Flexible and high performing polymer electrolytes obtained by UV-induced polymer–cellulose grafting

Abstract

A novel UV-induced procedure, in the presence of glycidyl acrylate and proper radical/cationic photoinitiators, is here proposed to promote the in situ grafting of a methacrylic polymer network to the surface of natural cellulose fibers. As a result, composite polymer membranes reinforced by cellulose handsheets are obtained and demonstrate excellent mechanical features. The reactivity of the monomers is studied by FT-IR in real time measurements and the successful grafting is confirmed by surface spectroscopies. The polymer membranes prepared show high elastic modulus and tensile resistance and maintain high flexibility even after activation by swelling into a standard liquid electrolyte solution. The swelling procedure allowed us to obtain high ionic conductivity and remarkable electrochemical behavior when tested in laboratory scale lithium polymer cells. The system shows attractive features such as intrinsic safety, eco-compatibility, and low production cost and industrialization potentials, and is highly suitable for the rapidly expanding field of Li-based flexible batteries.

---

Introduction

Due to their remarkable characteristics, lithium ion batteries are the system of choice for energy storage.^1–3 Although these batteries are nowadays well established commercial products, further research and development are always required to improve their performance in order to meet the stringent market requirements. The search for new and optimized materials for both the electrodes and the electrolyte is the main focus, in view of enhancing safety, energy density and cost.^4–6

A big effort in the research and optimization of existing battery technologies is put in the development of light, flexible and portable devices. If the essential components (electrodes, spacer and electrolyte) of the electrochemical cells can be made intrinsically safe and flexible, this would enable to embed them in a wide range of innovative products such as smart cards, displays and implantable medical devices.^4,7,8 Considering the electrolyte, all-solid and/or quasi-solid polymer membranes are nowadays proposed: they are flexible and at the same time they guarantee safety with respect to the liquid electrolytes of common Li batteries.^2,9,10

Different techniques have been used to produce these membranes, among them UV-curing. This technology is used in many industrial fields such as inks and coatings, optical and electronics, being fast, low cost and versatile and it is well established also in paper-making industry as printing technique.^11,12 In particular it has been chosen here because fully cured membranes can be obtained in seconds at room temperature irradiating a proper mixture of reactive molecules and photoinitiator.¹³

Additives, salts and, eventually, proper solvents may be directly added along with the reactive molecules. Alternatively, the solid membranes can be prepared, and then be activated by swelling them by means of a lithium salt solution in a dedicated step. This strategy is more convenient at the industrial level as it should be preferable to manipulate the lithium salt in inert atmosphere.¹⁴

Acrylic and methacrylic based gel polymer electrolytes obtained by swelling photopolymerized membranes in a liquid electrolyte¹⁴ show low glass transition temperatures and good electrochemical behaviors, comparable to those of other swelled polymers,^15,16 above all porous PVFD-HFP membranes.¹⁷ When compared to these latter, which are among the most common gel polymer electrolytes, the main advantage of a photocured membrane is the extremely fast preparation which, furthermore, does not involve the use of solvents.

In order to improve the limited mechanical properties of the UV cured membranes cellulose fibers^18,19 or cellulose paper²⁰ can be proposed as reinforcing agents. Cellulose is the most abundant natural polymer and its fibers present high mechanical properties and chemical stability. Their use and treatments are deeply known in papermaking industry, which allows to obtain materials and substrates having relatively low costs and properties easy to be tailor-made.^21,22 Paper as a flexible material is also already in use for the development of electrodes²³ and paper-based batteries.^20,24,25

Cellulose fibers and paper are very promising, however in the case of UV-cured membranes, the swelling step (described above) may seriously affect the adhesion between the polymer network and the cellulose fibers. Therefore, for enabling the swelling procedure, the interface between the polymer matrix and the cellulose reinforcement must be properly modified. An attempt is here made to obtain a covalent bonding between the polymer matrix and the cellulose substrate via a grafting procedure using glycidyl acrylate. Indeed, both glycidyl acrylate (GA) and methacrylate (GMA) have been largely used to modify polymer surfaces.^26–29 Such monomers presenting a (meth)acrylic group can be grafted onto cellulose and/or copolymerize with the photocurable formulation. They also present a highly reactive epoxy group; such a group, upon opening, can give a chain transfer reaction to the hydroxyl groups of cellulose.³⁰

In the present work, GA is added to a photo-curable acrylic formulation for the preparation of composites polymer electrolytes reinforced by cellulose handsheets. The grafting of the polymer matrix to the cellulose substrate is thoroughly evaluated and characterized. The resulting membranes are then activated by swelling in a standard liquid electrolyte solution and tested as quasi-solid polymer electrolytes in lab-scale lithium cells showing improved performances.

Experimental section

Materials

Cellulose fibers derived from Hardwood (HW, eucalyptus) and Softwood (SW, pine) plants were used to produce paper sheets. Bisphenol A ethoxylate (15 EO/phenol) dimethacrylate (BEMA, avg. M_n = 1700, Aldrich), poly(ethylene glycol) methyl ether methacrylate (PEGMA, avg. M_n = 1100, Aldrich), glycidyl acrylate (GA, Aldrich) were used as reactive monomers. 2-Hydroxy-2-methyl-1-phenyl-1-propanone (Darocure 1173, Ciba Specialty Chemicals) was used as the free radical photoinitiator (RPI); triarylsulphonium exafluoro phosphate (Aldrich) was used as cationic (CPI). Lithium bistrifluoromethane-sulfonimide (CF₃SO₂NLiSO₂CF₃, LiTFSI, Aldrich) salt was dissolved in a 1 : 1 w/w ethylene carbonate–diethyl carbonate (EC–DEC, Fluka) solution in order to obtain a 1 M solution of standard liquid electrolyte.

Before their use, the liquid electrolyte components were kept open in the inert atmosphere of a dry glove box (MBraun Labstar, O₂ and H₂O content <0.1 ppm) filled with extra pure argon for several days and also treated with molecular sieves (molecular sieves, beads 4 Å, 8–12 mesh, Aldrich) to ensure the complete removal of traces of water/moisture.

Methods

Handsheet preparation. Untreated cellulose fibers (Softwood and Hardwood fibers mixed in the 40 : 60 ratio) stored in the laboratory in the form of thick sheets were re-pulped and blended using a high speed blender. The resulting suspension of fibers was then submitted to the refining treatment to reach a degree of refinement of 35° SR (Schopper-Riegler).³¹ This mechanical modification was done by a Valley beater (equipment for pulp refining), according to ISO 5264-1 standard, in order to beat the pulp in a uniform and reproducible way.

To obtain the handsheets, a suspension of fibers (1 L) having a concentration of 1.5 g L⁻¹ was prepared, it was introduced into a sheet-former and stirred by bubbling for about 3 minutes. The filtrate laying on copper wires was then dried at 90 °C for 7 minutes under high vacuum to give handsheets (hereafter named SW40) of about 1.5 g in weight corresponding to a grammage of about 51 g m⁻² and thickness of 110 ± 2 μm.

Sample preparation for grafting evaluation. A GA (5 wt%)–acetone (95 wt%) solution was firstly prepared, then 2 wt% of radical photoinitiator and 1 wt% of cationic photoinitiator were added. The SW40 handsheet was soaked for 5 minutes into the mixture and irradiated 1.5 minutes at each side. The sample (named G-SW40) was strongly washed several times with acetone and dried until a constant weight was confirmed.

Reinforced polymer electrolyte membrane preparation. The reactive formulations were prepared in the dry glove-box, by mixing BEMA and PEGMA together with EC–DEC and the radical photoinitiator; when needed, 5 wt% of GA was added together with the cationic photoinitiator. Different formulations were then casted onto the SW40 cellulose handsheet and spread with a bar. The impregnated handsheet was placed in a quartz tube sealed in dry box and UV irradiated. The irradiation was done for 3 minutes by using a medium vapor pressure Hg lamp (Helios Italquartz, Italy), with a radiation intensity of 30 mW cm⁻². The composition of the samples prepared is shown in Table 2. The resulting composite membranes were activated by soaking for 15 minutes in a 1 M solution of LiTFSI in EC–DEC.

Table 1 Relative areas of the C1, C2, C3 and C4 peaks from high resolution C1s peak XPS analysis on the different samples

Peak	Pristine SW40 (%)	Pure GA (%)	G-SW40 grafted sample (%)
C1 at 283 eV	30	54	36
C2 at 285 eV	44	35	42
C3 at 286 eV	26	—	19
C4 at 287 eV	—	11	3

---

Table 2 Composition of the samples reinforced with SW40 handsheet (SR 35°, 52 g m⁻²)

Sample	BEMA (wt%)	PEGMA (wt%)	GA (wt%)	EC–DEC (wt%)	Radical photoinitiator (wt%)	Cationic photoinitiator (wt%)
A	50	50	—		+3	—
B	47.5	47.5	5		+3	—
C	47.5	47.5	5		+3	+3
D	25	25	5	45	+3	+2

---

Characterization methods. The quantity of polymer present in the reinforced membrane with respect to the total weight, called active swelling percentage (AS%), was calculated by the formula:

AS% = [(W_GPE − W_pap)/W_GPE] × 100

where W_GPE is the weight of the final reinforced membrane and W_pap is the weight of the reinforcing substrate.

The kinetics of the photo-polymerisation process was investigated using a NICOLET-5700 Real Time FT-IR instrument, which collects spectra in real time while the sample is irradiated by UV light, following the decrease of the band attributable to the acrylate and epoxy groups. The tests were carried out at ambient temperature on an UV transparent SiC wafer by irradiating the mixtures of monomers for 3 min. The UV lamp used was Lightning Cure LC-8 with an intensity of 15–16 mW cm⁻² (intensity measured by an ORIEL photometer).

To investigate the presence of grafted polymer at the surface of the cellulose handsheet, a FT-IR spectrophotometer (NICOLET-5700) was used in ATR mode, by using a diamond prism. Analyses were performed on the handsheets before and after the grafting procedure. Each spectrum was collected by cumulating 64 scans at a 1 cm⁻¹ resolution. Spectra were collected in 5 different points of the sample.

X-ray photoelectron spectroscopy (XPS) analyses were performed by a scanning ESCA microprobe PHI 5000 Versaprobe, with a monochromatic X-ray beam, with an Al Kα source (1486.6 eV). All samples were analyzed with a combined electron and argon ion gun neutralizer system in order to reduce the charging effect during the measurements. The survey scans were acquired with pass energy of 187.85 eV, whilst the high resolution scan with 23.50 eV. The X-ray beam size was settled at 100 μm for high sensitivity. The semi-quantitative atomic compositions were obtained using Multipak 9.0 dedicated software. In the adopted procedure the peak area of the element was corrected by the respective sensitivity factors.

The electrolyte uptake, indicating the quantity of liquid electrolyte present in the final composite polymer electrolyte is here called swelling percentage (SP%). It was calculated by the formula:

SP% = [(W_f − W_i)/W_f] × 100

where W_f is the weight of the membrane after swelling and W_i is the weight before swelling. The mechanical properties of the gel polymer electrolytes after swelling were evaluated by bending test: samples were consecutively rolled up around cylindrical hoses with radii ranging between 3 and 32 mm.

The glass transition temperature (T_g) was evaluated by differential scanning calorimetry (DSC) with a METTLER DSC-30 (Greifensee, Switzerland) instrument, equipped with a low temperature probe. Samples were put in aluminum pans, prepared in a dry glove box. In a typical measurement, the electrolyte samples were cooled from ambient temperature down to −80 °C and then heated at 10 °C min⁻¹ up to 120 °C. For each sample, the same heating module was applied and the final heat flow value was recorded during the second heating cycle. The T_g was defined as the midpoint of the heat capacity change observed in the DSC trace during the transition from glassy to rubbery state. The thermal stability was tested by thermo-gravimetric analysis using a TGA/SDTA-851 instrument from METTLER (Switzerland) over a temperature range of 25–800 °C under N₂ flux (60 mL min⁻¹) at a heating rate of 10 °C min⁻¹.

The ionic conductivity tests were carried out on a heating stepped ramp from 20 to 80 °C. The cells (Electrochemical Test Cell model EL-Std purchased from EL-Cell GmbH), assembled by sandwiching the electrolyte sample within two stainless-steel blocking electrodes, were housed in a Memmert GmbH oven model UFE-400 with a temperature control of ±1 °C. The resistance of the electrolyte was given by the high frequency intercept determined by analyzing the impedance response using a fitting program provided with the Electrochemistry Power Suite software (version 2.58, Princeton Applied Research). Each sample was equilibrated at the experimental temperature for about 1 h before measurement, to allow thermal equilibration of the cells. All measurements were carried out on at least three different fresh samples in order to verify the reproducibility of the obtained results.

The electrochemical stability window (ESW) was evaluated at ambient temperature by linear sweep voltammetry (Li metal as the reference electrode, Cell model: El-Ref) using an Arbin Instrument Testing System model BT-2000. Separate tests were carried out on each polymer electrolyte sample to determine the cathodic and anodic electrochemical stability limits. The measurements were performed by scanning the cell voltage from the open circuit voltage (OCV) towards 0.0 V vs. Li (cathodic scan) or 5.5 V vs. Li (anodic scan). In both cases, the potential was scanned at a rate of 0.100 mV s⁻¹. The current onset of the cell was associated with the decomposition voltage of the electrolyte. Cell configuration adopted for anodic scan: acetylene black (Shawinigan Black AB-50, Chevron Corp., USA.) over Al current collector and Li metal as electrodes and the given membrane as electrolyte (active area equal to 2.54 cm²); cathodic scan: Cu foil and Li metal as electrodes and the given membrane as electrolyte (active area equal to 2.54 cm²).

The laboratory scale lithium polymer test cells were assembled by contacting in sequence a lithium metal disk anode, a layer of the reinforced polymer electrolyte and a LiFePO4/C disk composite cathode (electrode area: 2.54 cm²). The latter was prepared by a quick and low cost mild hydrothermal synthesis.³² The electrodes/electrolyte assembly was housed in the EL-Std cell. Both electrode fabrication and cell assembly were performed in the environmentally controlled Ar-filled dry glove box. The lithium cell was tested for its electrochemical performance at ambient temperature in terms of charge/discharge galvanostatic cycling using an Arbin Instrument Testing System model BT-2000. The potential cut-offs were fixed at 4.0 V vs. Li/Li⁺ (charge step) and 2.5 V vs. Li/Li⁺ (discharge step), respectively.

Results

Study of the GA–cellulose grafting and membrane preparation

A preliminary study on the reactivity of the functionalities of GA (acrylic and epoxy groups), was done by FT-IR spectroscopy in real time and the influence of both the radical and the cationic photoinitiator was evaluated. The instrument collected IR spectra while a sample of pure GA, with the desired initiator, was irradiated by UV light; kinetics of the conversions was determined by following the decrease of the band attributable to the double bond of the acrylate groups at 1640 cm⁻¹ and the one of the epoxy groups at 760 cm⁻¹. The C=O bond at 1730 cm⁻¹ was used as an internal standard. The conversion of the reactive functions was calculated as follows:where A_{reactive f,t} represents the peak areas under the reactive functionality bands at the time t, A_{reactive f,0} is the area at the beginning of the reaction and A_ref represents the area under the band of the internal reference at the time = t and at the beginning of the reaction (this area should not vary). The rate of reaction and the polymerization conversion of GA with different kinds of photoinitiators is shown in Fig. 1.

Fig. 1 Kinetic plots of the conversion of the acrylic and epoxy groups of GA: (a) in the presence of a radical photoinitiator 2% w/w, (b) in the presence of a cationic photoinitiator 1% w/w and (c) in the presence of both photoinitiators.

Plot (a) in Fig. 1 shows the conversion of the double bond of the acrylate (red line) as well as the ring opening reaction of the epoxy group of GA (black line) in the presence of the radical PI (RPI). It is possible to observe that both functionalities react. The acrylic double bond polymerization is initiated by the radicals formed in the photodissociation of the RPI (Scheme 1).

Scheme 1 Photodissociation of the radical initiator.

By exploiting the exothermicity of the radical double bonds addition, the conversion of the epoxy group is activated by a thermal process. Both processes are fast and reach their maximum conversion in few seconds. Nevertheless, the conversion of the groups is not complete, that is slightly above 70 and 50% for the acrylic and the epoxy group, respectively.

Both the epoxy and the acrylic group react also in the presence of the cationic photoinitiator (CPI), as shown in plot (b) of Fig. 1. The epoxy ring polymerizes by the protons generated upon photodissociation of the onium salt.

As shown in Scheme 2, radicals are generated as well, and they initiate the polymerization of the acrylate functionality.

Scheme 2 Photodissociation of the onium salt.

Plot (c) in Fig. 1 shows that both a faster reaction of GA and a higher conversion of both acrylic and epoxy functions are reached by using the two initiators together. Indeed, the acrylic groups reach in a few seconds a percentage of conversion of about 70%, which then remains constant regardless of the continuous irradiation; a percentage of about 75% is obtained for the epoxy ring conversion, which keeps growing finally reaching the 98% after 2 minutes of irradiation.

Based on these results, the grafting of cellulose by GA was done in the presence of both PIs. The interaction between GA and cellulose after UV irradiation was firstly evaluated by analysis of the G-SW40 sample specifically prepared as previously detailed. The successful grafting was firstly monitored by measuring the increase of weight of the handsheet treated with GA. The weight gain measured was about 5%, corresponding to 0.39 mmol of GA per gram of cellulose. FT-IR ATR analysis spectra were then collected (see green line in Fig. 2) and also compared to the spectra of pristine SW40 paper (blue line) and pure GA (red line) before polymerization. The main peaks present in the GA profile are due to the acrylic group (1735, 1640 cm⁻¹) vibrations and to the asymmetric and symmetric epoxide stretches (913, 844 and 760 cm⁻¹).³⁰ Cellulose, due to its structure, shows several characteristics absorption bands; the peaks present between 895–1125 cm⁻¹ are due to C–O stretching and ring vibration modes; at 1162 cm⁻¹ is visible the antisymmetrical bridge oxygen stretching while peaks between 1500–1300 cm⁻¹ are attributable to the C–H bending.³⁴ Upon grafting, the disappearance of the peaks at 913, 844 and 760 cm⁻¹ related to the epoxy ring and the simultaneous disappearance of the peak at 1640 cm⁻¹ due to the double bond in the acrylic group are observed. This evidences that the reaction of both groups has occurred, as expected from the previously shown FT-IR analysis. The peak at 1735 cm⁻¹ characteristic of the ester group of GA is present also after irradiation, thus confirming the effective presence of the monomer on the paper surface.

Fig. 2 FT-IR analysis performed in ATR mode on pure GA (red line, top), pristine SW40 paper (black line, middle) and G-SW40 grafted sample after washing in acetone (green line, bottom).

The grafting of GA onto cellulose was further evaluated by XPS analysis (see Fig. 3 and also Fig. S1 in ESI†).³³ XPS was performed on the G-SW40 grafted sample (green line in Fig. 3) and the results compared to those obtained for the pristine SW40 paper (black line) and the pure GA after irradiation in the presence of the two PIs (red line). The survey spectra are shown in Fig. S1.† As expected, they show the presence of oxygen (peak at 531 eV) and carbon (peak at 285 eV). The relative intensity of the oxygen signal and, correspondingly, the O/C ratio decrease in the GA grafted sample, thus accounting for the presence of the monomer onto the cellulose substrate. The core line spectra of carbon were then thoroughly analyzed. The C1s binding energy depends on the number of bonds between carbon and oxygen atoms. Plot (a) in Fig. 3 shows the C1s peaks of the three samples at high resolution, where each peak gives an indication of the kind of bonds between the atoms. Generally, the signal of carbon can be resolved into four different peaks: C1 (283 eV), that originates from C atoms bonded only to other C atoms or H atoms and the C2 (285 eV), C3 (286 eV) and C4 (287 eV) peaks that originate from C atoms bonded to one, two (or one double) and three O atoms, respectively.³⁴ Plots (b)–(d) of Fig. 3 show the same spectra with the deconvolution of the C signal and the fitting of each peak. The relative intensities measured for each sample from the area of the peaks are listed in Table 1.

Fig. 3 High resolution C1s XPS analysis. (a) Comparison of the C1s spectra for the different samples, (b) high resolution XPS spectra of the C1s peaks of pristine SW40 paper (glucose ring molecular structure in the inset), (c) pure GA sample (glycidyl acrylate molecular structure in the inset) and (d) G-SW40 grafted sample.

According to its molecular structure, the glycosidic ring of cellulose contains only C atoms bonded to oxygen along with a negligible amount of carboxylic groups; therefore, it mainly shows C2 and C3 peaks. In our case, the C1 peak is also visible; even if this peak is not expected, a trimodal C(1s) peak with a C1 component in addition to C2 and C3 contributions is universally observed.^35–37 When pure cellulose is analyzed, the presence of this peak is often associated with hydrocarbon contamination on the sample surface or radiation damage of the cellulose surface.³⁴ Alternatively, Hua et al.³⁸ suggested that part of the C1 peak intensity arises from C–C–O contributions that have a lower binding energy than the hydroxyl carbon. Here, the presence of the C1 peak may also indicate the presence of lignin. In plot (c) of Fig. 3, the spectrum obtained from pure glycidyl acrylate shows C1, C2 and C4 peaks, C3 being hardly detectable. Again, the peaks can be easily correlated to the GA molecular structure shown in the inset. The GA structure presents an ester bond (O–C=O), while no ether–ether bonds or hydroxyl groups are present. Two C atoms single-bonded to other C atoms are also present (C–C bond). When GA is grafted onto the cellulose handsheet surface the spectrum contains signals corresponding to C1, C2, C3 and C4, as shown in plot (d) of Fig. 3. Based on the GA structure, an increase in the intensity of the C1 and C4 peaks is expected while the relative intensity of the C2 and C3 peaks is expected to decrease in the grafted sample when compared to the pristine cellulose profile. This is in fact confirmed by the experimental results. The presence of the C4 peak related to the ester group in the G-SW40 grafted sample demonstrates the effective presence of the monomer on the paper surface, which remains even after intensive washing with acetone, thus accounting for the successful grafting.

Considering that in the complete system composed of cellulose, GA and the two PIs several reactions may take place at the same time, the following hypotheses can be proposed concerning the grafting mechanism. Firstly, the simple irradiation of the cellulose substrate may create active sites (radicals) at the surface of the handsheet³² that can react with the double bonds of the acrylic group of GA, thus resulting in a “grafted radical” at the cellulose surface. Upon UV irradiation, RPI may create free radicals that can react with the double bonds of the GA acrylic moieties. This may promote their reaction either with other free GA molecules, thus forming a homopolymer, or with the “grafted radical” previously described, thus anchoring the GA molecule at the cellulose surface. At the same time, upon UV irradiation, CPI may create Brönsted acids and free radicals. In particular, Brönsted acids may promote the opening of the epoxy rings; as a result, the epoxy functionality may have the possibility to react either by creating homopolymers or by attacking the –OH groups at the cellulose surface, grafting the molecule on it by the chain transfer reaction described in Scheme 3.

Scheme 3 Chain transfer reaction for GA grafting onto the cellulose surface.

In the case of sample G-SW40 however the homopolymer created during the preparation of the samples was eliminated during the washing step.

During the preparation of the polymer electrolyte membrane, the system becomes even more complicated due to the simultaneous addition of both the oligomer BEMA and the monomer PEGMA. The concurrent grafting and polymerization reactions take place, along with the formation of homopolymer thus creating a complex network; nevertheless, also for the complex network, an improved adhesion of the polymer to the cellulose substrate is demonstrated by the experimental observations reported below.

Different polymer membranes composed of ethylene-glycol-based methacrylic oligomers and reinforced with the SW40 handsheet were prepared in order to observe the influence of GA and the different photoinitiators on the complete system. The samples prepared are summarized in Table 2. The active swelling percentage, meaning the weight of polymer present in the membrane with respect to the total weight (polymer and cellulose) resulted to be of about 70% for all the samples.

A preliminary investigation devoted to study the polymerization of the various formulations was carried out (see Section S2 in ESI†), confirming also for the complete system that the concurrent presence of the two photoinitiators led to the highest conversion of both the acrylic and the epoxy groups (i.e., 95 and 70%, respectively, after 3 minutes of irradiation).

The UV cured reinforced samples underwent a swelling test which gave the qualitative but direct evidence of the efficacy of the grafting of GA onto the cellulose substrate, fundamental for the production of swelled-reinforced membranes. After preparation, the reinforced membranes were immersed in 20 mL of an organic solution composed of ethylene carbonate and diethyl carbonate (EC–DEC) in the 50 : 50 ratio and evaluated for their integrity every 15 min. Fig. 4 shows the appearance of the different samples upon different swelling times.

Fig. 4 Appearance of the different samples upon different swelling times in a EC–DEC LiTFSI 1M solution: (A) sample A after 5 min of swelling, (B) sample B after 1 h of swelling, (C) sample C after 2 h of swelling, (D) sample D after 2 h of swelling.

As clearly visible, sample A (prepared without GA) seriously degraded after only 15 min of swelling; same for sample B (prepared without CPI) after 1 hour of swelling. On the contrary, both sample C and D (prepared with GA and both PIs) maintained their physical characteristics even after 2 hours of swelling and their swelling percentage was of 30 and 43 wt% respectively. In particular, a prolonged test showed that sample D can be left even 24 hours in the organic solution without compromising its structure, being still flexible and easy to be handled as shown in image D of Fig. 4. These observations demonstrate that the presence of GA can really improve the stability of the reinforced membranes. More important, the good stability of both samples C and D demonstrated also that the cationic photoinitiator is fundamental for the successful grafting of the polymer to the cellulose substrate. This indicates that the chain transfer reaction sketched in Scheme 3, which occurs only in the presence of both the PIs, must play the most important role in the grafting procedure.

Characterization of the reinforced composite polymer electrolyte (CRPE)

Based on the above discussed results, particularly the swelling test, sample D was prepared by the addition of 45 wt% of EC–DEC solution to the reactive formulation and further completely characterized in terms of its thermal–mechanical, conductivity and electrochemical properties. Such a composition allowed to obtain polymers having lower T_g and, particularly, increased liquid electrolyte swelling ability.¹⁶ The composite reinforced membrane was activated by swelling in the liquid electrolyte solution (max uptake of 43 wt% after 15 min), resulting in a composite reinforced polymer electrolyte (namely, CRPE-D). The thermal and mechanical properties of the membrane after swelling are summarized in Table 3.

Table 3 Properties of CRPE-D sample

Sample	Tg (°C)	TGA 1^st degradation step (°C)	TGA 2^nd degradation step (°C)	Young's modulus (MPa)	Tensile resistance (MPa)	Swelling percentage (%)
CRPE-D	−54 °C	80	250	1970	15.6	43

---

The DSC analysis revealed a low T_g which is ideal for the envisaged application. The thermal stability of the membrane was assessed by TGA measurement (thermogram reported in Fig. S3 in ESI†) which showed a first weight loss inevitably linked to the presence of the organic solvent, nevertheless demonstrating the stability of the polymeric matrix up to >250 °C. The mechanical behavior of CRPE-D was evaluated by means of tensile test and results are listed in Table 3. Values having the same order of magnitude of those measured for pure cellulose handsheets were obtained.¹ Furthermore, a bending test performed after swelling showed that the CRPE-D can be easily bent and rolled, withstanding all the bending radii till 3 mm without compromising its elasticity and overall integrity.

The ionic conductivity of CRPE-D was then tested by impedance spectroscopy. Independent cells were formed by sandwiching the given sample between two blocking stainless-steel electrodes, and sealed in the Ar filled dry glove box. The behavior of the samples was monitored in the temperature range between 25 and 80 °C and results are shown in Fig. 5(a) and (b). As shown in plot (a), all impedance spectra obtained in the selected temperature range were linear, with no sign of high-frequency semicircles which could indicate lack of homogeneity. The relationship between ionic conductivity (σ) and temperature is shown in the Arrhenius plot (b) of Fig. 5.

Fig. 5 Ionic conductivity results for sample CRPE-D: (a) Arrhenius plot and (b) representative EIS curves.

CRPE-D showed an ionic conductivity of about 1.6 × 10⁻⁴ S cm⁻¹ at ambient temperature which increased by increasing the temperature to above 1.2 × 10⁻³ S cm⁻¹ at 80 °C, a noticeable value at the level of the best results reported in the literature¹⁶ for similar quasi-solid systems. The electrochemical stability window (ESW), being a fundamental parameter regarding cycling reversibility, was then evaluated at ambient temperature. The cathodic (A) and anodic (B) breakdown potential curves obtained are shown in Fig. 6(A) and (B). Measurements were performed at ambient temperature by means of linear sweep voltammetry scanning the cell potential from the OCV towards −0.2 V (vs. Li/Li⁺) in the cathodic scan or from the OCV towards 5.2 V (vs. Li/Li⁺) in the anodic scan. As clearly evident, in addition to high ionic conductivity, the reinforced composite polymer electrolyte showed an appreciably wide electrochemical stability window, which makes it particularly valuable in view of a practical battery application. Indeed, the onset of the current, which is representative for the electrolyte decomposition, indicates an anodic breakdown potential of about 4.5 V vs. Li/Li⁺. During the first scan, a faint peak at about 4.2 V vs. Li/Li⁺ is observable, which results in a small increase of the residual current during the following scans. This behavior may be ascribed to the presence of impurity/moisture in the handsheet or in the monomers, or even to side reactions between the polymeric and/or the liquid electrolyte components which result in a passivation phenomenon which extends the anodic stability up to higher potential during the following scans. The anodic scan showed very low residual current before breakdown, confirming that the membrane can be safely used within its working limits. The cathodic scan shows a very low current during the whole potential scan range; a reversible peak at about 0.75 V vs. Li/Li⁺ during the first cycle, which then decreases during the followings, and a clear lithium plating and stripping process around 0.0 V vs. Li/Li⁺ are visible. Particularly, this latter is very important and indicates the good performance of the newly elaborated CRPE-D as electrolyte separator that allows a reversible passage of the lithium ions through the interface and their deposition on the electrode.

Fig. 6 Electrochemical stability window (ESW) of CRPE-D at ambient conditions: (a) cathodic and (b) anodic scans. Potential scan rate of 0.100 mV s⁻¹.

Finally, the reinforced polymer electrolyte membrane was assembled in a complete lithium polymer cell laboratory prototype and its electrochemical behavior investigated by means of galvanostatic charge/discharge cycling at ambient temperature and different current regimes. The response of the prototype, assembled by combining a lithium metal anode, a LiFePO₄/C composite cathode and using CRPE-D as separator, is shown in Fig. 7. The plot shows the percentage of specific discharge capacity of the cell as a function of the cycle number and of the different current regime applied (from C/20 to 1 C); moreover, in the inset the representative charge/discharge profiles at moderately high 1 C rate after 150 cycles are shown. The initial reversible specific capacity upon charge at C/20 was 120 mA h g⁻¹; after 10 charge/discharge cycles at C/20, the C-rate was progressively increased upon discharge up to 1 C. The cell delivered a percentage of specific discharge capacity approaching 95% during the initial cycles when using a low current density of C/20. As the current density increases, the specific discharge capacity decreased only slightly and, at a relatively high 1 C-rate, the cell was still able to deliver almost 80% of the initial capacity, thus accounting for a good rate capability considering a quasi-solid system cycled at ambient conditions. Good performance at high current rate may be ascribed to the efficient ionic conduction in the composite polymer electrolyte separator and the favorable interfacial charge transport characteristics.^14,19,32

Fig. 7 Results of the ambient temperature galvanostatic charge/discharge cycling test of the quasi-solid lithium polymer cell assembled by sandwiching CRPE-D between a LiFePO₄/C cathode and a Li metal anode, at different C-rates (from C/20 to 1 C). Percentage of specific discharge capacity, calculated with respect to an initial reversible capacity of 120 mA h g⁻¹ (in the inset: typical charge/discharge potential vs. time profiles at 1 C rate).

The extraction and insertion of lithium ions into the structure of LiFePO₄ was found to be highly reversible upon repeated cycling and no abnormal drift was observed even at high current regimes, as indicated by the full restore of the specific capacity when the C-rate was reduced to C/10 (see the specific capacity values from 1 C to C/10 after 60 cycles). The cycling response was found to be highly stable even upon prolonged testing, with Coulombic efficiency approaching 100% at higher current regimes (see Fig. 7), which accounts for the possible practical implementation of the reinforced composite electrolyte in advanced safe quasi-solid lithium polymer batteries.

Conclusions

Composite quasi-solid polymer electrolyte membranes reinforced by specifically designed cellulose handsheets were here prepared by UV-induced free radical polymerization. They demonstrated excellent mechanical integrity even after the activation process by means of swelling in a standard liquid electrolyte solution and remarkable electrochemical performance.

The mechanical integrity upon swelling was achieved by the addition of glycidyl acrylate (GA) to the reactive mixture which was successfully grafted at the surface of the cellulose handsheets by the simultaneous use of both a radical and a cationic photoinitiator. Indeed, by means of FT-IR analysis in real time and XPS surface spectroscopy, we demonstrated that the reactive groups of GA may react in the presence of the two photoinitiators, also leading to the grafting reaction of the epoxy group via chain-transfer.

In addition to the extremely good mechanical properties due to the presence of the cellulose handsheet reinforcement, the composite electrolyte demonstrated a good ambient temperature ionic conductivity, wide electrochemical stability window and stable cycling characteristics particularly in terms of capacity retention and durability. Besides these properties, the system shows attractive features such as intrinsic safety, eco-compatibility, low production cost and industrialization potentials, highly suitable in the field of advanced safe Li-based batteries.

Notes and references

1. M. Perrin, Y. M. Saint-Drenan, F. Mattera and P. Malbranche, J. Power Sources, 2005, 144, 402–410.
2. B. Scrosati and J. Garche, J. Power Sources, 2010, 195, 2419–2430.
3. M. Winter and R. J. Brodd, Chem. Rev., 2004, 104, 4245–4270.
4. A. Patil, V. Patil, D. W. Shin, J.-W. Choi, D.-S. Paik and S.-J. Yoon, Mater. Res. Bull., 2008, 43, 1913–1942.
5. J. W. Fergus, J. Power Sources, 2010, 195, 939–954.
6. R. Marom, S. F. Amalraj, N. Leifer, D. Jacob and D. Aurbach, J. Mater. Chem., 2011, 21, 9938–9954.
7. M. Tarascon and J.-M. Armand, Nature, 2008, 451, 652–665.
8. J.-M. Armand and M. Tarascon, Nature, 2001, 414, 359–367.
9. D. Baril, C. Michot and M. Armand, Solid State Ionics, 1997, 94, 35–47.
10. J. B. Bates, N. J. Dudney, B. Neudecker, A. Ueda and C. D. Evans, Solid State Ionics, 2000, 135, 33–45.
11. J. P. Fouassier, Photoinitiation, Photopolymerization, and Photocuring Fundamentals and Applications, Hanser Publishers, New York, 1995.
12. M. A. Khan, M. M. Rahman, M. K. Gosh and T. A. Chowdhury, J. Appl. Polym. Sci., 2003, 87, 1774–1780.
13. C. Decker, Polym. Int., 1998, 45, 133–141.
14. J. R. Nair, C. Gerbaldi, M. Destro, R. Bongiovanni and N. Penazzi, React. Funct. Polym., 2011, 71, 409–416.
15. J. Y. Song, Y. Y. Wang and C. C. Wan, J. Power Sources, 1999, 77, 183–197.
16. A. Manuel Stephan, Eur. Polym. J., 2006, 42, 21–42.
17. S. Ferrari, E. Quartarone, P. Mustarelli, A. Magistris, M. Fagnoni, S. Protti, C. Gerbaldi and A. Spinella, J. Power Sources, 2010, 195, 559–566.
18. A. Chiappone, J. Nair, C. Gerbaldi, R. Bongiovanni and E. Zeno, Cellulose, 2013, 20, 2439–2449.
19. A. Chiappone, J. R. Nair, C. Gerbaldi, L. Jabbour, R. Bongiovanni, E. Zeno, D. Beneventi and N. Penazzi, J. Power Sources, 2011, 196, 10280–10288.
20. J. R. Nair, A. Chiappone, C. Gerbaldi, V. S. Ijeri, E. Zeno, R. Bongiovanni, S. Bodoardo and N. Penazzi, Electrochim. Acta, 2011, 57, 104–111.
21. S. J. Eichhorn, C. A. Baillie, N. Zafeiropoulos, L. Y. Mwaikambo, M. P. Ansell, A. Dufresne, K. M. Entwistle, P. J. Herrera-Franco, G. C. Escamilla, L. Groom, M. Hughes, C. Hill, T. G. Rials and P. M. Wild, J. Mater. Sci., 2001, 36, 2107–2131.
22. H. Haslach Jr, Mech. Time-Depend. Mater., 2000, 4, 169–210.
23. L. Jabbour, C. Gerbaldi, D. Chaussy, E. Zeno, S. Bodoardo and D. Beneventi, J. Mater. Chem., 2010, 20, 7344–7347.
24. L. Hu, H. Wu, F. La Mantia, Y. Yang and Y. Cui, ACS Nano, 2010, 4, 5843–5848.
25. L. Jabbour, R. Bongiovanni, D. Chaussy, C. Gerbaldi and D. Beneventi, Cellulose, 2013, 20, 1523–1545.
26. K. Kato, E. Uchida, E.-T. Kang, Y. Uyama and Y. Ikada, Prog. Polym. Sci., 2003, 28, 209–259.
27. H. Kubota and S. Ujita, J. Appl. Polym. Sci., 1995, 56, 25–31.
28. K. Allmér, A. Hult and B. Rånby, J. Polym. Sci., Part A: Polym. Chem., 1989, 27, 1641–1652.
29. A. Bhattacharya and B. N. Misra, Prog. Polym. Sci., 2004, 29, 767–814.
30. H. A. Krässig and V. Stannett, in Fortschritte Der Hochpolymeren-Forschung, Springer, Berlin Heidelberg, 1965, pp. 111–156.
31. Scandinavian Pulp Paper and Board Testing Committee, Drainability of Pulp by the Schopper-Riegler Method, in SCAN-C19:65, 1964.
32. G. Meligrana, C. Gerbaldi, A. Tuel, S. Bodoardo and N. Penazzi, J. Power Sources, 2006, 160, 516–522.
33. P. Stenstad, M. Andresen, B. Tanem and P. Stenius, Cellulose, 2008, 15, 35–45.
34. P. Yu, H. Block, Z. Niu and K. Doiron, J. Synchrotron Radiat., 2007, 14, 382–390.
35. L. Fras, L. S. Johansson, P. Stenius, J. Laine, K. Stana-Kleinschek and V. Ribitsch, Colloids Surf., A, 2005, 260, 101–108.
36. J. Gustafsson, L. Ciovica and J. Peltonen, Polymer, 2003, 44, 661–670.
37. J. C. Bastidas, R. Venditti, J. Pawlak, R. Gilbert, S. Zauscher and J. F. Kadla, Carbohydr. Polym., 2005, 62, 369–378.
38. Z. Q. Hua, R. Sitaru, F. Denes and R. A. Young, Plasmas Polym., 1997, 2, 199–224.

---

Footnote

† Electronic supplementary information (ESI) available: S1 XPS survey spectra; S2 kinetic plot of the reactive groups conversion in the final formulation; S3 TGA analysis. See DOI: 10.1039/c4ra07299e

---