AC conductivity, dielectric and thermal properties of three-phase hybrid composite: PVA/Bi2O3/cotton microfiber composite

The impact of bismuth (III) oxide (Bi2O3) on the characteristics of the cellulose/polyvinyl alcohol (PVA) blend was reported for a high weight ratio of the oxide (15wt%). Composite samples were made with 15wt% oxide and 2:1 weight ratio of PVA to cellulose using a hot hydraulic press technique (5 MPa and 175 °C), which led to samples in the form of a disk. The thermal stability of the composite was illustrated using the thermal gravitational analysis (TGS) at a heating rate of 10 °C min−1 in N2 environment. The results show that the thermal stability of the composite sample was greater than that of the blended sample in the high-temperature region. The blend and composite samples exhibited two weight-loss stages throughout the thermal decomposition process. These two stages correspond to the slow decomposition (200 to 400 °C) and fast decomposition stages (400 to 450 °C for blend and from 430 to 460 °C for composite). Only 5% mass loss for both samples was detected due to heating from 50 °C to 200 °C. Dielectric spectroscopy (from 100 Hz to 1 MHz) was used to investigate the effects of Bi2O3 on the relaxation and conduction mechanisms of the composite samples at different temperatures. Dielectric permittivity, AC conductivity, electrical modulus, and complex impedance were investigated. Jonscher’s equation was applied to the blend and composite samples. The modified Jonscher’s equation fit well at low temperatures. As the temperature increases, the deviation from the normal Jonscher equation decreases. The activation energies of the blend and composite were calculated by determining the bulk resistance (RB) from the Nyquist plots. The activation energy of the blend was increased by adding the filler (Bi2O3).


Introduction
In the past ten years, scientists have become increasingly interested in the utilization of natural resources to produce new materials for industrial and technological applications. Plastics and polymers have contributed to the changes in human life since 1907. The adaptability and biodegradability of inexpensive materials make it ideal for many applications, once scientists can adapt them. If bio-plastics and bio-polymers can be adapted to replace industrial polymers or reduce their use, they can serve humanity by preserving the environment. Many observations have indicated the existence of this trend in various scientific research fields. For example, researchers typically use biomass as a replacement energy source for cleaner and more sustainable energy [1][2][3]. However, there is a tendency for researchers to substitute plant extracts with conventional chemical processes for the production of nanomaterials. This is referred to as the green synthesis technology [4][5][6].
In the medical field, there are different research trends to benefit from natural products [7], whether for treatment or control of drug delivery, and other purposes. Several types of cellulose represent the main product in medical applications [8,9].
Cotton fibers are highly suitable for use in many technological industries. Owing to their inherent biodegradability, cotton fibers and nanofibers are frequently used to strengthen various types of polymers. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Cotton fibers gain strength from their crystalline and fibrous structures, in addition to exhibiting a high degree of thermal conductivity and outstanding mechanical properties [10].
Many industries depend on reclaimed and recycled cotton fibers, such as the manufacture of low-quality yarns, fabrics, and non-woven materials, which are used in the manufacture of furniture and thermal insulation of buildings, in addition to the automobile industry [11][12][13]. Also cotton cellulose fiber was use in the field of water purification technologies. Different types of polymers were used to coat and treat cotton fibers to be used for the simultaneous removal of oils, dyes, and metal ions from water [14,15]. Ni et al synthesized cotton/ thermoplastic polyurethane compounds for use in flexible and wearable electronics [16]. Li electronic applications, conductive films based on functionalized cotton fibers with suitable strength and durability [17]. In their study, the fiber ratios, size, distribution, and interaction between the cotton fibers and composite represent the control parameters. Pascu et al investigated the effect of a magnetic field on an active magnetic compound based on cotton fibers, fine iron carbonyl particles, and barium titanate (BaTiO 3 ) particles [18]. They found that the magnetic field had a clear effect on the compound, as it strongly depended on the composition of the compound and the preparation conditions. Bismuth (III) oxide (Bi 2 O 3 ) was used to improve the insulating properties of ceramics and glass [19,20], and there are many articles in this regard but no articles for improving the dielectric properties of cotton cellulose fiber. Also, the addition of Bismuth (III) oxide to improve the dielectric properties of cotton cellulose represent the first step for the development of a composite that can be used in water treatment as future work. Bismuth (III) oxide is very useful and important of this oxide in the field of water treatment [21,22].
In this study, the effect of Bi 2 O 3 on the properties of a cellulose/polyvinyl alcohol (PVA) blend prepared using hot-press techniques was reported for a concentration of 15wt% of Bi 2 O 3 . The blend ratio is 2 to1 of PVAcellulose. Both thermal gravimetric analysis and dielectric studies were performed. The results for these two-and three-phase hydride materials are explained in detail in this study. The composite was prepared at a high weight ratio (15wt% Bi 2 O 3 ) because a low concentration did not contribute to the improvement of the composite. Enhancing the dielectric properties of such composites can help produce low-cost components in the field of electronics.

Materials
In the present study, the polymer host or binder was (PVA) with an average molecular weight of 130,000, 99% degree of hydrolysis, and a degree of polymerization of approximately 1200. PRIVATE (Egypt) provided cotton cellulose microfibers with aspect ratios of 2-4. Bismuth (III) oxide (Bi 2 O 3 ) with a molecular weight of 465.96 was purchased from Sigma Aldrich. The chemical structures of cotton cellulose fibers (left) and PVA are shown in figure 1.

Preparation of three-phase composite samples
The blended sample was prepared by mixing the ratio (1:2) of the cotton fiber to the polymer (PVA) and mechanically stirring at 80°°C for 2 h. The samples were exposed to microwave waves for 20 s and left to cool before the second exposure (five periods). Microwave exposure evaporates water and can create interactions between the polymer and cellulose chains. A three-phase sample was prepared by adding 15wt% bismuth oxide to the two-phase composite. The three-phase sample was prepared following the same protocol as that used for the two-phase sample. A conventional microwave device was used and the samples were placed at the center of the microwave to fix the preparation conditions. The sample was left to dry until it was hardly deformed, and then pressed into discs by mechanical pressing. The blend and composite samples were hot-pressed into disks at 220°C for 1 min under 0.5-ton pressure. Figure 2 illustrates the sample-preparation process.

Dielectric measurements
The impedance (Z) and AC conductivity (σ ac ) of the composites were measured using a HIOKI 3532-50 -LCR measuring bridge (5 MHz, Hioki, Japan) at frequencies varying from 100 Hz to 1 MHz. Both Z and δ (phase angle) were collected automatically after averaging, and Z' (real part of the impedance), Z' (imaginary part of the impedance), and σ ac (AC conductivity) were calculated. Copper conductive plates were used on both sides of the sample as the conductive electrodes.

Thermal analysis
Thermogravimetric analysis (TGA) was performed on a Shimadzu DTG-60H instrument. The degradation of two-and three-phase composite samples in N2 gas (30 ml min −1 flow rate) and a temperature range of ambient to 700°C was studied using a constant scan rate of 10°C min −1 .

Thermal analysis Figures 3(A) and (B)
show the TGA and first-derivative thermogravimetric (DTGA) results for the blended and composite samples. TGA was used to examine the material's thermal degradation and stability from 30°C to 600°C at a rate of 10°C min −1 . The blend and composite samples exhibited two weight-loss stages throughout the thermal decomposition process. These two stages corresponded to the slow and fast decomposition stages, respectively. The mass of the blend and composite samples was reduced by 5% owing to heating from 50°C to  200°C. This loss was attributed to the evaporation of water. Water volatilization and evaporation represent the main processes in the slow decomposition stages (from 200°C to 400°C). Meanwhile, fast decomposition occurred from 400°C to 450°C for the blend sample and from 430°C to 460°C for the composite sample. The mass loss during the slow decomposition stage was approximately 66%, whereas only 28% mass loss occurred during the fast decomposition stage. The slow decomposition stage be attributed to the decomposition of the non-cellulosic materials of cotton fibers. There are mostly waxes, pectin, organic acids, sugars, etc, among these non-cellulosic substances. It has been reported that these materials decompose at a lower temperature than cellulose. During this stage (fast), the blend and composite samples were almost completely burned owing to thermal degradation and destruction of the crystalline structure.
As shown in figure 3(B), the slow decomposition stages of the blend and composite samples were represented by broad peaks. Each of these peaks could be deconvoluted into two separate peaks (figures 4(A) and (B)). The peak positions of the separated peaks are shifted to higher temperatures (more stability) for the composite samples (284°C and 328°C for blend sample and 293°C and 340°C for composite sample).
It should also be noted that the DTGA curve for the fast decomposition stage is composed of a superposition of two or more reactions that can be separated. Figures 4(C) and (D) illustrate the peak separation in the fast stage for the blend and composite samples, respectively. The fast decomposition stage of the blend sample can be separated into two overlapping peaks (figure 4(C)), whereas the composite sample is composed of several (complex) overlapping peaks (figure 4(D)). For the blended sample, the two separated peaks (fast decomposition stage) can be attributed to the decomposition of cellulose and PVA. However, in the case of the composite sample, the heterogeneity of the sample increased owing to the presence of fillers. This, in turn, causes the constituents to decompose independently according to their status and relationship to the filler.
Accordingly, the addition of Bi 2 O 3 produces an increase in the thermal stability of the cellulose/PVA blend sample. In addition, the sample heterogeneity increases causing more complex decomposition especially for the fast decomposition stage. 3.2. Impedance spectroscopy 3.2.1. Dielectric permittivity Impedance spectroscopy is one of the most crucial methods for identifying the dielectric relaxation processes in polymers, polymer blends, and polymer composites. More information about the physical process occurring inside the material can be obtained from the complex impedance analysis and anticipated equivalent circuit based on its data. In addition, the relaxation process depends on the sample constituents and temperature. Figures 5 and 6 show the temperature dependence of the complex dielectric permittivity's (ε′ & ε′) of the blend and composite samples, respectively. The measurements were performed at various frequencies (100 Hz to 1 MHz) and temperatures. As can be seen, as the frequency increased, ε′ and ε′ of the sample decreased. This could be due to a decrease in the number of dipoles engaged in polarization. The dipoles (such as the OH groups of cotton cellulose) can easily follow the electric field at low frequencies. However, at high frequencies, the electric dipoles begin to lag behind the applied electric field, resulting in a decrease in the permittivity (ε′). It is observed that ε′ decreases with increasing frequency. With increasing frequency, the imaginary part (ε′) continued to decrease, and there was no indication of relaxation in this frequency region. It is also noted that the addition of bismuth oxide (15wt%) increased the value of the real part (ε′), whereas the imaginary part (ε′) was not significantly affected. Figures (7(A)-(D)) represent the dependence of ε and ε′ on the temperature at different frequencies (0.5, 1, 10, 100, and 10 3 kHz). At low frequencies, the real permittivity (ε′) increases with increasing temperature, accompanied by a peak for both samples (pure blend and blend sample loaded with 15wt% BO). As the frequency increases, the peak diminishes and shifted to the higher temperatures, indicating a dielectric relaxation in this temperature range. A shift in the dielectric peak was easily detectable for the composite sample than for the pure sample. The dependence of the real part (ε′) on temperature decreases at high frequencies, which can be attributed to the weak response of the electrical dipoles to high frequencies. By computing the relaxation time using τ = 1/ω and the temperature of the peak position (T p ), the activation energy can be predicted using the Arrhenius equation τ = τ 0 exp (E a /K b T p ).  contribution of ionic conduction by increasing the number of ionic impurities, besides the creation of free radicals on the surface of the oxide. These free radicals will interact with the O-H available in cellulose, and the ionic impurities will increase from the calculated activation energy.
As a result, the addition of Bi 2 O 3 increased the real permittivity but did not significantly affect the imaginary permittivity. Furthermore, it was observed that the presence of interfacial polarization at the interface between the oxide particles and polymeric blend increased the activation energy with the addition of Bi 2 O 3 .

AC conductivity
Owing to the lack of free charge carriers, pure polymers (other than conductive or semiconductive polymers) typically fall under the category of insulating materials [23]. As a result, their response to the applied AC field depends on dielectric relaxation, which can be attributed to the transfer of space charges, rotation of both permanent and induced dipoles, segment mobility of polar groups, interfacial charge, and relaxation caused by the change in the material state from glassy to rubber [24]. Cotton cellulose is a macromolecule with a 99% cellulose content after scouring and bleaching. It is a polymer composed of a long chain of glucose molecules joined by C-1-C-4 oxygen bridges [25]. The relaxation process through the host polymer may be affected by the addition of fillers. Additionally, the interaction of the polymer chains and other fillers may result in a complex behavior during the relaxation process of the composite. The relaxation process may also be hampered by induced and interfacial polarization [26].
From impedance spectroscopy measurements, the AC conductivity (σ ac ) was calculated using the following equation; where ε 0 and ε′ are the free space and material dielectric constants, respectively, and tan(δ) is the loss tangent or dissipation factor. The AC conductivity was measured as a function of frequency at different temperatures (40°C-200°C).  at low to intermediate frequencies, which usually correlates to the sample's DC conduction (dc). With increasing temperature, this flattened zone extended and shifted toward higher frequencies. With an increase in frequency, a dispersion in the AC conductivity was observed. This is a well-known conduction mechanism in polymers aided by polaronic hopping between various states [27].
In the case of the composite sample, the plateau section was small at low temperatures and then widened significantly as the temperature increased, covering most of the frequency range. The data indicate that σ ac increases as the temperature increases (typical response for the majority of insulating polymers) [28]. This can be attributed to the fact that the number of transit sites and the mobility of charge carriers in the polymer matrix and polymer composite increase as the temperature increases [29]. Therefore, charges can overcome these barriers and contribute to the electrical-conduction mechanism. It should be noted that when the temperature increased, the frequency at which dispersion occurred (hopping frequency ω H ) shifted toward a higher frequency. Jonscher's power law (JPL) may be used in this situation [29]. where ω is the angular frequency, σ dc is the dc-conductivity (i.e., independent of frequency at ω ≈ 0), A is a temperature-dependent constant, and n is an exponent (0 n 1). Both values (A and n) depend on the temperature. The constant A is correlated to the sample polarizability, whereas n represents an indicator of the reactivity of the sample ingredients (filler-polymer, filler-filler, and/or polymer-polymer interactions). The AC conductivity behavior strongly depended on the type of conduction mechanism present in the sample. Different theoretical models (such as quantum mechanical tunnelling (QMT), correlated barrier hopping (CBH), non-overlapping small polaron tunnelling (SPT), and overlapping long polaron tunnelling (OLPT)) can be utilized to determine the type of conduction mechanism [30] for the sample under test.
The experimental and theoretical data for the blended sample were consistent over the frequency range and temperature range, as shown in figures 8(A) and (B). The dependence of the parameters n and A on temperature is illustrated in figure 8(C). There are critical variations in both parameters at approximately 50°, 100°, and 160°C. By applying the normal Jonscher's equation to the blended composite sample, it is noticed that for the low-temperature stages (up to 80°C), the experimental and theoretical data are inconsistent in the lowfrequency range.
The modified Jonscher's equation giving by the following expression; where ω is the angular frequency, σ dc is the dc-conductivity, A1 and A2 are the pre-exponential factors, and n 1 and n 2 are the Jonscher's law exponents. The low-frequency region is characterized by the exponent n 1 (0 < n 1 < =1), which corresponds to translational ion jumping; the high-frequency region is characterized by the exponent n 2 (1 < n 2 < =2), which indicates well-localized relaxation and reorientation [30]. By applying the modified Jonscher's equation, it was found that there is a good fitting in both frequency bands (low and high). As the temperature increases, the deviation from the normal Jonscher's equation decreases. As an example, figure 9(A) illustrates the application of both the normal and modified Jonscher's equations to composite samples at 60°C and 90°C. The red and blue lines represent the normal and modified equations at 60°C, respectively, while the green line represents the application of the normal JPL at 90°C. Figure 9(B) shows the experimental data (symbols) and the corresponding theoretical curves for the normal Jonscher's equation (continued lines) at different temperatures, which were used to calculate the parameters σ dc , n, and A.
The dependence of parameters n and A on the temperature of the composite sample is illustrated in figure 9(C). It is clear that their behaviors are opposite. Both parameters (n and A) for the pure cellulose samples were characterized by the presence of a peak and a minimum, which were in exchange. The value of the parameter n decreases from 40°C to near 100°C (given a minimum) and then begins to increase to give a peak of near 160°C. The value of parameter A increased from 40°C to a peak around 100°C, then gradually decreased to the bottom around 160°C. This behavior will be discussed later in this paper.

Electrical modulus
The real (M′) and imaginary (M′′) components of the electric modulus (M * ) were calculated from ε′ and ε′ respectively using the following equations: The frequency dependences of M′ and M′′ for the blended and composite samples are shown in figures 10(A)-(F). In both cellulose/PVA blends and composite blends, M′ and M′′ increased uniformly with frequency. M' exhibited a plateau region at high frequencies, whereas M′′ had a flattened peak.  Figure 11 shows the dependence of M′ and M′′ on temperature for the blend and blend-composites at different fixed frequencies (1, 10, 100 and 350 kHz). The temperature dependence of the electrical modulus of the pure blend sample can be interpreted by considering the glass transition temperature (T g ) of both polymers (PVA and cellulose). The glass transition of PVA is approximately 90°C, while for cellulose, T g , is approximately 170°C. During microwave irradiation, cellulose and PVA may form physical interactions, thereby affecting their glass transition. The values of the electrical modulus decrease gradually with temperature, passing through two turning points around 100°C and 170°C, corresponding to the glass transition of PVA and cellulose respectively.
For the three-phase (composite) sample, the general trend of M' and M' tends to decrease with increasing temperature especially at lower frequencies (1 khz and 10 khz). Whereas the dependence of M' and M' on temperature decreases at high frequencies (slight increase and slow decrease with temperature). With increasing temperature, M' relaxation peaks shifted to higher frequencies. In the composite sample, additional polarization (interfacial polarization) was generated due to the addition of Bi 2 O 3 . Figure 12 shows the frequency dependence of the real and imaginary parts of the complex impedance of the blended sample. Z' exhibits a frequency-independent region extending from 100 Hz to the temperaturedependent critical frequency (fc), which shifts to a higher frequency with increasing temperature. The values of Z' and Z' decreased as the temperature increased up to 100°C, then their values increased with temperature up to 140°C. Above 140°C, the values decreased again (see figures 12(E) and (F)). At high frequencies, Z' and Z' exhibited very weak temperature dependence. The linear trend of Z'(f) at the high frequency indicates the capacitive behavior of the sample in this range of frequency. There is a high-frequency peak associated with the imaginary part (Z') which is shifted to higher frequencies with increasing temperature.

Impedance analysis
To explain the impedance behavior of the blend sample with temperature, the glass transition of both polymers must be taken into consideration. It is known that the glass temperature (Tg) of PVA is about 86°C [31], while that of cotton cellulose is about 160°C [32]. As the temperature of the blend sample increase, the chain mobility will increase and the impedance decreases. When T is higher than Tg for PVA (T > 100°C), phase separation occurs and the motion of PVA chains will be confined by the cellulose chains. This chain restriction will cause a reduction in the conductivity of the sample (Z' increase). As the temperature approached 140°C, the cellulose chain mobility will be increased and the sample impedance decrease with increasing temperature (T > 140°C).This is schematically represented in figure 13 for the blended samples.
For the composite sample, figures (14(A) and (B)) show the frequency dependence of the real and imaginary parts of the complex impedance (Z * ). The set of curves of the real part (Z') was characterized by a small plateau region, which expanded to higher frequencies as the temperature increased. This indicates that there is competition between the resistive and capacitive components inside the sample and the resistive component increased with temperature. For the imaginary component of Z * (Z″), the frequency dependence curves are characterized by a peak, indicating the presence of a type of relaxation. This confirms the behavior of M″. The relaxation peaks suffer from a shift in the direction of higher frequencies with increasing temperature.
The Cole-Cole plot (Nyquist plot or complex impedance plot) and Cole model equivalent circuits for the blend and composite samples at different temperatures are illustrated in figure 15(A)-(D). For the blended sample (figures 15(A) and (B)), the equivalent circuit is composed of a series resistance (R 1 ) connected to a parallel set of resistances (R 2 ) and a constant phase element (CPE 1 ), as shown in figure 15(D). For the composite samples ( figure 15(C)), the equivalent circuit was composed of a series resistance (R 1 ) connected to a parallel set composed of (CPE 1 ) in parallel with a series set composed of (CPE 2 ) and a resistor (R 2 ), as shown in figure 15(D). For the composite sample, the diameter of the semicircle decreased with increasing temperature at all temperatures.   The difference in the equivalent circuit composition explains the effect of bismuth oxide. For the blend sample, the resistance R 1 represents the electrode resistance, whereas CPE 1 represents the interfacial polarization between the heterogeneous immiscible blends. The resistance R 2 represents the internal resistance of the blend. Similarly, the equivalent circuit of the composite sample is explained. Resistance R 1 represents the electrode resistance, whereas CPE 1 represents the interfacial polarization between the heterogeneous immiscible blend, which is connected in parallel with two series components R 2 and CPE 2 . The constant phase element CPE 2 represents the interfacial polarization between the filler and blend, while R 2 represents the internal resistance of the mixture.  The bulk resistance R bulk of the samples was determined from the intersection of the semicircles on the real axis of impedance (Z'), which corresponds to the direct current (DC) resistance [33]. DC electrical conductivity was calculated using the following equation: s = Where 'σ o ' is a constant, 'k B ' is the Boltzmann constant (k B = 8.617 × 10 -5 eV/k), E a is the activation energy, and T is absolute temperature. The activation energies of the blend and composite samples in both temperature regions are listed in table (1). The increase in the activation energy of the composite system can be attributed to the formation of interfacial polarization between the filler and blend, which confines the motion of the polymeric chains. In addition, the  penetration of fillers between the polymeric chains may increase the crystalline phase of the polymer, which increases the activation energy [22].

Conclusion
From the obtained results, the conclusions can be summarized as follows: I. The addition of Bi 2 O 3 produces an increase in the thermal stability of the cellulose/PVA blend sample. The complex decomposition for the composite sample was attributed to the increase of the sample heterogeneity due to addition of Bi 2 O 3 powder.
II. The addition of Bi 2 O 3 increased the real permittivity and did not significantly affect the imaginary permittivity. Furthermore, it was observed that the presence of interfacial polarization at the interface between the oxide particles and polymeric blend increased the activation energy with the addition of Bi 2 O 3 .
III. When applying Jonscher's equation to identify the conduction mechanism, we did not find a great match with the practical values, so the modified Jonscher's equation was applied. The values of the exponent n ranged between 0.5 and 0.6 for the blend sample, while for the composite sample the value changed between 0.5 and 0.8 which indicates a change in the conduction mechanism as a result of the addition of Bi 2 O 3 .
IV. Both the impedance and the electrical modulus showed the same behavior, with the appearance of a relaxation mechanism as a result of the interfacial polarization of the oxide-laden sample.
V. Cole model equivalent circuits for the blend and composite samples at different temperatures are predicted using ZSimWin software. The incorporation of Bi 2 O 3 to the blend sample affects the model equivalent circuit due to the interfacial polarization.
VI. The activation energies of the samples were calculated using the Cole-Cole diagram and the Arrhenius equation. The activation energy for the composite sample is greater than the blend sample and this was attributed to the formation of interfacial polarization between the filler and blend.