Introduction

Electrospinning is a highly versatile and widely used technique employed in the preparation of micro and nano polymer fibres with tuneable properties such as high porosity and high specific surface area (Li and Yang 2020; Wable et al. 2021).

This technique consists of injecting the polymer solution through a syringe, applying an electric potential difference between the needle and the fibre collector, which can be stationary or rotating (Luraghi et al. 2021). The spinnability of the fibres can be modified varying the different technique parameters, such as the applied voltage, the distance between the needle tip and the collector, the polymer concentration in the solution and the polymer feed rate (Angel et al. 2020), which affect their final morphology and properties.

Moreover, the spinnability can be improved by adding low contents of additives, such as poly(ethylene oxide) (PEO) (Li and Yang 2020; Santamaria-Echart et al. 2017). Thus, taking into account the possibilities offered by electrospinning, the polymer mats fabricated by this technique show promising utility as biosensor-integrated drug delivery systems, and membrane-based systems for tissue engineering, among others (Cai et al. 2016; Yoon et al. 2009).

The preparation of hybrid inorganic/organic polymeric materials via the electrospinning technique is a fast-growing field of research (Angel et al. 2022; Rodríguez-Tobías et al. 2019; Vahabi et al. 2021) due to the possible applications they can offer, from piezoelectric materials (Chamankar et al. 2020; Mimura et al. 2010) to materials with improved antibacterial properties (Chong et al. 2022; Morselli et al. 2017).

Moreover, electrospun fibres with TiO2 nanoparticles (Gosal et al. 2019) were studied owing to their high thermal and chemical stability, and good optical properties (Zhang et al. 2018), by employing commercial titanium dioxide (TiO2) nanoparticles with different polymeric matrices such as poly(lactic acid) (PLA) (Toniatto et al. 2017), polycaprolactone (PCL) (Nandagopal et al. 2016), PEO (de la Cruz et al. 2016) and polystyrene (PS) (Neppalli et al. 2014), while also by employing the sol–gel synthesised TiO2 method with poly(vinyl alcohol) (PVA) (El-Aassar et al. 2016) and polyvinylpyrrolidone (PVP) (Passaro et al. 2022).

Attractive features of cellulose acetate (CA) such as biodegradability, low cost, high availability and good biocompatibility (Gorbacheva et al. 2021; Zhou et al. 2021) ensure its utilization as an electrospun matrix for different nanocomposites, especially for the development of membranes. In this context, Goetz et al. (2016) achieved CA electrospun mats impregnated with chitin nanocrystals (ChNC) for water filtration membranes with improved mechanical properties, lower biofouling and high hydrophilic character. Membranes for photodegradation of organic dyes [rhodamine B (RhB), methylene orange (MO) and methylene blue (MB)] were also prepared using recycled cellulose triacetate (rTAC) modified with commercial graphene and synthesised TiO2 and silver vanadate (Ag3VO4) nanoparticles (Motora et al. 2020). In the same way, Tian et al. (2015) and Nair and Mathew (2017) fabricated CA membranes with commercial hydroxyapatite (HAp) and cellulose nanocrystal (CNC) for Bovine serum albumin (BSA) and Victoria Blue adsorption, respectively. Besides their use as membranes, CA electrospun nanocomposites were also employed for biomedical applications after being modified with commercial ferric oxide (Fe2O3) nanoparticles (Tsioptsiasa et al. 2010), silk fibroin, silver (Ag) and gold (Au) nanoparticles (Arumugam et al. 2021), among others.

The improvement of the mechanical properties is essential for the development of competitive electrospun materials. In this regard, Wang et al. (2020) achieved significant improvement in the Young’s modulus (13%) and the tensile strength (11%) for CA-PAN (polyacrylonitrile) polymer blend with 33 wt% of CA, if compared with the mechanical properties of neat PAN mat. Similarly, Arrieta et al. (2020) improved by ~ 20% the elongation at break of PLA/PHB (poly(lactic acid)/polyhydroxybutyrate) blends adding 15% of oligomeric lactic acid (OLA), without affecting the Young’s modulus. These sustainable flexible blends are good candidates to be used in agricultural applications. In the case of nanocomposites, the addition of 1% of graphene nanoplatelets (GNP) drastically increased the tensile strength (229%) and elongation at break (131%) of chitosan/PVA blends (Bazzi et al. 2022). Furthermore, the introduction of 1.5% of graphene oxide (GO) increased the Young’s modulus and tensile strength of CA up to 75% and 73%, respectively (Aboamera et al. 2019).

In this study, the cellulose acetate based mats modified with poly(ethylene oxide-b-propylene oxide-b-ethylene oxide) triblock copolymer and sol–gel synthesised TiO2 nanoparticles were fabricated by the electrospinning technique. This investigation work was done to verify the hypothesis that the addition of EPE block copolymer increased the spinnability of cellulose acetate (CA). Moreover, the preparation of nanocomposite based on CA or CA/EPE with sol–gel synthesized TiO2 nanoparticles by using the electrospinning process was proved with the aim to enhance the mechanical properties of CA membranes. The thermal and morphological properties of the prepared mats were deeply studied. Moreover, the wettability was investigated by the water contact angle, and mats were immersed in different solvent to study their solvent resistance. In addition, the self-cleaning properties of the mats were also analysed.

Experimental section

Materials

Cellulose acetate (CA, Mn 50,000 g mol−1, DS = 2.5), poly(ethylene oxide-b-propylene oxide-b-ethylene oxide) triblock copolymer (PEO-b-PPO-b-PEO or EPE, Mn 12,600 g mol−1, 70 wt% PEO) and titanium (IV) isopropoxide (Ti[OCH(CH3)2]4) (TTIP) were supplied by Sigma Aldrich. Toluene and hydrochloric acid (37%) were purchased from Scharlau and isopropanol (IPA) from Panreac. All the reagents were the analytical grade and were used as received.

Preparation of mats

Solutions of CA and EPE/CA (20:80 weight ratio) were prepared in acetone, maintaining a total concentration of 12.5 wt%. On the other hand, IPA, toluene and TTIP were mixed (7:7:1 volume ratio), acidifying the solution with hydrochloric acid (approx. pH 2), to prepare the titanium sol–gel solution. This mixture was stirred for 1 h. After that, 40 vol% sol–gel solution was incorporated to CA or EPE/CA solution to obtain TiO2/CA or TiO2-EPE/CA solutions, respectively, and finally were kept stirring 1 h more.

FLUIDNATEK LE-10 electrospinning machine from Bionicia was employed to spin prepared solutions at ambient conditions, using a blunt needle tip (Model: 21G Sterican, B Braun) of 0.8 mm diameter and 22 mm length. Feed rate was kept constant at 4 mL h−1 and the distance between the needle tip and the stationary collector was fixed at 7 cm. The voltage applied on the needle was equal to 9 kV, while the stationary collector voltage was − 0.8 kV. Fabricated mats were collected on the surface of aluminium foil (20 cm × 20 cm) and electrospinning was carried out during 5 h for CA and EPE/CA, and 8 h for TiO2/CA and TiO2-EPE/CA, in order to ensure a similar weight (~ 2 g) of the prepared CA and EPE/CA mats. The thickness of developed mats was ~ 0.15 mm for CA and EPE/CA mats and ~ 0.50 mm for TiO2/CA or TiO2-EPE/CA mats.

Characterisation

Mettler Toledo DSC3+ was employed to carry out differential scanning calorimetry (DSC) measurements. Sealed aluminium pans containing a mat weight of 2–3 mg were used. Scans were performed at a heating rate of 20 °C min−1 and under N2 atmosphere. Pans were stabilized at 0 °C for 10 min, then heating to 250 °C, followed by a cooling to 0 °C and subsequent stabilized for 10 min, and finally heating to 250 °C. The second heating was plotted in order to avoid the thermal history of investigated mats. The degree of crystallinity of CA phase and PEO block was calculated with the following equation, with the value of the melting enthalpy of 100% crystalline CA (Cerqueira et al. 2006) and PEO (Zheng et al. 2003).

$${\text{X}}_{{{\mathrm{c}},{\mathrm{i}}}} = \frac{{\Delta {\text{H}}_{{{\mathrm{m}},{\mathrm{i}}}} }}{{{\text{w}}_{{\mathrm{i}}} \cdot\Delta {\text{H}}_{{{\mathrm{m}},{\mathrm{i}}}}^{^\circ } }}\cdot100$$

where Xc,i is the degree of crystallinity of component i, \(\Delta {\text{H}}_{{\text{m ,i}}}\) is the melting enthalpy of component i, \({\text{w}}_{\mathrm{i}}\) is the weight fraction of component i and \(\Delta {\text{H}}_{{\mathrm{m}}}^{^\circ }\) is the melting enthalpy of 100% crystalline component i.

Thermogravimetric analysis (TGA) were performed in a Mettler Toledo TGA/DSC3+ equipment, from 25 to 900 °C under air atmosphere at a 10 °C min−1 heating rate, with a sample weight of ~ 3 mg.

Scanning electron microscope (SEM) images (surface and cross-section) were collected using a Hitachi S-4800. Fibre mats were analysed with an accelerating voltage of 5 kV. Investigated mats were cut using liquid nitrogen. The ImageJ software programme was used in order to measure the size of the fibres and the pores of each fabricated mat. The average diameter of each entity was calculated by analysing around 20–25 randomly selected fibres or pores using for this purpose 3 different SEM images.

Mechanical test specimens were cut with a dog-bone shape with a pneumatic sample cutting press (Metrotec) according to ASTM D 1708-02 standard and conditioned for 48 h under vacuum prior to testing. The mechanical properties were evaluated at room temperature on an Instron 5967 testing machine, with a load cell of 50 N and a strain rate of 3 mm min−1. The length between grips was set at 18 mm. The Young’s modulus was calculated as an average out of five test specimen data.

The water contact angle (WCA) was studied with DataPhysics contact angle system (Model OCA 20), repeating the measurements six times for each mat.

Self-cleaning experiments were carried out fixing a square piece of developed mat in a glass plate, placing it with a 13° inclination, following the procedure describe elsewhere (Wang et al. 2022a, 2022b). Pink chalk powder was deposited on the surface of each investigated mat and removed with 1 mL of water. The surface of each mat was studied by optical microscopy (OM), Nikon Eclipse 80i in transmission mode, before and after the self-cleaning process.

Solvent resistance of the fabricated mats was also studied immersing in distilled water, IPA, acetone and toluene for 7 days, weighting them before and after that time.

Mats were also characterized by a colorimeter and transmission electron microscopy (TEM). Details are given in the Supplementary Material.

Results and discussion

Appearance of fabricated mats

CA and EPE/CA mats displayed very similar appearance and colour, without any significant differences (Fig. 1). Both mats presented L*, a* and b* colour parameters according to their white colour (Table S1 of Supplementary Material). The addition of sol–gel synthesised TiO2 nanoparticles into these mats resulted in slight increase of the L* value and as a consequence of ΔE*, which is attributed to the bleaching effect of TiO2 nanoparticles (Cao et al. 2022; El-Aassar et al. 2016).

Fig. 1
figure 1

Digital images of fabricated mats (2 cm × 2 cm)

Full size image

Deeper analysis of the investigated mats allowed to distinguish a micrometric multilayer structure for all of them, especially visible for TiO2/CA and TiO2-EPE/CA mats, as clearly shown in the digital images and cross-section SEM images in Fig. S1 of the Supplementary Material. The electrospinning process and the employed conditions led to mats being organised into successive micrometric layers, which can be removed individually from each fabricated mat and the incorporation of TiO2 nanoparticles promoted this structure. The thickness of each individual layer was ~ 0.04 mm for all the mats.

Thermal behaviour and stability

Figure 2 shows DSC thermograms and TGA curves of fabricated mats. CA mat showed a Tg and Tm at 193 and 228 °C, respectively, and a degree of crystallinity equal to 2.6% (Table S2). After the addition of EPE triblock copolymer, the Tg decreased to 187 °C and the Tm to 223 °C, indicating the partial miscibility of PEO block of EPE triblock copolymer and CA through the formation of intermolecular hydrogen bonds between the hydroxyl groups of CA and the ether oxygen of PEO blocks (Gutierrez et al. 2017). Moreover, the Tm of PEO block appeared at 53 °C, with a degree of crystallinity of ~ 13%.

Fig. 2
figure 2

A DSC thermograms and B TGA curves of fabricated mats (zoom in the range of 100–300 °C is shown as inset)

Full size image

For TiO2/CA mat, the Tg of CA phase shifted 5 °C to a lower temperature if compared to the Tg of CA mat. Furthermore, the melting enthalpy was difficult to calculate, which was indicative of the low degree of crystallinity of CA phase of TiO2/CA mat. Therefore, the CA was compatible with the TiO2 nanoparticles, most likely due to the formation of hydrogen bonds between the hydroxyl groups of sol–gel nanoparticles and the ether or carbonyl group of CA. The possible hydrogen bond formation was study by FTIR. FTIR spectra of fabricated mats are showed in Fig. S2 of Supplementary Material. The addition of sol–gel synthesised TiO2 nanoparticles into CA and EPE/CA mats resulted in a slight displacement of –OH stretching band region to lower wavenumbers and peaks became broader confirming hydrogen bond formation.

For TiO2-EPE/CA mat, the Tg of CA phase appeared at 14 °C, which is a lower temperature if compared with the Tg of CA mat and 8 °C if compared with EPE/CA mat.

This behaviour confirmed that incorporated sol–gel synthesised TiO2 nanoparticles were compatible not only with CA but also with PEO block of EPE triblock copolymer. The Tm of PEO block remained at 53 °C, however the degree of crystallinity decreased to ~ 11% if compared with EPE/CA mat, which was in accordance with the above mentioned compatibility.

As for the thermal stability, CA mat showed a three-step thermal degradation process (Fig. 2B) (Gutierrez et al. 2013). The first one, a minor decrease ~ 100 °C, was ascribed to the evaporation of the water adsorbed from the environment. The second (270–420 °C) and the third (440–560 °C) weight loss corresponded to the degradation of CA (mainly dehydration and decarboxylation reactions) and the degradation of generated char (Gaan et al. 2009; Gutierrez et al. 2012). For EPE/CA mat, the second weight loss started at lower temperatures if compared with CA mat, at around 220 °C. Moreover, a weight loss around 180 °C was detected, corresponding to the degradation of EPE triblock copolymer (Gomez-Hermoso-de-Mendoza et al. 2019).

In the case of TiO2/CA and TiO2-EPE/CA mats, the second and third degradation steps occurred at slightly lower temperatures if compared with CA mat (220–360 °C for second and 370–470 °C for third one), indicating the considerable lower thermal stability of mats modified with sol–gel synthesised TiO2 nanoparticles. This behaviour was also reported in the literature and can be attributed to the presence of traces of remaining water, isopropanol and the unreacted isopropoxide groups in formed sol–gel TiO2 nanoparticle network (Gutierrez et al. 2013; Zeng et al. 2010). For TiO2-EPE/CA mat, a slight decrease can be observed at 190 °C corresponding to the degradation of EPE triblock copolymer (Fig. 2B zoom inset).

Furthermore, the remaining final inorganic residue confirmed that the content of TiO2 nanoparticles was 10 wt% for both mats with this being indicative of the reproducibility of the preparation protocol.

Despite the differences observed in the degradation steps, it can be concluded that the addition of EPE and sol–gel synthesised TiO2 nanoparticles did not significantly affect the thermal stability of mats and can be used in the same temperature range as CA mat.

Morphology

Figure 3 shows the surface morphologies of CA based mats investigated by SEM. First of all, it should be mentioned that all investigated CA based mats presented uniformly distributed, smooth on the surface fibres clearly visualized in higher magnification SEM images.

Fig. 3
figure 3

SEM images of fabricated mats at different magnifications: A CA, B EPE/CA, C TiO2/CA and D TiO2-EPE/CA

Full size image

These higher magnification SEM images affirmed also the lack of both the macrophase separation of PPO block of EPE triblock copolymer for EPE/CA mat and the sol–gel synthesised TiO2 nanoparticles agglomerations for TiO2/CA and TiO2-EPE/CA mats. The fibres of CA mat (Fig. 3A) were randomly oriented with a diameter size of 5.2 ± 1.3 µm (Fig. S3). This mat consisted of homogeneously distributed fibres with very few visible beads, and with highly smooth surfaces. Moreover, the pore size was 2.4 ± 1.1 µm for this mat (Fig. S3). EPE/CA mat had very similar distribution (Fig. 3B) and fibres diameter (3.8 ± 1.7 µm), as well as similar pore size (3.2 ± 1.3 µm). However, here it should be pointed out that a better spinnability was achieved for this mat if compared with CA.

This is ascribed to the addition of EPE triblock copolymer with high PEO block content since, as is well known from the literature (Li and Yang 2020; Santamaria-Echart et al. 2017), PEO homopolymer can enable the electrospinning process.

In the case of TiO2/CA and TiO2-EPE/CA mats, the fibres presented higher stretching and as a consequence lower diameter if compared with fibres of CA and EPE/CA mats. This is due to the addition of sol–gel solution, which provoked a dilution of CA or EPE/CA solutions, which decrease the amount of electrospun mass per minute. The diameter of the fibres was 1.8 ± 0.6 µm for TiO2/CA and 1.4 ± 0.4 µm for TiO2-EPE/CA, indicating higher fibres diameter uniformity of mats without sol–gel synthesised TiO2 nanoparticles.

Moreover, the smaller diameter of the fibres led to a larger number of fibres per unit area, therefore the density of these mats seems to be higher than those without TiO2 nanoparticles. As a consequence, the pore size was lower for mats with sol–gel synthesised TiO2 nanoparticles (1.2 ± 0.4 µm for TiO2/CA and 0.9 ± 0.4 µm for TiO2-EPE/CA) if compared with CA and EPE/CA mats. As for the bead formation, in both cases some small beads were detected, with a size of 0.7 ± 0.2 µm. However, these beads did not affect mat formation and fibre morphology. As in case of EPE/CA mat, the electrospinning process was more efficient for TiO2-EPE/CA mat than for TiO2/CA as a consequence of the addition of PEO block of EPE triblock copolymer.

In order to study more deeply the morphology of fabricated mats, TEM measurement were performed. CA and EPE/CA fibres (Fig. S4A and B of Supplementary Material) displayed smooth surfaces whereas for EPE/CA some black circular (~ 14 nm) and worm-like domains (~ 50 nm) were detected (marked in white in Fig. S4B). These microseparated domains could correspond to PPO block phase separated from the PEO block/CA phase (Gomez-Hermoso-de-Mendoza et al. 2021). Only a few domains were detected on the analysed surface of the fibres due to the low amount of PPO block in the mat (~ 6 wt%).

On the other hand, the surface of fibres completely changed to mopped surface when sol–gel solution was added (Figure S3C and D). This can be easier distinguished if compared to TEM images of mats without and with sol–gel synthesised TiO2 nanoparticles (CA mat with TiO2/CA mat and EPE/CA mat with TiO2-EPE/CA mat). Small in diameter (~ 8 nm) and almost interconnected between each other, TiO2 nanoparticles were well-dispersed along the fabricated fibres. Moreover, the fibres maintained their smooth surface, in which no apparent modification of their surface was observed. For TiO2-EPE/CA mat some PPO block domains (white circles) were also detected as for EPE/CA mat, which indicated that the microphase separation of the PEO block of EPE triblock copolymer took place even in the presence of TiO2 nanoparticles.

Mechanical properties

The Young’s modulus of CA mat was ~ 7 MPa (Table 1), slightly higher if compared with the values of electrospun cellulose acetate mats previously published elsewhere (Ghorani et al. 2013; Sinha et al. 2020; Wang et al. 2020), with values between 1 and 3 MPa. The electrospinning conditions affects fibre diameter, mat porosity and bead formation, among others (Angel et al. 2020; Can-Herrera et al. 2021), which have a strong impact on the mechanical properties. Thus, the conditions employed in this work led to mats with improved stiffness. After the incorporation of EPE triblock copolymer, the Young’s modulus decreased to 1.5 MPa due to the plasticising effect of PEO block of EPE.

Table 1 Young’s modulus of fabricated mats
Full size table

TiO2/CA mat showed a drastic increase of the Young’s modulus (from 7 to 83 MPa, 12 times higher) if compared with CA mat that implies a significant increment in the stiffness of this mat. This phenomenon can be correlated with the reinforcement effect of sol–gel synthesised TiO2 nanoparticles (Zhang et al. 2022). It can be also appreciated in their handling, since mats with sol–gel synthesised TiO2 nanoparticles were stronger to touch. This increase was also high for TiO2-EPE/CA mat, with a value of ~ 78 MPa if compared to 1.5 MPa of EPE/CA mat, that is, 52 times higher. Therefore, TiO2-EPE/CA mat displayed very similar value of Young’s modulus than TiO2/CA mat despite the plasticising effect of PEO block. This behaviour can be related to a more homogeneous distribution of the nanoparticles along the fibre surface as concluded before from TEM measurement and achieved due to the self-assembly ability of the EPE triblock copolymer which is able to microphase separated from CA matrix during the electrospinning process.

Contact angle measurements

The digital images and WCA values are collected in Fig. S5. CA mat presented a WCA of 133° ± 7°, which is higher than those reported in the literature (Wang et al. 2015; Elsherbiny et al. 2022). Moreover, this WCA was 8% higher than that of CA mat prepared in different electrospinning conditions (feed rate of 7 mL/h, 7 kV and − 0.8 kV voltage and a distance of 12 cm), which displayed a 123° ± 5° value. This confirmed the strong influence of the electrospinning conditions on the designed mat properties. Taking this into account, in this work the electrospinning conditions that allow to achieve the highest hydrophobicity for CA mat were chosen for the preparation of all investigated mats.

The incorporation of sol–gel synthesised TiO2 nanoparticles led to an increase of the hydrophobicity, increasing by 5% the WCA for TiO2/CA mat (140° ± 6°), since this value was almost one for the superhydrophobic surfaces (very near to 150°) (Wang et al. 2022c). As previously reported, the hydrophobicity of surfaces is closely related to the surface energy and roughness of the material (Yang et al. 2021). As observed by SEM images, mats with sol–gel synthesised TiO2 nanoparticles possessed a low fibre diameter and pore size and with high fibre density, leading to a higher hydrophobicity.

As shown in Table S3, there were differences in the WCA for CA and TiO2/CA mats after the layer-by-layer peeling process. CA mat displayed a decrease of the WCA after peeling. For TiO2/CA mat, after the first layer was taken off, the WCA remained constant. However, consecutive peeling led to a slight decrease of the WCA. Every time that a peeled layer is removed, the mat loses density and an increase in pore size is more likely, which facilitates the wettability of the water droplet and resulted in a decrease of the WCA.

Due to the hydrophilic nature of PEO blocks in EPE/CA and TiO2-EPE/CA mats the water droplet was absorbed a priori the contact angle measurement.

Self-cleaning

The high hydrophobic character of CA and TiO2/CA mats can endow them with self-cleaning ability. For CA mat, the surface deteriorated after the water droplet was poured on it, and residual chalk was clearly observed on its surface (Fig. 4). Moreover, after peeling process, some pink colour remained on the surface of this mat. On the contrary, the water droplet removed the chalk contaminant from the surface of TiO2/CA mat, resulting in an almost clean and dry surface with only a few fibres impregnated with chalk powder. After the peeling process, the surface was completely clean.

Fig. 4
figure 4

OM micrograph and digital images (inset of each image) of fabricated mats before self-cleaning, after self-cleaning and after peeling

Full size image

Although the contact angle of EPE/CA and TiO2-EPE/CA mats was not measured, a self-cleaning test attempt was carried out. EPE/CA mat did not present any self-cleaning ability, since the mat absorbed the water droplet and was not able to remove the chalk powder (Fig. S6). Additionally, peeling was more complicated. Therefore, this mat did not possess self-cleaning properties. Nevertheless, the addition of TiO2 nanoparticles provided self-cleaning ability to TiO2-EPE/CA mat. After pouring the water, the surface did not deteriorate as for CA mat, and although not all of the chalk powder could be removed as for TiO2/CA mat, some of it was erased. Moreover, after peeling process, the surface was clean, but still somewhat wet due to the absorption of water by the EPE triblock copolymer with high PEO block content.

Here it can be concluded that the addition of sol–gel synthesised TiO2 nanoparticles improved the hydrophobicity of mats regarding waterproof materials. This phenomenon is probably related to both the well-known self-cleaning properties of the TiO2 nanoparticles and the change in the morphology (different fibre diameter and pore size and, as a consequence, the density of the mats) as observed by SEM and TEM images.

Solvent resistance

The solvent resistance of the fabricated mats was studied after immersion during 7 days in water, IPA, acetone and toluene (Fig. 5).

Fig. 5
figure 5

Digital images of fabricated mats in different solvents after 7 days

Full size image

CA mat was deteriorated in water and some small fibres were detected, slightly clouding the solution. The water molecules, due to their high dipole moment (1.87) and polarity index (10.2), are very prone to form hydrogen bonds.

Consequently, they can weaken to some extent the intermolecular hydrogen bonds of CA itself for CA mat, resulting in the detaching of small fibres from the main mat. However, the mat did not dissolve because, as is well known, the CA molecules are strongly bound by a large number of hydrogen bonds that prevent the penetration of water (Alves et al. 2016; Rincón-Iglesias et al. 2019). Thus, the weight loss observed for CA mat (Table S4) was due to the separation of these fibres. On the contrary, CA mat integrity was better maintained in IPA and not as many fibres were seen in the solution. This minor difference can be ascribed to a lower dipole moment (1.66) and polarity index (3.9) of IPA if compared with water, which set the lower hydrogen bonding capacity of IPA. Additionally, as it is well known, CA mat was soluble in acetone. In the case of toluene, it turned transparent when immersed in the solvent, recovering its initial white colour after drying. This phenomenon is explained by taking into account the refractive index of CA mat and solvents. CA presented a value of 1.47 (Guo et al. 2021; Sultanova et al. 2009), whereas the solvents have a value of 1.33 for water, 1.38 for IPA, 1.36 for acetone and 1.49 for toluene (Akagi et al. 2012; Volk et al. 2005). Therefore, only toluene has a value similar to CA, which leads to similar refraction of light in both mediums resulting in the high transparency of CA mat in toluene.  It is worth nothing that, the weight loss was very low, indicating the high stability of the CA mat in this solvent with low dipole moment (0.31) and polarity index (2.4).

As for CA, EPE/CA mat fibres were released after soaking in water and IPA, and they were completely dissolved in acetone. EPE/CA mat was also transparent in toluene since the refractive index of EPE was ~ 1.46. Moreover, it displayed higher weight loss than CA mat, probably due to the solubility of the EPE triblock copolymer in this solvent. Nevertheless, the TiO2 nanoparticles provoked changes in the solvent resistance of the mats. In water, as well as in IPA and toluene, TiO2/CA mat retained its mat integrated structure during all 7 days without any detaching, as well as preserving its characteristic white colour in both water and IPA. In toluene, the mat was transparent when immersed in the solvent, however, some opaque domains were distinguished. Those areas could correspond to TiO2 nanoparticles, which have higher refractive index (2.5–2.9) than EPE and CA (Kischkat et al. 2021; Yovcheva et al. 2012). After drying it recovered its initial white colour, similar to CA and EPE/CA mats.  Besides, the TiO2/CA mat did not show weight loss in any of the three solvents. In the case of acetone, a 20% weight loss was observed, being this value very low if compared with the 100% weight loss for CA mat. This can be related to the fact that in this solution, CA fibres were coated by TiO2 nanoparticles.

Thus, the addition of sol–gel synthesised TiO2 nanoparticles prevented CA from being dissolved in acetone and enhanced the mat integrity in polar solvents such as water by preventing the formation of hydrogen bonds between the solvent and the CA fibres. Obtained solubility results indicated that TiO2 nanoparticles were well-attached to CA fibres, since they did not separate from the mat in the presence of solvent, contrary to neat sol–gel solution, in which the nanoparticles precipitated in it.

Similar behaviour was observed for TiO2-EPE/CA mat, which stability was very similar in IPA and the weight loss in acetone and toluene was lower if compared with EPE/CA. Consequently, it can be concluded that there was a good compatibility between the components of this mat, as detected from DSC, and that sol–gel synthesised TiO2 nanoparticles improved the solvent resistance of investigated mats.

Conclusions

Mats based on cellulose acetate (CA), poly (ethylene oxide-b-propylene oxide-b-ethylene oxide (EPE) and sol–gel synthesised TiO2 nanoparticles were fabricated by the electrospinning process, being this process more effective for mats with EPE due to the higher spinnability of PEO block. The addition of 10 wt% TiO2 nanoparticles led to mats with a whiter colour, as well as an easily detected micrometric multilayer structure and, as a consequence, a better layer-by-layer peeling ability. DSC results confirmed a partial miscibility of the components of the mats. High magnification SEM images confirmed the lack of macrophase separation of PPO block of EPE triblock copolymer in EPE/CA mat and no agglomeration of TiO2 nanoparticles in TiO2/CA and TiO2-EPE/CA mats. SEM results also indicated a decrease in the fibre diameter and pore size when TiO2 nanoparticles were added. Therefore, this led to increased mat hydrophobicity and improved their self-cleaning ability. These properties were better for TiO2/CA than for TiO2-EPE/CA, due to the absorption of water by the EPE triblock copolymer. On the other hand, TEM results proved that EPE triblock copolymer self-assembled during electrospinning process leading to microphase separation of PPO block of EPE triblock copolymer from the PEO block/CA matrix and formation of TiO2 nanoparticles with the size of ~ 8 nm in diameter. The addition of TiO2 nanoparticles also improved the mechanical properties, being the Young’s modulus of TiO2/CA and TiO2-EPE/CA mats higher than CA and EPE/CA mats. In addition, the Young's modulus of TiO2-EPE/CA mat was similar to that for TiO2/CA mat despite the plasticising effect of PEO block. Thus, the nanostructuration of the fibres by self-assembly of EPE triblock copolymer can be responsible for a better distribution of TiO2 nanoparticles. At the same time, the addition of sol–gel synthesised TiO2 nanoparticles resulted in the improvement of the solvent resistance of these mats against water, IPA, acetone and toluene. In summary, the addition of EPE triblock copolymer allowed to achieve nanostructured mats even in presence of TiO2 nanoparticles. Furthermore, the presence of TiO2 nanoparticles resulted in significant improvement of Young’s modulus, solvent stability and self-cleaning properties with a very easy layer-by-layer peeling process, which make them ideal as waterproof and disposable dressings.