Exploring wettability difference-driven wetting by utilizing electrospun chimeric Janus microﬁber comprising cellulose acetate and polyvinylpyrrolidone

S.W


Introduction
As a fundamental property, wettability is ubiquitous in nature and our lives, where water is involved. For some plants and animals, they evolve hydrophilic or hydrophobic parts of themselves for their benefit. Some water-loving surfaces have been found from nature, such as fish scale, shark skin, spider silk and nepenthes alata. In addition, some water-repelling surfaces from animals and plants have been discovered such as duck feather, lotus, legs of water striders and butterfly wings [1]. For human interests, various artificial materials with different wettability have been designed, fabricated and applied in many fields, such as sewage treatment [2], drug delivery [3,4], sensors [5,6], lithium-ion battery [7,8], environmental remediation [9], filters [10,11] and biomedical scaffolds [12,13]. In general, with regard to drug delivery systems, a hydrophilic surface is required because water is the environment, wherein medicine can be absorbed easily.
A positive correlation between drug release rate and wettability has been proven by many reports; thus, various strategies are proposed to improve wettability of drug delivery systems [14]. When mixed into the drug carrier matrix, surfactants serve as effective small molecules in modulating the hydrophilic property [15,16]. Some active substances can also play the same role. Siranjeevi Ravichandran et al. mixed Clerodendrum phlomidis leaf extract into electrospun polycaprolactone (PCL) nanofibers, thereby reducing the water contact angle (WCA) from 137.4°± 5.2°for pure PCL to 72.41°± 2.4°for complex fiber [17]. Jaganathan et al. added Mahua oil and propolis into electrospun polyurethane scaffold and obtained a greatly improved complex wettability material and they explored its application in bone tissue engineering, which obtained good cell viability of 206.3% ± 14.36% [18]. Incorporating a hydrophilic polymer to adjust hydrophobicity/ hydrophilicity is a common and handy method [19,20]. Thus, electrospinning is usually applied because it can leave abundant and easily wetted groups on the surface of fibers. Some groups wetted with ease can vastly increase the wettability, although they were implanted on metal [21]. In some other controlled-release drug delivery systems, wettability can be changed by modulating the pH [22,23] and temperature [24,25].
Electrospinning, a simple and handy technology in preparing micro/nano fibers, has been developed for tens of years and applied to various fields [26][27][28][29][30] since Reneker's group carried it forward again in 1990s [31]. It has also contributed to the preparation of complex structures of micro/nano fibers [32][33][34]. As two fundamental structures, core-shell and Janus are basic structures in nature. Although core-shell spinneret was proposed with a traditional uniaxial spinneret by Cooley in 1902 [35], Janus spinneret and corresponding electrospinning were raised and carried out later by Gupta and Wilkes in 2003 [36]. It is separated among working solutions during electrospinning, which is difficult to be solved. Side-by-side electrospinning technology is developed slowly than coaxial electrospinning. A set of data from Web of Science indicates that the item numbers of ''side-by-side electrospinning or Janus" and ''coaxial electrospinning or core-shell" are 36,974 and 104,665 until July 2022. In addressing this problem, key points of this challenge should be refined. Among multilayer Janus electrospinning, more than one working solution is gathered on the tip of the structured spinneret. When solutions are not fully contacted or their contacted area is not large enough, the working fluid formed by solutions will be separated by electrostatic repulsion force. Thus, diverse strategies have been proposed. In 2017, Yu et al. designed a special structure spinneret, which has two acentric needles nested into a third metal capillary. Utilizing this spinneret, Janus nanofiber has been prepared with a smooth surface. In another experiment, an extra solvent used as a sheath to protect solutions from separation [37]. Therefore, increasing the contact area between working solutions or separating charge from working solutions can enhance the stability of side-by-side electrospinning. Otherwise, in 2019, Cai et al. proposed a dual spinneret penetrating through the center of a small round metal sheet, which can bind working solutions [38]. An additional electric field is used to allow working solutions to contact sufficiently. Therefore, sufficient contact among working solutions is the key point to conduct side-byside electrospinning.
Polyvinyl pyrrolidone (PVP), a non-toxic, biocompatible, chemically inert and non-ionic polymer, is widely used in drug delivery systems [39,40], cosmetics [41] and nanoparticle carriers [42]. Cellulose acetate (CA), the acetate ester of cellulose, has various advantages such as biocompatibility, biodegradability, regenerative property and high affinity. It has been broadly applied in many fields [43][44][45]. PVP and CA are frequently used hydrophilic materials. In addition, they are both electrospinnable with appropriate molecular weight and concentration. The combination of PVP and CA applied to the drug delivery system through electrospinning has been explored for many times. Edikresnha et al. electrospun PVP/CA composite nanofiber without loading glycerin and reported that garlic extract performed good wettability [46]. Tsekova et al. prepared a curcumin-loaded CA/PVP fibrous membrane to promote drug release. The WCA for different prepared fibers was recorded simultaneously with water droplet deposition, 129.4°± 3.8°for pure curcumin CA and 14.8°± 1.7°for curcumin-loaded CA/PVP fibrous mat [47]. Castillo-Ortega et al. fabricated an amoxicillin/ CA@CA/PVP core-shell fiber utilizing coaxial electrospinning. This membrane performed a fast release in tens of minutes indicating its good wettability [48]. As pure CA fiber mats usually possess lower hydrophilic property and need longer wetting time than PVP, they can easily dissolve into most inorganic and organic solvents. PVP is a user-friendly material that can be mixed with CA to improve its wettability. However, there are few works on building a wettability difference fiber surface. As uniaxial electrospinning technology was not easy to realize this special structure. Thus, Janus structure, especially multilayer Janus structure, should be applied for this design. Otherwise, the combination of mixing easy hydrophilic materials and constructing wettability gradient surface will further improve wettability of fibrous membrane.
In this work, for addressing the separation among working solutions, a specially designed complex structured spinneret was used. In this spinneret, a metal tube netted with a side-by-side structure protrudes a certain length to ensure that working solutions can be contacted adequately. In addition, the out-most metal tube can serve as an additional electro field [38]. KET, a commonly used nonsteroidal anti-inflammatory drug [49], was applied as a model drug in exploring release behavior of different kinds of fiber membranes. PVP and CA were used as a polymer matrix. By dissolving the different concentrations of these two polymers, different working solutions can be prepared and utilized for fabricating uniaxial, Janus and chimeric Janus microfibers through the special designed spinneret. Thus, wettability difference can be constructed on the fiber surface, particularly for chimeric Janus fiber comprising three sections with various wettability. This wettability difference will play a similar role as the wettability gradient on the cactus spines, which can promote the self-motion of water and decrease the WCA [50]. Wetting processes were recorded with the results showing that Janus and chimeric Janus fiber membranes performed better wettability than the PVP-CA uniaxial fiber. Driven by wettability gradient, Janus and chimeric Janus fibers showed improved wetting performance. As electrospun nanofibers have been widely explored in drug delivery systems [51][52][53][54], in vitro drug release experiments were further conducted to explore the effect of wetting speed on drug release profiles.

Fiber preparation
The whole process of this research is shown in Fig. 1. Working solutions were prepared on the basis of a diluted drug solution, dissolving 1%(w/v) KET into a mixed solvent: acetone/alcohol/DMAc = 4:1:1(v/v/v). Fluid 3 was obtained by adding 10% (w/v) CA into the diluted drug solution. If 8% (w/v) CA and 4% (w/v) PVP K60 or 8% (w/v) CA and 2% (w/v) PVP K60 were added into the prior diluted solution, then Fluid 1 or Fluid 2 would be prepared. Their flow rate in each process, average diameters, structures and theoretical contents of different components of fiber products are listed in Table 1.
The spinneret used in this research is shown in Fig. 2(a-c). Three tiny metallic capillaries were embedded into a 14G metallic needle. A 3 * 0.15 mm (external diameter * wall thickness) metallic tube was set at the outer most to provide an appropriate electrostatic repulsion to constrain working solutions from separation and stretch working fluid. When three appropriate solutions flow past suitable channels, core-shell fibers with a chimeric Janus core were obtained under electrospinning. In this research, the connection modes in which different solutions flow through corresponding channels for various electrospinning processes are shown in Fig. 2(d-f).
In every electrospinning, four equipment were used. Pumps (KDS100, KDS200, Cole Parmer, USA) drive working solutions to gather on the tip of the specially designed spinneret. Under electric field force, which is generated by a high-voltage electrostatic generator (ZGF-60 kV/2 mA, Electric Power Automation Co. Let., China), the working fluid was ejected from the Taylor cone formed at the tip of the spinneret when electrostatic repulsion had exceeded the interaction among working solutions (Fig. 2g). Afterward, a straight jet was followed by. Then, working fluid would experience bending and swing stage to stretch working fluid into nano/micro scale (Fig. 2h). Finally, the working fluid was solidified into fiber and collected by a collector. For all electrospinning processes, collected distance and high voltage were kept at 20 cm and 8.5 kV, respectively.

Characteristics
Among electrospinning processes, a digital camera (Sony Alpha 6400) was used to picture different processes. The morphology and inner structure of microfibers were observed by using a fieldemission scanning electron microscope (FESEM; Quanta FEG450, FEI Corporation, USA) and transmission electron microscope (TEM; JEM 2200F, JEOL, Japan). The diameter of microfibers was measured by using ImageJ (National Institutes of Health, USA). In obtaining the average diameter of fibers, 100 places were ran- domly selected in each SEM image. The results were recorded as mean ± S.D. Surface tension and viscosity of three working solutions were tested by surface tension meter (Kruss K100C-MK2, Germany) and rotary viscosity instrument (SNB-2, China), respectively. The surface tension tests were referred to GB/T 22237-2008 and viscosity tests were referred to GB/T 2794-2013. Otherwise, the viscosity tests were conducted under room temperature and 3# rotor with 6 rpm speed.
The physical states and interactions among molecules of raw materials and products were discovered using an X-ray powder diffractometer (XRD; AXS, Bruker, Germany) and reflectance-Fourier transform infrared (ATR-FTIR; Spectrum 100, Perkin-Elmer, USA), respectively. In XRD tests, data were collected among 10°-60°and 5°per minute. As for ATR-FTIR tests, data were gathered from 400 cm À1 to 4000 cm À1 in a step of 4 cm À1 . Differential scanning calorimetry (DSC, MDSC 2910, TA Instruments Co., DE, USA) was used to detect the thermodynamic property of raw materials and F3 microfiber. DSC data was collected from 70 ℃ to 220 ℃, 10 ℃ per minute.
In wettability tests of fiber membranes, an interfacial tension measuring apparatus (JC2000C1, Shanghai Zhongchen Digital Technology Apparatus Co., Ltd., China) was applied. For each sample, five different positions were tested and the results were recorded as mean ± S.D. (n = 5). For every test, approximately 5 lL of deionized water was dropped on the fiber membrane at room temperature. The reaction was observed for 3 min and recorded by using a built-in camera.
Studies on in vitro drug release were carried out in a constanttemperature water bath shaker. Based on Chinese pharmacopoeia (2015th), 0.1 g of membrane was added to 450 mL of PBS (pH 7.4). After drawing 4 mL of solution at a series of predetermined time, blank PBS with the same volume was injected into the solution. The percentage (P) of drug released from fibers can be calculated using Eq. (1).
where V 0 is the volume of the solution equals to 450 mL; V is the volume collected at each time, which is equal to 4 mL. Q is the theoretical amount of drug in 0.1 g of fiber membrane. In addition, c n is the drug concentration at predetermined time, which can be calculated using Beer's law as shown in Eq. (2).
where A is the absorbance that can be detected using an ultraviolet-visible spectrophotometer (UV-Vis; Unico Instrument Co., Ltd., China) and c is the concentration. e and l are the absorption coefficient and optical length, respectively. In this research, they are both constants.

Morphology and inner structure of microfibers
In traditional Janus electrospinning, caused by electrostatic repulsion, working fluid integrated by different working solutions is easily bifurcated. Given the special design of the spinneret, Janus and chimeric Janus fibers were fabricated. On the one hand, the middle tube of the spinneret protrudes approximately 0.2 mm than the inner capillary; this design can facilitate contact among working solutions. On the other hand, an outer-most metallic tube was set to constrain working solutions from separating. By using this specially designed spinneret, working solutions can contact one another more adequately than those in traditional tri-fluid sideby-side electrospinning.
The inner structures of Janus fiber can be observed from TEM images. As shown in Fig. 4a, the two sides of a Janus fiber measured 418.137 and 246.480 nm. The boundary line formed by different gray levels of two sides is easy to distinguish. As the flow rates  are equal, their size difference may be due to different concentrations. The chimeric Janus fiber can be observed in Fig. 4b with sides measuring approximately 278.170, 267.826 and 289.655 nm. As the chimeric Janus structure is not easily identified in Fig. 4b, a contrast-changed image (Fig. 4c) was provided for better observation. A gray gap closed to the center is displayed in the dark fiber. This structure might have three sides, two at the front and one at the back.
In tri-fluid side-by-side electrospinning, three sections with different ratios of polymers were integrated into one microfiber. Thus, forming a clear boundary from an overhead view is difficult, as different sections are overlapped. Different sections (A, B, C and D) marked on a chimeric Janus microfiber (Fig. 4c) are highlighted in the inset. As for the gap between B and C sections, the combination among them can be weak, thereby forming an area through which electron beam can pass. The weak combination is due to fast solidification, which limited working solutions flowing into one another. Thus, by utilizing the specially designed spinneret and under appropriate parameters, electrospinning was successfully carried out, then a series of uniform microfibers was obtained.

Details on chimeric tri-layer side-by-side electrospinning
By combining a traditional tri-layer side-by-side electrospinning and strategies on avoiding fluid splitting, chimeric tri-layer side-by-side electrospinning was carried out in preparing chimeric Janus microfibers composed of three different parts. Given the middle metallic capillary, which packs three 22G needles flowing through working solutions and protrudes a short length, after working solutions flow out from inner-most needles, they will effectively contact before forming a Taylor cone. As the protrude length is short, working solutions cannot be mixed. Along with the three stages of electrospinning, namely, Taylor cone, straight jet and bending and swing, solvent was gradually removed from the working fluid, which finally transformed into a microfiber (Fig. 5a). In the Taylor cone stage, charges distributed on the edge of the out-most metallic tube are similar to those distributed on the surface of the working fluid. The electrostatic repulsion generated has two effects (Fig. 5b): constraining working solutions from splitting and stretching working fluid to promote Taylor cone to straight jet. As electrospinning develops, the working fluid moves  farther from the spinneret. Influences, which acted on the working fluid, from electrostatic repulsion are decaying. Consequently, working fluid composed of working solutions integrated by condensed solutions will be partly removed from the preceding course. Therefore, condensed working fluid with a robust connection among working solutions cannot be departed by electrostatic repulsion before the formation of chimeric Janus fibers. Surface tension and viscosity of three working solutions were listed in Table S1. By comparing traditional side-by-side electrospinning with trifluid electrospinning, the traditional one is often severely affected by electrostatic repulsion, which will split working fluid up before it solidifies (Fig. 5c). Thus, a traditional process cannot effectively form tri-layer Janus fibers. However, in this work, a St20;''pecially designed spinneret was used to carried out chimeric tri-layer sideby-side electrospinning, thereby preventing the splitting of the working fluid. Along with solidification, working fluid removes solvent gradually and finally forms fibers with a chimeric Janus structure.

Chemical and physical states of microfibers
As a classic crystal medicine, KET has various sharp peaks that can be observed from its XRD pattern (Fig. 6a). On the contrary, polymers, such as CA and PVP, who do not have a long-range order structure, serve as humps on their patterns. Electrospinning can transfer matter from crystal into amorphous state; thus, the XRD pattern of microfiber membranes shows an amorphous form.
FTIR curves indicate compatibility among materials. As shown in Fig. 6b, the PVP spectra that peaks at 1651 and 1423 cm À1 correspond to the stretching vibration of the carbonyl group in amide and C-H bending vibration of CH 2 . In addition, C-N stretching peaks at 1284 cm À1 . A broad band at approximately 3427 cm À1 represents the O-H bond [55]. In the CA spectrum, several peaks related to the acetyl group emerge, such as C=O stretching (1727 cm À1 ), C-H bending of -CH 3 (1365 cm À1 ) and C-O stretching (1219 cm À1 ). Otherwise, as the backbone bond, the absorbance peak of C-O-C arises at 1035 cm À1 [56,57].
Based on the FTIR spectrum of KET, two absorbance peaks of stretching vibrations, 1694 and 1654 cm À1 , from C=O can be easily observed: one from the carboxyl group and the other from the ketonic group [58,59]. However, in membranes' spectra, index peaks cannot be well distinguished because the model drug was dispersed uniformly in fiber after having been electrospun. Some red shifts are also observed compared with raw materials, for example, 1651 cm À1 in PVP transfers to 1658 cm À1 ; 1727, 1365 and 1219 cm À1 in CA transfers to 1743, 1370 and 1227 cm À1 , respectively. The findings revealed that some proton acceptors as C=O, C-O can interact with proton donors such as O-H [60]. DSC data also supports the results form XRD and FTIR (Fig. S1).

Wetting ability of different microfibers
Wettability is an important performance for medicated fibers. When medicine-loading fiber membranes were soaked in water, this property will affect drug release profiles. Therefore, before conducting in vitro experiments, wetting tests had been carried out to investigate wetting time for F1 to F3 microfibers. The results are shown in Fig. 7a. Compared with F1, F2 and F3 exhibit higher WCA at 135.6728°± 3.3694°and 135.7270°± 6.2137°and quickly decrease to 8.1438°± 14.5002°and 19.5269°± 18.3367°, respectively, in 90 s. Uniaxial microfiber, F1, had the lowest WCA (120.7271°± 3.2825°) at the beginning. Its hydrophobic performance can last 70 s, which is two times longer than that of F2 and F3. After 3 min, it will decrease to 22.9380°± 4.4291°.
After the first model proposed by Young in 1805, the second model, which considered solid surface roughness, was raised by Wenzel in 1936. Cassie proposed the third wetting model, Cassie's model, which introduce air into the gap between liquid and solid [61]. Apart from the two previous models, Cassie's model is static (Fig. 7b). In this work, the WCAs are not the same all the time. In the first tens of seconds, Cassie's model works well because electrospun microfiber membranes have many pores. Considerable air is found between liquid and solid. The WCA of F2 and F3 was slightly higher than that of F1 probably because of the high CA content on the fiber surface. The reason of low WCA should be slightly different from that in the F1 membrane. In the presence of a wettability gradient, F2 and F3 have two and three sections with different wettability, respectively. On the surface of the Janus fiber, the effective gradient from hydrophobic to hydrophilic is the same in different directions. However, it is various in different directions on the surface of the chimeric Janus fiber (Fig. 7c). Based on published reports, the wettability difference could promote wetting [62]. Furthermore, some published articles demonstrated that water movement can be driven by wettability gradient [63]. Thus, for Janus and chimeric Janus fibers, wetting rather than water absorption should be considered. The wettability gradient-driven wetting will spread water, thereby covering the surface of the microfiber. A considerable amount of water is needed to wet fibrous membranes because of the huge specific surface area granted by electrospinning, thereby causing a quick reduction in WCA.
The other reason is water absorption based on the surface energy theory. With a high binding energy, the carbonylic group originated from pyrrolidone in PVP is an excellent hydrophilic group. Although the carbonylic group originated from ester, CA has a low binding energy group, methyl, as its end group. As electrospinning can keep various groups on the surface of fibers. In this work, for uniaxial microfiber, abundant methyl groups will decrease its surface energy, thereby lowering the speed of wetting [21]. However, for F2 and F3, they will take more time to absorb water because their surfaces have a larger percentage of CA than F1. Therefore, for uniaxial fiber membrane, wetting and water absorption could decrease the WCA. However, for Janus and chimeric Janus fiber membranes, wetting dominates the quick reduction of WCA, after that water absorption decreases.

In vitro drug release profiles
In vitro drug release experiments were carried out by utilizing the standard curve shown in Fig. 8a, which is fitted by a series of data gathered by using an UV-vis spectrophotometer at approximately 260 nm with a set of determined KET concentrations. Drug release profiles in 6 h and 60 min have been shown in Fig. 8b and 8c, respectively. Drug release rates for all three microfiber membranes can nearly complete release. It only takes 60 min to release  about 80% of the three kinds of membranes. Furthermore, 5 h was needed to release the rest of the membranes. As shown in Fig. 8c, during the first 20 min, F2 and F3 performed faster release than F1 because of the different wetting time of microfiber membranes. F2 and F3 were wetted faster than F1; thus, drug is released quicker. In addition, the release rate of F1 surpassed F2 and F3 at around 60 min and sustained to the end. When three kinds of fibers are all immersed in water, F2 and F3 fibers performed slower release than F1 because they possess a lower content of PVP compared with F1, which is a classic fastrelease excipient.
Four drug release models, including Zero-order, First-order, Higuchi and Ritger-Peppas models, were used to match the data from drug release profiles of three kinds of fiber membranes and further explore the drug release. Their equations are described as follows.
Zero-order model: First-order model: This model can transform into another format that is uniformed with others.
Higuchi model: Ritger-Peppas model: where M t and M 1 are the cumulative release amount at t and 1 time, respectively; Z(t), F(t), H(t) and RP(t) are the functions of t, representing the cumulative release rate; k 0 , k 1 , k 2 , k 3 , Z 0 , F 0 , H 0 and n are all constants.
The obtained fitting functions and their standard deviations are listed in Table 2. In the data used, the first 2 h was cut out from release profiles, whereas the release rates at 2 h nearly reached the maximum. As the three kinds of microfiber membranes are not drug controlled-release systems, the zero-order model did not perform well. First-order and Higuchi models were designed for the sustained-release system; thus, they had two highest standard deviations. This result indicates that the three fiber membranes belong to sustained-release systems. Furthermore, based on Ritger-Peppas functions, n value revealed the drug release method. For all three release profiles, n values, F1 with 0.6573, F2 with 0.6337 and F3 with 0.5038, are all between 0.45 and 0.89, indicating that drug diffusion and corrosion backbone effects are involved in drug release. It can be easily understood that PVP is a frequently used water-soluble material and CA is a well-known hydrophilic swelling polymer.
Based on the wetting data, drug release profiles can be clarified. Different from Janus and chimeric Janus fibers, as a homogeneous fiber, F1 fabricated by uniaxial electrospinning was wetted gradually. Water was absorbed by a fibrous membrane slowly in the first several minutes. Meanwhile, drug was released through a diffusion and backbone corrosion mechanism. By contrast, F2 and F3 membranes preferred to be wetted initially by a wettability gradient among different sections, accompanied with drug release. Similar to the Janus microfiber, in chimeric Janus microfibers, drug was gradually released into PBS from 4%PVP-CA, 2%PVP-CA and CA sections in sequence (Fig. 8g).

Conclusions
In this research, by using a specially designed complex spinneret, three electrospinning processes have been carried out and uniaxial, Janus and chimeric Janus microfibers were obtained. Under the same high voltage, collected distance and total flow rate, microfibers with different structures have similar average diameter. XRD and FTIR spectra show fibers made from raw materials with good compatibility and hydrogen bond formed among them. As the newly prepared microfiber, chimeric Janus fiber has three sections from three different working solutions. With different ratios between PVP and CA, the wettability gradient is formed within a microfiber. Wettability gradient-driven wetting and water movement theories were used to explain the difference on decreasing the WCA of various fiber membranes. Given this wettability gradient, the chimeric Janus fiber performed quick decrease on WCA and fast drug release at the earlier stage. Four fitting models were used to match drug release profiles, finding that all of the three membranes belonged to sustained-release systems. Furthermore, the Ritger-Peppas model revealed that they were in accordance with the release, which is affected by backbone corrosion and drug diffusion effects.
For a long-term perspective, two strategies for wettability improvement including constructing wettability-difference surface and easy hydrophilic material addition can be incorporated into chimeric Janus fibers. Thus, this tri-fluid electrospinning technology should be a robust platform for other applications, such as fog harvesting membranes and humidity sensors.

Data availability
Data will be made available on request.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.