In situ polymerization and electrical conductivity of polypyrrole/cellulose nanocomposites using Schweizer's reagent

Cellulose-based composites have attracted interest given the shift towards ‘green’ materials, but achieving uniform dispersions of cellulose in polymer matrices and/or enhancement of interfacial interactions between components remains challenging. Herein we report the preparation of polypyrrole/cellulose nanocomposites in [Cu(NH3)4(H2O)2](OH)2 (Schweizer's reagent/cuoxam)-based reaction media via in situ polymerization. The effect of cellulose template morphology and reaction media on the microstructure, electrical conductivity, and surface wettability was studied. Aqueous reaction media favored the formation of a uniform polypyrrole coating encapsulating the cellulose fibers; concentrated cuoxam solutions promoted inhomogeneity and exhibited a progressive decline in conductivity. The maximum conductivity attained was 3.08 S cm−1 from a bacterial cellulose-templated composite prepared in aqueous reaction media and afforded an approximately threefold increase in conductivity when compared with pure PPy at 1.14 S cm−1. Generally, the composites resembled wetting surfaces – with highly concentrated cuoxam solutions yielding improved hydrophilicity, while substitution of bacterial cellulose with nanocrystalline cellulose engendered a shift towards hydrophobicity. Most composites displayed a contact angle of less than 90° suggesting PPy/cellulose composites tended towards hydrophilic behavior. This study highlights investigations into the viability of cellulose solvents as a facile means to control the structure and performance of in situ functionalized cellulose nanocomposites.


Introduction
Intrinsically conducting polymers (ICPs) have attracted widespread interest due to their unique optical properties, exceptional electroconductivity, electrochemical activity, and biocompatibility. 1 Polypyrrole (PPy) is a particularly appealing candidate on account of its non-toxicity, resistance to oxidation, and environmental stability. 2,3 Application of this polymer, however, is limited by poor processability and inadequate mechanical properties. 2 Currently, there is growing interest in the hybridization of PPy with other materials to generate functional nanocomposites. [4][5][6][7] Coating a support matrix with ICPs is a ubiquitous approach used to prepare exible ICP lms or membranes. 8,9 Deposition of the polymer on templating bers (e.g. cotton, 10,11 silk, 12 polyamide, 13 polyester 14 ) to form conductive brous matrices has primarily been achieved through in situ oxidation. 2 Moreover, natural bers -particularly celluloseoffer a 'greener' alternative to conventional templating agents. 15 Cellulose is the most abundant biopolymer on Earth and comprises b-1,4-linked anhydroglucose units. 16 Consolidation of the unique properties of cellulose (biocompatibility, biodegradability, polyfunctionality and hydrophilicity) with the aforementioned electrical properties of PPy can yield synergistic effects. 17 Bacterial cellulose (BC) is a specic type of cellulose produced via bottom-up synthesis. A variety of Acetobacter and Gluconobacter bacteria orchestrate the conversion of biochemically activated glucose monomers into a 3-D network of ribbon-like BC bers with a cross-sectional width of 20-100 nm. [18][19][20][21][22] BC serves as a more sustainable alternative to plant cellulose because, unlike cellulosic biomass which undergoes extensive treatment and processing (to remove hemicelluloses and lignin), BC can be easily biosynthesized in diverse media to meet demand without post-treatment. [23][24][25] As a consequence of its unique architecture, BC possesses high purity, crystallinity and water-holding capacity, as well as a high degree of polymerisation. 16,23,26 Thus, BC also boasts much of the benets associated with nanocrystalline cellulose (NC), whilst avoiding time-consuming and costly processing associated with ner nanostructures. In the context of this study focusing on conductive composites, BC makes for an ideal scaffold. Generally, BC bers possess a relatively high aspect ratio coupled with School of Chemistry and Physics, University of KwaZulu-Natal, Westville Campus, Durban, 4000, South Africa. E-mail: vanzylw@ukzn.ac.za † Electronic supplementary information (ESI) available: Thickness measurements, TEM images, FT-IR and XRD spectra of PPy/cellulose; mechanisms of pyrrole polymerization and cellulose dissolution; synthesis of Schweizer's reagent; experimental details for preparation and characterization of cellulose; TEM, FT-IR and XRD analysis of cellulose; STEM-EDS analysis of PPy/cellulose composites (PDF). See https://doi.org/10.1039/d2ra04320c. a 3-D network structure. 27 Hence, conductive BC composites benet from a low percolation threshold, such that less material is required to achieve substantial conductivity which is hugely advantageous from both an economic and environmental perspective. Thus, we narrow our focus to PPy/BC nanocomposites. Several studies have also demonstrated the polymerization of pyrrole in a variety of reaction media, including ionic liquids, protic and aprotic solvents, and polymeric solutions. [28][29][30] Additionally, various solvents, including chloroform, methanol and tetrahydrofuran, facilitate cellulose swelling to a greater extent than water. Implementation of these solvents in the preparation of PPy/cellulose composites is reported to improve homogeneity and conductivity. 31 In this study we report a novel approach for the synthesis of polypyrrole/cellulose nanocomposites, using cuoxam-based solvent systems as reaction media. This solventalso referred to as Schweizer's reagentbears the chemical formula [Cu(NH 3 ) 4 (H 2 O) 2 ](OH) 2 . It has been demonstrated that this ammoniacal solution of copper(II) hydroxide can completely dissolve cellulose (and other polyols) under highly alkaline conditions. 32,33 In previous work we have successfully utilized this solvent for the in situ functionalization of cellulose containing a luminescent inorganic cluster, and we also demonstrated the formation of Kombucha-based bacterial nanocellulose embedded in a polypyrrole/PVA composite. 34,35 Due to the solubility of the template (i.e. cellulose) in cuoxam, it was anticipated that the resultant composites, prepared via a classic 'bottom-up' synthesis approach, would exhibit improved homogeneityand hence enhanced electroconductivity relative to analogous composites prepared in aqueous media. Furthermore, two distinct templating agents, namely (i) bacterial cellulose (BC) and (ii) nanocrystalline cellulose (NC), were implemented in the preparation of these composite materials. The unique structural arrangement of these materials in solution was expected to cause microstructural differences between the resultant PPy/BC and PPy/NC composites. It was predicted that this would also inuence electroconductivity and surface wettability. The effect of reaction duration on these properties was also investigated.

Materials
All commercially available chemicals were reagent grade and used without further purication. Pyrrole (C 4 H 5 N, $98%, M w ¼ 67.09 g mol À1 ) and copper(II) sulfate pentahydrate (CuSO 4 -$5H 2 O, $98.0%, M w ¼ 249.68 g mol À1 ) were obtained from Sigma-Aldrich Co. Ltd. Iron(III) chloride hexahydrate (FeCl 3 -$6H 2 O, $99%, M w ¼ 381.37 g mol À1 ) was obtained from ACE (Pty) Ltd. Potassium hydroxide (KOH, $88.7%, M w ¼ 56.11 g mol À1 ) was obtained from Set Point Laboratories. Hydrochloric acid (HCl, 30-34%, M w ¼ 36.46%) was purchased from Merck (Pty) Ltd. Ammonia gas (NH 3 ) was purchased from Afrox (South Africa). Deionized water was used in the preparation of all aqueous solutions. Suspensions of bacterial cellulose and nanocrystalline cellulose were prepared as described in the ESI. † Powder X-ray diffraction XRD measurements were obtained using a Bruker multipurpose X-ray diffractometer D8-Avance operated in a continuous q-q scan in locked coupled mode with copper radiation. The sample was mounted in the center of the sample holder on a glass slide and levelled up to the correct height. The measurements were obtained at a typical step size of 0.034 in 2q. A position-sensitive detector, Lyn-Eye, was used to record diffraction data at a speed of 0.5 s/step. The data was processed using EVA soware from Bruker.

Fourier transform infrared spectroscopy
Spectra were recorded on a PerkinElmer Spectrum 100 FT-IR spectrometer tted with a universal attenuated total reectance sampling accessory. The OMNIC™ 7.0 Professional Soware Suite (Thermo Scientic) was used to process the raw spectra.

Transmission electron microscopy
A tiny fragment of each solid composite was sonicated in water to produce a suspension. A small volume of the suspension was transferred onto a copper grid and le to dry for approximately ten minutes under a heat lamp. All samples were stained with a drop of uranyl acetate (1% w/v solution) to enable visualization of cellulosic material. TEM images were captured using a JEOL 2100 (Japan) high-resolution transmission electron microscope (HRTEM) operated at 20 kV. Composite thickness measurements were made using a digital Vernier caliper instrument with a resolution of 0.01 mm. The results are reported in Table S1. †

Synthesis of polypyrrole/cellulose composites
A series of polypyrrole/cellulose composites were prepared in two types of reaction media: (i) H 2 O (A) and (ii) cuoxam solutions of varying concentration (B, C, D). Additionally, the composites incorporated one of two cellulose starting materials: (i) bacterial cellulose or (ii) nanocrystalline cellulose, denoted BC and NC, respectively. Reaction duration was xed at 50 minutes or 20 hours, denoted 1 and 2, respectively. Preparation and characterization of the cellulosic materials is described in the ESI. † The various chemical compositions and reaction conditions implemented for the synthesis of these materials are recorded in Table 1. The modied procedure outlined below is adapted from the methodology reported by Wang et al. 36 Preparation of the oxidant/dopant solution

Synthesis of composites in H 2 O (A)
A mass of BC (1) or NC (2) (50 mg) was diluted to a nal volume of 150 mL. The aqueous cellulose suspension was stirred for 20 minutes to facilitate dispersion of the cellulose. Thereaer, a volume of pyrrole (500 mL) was added all at once and stirring was continued for a further 20 minutes. The reaction vessel was then transferred to an ice bath. Once cooled to 0 C, the oxidant/ dopant solution was added dropwise under vigorous stirring. The reaction was allowed to proceed under dark conditions for a duration of 50 minutes (X) or 20 hours (Y) while maintaining a constant temperature. The resulting grey/black suspension was ltered in vacuo and washed with water until the ltrate became colorless. This was followed by further washing with aliquots (10.0 mL) of acetone and HCl (1 M). The composites were then transferred to a desiccator to dry. Thereaer, the composites were separated from the lter paper, transferred to glass vials, and stored in a desiccator at room temperature.

Synthesis of composites in cuoxam solutions (B-D)
The feeding mass ratios of cellulose to CuSO 4 $5H 2 O were 1 : 0.5, 1 : 3 and 1 : 20 and the corresponding composites were labelled B, C and D, respectively. A typical procedure for the preparation of samples labelled B involved the dissolution of CuSO 4 $5H 2 O (25 mg, 0.10 mol) in water. An aqueous KOH solution (0.23 g, 4.1 mmol) was added to the CuSO 4 $5H 2 O solution under vigorous stirring to form a pale blue precipitate (i.e., Cu(OH) 2 ). Ammonia gas was bubbled through the precipitate under stirring, resulting in a dark blue solution (i.e., [Cu(NH 3 ) 4 (H 2 O) 2 ](OH) 2 or Schweizer's reagent). A mass of BC (1) or NC (2) (50 mg) was added to the cuoxam solution and diluted to a nal volume of 150 mL. The cellulose/cuoxam solution was stirred for 20 minutes to facilitate dissolution. Thereaer, a volume of pyrrole (500 mL) was added all at once, and stirring was continued for a further 20 minutes. The reaction vessel was then transferred to an ice bath. Once cooled to 0 C, the oxidant/ dopant solution was added dropwise under vigorous stirring. The reaction was allowed to proceed under dark conditions for a duration of 50 minutes (X) or 20 hours (Y) while maintaining a constant temperature. The resulting turbid brown solution was poured into an excess of 32% HCl solution, resulting in the gradual precipitation of black brous material in a clear yellow solution. The black solid was ltered in vacuo and washed with water until the ltrate became colorless. This was followed by further washing with aliquots (10.0 mL) of acetone and HCl (1 M). The composites were then transferred to a desiccator to dry. Thereaer, the composites were separated from the lter paper, transferred to glass vials, and stored in a desiccator at room temperature.

Sheet resistance measurement
Sheet resistance measurements were made using a four-point probe (FPP) head (Jandel Cylindrical Four Point Probe) mounted on a height-adjustable stand. The probe was connected to a source measure unit instrument (Keithley 2450 SourceMeter), and measurements for each composite were recorded at an operating current of 1 mA. Multiple readings for each sample were obtained by adjusting the position of the probe across the surface of the lm. The average of these results was recorded.

Contact angle measurement
The contact angle measurements were obtained by the following procedure. The composite lm was xed to a at surface. A drop (10 mL) of double distilled water was placed on the surface of the lm being tested. Aer 10 seconds, a digital photograph of the interface between the drop and the specimen surface was captured using an Apple iPhone X mobile phone with a 12 MP (megapixel) wide-angle camera tted with a detachable magnication lens. The photographs were digitally processed to meet soware requirements. The contact angle was determined using ImageJ soware in conjunction with the drop analysis soware plug-in developed by Stalder et al. 37 For each specimen, several measurements were made at several sites on the sample surface to obtain an average, reported as the nal result.

Synthesis of polypyrrole/cellulose composites
The proposed mechanisms for PPy chain-formation are described in Scheme S1. † Regardless of the true polymerization mechanism, it is evident that positively charged intermediates are involved in PPy synthesis. Therefore, it was hypothesized that templating agents with anionic moieties would improve the binding of the nal PPy product to the template through the presumed coulombic attraction of intermediates during polymerization. This, along with the ability of cuoxam to uniformly disperse the cellulose template, was the rationale behind the selection of this cellulose solvent as the reaction medium. As described by the complexation mechanism proposed by Burchard et al., the extraction of protons at O-2 and O-3 (Scheme 1) drives the formation of a stable polyolatocopper complex cellulose under highly alkaline conditions. 38 Accordingly, the polymeric chain units develop a negative charge. The generalized fabrication process for the preparation of PPy/BC composites is illustrated in Scheme 2. The procedure followed for the formation of PPy/NC composites was identicalexcept Feeding mass ratio of cellulose: The reaction duration was either 50 minutes or 20 hours. b BC and NC denote bacterial and nanocrystalline cellulose, respectively.
for the substitution of NC for BC. Initially, the BC material was introduced into the solvent system and stirred thoroughly to promote dispersion (or dissolution in cuoxam-based reaction media). This was followed by the addition of the pyrrole (Py) monomer solution and further mixing at room temperature. Sufficient time needed to be allotted to this step for maximal adsorption of Py onto the BC surface. The system was then cooled (to 0 C) prior to dropwise addition of the FeCl 3 /HCl solution. Low temperatures and controlled addition of the oxidant/dopant solution encourage more orderly growth of PPy and reduce aggregation. 39 As the reaction proceeded and PPy formed, the colorless mixture (aqueous reaction medium) transformed into a clear grey/black suspension of lamentous material. In contrast, the cuoxam-based solutionswhich were initially dark blue/black in colortransformed into turbid brown suspensions. Acidication of the cellulose-cuoxam chelate facilitates regeneration of the dissolved cellulose. The dative covalent bonds (Cu-O-2 and Cu-O-3, Scheme 1) can be broken via protonation of the cellobiose units (i.e., the addition of acid) to reprecipitate the cellulose bers. Hence, addition of these suspensions to a concentrated HCl solution was necessary to quench the excess cuoxam and simultaneously regenerate the PPy-coated BC bers. This transition was marked by the precipitation of black brous material in a clear yellow solution. This step was obviously unnecessary for composites prepared in aqueous reaction media, and thus omitted. Finally, the composite suspensions were ltered in vacuo and washed with water, acetone and HCl. Water and acetone were used to remove contaminants (e.g., unreacted Py or FeCl 3 ). In situ p-doping (oxidation) of the PPy chains was achieved through the nal washing step with HCl (dopant). Simultaneous diffusion of chloride anions (Cl À ) into the polymer maintains electroneutrality. 40 The delocalized positive charges (electron holes) enable charge transport between and along PPy chains to generate bulk conductivity. 41 Since humidity can diminish performance, long-term storage of the paper-like PPy/BC composites in a desiccator was essential.

Establishing morphological differences with TEM imaging
Effects of reaction media on morphology. The effect of reaction media on the morphology of the PPy/cellulose composites was studied by comparing the TEM micrographs obtained for BC-A1, BC-B1, BC-C1 and BC-D1. In BC-A1, a coresheath morphology was observedthe core being BC bers and the sheath being composed of fused PPy particles (Fig. 1a). The increase in ber diameter (D) from 40 nm (pristine BC) to approximately 78 nm (composite bers) corresponds to a relatively uniform PPy coating with an average thickness of 19 nm. This result is in keeping with previous studies on the preparation of PPy/BC composites in an aqueous environment. 42,43 In BC-B1, a sparse distribution of isolated PPy particles (D ¼ 44 nm) on the BC template was observed (Fig. 1b). Hence, even the presence of minute quantities of cuoxam in the solvent system signicantly affected the polymerization process. These ndings were most likely caused by reduced adsorption of pyrrole monomers on the BC bers due to a signicant portion of these surface sites already being occupied by cuoxamcomplexed with the cellulose backbone. Samples BC-C1 ( Fig. S1a and b †) and BC-D1 ( Fig. S1c and d †) both exhibited substantial inhomogeneity. Preparation of these samples occurred in highly concentrated cuoxam solutions. One would have expected that the improved dispersion of BC that occurs in this solvent environment would yield evenly coated and well separated PPy/BC bersas observed in other dispersing solvent systems. 36 Instead, these samples featured dense aggregates of PPy accompanied by regions of largely uncoated or virtually pristine BC bers. In these highly concentrated systems, the problems encountered with BC-B1 would most likely be further exacerbated. As the concentration of cuoxam increases, competition between Py monomers and cuoxam for access to the BC surface would likely intensify. Consequently, all of the Py monomers may end up distributed over a very limited (or at least reduced) surface area, resulting in the formation of dense aggregates upon polymerization. Concomitantly, once the solvent has been neutralized, the vast majority of the BC template would emerge in effectively pristine condition. Additionally, the strongly basic conditions associated with these solvent systems could have contributed to this morphology. Deprotonation of the -OH groups on both the BC and the Py monomers yields anionic derivatives. Coulombic repulsion between these species would have severely inhibited the adsorption of Py monomersprobably leading to the formation of unbound PPy aggregates. Following cellulose regeneration, adsorption of these aggregates onto the BC template is likely to have ensued.
Effects of reaction duration on morphology. The effect of reaction time was studied by comparing the sample series labelled 1 (50 minute reaction duration) with their equivalents (in terms of the solvent system) labelled 2 (20 hours reaction duration). Evaluating the micrographs of BC-D1 (Fig. S1c and  d †) and BC-D2 ( Fig. 2a and b), it became apparent that reaction duration affects the microstructure of the PPy deposits. Aer 50 minutes, a sparse distribution of aggregated PPy was observed. Meanwhile, aer 20 hours a more uniform and even distribution of PPy on the BC template was observed. Thus, protracted polymerization appeared to improve the homogeneity of the PPy/BC composite. This result is supported by ndings by Müller et al. 2 Effects of template structure on morphology. The effect of structural variation of the cellulose template on the PPy/ cellulose composites was studied by comparing samples: BC-A2 with NC-A2, and BC-C2 with NC-C2 (where bacterial cellulose-and nanocrystalline cellulose-templated samples are denoted BC and NC, respectively). The pristine BC sample existed as a ribbon-like network of long nanobers, with an average diameter of 40 nm. By contrast the pristine NC sample, produced via sulfuric acid-catalyzed hydrolysis of the BC sample, was comprised of rod-like cellulose nanocrystals. The NC particles possessed an average diameter and length of 10 nm and 0.3 mm, respectively. TEM, FT-IR and XRD analysis of the cellulose templates is contained in the ESI. † In water, bacterial cellulose forms an entangled net-like structure, while unfunctionalised cellulose nanocrystals tend to agglomerate (especially at high concentrations). These discrepancies affect template accessibility for pyrrole adsorption during in situ polymerization. Solution behavior in cuoxam is discussed later. In sample BC-A2, deposition of PPy was evident, and the BC bers appeared to be thoroughly entangled (Fig. 3a).
This could be due to additional hydrogen bonding interactions between proximate polypyrrole depositsas well as nearby cellulose chainsleading to greater cohesion of the composite bers. While in NC-A2, some nanobres appeared to be uniformly coated with PPy, while others remained uncoated ( Fig. 3b-d). Residual sulfate ester moieties on the NC surface improve the colloidal stability (i.e., dispersibility) of these nanoparticles in aqueous solutions. This effect could have been responsible for the observed uniformity of the PPy coating of NC-A2. Nonetheless, great efforts were made to eliminate this surface functionality (0.20 at%S, Table S2 †). Thus, localized aggregation of these nanoparticles via hydrogen bonds would still be conceivable, thereby limiting Py adsorption. This could account for the emergence of uncoated nanobres. Samples BC-C2 and NC-C2 were prepared in concentrated cuoxam reaction media. A sparse distribution of PPy particles anchored on the surface of the BC network was observed for BC-C2 (Fig. 4a). A much more uniform and even coating of the NC bers was observed for NC-C2 (Fig. 4b).   Fig. 3b-d). Both composites were prepared in water with a 20 hours reaction duration.
The effect of molecular weight on polymer chain behavior in cuoxam could potentially explain these observations. Seger et al. described how intramolecular crosslinking and a reduction in chain stiffness (caused by imperfections in copper binding to cellulose segments) can lead to an entangled or coiled solution structure in high molecular weight samples (Fig. 5). 44 Expression of this solution behavior could sterically hinder the adsorption of pyrrole onto the BC template. This phenomenon is not observed in low molecular weight cellulose which explains why the NC-templated sample featured a more homogeneous PPy coating.
Characterization using FT-IR and XRD. FT-IR spectroscopy is a useful tool for the characterization of surface bonding in materials. Additionally, covalent interactions between composite components can be evaluated. The spectra obtained for composite series 1 are presented in Fig. 6. Characteristic bands at 3342 cm À1 (O-H stretch) and 2915 cm À1 (asymmetric C-H stretch) are usually observed in a pure BC spectrum. 2 Similarly, N-H stretching vibrations are responsible for a broad band at 3338 cm À1 in pure PPy spectra. 45 In the composite spectra, these absorbance bands appeared as a single broad peak shied to lower energies. This red-shi was likely due to hydrogen bonding interactions between the N-H and O-H moieties in cellulose and PPy, respectively. 46 The composite absorbance patterns were largely comprised of overlapped absorbance bands contributed by each component. Nevertheless, a number of characteristic PPy bands (a-d) were apparent. The fact that these PPy peaks were well-dened in the composite spectra conrms that the polymerization of pyrrole was indeed accomplished. 47 The peak assignment 2,48-50 and measured energies of these absorbances are summarized in Table 2.
Relative to the other samples, the spectrum obtained for BC-A1 was red-shied. Furthermore, quenching of C-H (aliphatic) and O-H (hydroxyl) stretching vibrations, associated with cellulose, was evident. These ndings suggest that the BC bers in the BC-A1 sample were well-coated or insulated with a layer of PPy. 36 This agrees with the previously discussed results of TEM analysis. The energy shi could be caused by a varying degree of doping between samples and the consequent inuence of the counter-ion on bond strengths. 49,51 Similar observations were made for series 2 (Fig. S3 †). Notably, there was no discernible difference between the spectra of the NC-templated composites (NC-A2, NC-C2) and their BC-templated counterparts (BC-A2, BC-C2).
The XRD spectra obtained for series 1 and series 2 (along with pristine BC for reference) are presented in Fig. 7 and S4, † respectively. All of the diffraction peaks observed in the  composite spectra could be indexed to characteristic peaks of cellulose I. 52 The broad reection around 15.2 was attributed to the overlap of (110) and (110) reections. 53 The sharp shoulder peak at $22.5 corresponded to the (200) diffraction plane. The considerable background, in the range of 25-35observed only in the composite spectrawas likely caused by contributions from the broad diffraction peak of amorphous PPy. 54 Compared with pristine BC, signicant dampening of reections was observed in the PPy/BC spectra. In keeping with previous ndings, this indicates that the cellulose surface was at least partially coated with PPy. 45 Effects of morphology on electrical conductivity. According to percolation theory, electrical conductivity is primarily dependent on the microstructure of the composites and their ability to form a conductive network. 55 Improved connectivity is accompanied by a reduction in electrical resistance. This principle is central to the discussion that follows. The sheet resistance (R s ) values of the composite lms are reported in Table 3. Electrical conductivity (s) is also reported for ease of comparison with literature.
These values were obtained by eqn S(3), † where t denotes the measured thickness of each sample. The electrical conductivities vary over a fairly narrow range in the order of 10 À3 to 10 0 S cm À1 . The maximum conductivity was attained by BC-A2 (3.08 AE 0.37 S cm À1 ), a bacterial cellulose-templated composite prepared in aqueous reaction media for a duration of 20 hours. This compares well with the maximum of 7.34 S cm À1 obtained for similar PPy/BC composites prepared in DMF/H 2 O reaction media. 56 Table 4 presents a comparison of this work with similar conductive, nanocellulose-templated polypyrrole composites reported in the literature. Moreover, this composite afforded a roughly threefold increase in conductivity, when compared with pure PPy (1.14 S cm À1 ). 57 This conrms the hypothesis that, given an aptly chosen template, the microstructure associated with templated PPy composites facilitates improved electron transport within the material. Meanwhile, BC-D1, the BC-templated composite prepared in the most concentrated cuoxam medium for a duration of 20 hours, yielded the poorest conductivity ((1.26 AE 0.45) Â 10 À3 S cm À1 ). Yet, this s value is comparable with that obtained for similar composite materials (1.4 Â 10 À3 S cm À1 ) that are purported to be sufficiently conductive for application in electrodes, electronic devices and sensors. 43 The conductivity of composites from series 1 and series 2 have been presented graphically (Fig. 8) to establish the effect of reaction media. A clear trend emerged: as the solvent environment was modied from an aqueous (A) to a highly concentrated cuoxam solution (D), the conductivity decreased signicantly from values of order 10 0 to 10 À3 S cm À1 . This trend can be attributed to the morphological differences elucidated by TEM analysis. As previously discussed, the introduction of a minute quantity of cuoxam elicited a substantial transformation from a relatively uniform (BC-A1) to a sparse distribution (BC-B1) of PPy particles on the BC network (Fig. 1).   Cellulose itself is considered to be a non-conductive material (insulator). Consequently, the sparsely distributed PPy particles are separated by this insulating material, thereby inhibiting the formation of conductive pathwayswhich would inevitably impede electron ow. Hence, the observed decrease in s values from BC-A1 to BC-B1. The further increase observed for BC-C1 and BC-D1 could also be rationalized by comparing composite morphology (Fig. S1 †). In these samples, dense aggregates of PPy were found. These aggregates serve as localized regions with excellent conductivity. Nevertheless, these few conductive hotspots were insulated by the BC ber matrix which remained virtually uncoatedand hence non-conductive. Thus, macroscopically, electrical conductivity was very poor and the s values became very small. In BC-D1, this problem was exacerbated (due to an even higher concentration of cuoxam) which accounts for the further decrease in conductivity. It is also apparent that series 1 exhibited higher s values than series 2. TEM analysis revealed that the increase in reaction duration from 50 minutes (X) to 20 hours (Y) improved the homogeneity of the PPy coating. It follows that a more even distribution of PPy throughout the composite would lead to a greater density of conductive pathways, ultimately improving bulk conductance.
In order to establish the impact of the dimensionality (or aspect ratio) of the cellulose template on conductivity, the s measurements of BC-A2, NC-A2, BC-C2 and NC-C2 have been presented graphically (Fig. 9). In an aqueous environment (A), the BC-templated composite possessed a higher s value, albeit relatively similar in magnitude, than the NC-templated composite. Meanwhile, in a moderately concentrated cuoxam solution (C), the opposite was found to be true. Once again, microstructural variations can be used to rationalize these results. Although NC-A2 displayed more uniform coating of the cellulose template than BC-A2, the presence of pristine NC bers could impede the conduction of electrons due to their insulating nature (Fig. 3a). Furthermore, the NC sample was isolated as short rod-like particles (length << 300 nm). In contrast, ultrane BC bers formed an entangled network. It is easy to comprehend how the intrinsic connectivity of BC would improve conductance. From previous arguments, the dense, even polymer coating of NC (NC-C2) can easily rationalize the disparity in s values between BC-C2 and NC-C2 (Fig. 9).
Effects of morphology on surface wettability. The contact angle (CA) values obtained for the composite lms are reported in Table 5. For clarication, all reported measurements refer to the water contact angle. Except for NC-A2 (nanocrystalline cellulose-templated composite prepared in water; 20 hours reaction duration), all of the composites displayed a contact angle less than 90 . Hence, the PPy/cellulose composites tended towards hydrophilic behaviorwith the best wetting behavior (CA ¼ 23.09 AE 1.08 ) exhibited by BC-D1 (bacterial cellulosetemplated composite prepared in high cuoxam concentrations; 50 minute reaction duration).
The CA measurements of composites from series 1 and series 2 have been presented graphically (Fig. 10) to establish the effect of reaction media on wettability. A common trend was apparent: as the solvent environment changed from an aqueous (A) to a highly concentrated cuoxam solution (D), the CA increased initially and then gradually began to decline. A maximum was reached in system B (lowest cuoxam concentration of B-D) for both the series 1 and 2. Additionally, the minimum CA was attained at higher cuoxam concentrations (C or D), suggesting that very high concentrations of cuoxam diminished the CA to lower values than those initially obtained for composites prepared in aqueous media. By implication, highly concentrated cuoxam reaction media can be implemented in the design of PPy/cellulose composite materials with improved surface wettability (or hydrophilicity). This could provide new opportunities for these materials in biomedical, ltration, and anti-fogging applications. 58  To rationalize this behavior, one needs to look at the effect of the structural anisotropy of cellulose on wettability, studied by Yamane et al. 59 Cellulose is constituted of anhydroglucose units (AGUs) which adopt a chair conformation. The axial positions are occupied by hydrogen atoms of C-H bonds, which creates hydrophobicity in the axial direction. In contrast, AGUs are hydrophilic in the equatorial direction since all three hydroxyl groups are equatorially bonded (Fig. 11). In the unit cell of regenerate cellulose, the AGUs are orientated such that the plane on which these rings lie is perpendicular to the (1 10) crystallographic planes. Hence, the surface of the (110) plane is associated with a high density of equatorially positioned hydroxyl groups, creating an exceptionally hydrophilic surface.
Additionally, experiments have shown that the (1 10) planes constitute a large fraction of the surface area of regenerated cellulose lms. 59 It follows that regenerated cellulose lms exhibit extraordinary wetting (i.e. low CAs). This could explain the observed decrease of CA measurements in increasingly concentrated cuoxam solutions. A higher cuoxam concentration enables the dissolution of a larger fraction of BC, and hence   a higher fraction of regenerated cellulose would be present in the nal lm. The consequent improvement in lm hydrophilicity is concomitant with the measurement of a reduced CA. Doping effects can potentially explain the disparity in measured contact angles for composites prepared in system A and system B. TEM analysis revealed that solvent A induced a relatively uniform and even coating of PPy particles on the BC network (BC-A1, Fig. 1a). A sparse distribution was evident in solvent B (BC-B1, Fig. 1b). Therefore, given the greater quantity of PPy present, one would also expect more counterions (Cl À ) from the dopant to be adsorbed on the surface of composites prepared in solvent A. Since the dopant ions are hydrophilic, improved wettability should result. This could explain why lower contact angles for BC-A1 and BC-A2 were observed. Series 1 and 2, corresponding to a reaction duration of 50 minutes and 20 hours, respectively, displayed the same trend with similar magnitudes. Therefore, it seems that reaction duration did not play a signicant role in wetting behavior. In order to study the inuence of the dimensionality (or aspect ratio) of the cellulose template on wettability, the contact angle (CA) measurements of BC-A2, NC-A2, BC-C2 and NC-C2 have been presented graphically (Fig. 12).
It is apparent that the NC-templated composites exhibited signicantly higher contact angles (at least twice as large) than the BC-templated composites. Moreover, in the case of BC-A2 (48.5 ) and NC-A2 (101.6 ), a marked shi from hydrophilic to hydrophobic wetting behavior was observed. TEM analysis revealed that, relative to their BC-templated counterparts, the NC-templated composites possessed a more uniform coating of PPy nanoparticles. This can translate into reduced droplet contact area, as well as composites with reduced droplet permeability. 7,60 Hence, one can rationalize the signicantly higher contact angles measured for the NC-templated composites (NC-A2 and NC-C2). By implication, control over the morphology of the cellulose template enables manipulation of the wetting properties displayed by the resulting composite. Furthermore, an array of composites which span a broad range of contact anglesencompassing both wetting and non-wetting behaviorscan be produced via manipulation of this experimental parameter. The reason for the general decline in CA observed from system A to system C has been discussed.

Conclusions
A series of polypyrrole/cellulose composites were prepared in cuoxam (or Schweizer's reagent) solutions of varying concentration. A detailed account of the synthesis and characterization of these composites is presented. This study investigated the effect of reaction media and cellulose template morphology on electrical conductivity and surface wetting. In general, the templated polypyrrole composites demonstrated hydrophilic behaviorwith the exception of nanocrystalline cellulosetemplated composites, exhibiting a bias towards non-wetting behavior. Hence, microstructural variation of the cellulose template serves as a facile approach to tailor composite surface wettability. Substitution of aqueous reaction media with highly concentrated cuoxam solutions elicited a pronounced improvement in the wetting of the composites, while low cuoxam concentrations improved hydrophobicity. Hence, this work highlights the ability of Schweizer's reagent to control the wettability of conductive cellulose composites. The composites exhibited reasonably good conductivities (maximum of 3.1 S cm À1 ), with several composites showing a response Fig. 11 Anisotropic structure of cellulose I. Red, white, and grey spheres represent oxygen, hydrogen and carbon atoms, respectively. An overhead view (a) shows that equatorial positions are occupied by hydroxyl groups, forming hydrophilic planes (blue) at the sides of the anhydroglucose unit ring plane. A side view (b) shows that axial positions are occupied by aliphatic protons, forming hydrophobic planes (green) above and below the anhydroglucose unit ring plane. Adapted from Yamane et al. 59 Fig. 12 Contact angle measurements of PPy/cellulose composite films templated on bacterial cellulose (BC-A2 and BC-C2) and nanocrystalline cellulose (NC-A2 and NC-C2), prepared in different solvent systems with a 20 hours reaction duration. A denotes water; C denotes moderate cuoxam concentrations.
superior to pristine polypyrrole. Increasingly concentrated cuoxam reaction media yielded composites with progressively inferior conductivities. This contradicts the initial hypothesis which stated that preparation in cuoxam would favor the adoption of homogeneous morphologies (uniformly polypyrrole-coated cellulose) with improved electroconductivity. However, this study does indicate that composite microstructure, wettability, and conductivity are highly sensitive to the cuoxam concentrations in reaction media and, hence, to the extent of cellulose dissolutionhinting at the potential application of cellulose solvents as an economical and facile means to ne-tune the morphology and properties of in situ functionalized cellulose nanocomposites. Further work exploring other types of cellulose solvents to form structurally and functionally distinct composites is underway.

Author contributions
The manuscript was written through contributions of all authors. All authors have given approval to the nal version of the manuscript.

Conflicts of interest
There are no conicts to declare.