature of the chemicals used for the final negative block of the zwitterionic structure. The final negative charge is not favorable for an electroneutral ZW structure, as it was previously explained that negative charges trigger coagulation. However, it should be noted that the negative charge here is coupled with the positive moiety of quaternary ammonium within the structure of ZW. The water interaction of ZW moieties creates a more efficient hydration layer to protect blood proteins from reacting with the membrane. Accordingly, a more blood-compatible profile is expected.
3.11. Induction of Hemocompatibility Assessment and Its Stability
This section is assessing the hemocompatibility and the inflammatory biomarkers released in a patient’s uremic serum when interacting with newly coated membranes compared to the untreated one; in addition, we have evaluated the hemocompatibility of treated membranes after 10 days to evaluate the stability of the hemocompatibility improvement.
Figure 14 and
Table 13 reflect the clinical study results for hemocompatibility in terms of percentage change and the concentration in pg, respectively. Our measurements reflected a lower level of C5b-9 (71% less for PES-CB in comparison with PES). PES-SB resulted in a 44% decrease in the secreted concentration of C5b-9. Plasma terminal C complex C5b-9 complex is a stable and reliable marker of biocompatibility; measuring the concentration of this cytokine could reflect the hemocompatibility extent of the membrane regarding the complement cascade [
97]. The enhanced secretion of C5b-9 is due to the limited activation of the complement cascade. Another potential property of the ZW-modified membrane that will support a lower level of C5b-9 production is the stable hydration layer. Since a protected water layer is formed on the PES-CB membrane, the attachment of macromolecules as the initiation mechanism of C5b-9 is prohibited. The stability tests resulted in the growth of the C5b-9 factor to 2.5% for PES-CB-L in comparison with PES-CB and 25.5% for PES-SB-L in comparison with PES-SB.
Cytokines, such as interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), and IL-6, may induce an inflammatory state and are believed to play a significant role in dialysis-related morbidity [
98]. It has been proven that the basal expression of IL-6 and TNF-α is not a function of the membrane’s hemocompatibility, since they are more affected by the endotoxin content of the dialysate [
99]. Accordingly, the variations in the measured IL-6 inflammatory biomarker might not be considered a concern for the loss of hemocompatibility. The two cytokines which were negatively affected by the zwitterionization are IL-1 and IL-6. Commonly the trend of cytokine level change is similar when the blood touches an incompatible membrane [
100]. However, we do not see a similar trend between C5a and IL-1 in our measurements. The content of the IL-1a is increased in the PES-CB membrane (6.9% more in comparison with the control sample) while the PES-SB resulted in a 7.7% lower amount of the same interleukin. This could be due to the rich carboxyl and hydroxyl content of PES-CB (similar to the increased content of interleukins in blood-cellulose membrane contact [
101]). The PES-CB-L and PES-SB-L resulted in a 12% and 30% increase in the secretion content of IL-1a, respectively.
The trend for IL-1b was similar for both PES-CB and PES-SB samples (a 28% and 43% decrease, in comparison with the control sample). The stability test resulted in the loss of hemocompatibility for the same inflammatory factor to the extent of 41% and 81% for PES-CB-L and PES-SB-L in comparison with PES-CB and PES-SB, respectively. Neither of the PES zwitterionization approaches was effective in reducing the amount of IL-6. The content of IL-6 grew by 3.5% and 5.3% for PES-CB and PES-SB. After keeping the samples in deionized water, the secretion level of IL-6 reached back to 12.6 pg for both PES-CB-L and PES-SB-L (p < 0.01). By considering the content of IL-6 secretion, the stability test reflected the loss of hemocompatibility to 3.4% for PES-CB-L and 5.1% for PES-SB-L extents in comparison with PES-CB and PES-SB samples.
Proteins’ adsorption tendency to the dialysis membrane is mentioned to play a role in the release of inflammatory cytokines and vWF [
102,
103]. The exception to this general rule is C5a and IL-1b [
104]. Human C5a anaphylatoxin is a bioactive biomarker that has both spasmogenic and leukocyte-related characteristics. Various properties of C5a make it a critical component for the normal host defense mechanism of the human body. However, the increased level of the biomarker in dialysis sessions could promote the well-known complications of hemodialysis [
105]. It is proven that the generation of the complement anaphylatoxin C5a could lead to the expression of active tissue factor (TF) in ESRD. This will contribute to hemodialysis-induced thrombogenesis [
106]. The modified PES-CB membrane showed a higher C5a level (7.9% in comparison with the PES control sample). PES-SB membranes even resulted in a higher content of C5a in the blood (13.9%). Previous studies on PES membranes modified with polyvinyl pyrrolidone (PVP) reflect the same results. When the membrane is more negatively charged or when the membrane is rougher, more incompatible interactions are reported [
107]. The increased level of C5a for the zwitterionized membranes here could be due to more negatively charged surfaces. This agrees with our zeta potential measurements, as the PES-SB membrane was more negatively charged in comparison with PES-CB. The stability tests reflect that the content of C5a increased for PES-CB-L to 5.4% in comparison with PES-CB. PES-SB-L However, it had a decline in C5a content (4.7%) in comparison with PES-SB.
vWF is a glycoprotein taking part in hemostasis and an identifier for endothelial cell stimulation [
108]. Based on the literature, a pronounced inflammatory level of cytokines is mentioned to affect the increase of vWF [
109]. Our measurements reflected different patterns for C5a and vWF. We observed a lower concentration of vWF for the modified PES-CB (17% lower than PES) and PES-SB (43% lower than PES), while zwitterionization did not enhance the production of C5a, as discussed before. This does not agree with the exception made for C5a and vWF earlier [
104]. vWF is either created in endothelial structures and megakaryocytes or through the granules of platelets [
110]. Accordingly, vWF is a known predictor of cardiovascular shocks [
111] due to its “complement-thrombogenesis linking nature”. Since the level of C5a was not satisfactory for the PES-CB, the desired modified level of vWF could probably be due to the control of platelet activation. The different patterns of vWF and C5a for PES and PES-CB reflect that inflammation as a result of complement might not necessarily boost the level of vWF. The correlation between the adsorption of plasma proteins such as fibrinogen to the membrane and the amount of vWF has been mentioned elsewhere [
112]. The stability tests reflected an 8.4% and 109.4% increase in the secretion of vWF for PES-CB-L and PES-SB-L, in comparison with PES-CB and PES-SB, respectively.
The increase in the concentration level of properdin is linked to the prevalence of neutrophil leukocytes [
113]. An increased value was reported in the bloodstream in post-dialysis inflammatory biomarker measurements (the same trend for C5b-9 and C3a [
114,
115]). Accordingly, blood membrane interactions that lead to the degranulation of polymorphonuclear leukocytes could increase the level of properdin after the dialysis session [
116]. PES-CB resulted in a lower concentration of properdin (28% less in comparison with the control sample). The drop in the percentage of properdin was also observed for PES-SB (16% less in comparison with the control sample), which may be attributed to the modified PES-CB membrane provoking the neutrophils in the blood to a lower extent. The stability test reflects that the properdin level for PES-CB-L increased in comparison with PES-CB (5.3%), but the concentration of properdin was still lower in comparison with the PES control sample. PES-SB-L had a higher level of properdin in comparison with PES-SB as well (17.98%).
A comparison between the new approach of ZW implementation and the more known chemical structure of PVP could reflect on how zwitterions. PVP takes advantage of a heterocycle of carbon–nitrogen with double-bonded oxygen. The structure takes advantage of similarity to zwitterions with the two atoms and is negatively charged. Accordingly, the well-known hemocompatible agent is like the structures here. Using the same clinical framework of hemocompatibility assessment, Abdelrasoul et al. reported the extent of biomarker activations for the PES membranes dip-coated in PVP solution from 1 to 4 min [
107]. The modification resulted in improvements in C5b-9, IL-1β, IL-6, and Serpin. A higher coating time of PVP (3 and 4 min) led to improvements in C5a activation as well. Similar trends in terms of C5b-9 elimination were observed with the ZW modified membranes. Two out of the three interleukins were triggered by the ZW structures. IL-6 improved with PVP and not ZW structures, which means lower immune system provocation and inflammation could be expected from PVP coated membranes. In terms of complement control, PVP had better performance in comparison with ZW as it could reduce both C5a (which is a primary inflammatory mediator) and C5b-9 (which could result in cell lysis). ZW, on the other hand, could only lower C5b-9. The paper did not offer an analysis of other biomarkers studied in this research, i.e., properdin and vWF. A more generic comparison with more clinical aspects of hemocompatibility is required for a more comprehensive assessment.