Plasticized 3D-Printed Polymer Electrolytes for Lithium-Ion Batteries

Four polymer samples containing PLA:PEO(MW 5 × 10⁶):PEO(MW 2 × 10³) were fabricated at the following concentrations: 25:40:35, 40:30:30, 40:40:20, 0:50:50% (w/w), respectively. A commercial PLA filament was used for comparison. PLA pellets (PLA L-175 Purac^® Corbion) were dissolved in 1,3-dioxolane (Sigma-Aldrich) with stirring for 12 h at room temperature. The PEO powders of high molecular weight (_hPEO 5 × 10⁶) and low molecular weight (_lPEO MW 2 × 10³) (Sigma-Aldrich) were dissolved in acetonitrile with stirring for 24 h. Both solutions (PLA & PEO) were mixed with stirring for 12 h according to the desired composition mentioned above. After complete dissolution, the mixture was poured on a Teflon plate and dried for 12 h at room temperature. After drying, each cast was crushed to small pellets to be used for the fabrication of filaments. Each composite was extruded with a Felfil Evo Filament Extruder (Italy) to form a filament suitable for use as feedstock in a fused-deposition 3D printer. With appropriate choice of the nozzle diameter (in the range of 1.4–1.7 mm) and careful control of the nozzle temperature (typically in the range 175 °C–190 $^\circ {\rm{C}}$) and the extrusion speed, filaments were produced with a circular cross section of average diameter 1.75 mm and a typical standard deviation of 0.02–0.03 mm. Each filament was printed with the use of the Up-Plus² printer (Tiertime). Printed disc-shape samples with a diameter of 18 mm and a thickness of 200 μm were used as the separators/membranes. The printed samples were dried under vacuum at 100 $^\circ {\rm{C}}$ for 12 h in order to remove residual solvent and moisture. Calculation of %porosity and top-side surface area was made for pure PLA and PLA-PEO printed separators. This was done according to the separators' measured weight and calculated volume, considering the surface profile viewed in ESEM images. The length of protruded spiral was calculated according to:

$$\begin{eqnarray*}&&{\boldsymbol{L}}=\pi {\rm{n}}\left({{\rm{d}}}_{0}+h\left(N-1\right)\right)\end{eqnarray*}$$

N is the number of turns given by:

$$\begin{eqnarray*}&&{\boldsymbol{N}}=\displaystyle \frac{{{\rm{D}}}_{1}-{D}_{0}}{2{\boldsymbol{h}}}\end{eqnarray*}$$

where D₀ and D₁ are the inner and outer diameters respectively, of the printed circle, h is the thickness of the spiral.

For the conductivity tests, the cells, comprising 3D-printed polymer membranes sandwiched between two non-blocking lithium electrodes were assembled in 2032 coin-cell setup. The 3D printed separators were vacuum-impregnated with 0.3 M (or 1 M) LiTFSI–Pyr₁₄TFSI as IL electrolyte (20 μl) for five minutes prior to cell construction. The electrolyte was completely absorbed by the membranes, forming free-standing plasticized quasi-solid films. All handling of these materials took place under an argon atmosphere in a MBraun glove box containing less than 1 ppm water and oxygen.

Electrochemical-impedance spectroscopy (EIS) tests were carried out at 30 °C to 100 °C with a VMP3 potentiostat (Biologic Instrument) at a setting of 100 mV amplitude and a frequency range of 1 MHz to 0.01 Hz.

The differential-scanning-calorimetry (DSC) tests were carried out with a TA Instruments Research-Grade DSC with Modulation and Tzero cell module Q2000 MDSC. The samples, weighing 5 to 15 mg, were hermetically sealed in Alodine-Al pans in the glovebox and their spectra were recorded at a scan rate of 10 $^\circ {\rm{C}}$ min⁻¹ from −80 $^\circ {\rm{C}}$ up to 200 $^\circ {\rm{C}}.$

Surface morphology was evaluated with a Quanta 200 FEG Environmental Scanning Electron Microscope (ESEM). The samples were sputtered with a thin gold film (6–10 nm) prior to the scanning.

X-ray Photoelectron Spectroscopy (XPS) measurements were performed in UHV (2.5 × 10⁻¹⁰ Torr base pressure) with the use of a 5600 Multi-Technique System (PHI, USA). The sample was irradiated with an Al Kα monochromatic source (1486.6 eV) and the emitted electrons were analyzed by a Spherical Capacitor Analyzer with a slit aperture of 0.8 mm. The sample was analyzed at the surface only. C1s at 285.0 eV was taken as an energy reference for all the peaks.

NMR diffusion experiments on filaments were performed with a 300 MHz Varian spectrometer under a magnetic field of 7.05 T with a DOTY Z-gradient diffusion probe. The pulsed field gradient (PFG) NMR stimulated echo sequence was used to measure the self-diffusion coefficients of ⁷Li, ¹H, and ¹⁹F nuclei, corresponding to cations, Li⁺, PYR₁₄ ⁺ and TFSI⁻ ions, at variable temperatures, covering the 25°–85 $^\circ {\rm{C}}$ range. Gradient strength, g, was arrayed (32 values, linear increase, g = 0–1380 G cm⁻¹) for each experiment. Gradient-pulse duration was δ = 1–4 ms and diffusion delay was Δ = 40–300 ms. The self-diffusion coefficients, D, were then extracted by fitting the echo signal decay with the Stejskal–Tanner equation;

$$\begin{eqnarray}&&I={I}_{0}\exp \left[-D{\delta }^{2}{{\rm{g}}}^{2}{\gamma }^{2}\left({\rm{\Delta }}-\delta /3\right)\right]\end{eqnarray} \tag{ 1 }$$

where I is the amplitude of the attenuated echo signal, I₀ is the initial amplitude of the NMR signal, Δ and $\delta$ are delays and D is the diffusion constant, $\gamma$ is the gyromagnetic ratio and g is the gradient strength.

Bearing in mind that lithium-ion transport in polymer electrolytes is closely connected to the flexibility and relaxation of polymer chains, we blended PLA with high- and low-molecular-weight PEO. All four compositions of the PLA:_hPEO:_lPEO samples were successfully extruded at 160 $^\circ {\rm{C}}.$ Reducing the PLA content in the membrane resulted in a much more flexible filament than that of neat PLA, but printing of those filaments was more challenging. As can be seen in the ESEM micrographs (Fig. 1a), closely packed spherulites can be easily distinguished in a printed neat PLA sample. Blending of PLA with PEO destroys spherulitic morphology; instead, a needle-like structure with some flat surface inclusions, is formed (Fig. 1c). We attribute the appearance of these irregularities to the PEO-rich entities distributed in a continuous, PLA matrix plasticized by PEO in agreement with. ¹⁷ Previous work of Wittman and Manley ¹⁸ showed the needle-like morphology in the blend of a polyester and a crystallizable monomer (trioxane) at a hypoeutectic composition. Since there is no phase diagram for PLA-_hPEO-_lPEO films, known to us, we cannot for certain attribute this morphology to the formation of eutectic structure, but neither can we ignore this possibility. It is worth mentioning that the films do not exhibit phase separation. Zooming into the cross-section image of 100% PLA (Fig. 1b) reveals a smooth morphology with little plastic deformation. The PLA-_hPEO-_lPEO membrane (Fig. 1c) appears to be more ductile, as seen by the apparently more significant plastic deformation.

Zoom In Zoom Out Reset image size

Calculated porosity of the printed neat PLA and blended polymer membranes was very low, not exceeding 4% for all the compositions, indicating that the membranes are dense and contain little or no pores. On the other hand, the surface roughness differs significantly with PLA-_hPEO-_lPEO being more coarse, thus having more intimate polymer-IL contact area, which enhances the wettability and facilitates good absorption of IL 0.3 M LiTFSI Pyr₁₄TFSI electrolyte. Miscibility of polymer with IL also influences the absorption. According to Ref.,14 the PLA is not miscible with Pyr₁₄TFSI, which means that these compounds do not interact well with each other. Tsurumaki et al. ¹⁸ investigated the miscibility of 12 different ionic liquids with PEO and found that Pyr₁₄TFSI has moderate miscibility with PEO. This could also account for the enhanced absorption ability of the membranes containing PEO, expressed in improved uptake. The uptake of IL electrolyte (Pyr₁₄TFSI with 0.3 M LiTFSI) by printed membranes was tested by dripping 20 μl IL on each separator and monitoring the change in mechanical properties and liquid absorption for six days. The as-printed membranes that contained PEO absorbed the IL very well, while the absorption by the 100% PLA sample was poor. The separators that had absorbed the liquid became flexible.

The effect of IL impregnation on the thermal properties of membranes of different composition was measured by DSC. The DSC tests were conducted at a heating/cooling ramp of 10 deg min⁻¹. The results of the DSC tests shown in Fig. 2a and Table I are of the filaments. The DSC thermograms of neat high- and low-molecular-weight PEO and of neat PLA are displayed for comparison. The onset point of the _lPEO endotherm is at 56.6 $^\circ {\rm{C}}$ and that of _hPEO at 73.2 $^\circ {\rm{C}}.$ This higher melting temperature is caused by a longer macromolecule. Melting of PLA occurs at 180.7 $^\circ {\rm{C}}.$ The endothermic peak of _hPEO:_lPEO filament appears between the endotherms of neat polymers. Unexpectedly, small endothermic transition is detected in the DSC trace of this filament at 170 $^\circ {\rm{C}},$ which may indicate that a small amount of PLA remained inside the extruder. The higher the PLA content of the filament, the smaller the melting enthalpy of the low-temperature endotherm. The opposite behavior is observed for the high-temperature melting peak. Since there are no additional endothermic transitions in the filaments, as compared to neat polymers, it can be deduced that the original materials constituting polymer-blend samples have kept their properties, implying that no new inter-polymeric phases were created upon extrusion.

Zoom In Zoom Out Reset image size

The percentage of crystallinity of each polymer was calculated from the equation:

$$\begin{eqnarray*}&&{\chi }_{{\rm{c}}}=\displaystyle \frac{{\rm{\Delta }}{{\rm{H}}}_{{\rm{m}}}-{\rm{\Delta }}{{\rm{H}}}_{{\rm{cc}}}}{{\rm{w}}\,\cdot \,{\rm{\Delta }}{{\rm{H}}}_{100}}\,\cdot \,100\end{eqnarray*}$$

Where ΔH_m is the value of the melting enthalpy, ΔH_cc is the cold-crystallization enthalpy, ΔH₁₀₀ is the enthalpy of the completely crystalline polymer and w is the weight fraction of polymer in the sample. The values of ΔH₁₀₀ for PLA and PEO are 93.6 ¹⁹ and 196.0 ²⁰ J g⁻¹, respectively. The results obtained for the melting temperature (T_m), the enthalpy of melting (ΔH_m) and the degree of crystallinity (X_c) of each polymer, as well as the peak temperature (T_cc) and enthalpy (ΔH_cc) of cold crystallization of PLA are reported in Table I. The samples of filaments and printed membrane of the same composition (without IL) were tested and found to have similar DSC behavior. PLA domains in the blend filaments (PLA40-_hPEO40-_lPEO20 and PLA25-_hPEO40-_lPEO35) exhibit a higher crystallinity percentile (95.1 and 79.9% respectively) as compared to pristine PLA pellets (39.7%). Such behavior is described in the work of Y. Hu et al. ²¹ who studied the morphology of solidified PLA-PEG blends. It was found that blending with polyethylene glycol (MW 8,000) and slow cooling of the blend, accelerated crystallization of PLA in the form of large spherulites. Glass-transition temperatures (T_g) also differ between the different samples, from 67.9 °C of the PLA pellets to close to zero in the blended filament.

On the other hand, the crystallinity of the PEO phase is lower (23.4%) in the filament of high PLA content (40% wt). We relate this to the spatial disturbance and disorder induced by PLA, which gives rise to the creation of amorphous PEO regions.

Adding IL to the printed samples has a significant plasticizing effect on their thermal properties. As can be seen from the DSC traces in Fig. 2b, the addition of IL to pristine PLA causes a shift of the glass-transition and melting temperatures towards lower values by 10.4 $^\circ {\rm{C}}$ and 15.6 $^\circ {\rm{C}},$ respectively. The enthalpy of the endotherm decreases from 39.6 to 36.4 J g⁻¹, the full width at half maximum (FWHM) increases from 8.5 °C to 9.1 $^\circ {\rm{C}}$ and the height of the peak remains the same with a value of 0.6 W g⁻¹. The degree of crystallinity is lower by 27.2% as a result of the pronounced cold-crystallization process occurring before the melt. The enthalpy of the cold-crystallization exotherm increases upon addition of IL by almost one order of magnitude, from 2.5 to 24.7 J g⁻¹. Wetting by IL of printed polymer-blend membranes is followed by similar phase-transition trends (Table I). With the PLA40-PEO blend a shift of Tg by 8.4 $^\circ {\rm{C}}$ was detected as compared to the IL-free membrane. The melting temperature of the PEO phase was decreased by 13.4 $^\circ {\rm{C}},$ and its crystallinity was lower by 36.1%. The melting temperature of the PLA phase decreased by 4.2 °C and its crystallinity dropped by 47.9%.

The plasticizing effect of IL, which is reflected in the decrease of Tg and the widths and heights of the endothermic transitions of all the membranes, is likely to promote more homogeneous orientation of polymer chains in the crystalline regions of the separators upon printing. The lowering of the glass-transition temperature is interpreted as an increase in the polymer segmental motion, which is an important factor influencing the ionic conductivity of the polymer electrolytes. Such plasticizing effect of IL was previously reported by Osada et al. ¹¹ who incorporated Pyr₁₄TFSI-LiTFSI IL into a PLA matrix to obtain solid polymer electrolyte. It was demonstrated that the addition of IL causes a significant decrease in the glass-transition temperature. The plasticizing effect of IL was also demonstrated in the work of Desai et al. ²² who incorporated P₁₄TFSI (n-butyl-3-methylpyridinium trifluoromethanesulphonylimide) into PEO matrix to produce gel-polymer electrolytes (GPE).

To determine if there is a difference between the chemical environment at the surface of the plasticized PLA-PEO membrane and that in its bulk we carried out depth-profile XPS tests, i.e., XPS scan after a short sputter. In Fig. 3 we present the high-resolution spectra of pristine PLA in the C 1s and O1s binding energy regions. As expected, in the C 1s spectrum (Fig. 3a) we find prominent peaks, in most cases assigned to polymers: carbonyl groups (O–C=O) at 289.1 eV, C–O bonds at 287 eV, as well as a small contribution at 284.8 eV, typical of poly(lactide). ^22,23 After sputter, the PLA and polymeric C–O bonds of PEO (286.2 eV) remain but the carbonyl bond diminishes (Fig. 3e). The carbonyl bond reflects the C–O skeletal structure of both PEO and the PLA. ^24–26 This characteristic polymeric environment is also shown in the O1s spectrum (Fig. 3b) having typical oxygen-carbon bonds such as O–C–O at 533.6 eV, C=O at 532.3 eV, and polymeric bonds at 531.9eV. ²² The wetting of the membrane by the ionic liquid electrolyte is evident by the appearance of a C–N peak at 286.4 eV together with a TFSI⁻ peak at 292.5 eV in the C 1s region (Fig. 3c). ^27,28

Zoom In Zoom Out Reset image size

When we examine the variations induced in the O1s spectrum following the addition of the 0.3 M LiTFSI Pyr₁₄TFSI ionic liquid (Fig. 3d), we tend to conclude that Li ions form complexes not only with PEO, but with the PLA, as well, e.g., Li⤑:O = C–O bond. There is a low O1s peak at ∼530.9 eV that is not present in the neat PLA scan and can be associated with the interaction of a Lewis base (here the non-bonding electron pairs of oxygen-containing groups) with a metal atom or ion acting as a Lewis acid. ²⁹ A similar shift to a lower binding energy in the O1s region due to cation-oxygen Lewis-base interaction was reported by Tseng et al., for the bonding of Pt⁺ to a C–OH group. ³⁰

The charge-transport dynamics in the 3D-printed electrolytes have been investigated by pulsed-field gradient-diffusion nuclear magnetic resonance (PGSE-NMR) with the use of the probe nuclei ¹H, ⁷Li, and ¹⁹F, corresponding to the PYR₁₄ ⁺, Li⁺, and TFSI⁻ ions. The self-diffusion coefficients as a function of temperature are shown in Fig. 4. The values of diffusion coefficients are compiled in Table II. As expected, the diffusion coefficient increases as the temperature rises, and this trend is consistent through all the samples. A linear fit (R² > 0.98) to reciprocal temperature is in agreement with the Arrhenius Law. In spite of the fact that the Li⁺ cation is the smallest ion in the system, it exhibits the lowest diffusion coefficient, followed by TFSI⁻ and PYR₁₄ ⁺, whereas PYR₁₄ ⁺, constituting the largest ion, shows the highest diffusion coefficient over the whole temperature range (Table II). On the basis of ion sizes, PYR₁₄ ⁺ > TFSI⁻ > Li⁺, and on the Stokes−Einstein equation, one would predict that the order of the self- diffusion coefficients should be ${D}_{L{i}^{+}}\gt {D}_{TFS{I}^{-}}\gt {D}_{PYR{14}^{+}}.$ This indicates that, whereas the larger PYR₁₄ ⁺ cation diffuses as a single ion, neither the TFSI⁻ anions nor the Li⁺ cations diffuse as single ions. It is a clear indication that Li⁺ cations and TFSI⁻ anions are forming aggregates, as observed in many other Li-IL systems. ³¹ This trend is observed for all samples, except that in the PLA-IL-LiTFSI sample, all three diffusion coefficients become nearly equal at the highest temperature.

Zoom In Zoom Out Reset image size

Table II. Self-diffusion coefficients (D) of Li⁺, TFSI⁻ (F) and PYR₁₄ ⁺ (H) in IL, neat PLA + IL and PLA25:_hPEO40:_lPEO35 + IL at different temperatures.

T $^\circ {\rm{C}}$	Samples	${{\bf{D}}}_{1{\bf{L}}{\bf{i}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$	${{\bf{D}}}_{2{\bf{L}}{\bf{i}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$	${{\bf{D}}}_{1{\bf{F}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$	${{\bf{D}}}_{2{\bf{F}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$	${{\bf{D}}}_{1{\bf{H}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$	${{\bf{D}}}_{2{\bf{H}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$	${{\bf{D}}}_{3{\bf{H}}}$ $\left({{\bf{m}}}^{2}{{\bf{s}}}^{-1}\right)$
25	IL/Li	4.37 × 10−12	—	6.80 × 10−12	—	1.05 × 10−11	—	—
	Neat PLA/IL/Li	3.84 × 10−12	—	8.22 × 10−12 (87%)	8.87 × 10−13 (13%)	1.06 × 10−11 (66%)	2.46 × 10−13 (34%)	—
	PLA-hPEO-lPEO/IL/Li	1.37 × 10−12	—	2.13 × 10−12 (37%)	4.86 × 10−13 (63%)	5.66 × 10−12 (11%)	1.23 × 10−12 (49%)	1.14 × 10−13 (40%)
45	IL/Li	1.14 × 10−11	—	1.89 × 10−11	—	2.53 × 10−11	—	—
	Neat PLA/IL/Li	1.17 × 10−11	—	1.78 × 10−11 (87%)	2.23 × 10−12 (13%)	2.33 × 10−11 (57%)	1.16 × 10−13 (43%)	—
	PLA-hPEO-lPEO/IL/Li	2.58 × 10−12 (45%)	8.19 × 10−13 (55%)	7.82 × 10−12 (59%)	2.69 × 10−12 (41%)	1.12 × 10−11 (25%)	1.51 × 10−12 (50%)	1.80 × 10−13 (25%)
65	IL/Li	2.59 × 10−11	—	4.10 × 10−11	—	5.33 × 10−11	—	—
	Neat PLA/IL/Li	2.95 × 10−11	—	3.75 × 10−11 (72%)	1.48 × 10−12 (28%)	5.09 × 10−11 (54%)	8.50 × 10−13 (46%)	—
	PLA-hPEO-lPEO/IL/Li	5.88 × 10−12 (92%)	6.81 × 10−13 (8%)	3.11 × 10−11 (50%)	1.24 × 10−11 (50%)	2.33 × 10−11 (31%)	3.20 × 10−12 (39%)	3.96 × 10−14 (30%)
85	IL/Li	5.26 × 10−11	—	7.90 × 10−11	—	1.00 × 10−10	—	—
	Neat PLA/IL/Li	7.82 × 10−11	—	7.21 × 10−11 (55%)	2.25 × 10−12 (45%)	8.45 × 10−11 (49%)	1.75 × 10−12 (51%)	—
	PLA-hPEO-lPEO/IL/Li	1.45 × 10−11 (67%)	4.97 × 10−12 (33%)	5.19 × 10−11 (59%)	2.26 × 10−11 (41%)	4.49 × 10−11 (29%)	5.48 × 10−12 (40%)	3.50 × 10−14 (31%)

In the case of the IL-LiTFSI, classic single-component diffusion is apparent for all the ions. Diffusion results obtained for the polymer samples display two or three diffusing components, for all ¹H, ⁷Li and ¹⁹F nuclei. In the sample consisting of the IL electrolyte imbedded in the PLA matrix, ¹⁹F and ¹H spectra clearly show two diffusion components. According to the observed data for ⁷Li, it is less clear if ⁷Li has one or two components of diffusion, as a result of the lower signal-to-noise ratio. However, it was only possible to get an accurate value for the fast-diffusion component.

From observed data of the IL-plasticized PLA25- _hPEO40-_lPEO35 sample, ¹⁹F and ⁷Li nuclei clearly show two-exponential diffusion, but ¹H shows three-component diffusion. One component is diffusing rapidly, and is assigned to the IL cation inside the polymer. The other is considerably slower and is likely attributable to the lower MW PEO. This is most likely an indication of material structural heterogeneity on a scale of one to several microns (the length scale of the NMR measurement).

Satisfactory fitting of the two- and three-component diffusion, was obtained by adding one and two exponential terms respectively, to the standard Stejskal–Tanner expression;

$$\begin{eqnarray}&&\begin{array}{l}I={I}_{0}\left[{\rm{Aexp}}\left[-{D}_{A}{\delta }^{2}{{\rm{g}}}^{2}{\gamma }^{2}\left({\rm{\Delta }}-\delta /3\right)\right]\right.\\ \left.\,\,\,\,\,+\,{\rm{Bexp}}\left[-{D}_{B}{\delta }^{2}{{\rm{g}}}^{2}{\gamma }^{2}\left({\rm{\Delta }}-\delta /3\right)\right]\right]\end{array}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&\begin{array}{l}I={I}_{0}\left[{\rm{Aexp}}\left[-{D}_{A}{\delta }^{2}{{\rm{g}}}^{2}{\gamma }^{2}\left({\rm{\Delta }}-\delta /3\right)\right]\right.\\ \,+\,{\rm{Bexp}}\left[-{D}_{B}{\delta }^{2}{{\rm{g}}}^{2}{\gamma }^{2}\left({\rm{\Delta }}-\delta /3\right)\right]\\ \left.\,+\,{\rm{Cexp}}\left[-{D}_{C}{\delta }^{2}{{\rm{g}}}^{2}{\gamma }^{2}\left({\rm{\Delta }}-\delta /3\right)\right]\right]\end{array}\end{eqnarray} \tag{ 3 }$$

where ${D}_{A},$ ${D}_{B}\,\,$and ${D}_{C}$ are the diffusion constants for the diffusion components A, B and C.

Diffusion Arrhenius plots are displayed separately for each sample in Figs. 4a–4c and the numerical values of the diffusion coefficients estimated by least-squares fitting are summarized in Table II. These results show that diffusion coefficients of all components in IL-plasticized PLA25-_hPEO40-_lPEO35 electrolyte are considerably reduced compared to their values in the IL with 0.3 M LiTFSI, even though the ionic conductivity exhibited an acceptable value. Interestingly, in the PLA-IL-LiTFSI sample, the diffusion of all three nuclei is close to those in the IL/Li electrolyte. Also, ⁷Li diffusion in neat PLA-IL-LiTFSI exhibits a steeper increase with temperature than in the liquid electrolyte (Fig. 4d). Considering the observation of fast ⁷Li diffusion in neat PLA membrane plasticized by IL, we suppose that this results from poor wetting and low uptake of IL, which weakens their interaction and the signal, therefore, contains a significant contribution of cation diffusion in the "free" IL. Figure 4c displays the diffusion coefficients of the PLA25-_hPEO40-_lPEO35. While the ⁷Li values are the lowest of all samples, the ¹⁹F values actually exceed those of ¹H at the highest temperatures. This suggests that the Li⁺ ions are coordinated by the PEO segments, thereby freeing the TFSI⁻ ions to a certain extent.

Figure 5a shows typical Nyquist plots for the cells comprising plasticized by 0.3 M and 1 M IL PLA25-_hPEO40-_lPEO35 electrolyte, sandwiched between two non-blocking lithium electrodes. The plots are represented by two partially overlapping depressed semicircles. The interpretation of the impedance spectrum is often based on equivalent-circuit models that are used to approximate the physicochemical processes that occur in the cell. Interpretations of the plots is done in the following manner—R_bulk is extracted from the intercept of the fitted semicircle in the high-frequency branch with the x-axis (R₁). This semicircle is presumed to contain two overlapping arcs. The first arc, in the higher frequency range, with nF-capacitance is attributed to grain-boundary conduction in blended polymer electrolyte (R₂ = R_GB). The second (lower frequency) arc, with μF-capacitance values, represents SEI on lithium electrodes. ³² R₃ is extracted from the real part of the fitted semicircle and divided by two to calculate the R_SEI, which is the same for the two identical lithium electrodes. Another semicircle, evident at low frequencies and more pronounced in low-content PLA electrolytes (maximal frequency of 5 Hz and capacitance of 4 mF) can be related either to the formation of ion-pairs or to limited diffusion of the ions in the separator, as found by Strauss et al. ³³ Strauss observed this phenomenon in composite polymer electrolytes (CPEs) with EO:Li ratio ranging between 1.5 to 80 for the SPEs reported. W₅ in the equivalent circuit is the Warburg element associated with unlimited diffusion of ions in the excess of IL electrolyte. An arc, which appears in the high-frequency region for all the electrolytes containing PLA at all temperatures, diminishes with temperature in PLA-free electrolytes until it vanishes completely at 60 $^\circ {\rm{C}}.$ The electrolyte containing PLA alone exhibits about one order of magnitude lower bulk conductivity as compared to the one containing PEO. Increase of the PEO content of both high and low molecular weight (MW 5 * 10⁶ and MW 3000) at the expense of PLA polymer, leads to great improvement of bulk conductivity, increasing from 6 × 10⁻³ mS cm⁻¹ for neat PLA-IL to 0.15−0.18 mS cm⁻¹ for 25 and 40%wt. PEO (Fig. 5b) at 60°C. Grain-boundary conductivity increases from 0.1 to 1.9−2.2 mS cm⁻¹ (Fig. 5c). Increasing the concentration of lithium imide salt in PLA25-_hPEO40-_lPEO35 electrolyte from 0.3 M to 1.0 M leads to bulk-conductivity increase from 0.18 mS cm⁻¹ to 0.31 mS cm⁻¹ at 60 $^\circ {\rm{C}},$ which follows the three-fold increase in the number of charge carriers. In all probability, the conductivity could be further enhanced. However higher LiTFSI concentration causes increase of the viscosity of IL, which deteriorates wetting and absorbance of the printed membranes by IL, as reported by Nadherna et al. ³⁴ An inflection point appearing at about 60 $^\circ {\rm{C}}$ in Arrhenius plots is related to the melting of PEO. An apparent activation energy at above 60 $^\circ {\rm{C}}$ varies from 7 to 9 kcal mol e⁻¹; while below 60 $^\circ {\rm{C}}$ it is 14 and 9 kcal mole⁻¹ for 1 M IL and 0.3 M IL, respectively. The slope of the BC segment in the plot may be interpreted as the apparent activation energy of the solid electrolyte. However, the slope of the AB segment is associated with many processes: conduction in the solid and in the molten phases, ionic association in the molten phase, etc Fig. 5d exhibits SEI layer resistance vs temperature for the plasticized printed electrolytes. Again, it is clear that the SEI resistance values are much lower in the lithium cells containing IL-plasticized polymer membranes. This is likely to be due to the better wettability of the ductile PEO-containing membranes, allowing better contact of the printed films with lithium electrodes. The grain-boundary-conductivity values of printed electrolytes and the R_SEI were found to be scarcely affected by the concentration of lithium imide salt in the IL electrolyte.

Zoom In Zoom Out Reset image size

Lithium transference number is an important parameter for electrolytes, for the reason that the higher the t_Li+, the better it can mitigate the anion accumulation in the vicinity of the electrode/electrolyte interface by relieving the concentration polarization. The t_Li+ value of 3D printed plasticized electrolyte, measured by a combination of potential polarization and EIS at 60 °C, ³⁵ showed that it is close to 0.2, lower than the values reported for neat LiTFSI-Pyr₁₄ ionic liquid electrolyte. ³⁶ There are two possible reasons for a low t_Li+ value of t₊. One is the relatively low concentration of lithium cations in small amount of IL-electrolyte (20 μl) used for the plasticization, which promotes concentration polarization. The second reason could be the reduced molar ratio between Li⁺ and the functional groups of PLA and PEO, which participate in the complexation, and the slow relaxation of which influences ion transport. The interfacial stability of the electrolyte with Li anode was further elucidated by galvanostatic Li plating/stripping on symmetric Li∣PLA-PEO-IL∣Li cells. The tests were carried out at 25 and 50μA cm⁻² at voltage range of ±1 V. The prolong cycling shows that, the cell presents a stable profile with relatively low polarization with no indication of Li dendrite growth on the Li metal surface (Fig. 6). The developed plasticized polymer electrolyte was tested in the half cell, composed of metallic lithium anode and 3D printed LFP-based cathode. As one can see from the plots (Fig. 7), the operating voltage is similar to that of the batteries with commercial electrodes. The higher charge/discharge overpotential is attributed to the high impedance of the cell with printed cathode, which contained only poor-conducting PLA polymer, as the binder of LFP. In addition, since the amount of IL incorporated in the printed plasticized electrolyte (PPE) was very small, it was insufficient to provide low-resistant printed electrode/PPE interfacial contacts. Despite this, four times increase of the current density did enable reversible cycleability of the cell.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Finally, we would like to mention that we have recently presented for the first time all-solid-state LFP/LTO battery, in which all the components were printed by fused filament fabrication. ⁹ However, the electrochemical performance of these cells was poor due to high internal impedance. Aiming at the enhancement of conductivity of printed polymer electrolyte, we started the development of intrinsically-plasticized solid filaments prepared by one-shot dry-processing extrusion procedure. The research, presented in the current manuscript, is the first step of this development. At this point we studied the effect of vacuum infusion of LiTFSI-Pyr₁₄ IL electrolyte to the printed membrane and characterized this system by SEM, DSC, XPS, NMR and EIS methods. Of particular importance for us was using as small, as possible amount of IL in order to keep the printed samples in the easy-processable solid state and to enable further printing of intrinsically plasticized filaments. The study described in the manuscript enabled us to unravel the complex interplay between the IL uptakes (absorption of IL by PLA and PEO) and Li-PEO vs Li-PLA interactions, which influence phase transitions, conductivity and self-diffusion coefficients.

When comparing the electrochemical characteristics of all-solid ⁹ and quasi-solid electrolytes in Li/Li cells at 60 °C it is noticeable that bulk conductivity is one order of magnitude higher (1.1 × 10⁻² vs 0.18 mS cm⁻¹), and grain-boundary conductivity (1.4 × 10⁻² vs1.9 mS cm⁻¹) is two orders of conductivity higher in the cells with quasi-solid electrolyte. The significantly enhanced conductivity is attributed to the plasticizing effect of the ionic liquid Pyr₁₄TFSI, leading to higher molecular mobility and better ion transport. In addition, the R_SEI is lower in the cell with quasi-solid electrolyte, which is reflected in better electrochemical performance of Li/LFP cells.

The research of free-form-factor battery with all-fully printed components based on intrinsically-plasticized solid filaments is in progress and the results will be published elsewhere.

This study reports the fabrication of 3D-printed quasi-solid PLA-PEO electrolytes infused with LiTFSI ionic liquid. There is a complex interplay between the IL uptakes (absorption of IL by PLA and PEO) and Li-PEO—Li-PLA interactions, which influence phase transitions, conductivity and self-diffusion coefficients. The quasi-solid printed electrolyte showed higher bulk-conductivity values with increased content of PEO and of the concentration of lithium imide salt. Clear demonstration of the plasticizing effect of the ionic liquid is shown by DSC measurements, which are exhibited by the reduced melt and glass-transition temperatures of the polymers and diminished crystallinity. XPS data of 3D printed pristine PLA infused with LiTFSI solvated by ionic liquid show evidence of some kind of the interaction between of PLA- and lithium cation in addition to the well-known formation of PEO-Li complexes. By this is meant that PLA not only promotes thermal robustness up to 120 $^\circ {\rm{C}},$ but it may also induce additional conduction paths to the quasi-solid printed electrolyte. On the other hand, when comparing ionic-conduction properties of quasi-solid-state 3D-printed electrolyte plasticized by ionic liquid of pristine PLA and PLA-PEO printed blends, we can conclude that neat PLA exhibits poor Li ion conduction, yielding a bulk-conductivity value of 7 × 10⁻⁶ S.cm⁻¹. Whereas, being mixed with PEO the ion conduction increases by almost one-and-a-half orders of magnitude, approaching 0.2mS.cm⁻¹ at 60 $^\circ {\rm{C}}.$ Increasing the number of charge carriers causes the bulk conductivity to increase (4 × 10⁻⁴ S.cm⁻¹ at 60 $^\circ {\rm{C}}$). However, conduction through grain boundaries and solid-electrolyte-interphase resistance seem to be unaffected.

The self-diffusion coefficients in PLA-IL samples measured by PFG -NMR were found to be close to the neat IL electrolyte, but they were lower in the blended PLA-PEO-IL electrolytes. This disagreement can be attributed to the different ion-transport-length scales probed by NMR and AC impedance tests, with the former addressing micron-scale motion occurring during the several millisecond timeframes of the PFG experiment with the latter requiring minutes-long tests. Moreover, incomplete wetting of the printed PLA may leave small regions of IL electrolyte, which are invisible to the eye, but PFG detectable. And finally, the NMR measures all mobile species, including neutral ion pairs and higher aggregates, the former of which do not contribute to the conductivity. Cycling of symmetrical lithium cell with printed plasticized electrolyte for 140 h shows that, the cell presents a stable profile with relatively low polarization with no indication of Li dendrite growth. The operating voltage of LFP/Li cell was found to be similar to the to that of the batteries with commercial electrodes. The higher charge/discharge overpotential is caused by high impedance of the cell with printed cathode, which contained only poor-conducting PLA polymer, as the binder of LFP. In addition, since the amount of IL incorporated in the printed plasticized electrolyte (PPE) was very small, it was insufficient to provide low-resistant printed electrode/PPE interfacial contacts. Despite this, four times increase of the current density did enable reversible cycleability of the cell.

This research was funded by Israel Ministry of Science and Technology, grant 81485. Nishani Jayakody acknowledges financial support from the U.S Office of Naval Research.