Multiprocess 3D printing of sodium-ion batteries via vat photopolymerization and direct ink writing

In this work, the ability to print shape-conformable batteries with multi-process additive manufacturing is reported. Vat photopolymerization (VPP) 3D printing process is employed to manufacture gel polymer electrolytes (GPEs) for sodium-ion batteries (SIBs), while direct ink writing process is used to prepare positive electrodes. The sodium-ion chemistry has proven to be an adequate substitute to lithium-ion due to the availability of resources and their potential lower production cost and enhanced safety. Three-dimensional printing technologies have the potential to revolutionize the production of shape-conformable batteries with intricate geometries that have been demonstrated to increase the specific surface area of the electrode and ion diffusion, thus leading to improved power performances. This study shows the preparation of composite UV-photocurable resins with different polymer matrix-to-liquid electrolyte ratios, designed to act as GPEs once printed via VPP. The impact of the liquid electrolyte ratio within the GPEs is thoroughly examined through a variety of electrochemical techniques. The exposure time printing parameter is optimized to ensure adequate print accuracy of the GPE. Using the optimized resin composition as material feedstock, shape-conformable 3D printed GPE exhibiting an ionic conductivity of 3.3 × 10−3 S·cm−1 at room temperature and a stability window up to 4.8 V vs. Na0/Na+ is obtained. In parallel, a composite ink loaded with Na0.44MnO2 and conductive additives is developed to 3D print via direct ink writing positive electrodes. After demonstrating the functionality of the independent 3D printed components in SIBs, the last part of this work is focused on combining the 3D printed Na0.44MnO2 electrode and the 3D printed GPE into the same battery cell to pave the way towards the manufacturing of a complete 3D printed battery thanks to different additive manufacturing processes.


Introduction
Since commercialization in the 1990s, lithium-ion batteries (LIBs) have historically ruled the market as the preferred rechargeable batteries for portable electronics due to their high energy density and application heritage.However, concerns associated with the flammability of the liquid electrolyte [1] and the impact of mining scarce lithium on Earth [2] have led researchers to investigate alternatives such as sodium-ion batteries (SIBs) [3][4][5][6].Apart from terrestrial usage, SIBs are also gathering attention in space missions for NASA, as the components and precursors for SIBs have been shown to be relatively abundant in lunar and martian surfaces [7].In order to realize sustainable and efficient space exploration and habitats, such as those envisioned in NASA's Artemis program [8], the manufacturing at the point-of-demand sites using in-situ resource utilization is crucial.As part of the efforts for SIBs manufacturing, researchers are exploring additive manufacturing (also called 3D printing) as an efficient manufacturing tool to produce on-demand 3D batteries [9][10][11][12][13].With all commercial batteries being composed of an anode and a cathode separated by an electrolyte in a 2D stacked arrangement, 3D printing has the potential to revolutionize the manufacturing of batteries by allowing the production of complex and detailed structures, which can serve as energy storage devices and potentially as load-bearing structures [14].From the battery performance standpoint, 3D printing can revolutionize the production of shape-conformable SIBs as a dramatic increase in the electrode's surface area and ion diffusion in three dimensions theoretically lead to improved power performance [15].
One of the aspects to improve the safety in SIBs is focusing on replacing the classical flammable liquid electrolyte and separator by a polymer-based electrolyte, such as a solid polymer electrolyte, a hybrid polymer electrolyte, or a gel polymer electrolyte (GPE).In comparison with solid and hybrid polymer electrolytes, GPEs present the following advantages [16][17][18]: (i) flexibility and conformability, which can be particularly beneficial in applications where the electrodes are 3D printed or have certain irregularities in which the gel can provide better contact.(ii) GPEs can work at room temperature because the gel structure provides pathways for ion movement that are less restricted than those in a solid matrix.(iii) Enhanced safety, as GPEs can absorb and dissipate external mechanical stress better than their solid counterparts.(iv) Ease of manufacturing, meaning that they can be simply casted or 3D printed without having tortuosity as the main concern.The choice between GPEs and solid polymer electrolytes depends on the specific needs of the application.
Functional GPEs for SIBs have been successfully produced in recent years through conventional casting and electrolyte infusion techniques [19][20][21][22], but have not been widely produced via 3D printing due to limitations such as the need of an ionically-conductive, mechanically-stable and UV-curable polymer matrix.Among the few works that reported 3D printing of GPEs, Chen et al [23] 3D printed via vat photopolymerization (VPP) a zig-zag GPE from a poly(ethylene glycol) diacrylate (PEGDA)-based UV-curable resin containing 80 vol.% of a 1 M LiClO 4 -based liquid electrolyte.The registered ionic conductivity was 4.8 × 10 −3 S•cm −1 at room temperature.Rahman et al [24] developed a poly(vinylidene fluoride) with N,N-dimethyl acrylamide (PVDF/PDMAAm) UV-curable resin containing 50 wt.% of LiCl in ethylene glycol as electrolyte.The resulting 3D printed GPE delivered an ionic conductivity of 6.5 × 10 −4 S•cm −1 at room temperature.Lastly, Zehbe et al [25] mixed 80 wt.% of lithium sulfonate-based ionic liquids with a commercial UV-curable resin to print an ionogel.The authors registered an ionic conductivity of 0.7 × 10 −4 S•cm −1 at room temperature.As deduced from literature, the focus has been primarily 3D printing GPEs for LIB applications, but no effort has been published in the SIBs field yet.
In this work, two additive manufacturing processes (VPP and direct ink writing) are combined to 3D print shape-conformable SIBs.A composite UV-photocurable resin is prepared to act as a GPE in SIBs after printing via VPP.An optimization of the resin-to-electrolyte ratio is performed to maximize the ionic conductivity while ensuring adequate printability.For the most promising compositions, linear sweep voltammetry (LSV) measurements are performed on tape casted films to ensure the electrochemical stability within a wide voltage window of the subsequently 3D printed GPE.The 3D printing of GPE intricate geometries is shown, and the VPP printing accuracy and first layer resolution limitation are discussed.On the other hand, a composite Na 0.44 MnO 2 -based ink is developed to 3D print positive electrodes via direct ink writing.The functionality of the independent 3D printed components (GPE or positive electrode) is demonstrated in SIBs half-cell configuration.Finally, to demonstrate the ability of multiprocess additive manufacturing to prepare rechargeable batteries, the 3D printed Na 0.44 MnO 2 positive electrode and the 3D printed GPE, manufactured via direct ink writing and VPP, respectively, are combined into a single battery cell.

Design
The CAD software nTopology (nTopology, New York) was utilized to design planar discs measuring 15 mm in diameter and 100 µm in thickness.To demonstrate the design freedom, two other complex designs were created in discs of identical dimensions, one featuring a hexagonal honeycomb cylindrical volume lattice (referred to as mandala), and the other featured a square honeycomb rotated volume lattice (referred to as spiral).Moreover, a cube measuring 20 mm per side with a face-centered cubic volume lattice was designed with the same purpose.All of these designs are shown in the results and discussion section.After creating a mesh from the implicit bodies in nTopology, the designs were exported as standard tessellation language files, transforming the design bodies into a sequence of joined triangles describing the objects' geometries.
Tape casted GPE films with different resin-to-electrolyte ratios were prepared either in the glovebox (referred to in the text as 'Ar-GPE') or outside the glovebox (referred to in the text as 'air-GPE'), and subjected to the optimised UV-curing duration of 30 s with a 405 nm lamp (same wavelength as the UV light of the VPP 3D printer).GPE discs with diameter of 15 mm were cut from the tape casted films for electrochemical analysis.The prepared composite resins were also used as material feedstock for a commercial VPP digital light processing 3D printer (Bison 1000, Tethon 3D, USA) to produce GPE discs (planar, hexagonal honeycomb, square honeycomb rotated) and a cubic lattice.Printing parameters such as the UV light brightness and layer thickness were set to 400 mW cm −2 and 100 µm respectively, while the initial exposure time and heating feature of the printer tank were not used.The impact of the exposure time printing parameter was tuned between 5 and 100 s to investigate its impact on the printing accuracy.Composite resins were stored in a fridge and 3D printed in air at 10 • C.

Ink preparation and direct ink writing 3D printing
An ink with composition 56.1 wt.% Na 0.44 MnO 2 (NEI Corporation), 6.9 wt.% carbon black Timcal Super C45 (BET = 45 m 2 •g −1 and 20 nm particle size, MSE Supplies), and 3.7 wt.% N50 ethyl cellulose binder (technical grade, Sigma Aldrich) dispersed in 33.3 wt.% terpineol solvent (technical grade, Sigma Aldrich) was prepared for 3D printing of positive electrodes via the direct ink writing process.A NScrypt 3DN-500 (NScrypt, USA) was used to print full discs and a Voltera V-One (Voltera, USA) to print discs with varying path spacing.The substrate was a carbon-coated aluminum foil that serves as the current collector for the battery.After printing, the electrodes were left to dry naturally for 24 h followed by drying at 120 • C and then at 220 • C for 20 min each.The active material loading lies between 4 and 6 mg•cm −2 depending on the infill pattern.

Electrochemical characterization
Tape casted or 3D printed GPE discs were placed between two stainless steel plates inside of a coin cell-type battery for potentiostatic electrochemical impedance spectroscopy (PEIS) measurements.The measurements were recorded from an initial frequency of 2 MHz until 10 Hz at 20 points per decade, 0.05 V of AC amplitude, 0.005 V of DC Offset and 0.0707 V of AC peak value.Three GPE samples of each composition were tested to ensure reproducibility and obtain error bars.The resistance values were extracted from fitted Nyquist plots with Z-view ® software.Equation (1) was used to calculate ionic conductivity σ (in S•cm −1 ), where R is the resistance (in Ω), t is the thickness (in cm) and A is the area (in cm 2 ), For the LSV measurements, both the tape casted GPEs prepared in Ar with a diameter of 9.53 mm and thickness <400 µm, and the 3D printed GPE discs, were fitted between a stainless steel disc serving as the working electrode and a lithium metal disc serving both as a counter and reference electrode.The same LSV protocol was employed to analyze the liquid electrolyte contained within a glass fiber with known thickness (Whatman GF/D).The measurements were performed between 0.01 and 5 V at a scan rate of 0.1 mV•s −1 using an Interface 1010 potentiostat/galvanostat (Gamry Instruments, USA).The galvanostatic polarization tests were performed onto symmetric coin cells Na 0 /GPE/Na 0 at a current density of 1 mA•cm −2 by charging and discharging for one hour.

Materials characterization
Fourier transform infrared (FTIR) data was collected in the transmission mode using a PerkinElmer spectrometer on the wavelength range 600-4000 cm −1 .Scanning electron microscopy (SEM) was performed using a S-4800 (Hitachi, Japan) field emission SEM operating in high vacuum mode.A thin gold layer was deposited on the sample prior to analysis and secondary images were recorded with a 15 kV acceleration voltage at different magnifications.Dynamic viscosity testing was done using the Digital Battery Electrode Paste/Slurry Viscosity Tester model MSK-SFM-VT.The temperature was uniformly raised from 15 • C to 30 • C using a hot plate and intermittent magnetic stirring.Data measurement was recorded after one minute of a spindle (cylindrical LV-1) rotation for each degree raised.Three trials were done for both GPE 1:0 and 1:1, and the average standard deviation was calculated.

GPE composite resins formulation
GPEs were made of a liquid electrolyte composed of a sodium salt and solvents, and a photocurable resin composed of a polymer matrix and a photoinitiator.As for the electrolyte salt, NaClO 4 was chosen as it is not as hygroscopic as NaPF 6 [29], which will thus allow 3D printing in air without decomposing into hydrofluoric acid and other fluorinated compounds.This is done in anticipation of an eventual water uptake during the 3D printing step and upon transfer to the glovebox for further battery assembly.Moreover, NaClO 4 is particularly relevant for space-related applications for NASA, as it could be potentially obtained on the Moon through a simple synthesis involving recycling, and due to its immediate availability on the Martian surface [7,30,31].Table 1 summarizes the main physical properties of the electrolyte solvents and PEGDA polymer matrix that were considered.This analysis was done (i) to limit the volatilization of low-boiling point solvents during printing due to local heating; (ii) to ensure a high ionic Na + conductivity (linked to the solvent's dielectric constant); and (iii) to ensure an adequate viscosity for the VPP 3D printing process.In previous studies, NaClO 4 -based electrolytes have been prepared with any of the solvents displayed in table 1 [32].A combination of EC, PC and FEC appeared as the best fit for batteries containing Mn-based cathode (the material used in this study) [17,33].Fortunately, EC and PC are the solvents with the highest boiling point and dielectric constant among the options.FEC is considered an additive that promotes good solid electrolyte interphase formation given the little amount that is added to the electrolyte [34].A final composition of 1 M NaClO 4 in EC/PC 1:1 v/v with 3 wt.% of FEC was chosen as liquid electrolyte.PEGDA was selected as the polymer matrix owing to the polymer's low toxicity, good electrochemical stability and the presence of ethylene oxide side chains [35].The latter enhances Na + conduction due to the lone pair of electrons that coordinate with Na + , in a similar manner that non-photocurable poly (ethylene oxide) (PEO) has been reported to do for both lithium [36] and sodium salts [37].TPO was chosen as the photoinitiator as it promotes radical photopolymerization at a wavelength appropriate for the VPP 3D printers and UV lamp (405 nm).The high viscosity of PEGDA (57 cP) will increase the viscosity of the GPE so that the 3D printing process is facilitated.Furthermore, PEGDA exhibits a high boiling point (386.6 • C) and is not flammable, which is ideal to improve the safety of the battery.
Prior to the preparation of different resin:electrolyte ratios, the viscosity of a composite resin with ratio 1:1 v/v was compared to a 1:0 (pure resin) as a function of temperature, given the high difference in viscosity between the resin (≈57 cP) and liquid electrolyte (≈2 cP) at RT.The measurement of the viscosity as a function of temperature is particularly important for 3D printed GPEs for two reasons: (i) VPP 3D printers are equipped with a heating feature to decrease the viscosity of the resin-nevertheless, this feature is not used in this work to prevent the electrolyte solvents volatilization; (ii) while the resins are stored at 10 • C prior printing, the temperature of the resin increases during the GPE printing process upon prolonged UV light exposure.In figure 2(a) it can be observed that the viscosity of the 1:0 composite resin varies by 28 cps between 16 • C and 30 • C, whereas for a 1:1 resin the variation gap is only 8 cps.This means that the electrolyte addition allows the resins to be printed at a wider temperature range without significant viscosity change.Composite resins with higher amounts of electrolyte (ratios 1:2, 1:4, 1:5 and 1:6) were not tested because the viscosity will vary even less for these ratios and because with such an amount of electrolyte the measurement limit of the viscometer was reached.
In order to provide information about the nature of bonding and different functional groups present in the samples after the photopolymerization process, the vibrational energy levels were analyzed by FTIR.The FTIR spectra of photopolymerized resins 1:0 and 1:1 v/v are shown in figure 2(b).In the GPE 1:0 v/v spectrum the strong peaks at 1090, 1740 and 2870 cm −1 can be respectively associated with the C-O-C ether group, the C-H stretching of CH 2 and the C=O vibration of PEGDA predominantly [38].Due to the presence of these bonds it can be assumed that the UV lamp process at RT did not alter the chemical composition of the polymeric matrix.Since the photopolymerization process is done by opening the C=C bonds and crosslinking with the help of the photoinitiator, it is surprising to find the characteristic peaks of C=C bonds at 830 and 1440 cm −1 [38].Their presence could be the result of non-photopolymerized PEGDA, and of the photoinitiator's strong IR response (added in PEGDA:TPO 1:0.005 wt.).On the other hand, several differences exist between GPE 1:0 and 1:1 v/v, most of which can be attributed to the presence of liquid electrolyte in GPE 1:1 v/v.The peaks marked with green arrows at 720, 1060 and 1630 cm −1 in GPE 1:1 v/v can be attributed to the IR response of NaClO 4 [39].The spectra is nonetheless dominated by vibrations from the EC and PC solvents pointed by blue arrows at 780, 970, 1160, 1395, 1480, 1540, 1790, 2920 and 2990 cm −1 [40,41].The presence of these peaks indicates that both solvents are trapped within the polymeric matrix and they did not volatilize completely (quantification is not in the scope of this paper but the vola tilization is assumed to be minimal since the boiling points of EC and PC lie above 240 • C).While the PEGDA peaks are less visible on the GPE 1:1 spectra, PEGDA has been reported to be chemically stable in carbonate solvents [42].Overall, no evidence of salt complexation was found, although it might be due to the strong infrared response of EC and PC on the spectra.

Electrochemical testing of tape casted GPEs
The objective of the initial electrochemical testing of the tape casted GPE films was to find the most adequate ratio resin:electrolyte in terms of handleability and electrochemical performance, before preparing an important quantity of the composite resin feedstock for 3D printing.The ionic conductivity of GPEs is of particular importance for the cycling performance of the battery and therefore is the first test that is performed.GPEs resin:electrolyte 1:0, 1:2, 1:4, 1:5, 1:6 and 1:8 v/v ratio were prepared by tape casting and UV lamp photopolymerization inside of the glovebox (referred to as 'Ar-GPE') or outside the glovebox (referred to as 'air-GPE').PEIS testing was employed to calculate the ionic conductivity of each of the GPEs ratios and results are shown in figure 3(a).For the Ar-GPE samples prepared in the glovebox, the ionic conductivity increases regularly from GPE 1:2 (6.05 ± 0.3 × 10 −4 S•cm −1 ), to GPE 1:4 (3.6 ± 0.5 × 10 −3 S•cm −1 ), to GPE 1:5 (7.8 ± 0.8 × 10 −3 S•cm −1 ), to GPE 1:6 (1.0 ± 0.1 × 10 −2 S•cm −1 ) and to GPE 1:8 (2.4 ± 0.3 × 10 −2 S•cm −1 ).On the other hand, the conductivity values calculated for the air-GPE samples are in the same range (only a slight increase can be perceived, probably due to undesired water uptake during tape casting and photopolymerization in air).These values fall within the range considered as adequate ionic conductivity for battery applications: ∼× 10 -3 S•cm −1 at RT [17].The high Na + conduction in the presence of the polymer matrix can be explained by the ability of Na + to coordinate with the oxygen atoms in the carbonyl and ether polar groups in polymerized PEGDA that serve as electron donors, and is strongly influenced by salt concentration and temperature [38,43].Although it seems ideal to choose ratio 1:6 or 1:8 because of their high ionic conductivity, the VPP 3D printing of GPEs 1:6 and 1:8 was difficult due to the The inset corresponds to the equivalent circuit that was used for data fitting.Printability zone was defined according to experimental observations.(b) Cathodic LSV experiments from 0.01 to 5 V vs. Na 0 /Na + using only Ar-GPEs.The inset is a zoom to the region 2.5-5.0V vs. Na 0 /Na + .high quantity of electrolyte and low quantity of polymer matrix.The following electrochemical tests were thus performed on the promising ratios 1:4 and 1:5 to maximize the ionic conductivity while also having enough polymer matrix to allow printability.According to the PEIS results shown in figure 3(a), it is feasible to 3D print in air the ratios GPE 1:4 and 1:5 without perceiving a significant change in the ionic conductivity.
Electrochemical stability tests versus sodium metal in the form of LSV measurements were performed to tape casted Ar-GPEs 1:4 and 1:5.For comparison, this test was also performed to the classical NaClO 4 -based liquid electrolyte retained within a glass fiber separator.The results of the cathodic LSV in figure 3(b) not only show that both GPEs are stable up to 4.75 V vs. Na 0 /Na + , but also that GPEs are more stable than the liquid electrolyte that started to decompose moderately from 3.5 V vs. Na 0 /Na + and decomposed significantly from 4.5 V vs. Na 0 /Na + .This result agrees with PEGDA being electrochemically stable within the electrochemical window in which most SIBs cathode materials operate [44,45], while the photoinitiator is either electrochemically stable or the amount is small enough (∼0.1 wt.%) to not form noticeable redox peaks.Further studies may be performed to determine the role of the photoinitiator during cycling.The better electrochemical stability is nonetheless owed to the formation of a dense crosslinked structure retaining the liquid electrolyte, which restricts oxidative decomposition of the anions and organic solvents, in comparison to bare liquid electrolyte contained within a porouss glass fiber separator.In a similar study using PEGDA, Liu et al [38] found that a GPE composed of PEGDA and LiPF 6 electrolyte was stable up to 4.56 V vs. Li 0 /Li + , a higher potential in comparison with regular LiPF 6 -based electrolytes.

Battery testing using tape casted GPEs
After confirmation of the electrochemical stability and high ionic conductivity, tape casted GPEs 1:4 and 1:5 were assembled in half-cell batteries with Na 0.44 MnO 2 as the working electrode to evaluate their rate performance between 2.0 and 3.8 V vs. Na 0 /Na + (figure 4(a)).Two different configurations (figure 1) were tested: configuration #1 where both the working electrode and the GPE are tape casted (figure 4(a)), and configuration #2 where the working electrode is 3D printed and the GPE is tape casted (figure 4(b)).For comparison purposes (in figure 4(a)), a reference half-cell that uses a liquid electrolyte and glassy fiber instead of a GPE was used.The capacity delivered by the cell containing the Ar-GPE 1:5 (figure 4(a)) is very close to that of the reference, which reaches an specific discharge capacity of 122 mAh•g −1 at C/20, 120 mAh•g −1 at C/10, 100 mAh•g −1 at 1 C and 117 mAh•g −1 when coming back to C/20.On the other hand, the cell containing Ar-GPE 1:4 performs only 10-15 mAh•g −1 lower than Ar-GPE 1:5 at every C-rate, still relatively close from the theoretical specific capacity of Na 0.44 MnO 2 that is 121 mAh•g −1 [46].
Furthermore, the Ar-GPE 1:4 was tested using a direct ink writing 3D printed Na 0.44 MnO 2 electrode (3D-Na 0.44 MnO 2 ) as the working electrode (figure 4(b)).The 3D-Na 0.44 MnO 2 delivers higher capacity than the tape casted Na 0.44 MnO 2 electrode and it is very close to the performance of the reference cell.Both the tape casted and the 3D-Na 0.44 MnO 2 can be compared in terms of charge/discharge specific capacity because these values have been normalized by the electroactive material mass in each electrode.Moreover, the voltage-current curves of the cell containing Na 0.44 MnO 2 (tape casted or 3D-printed) and Ar-GPE1:4 look almost identical, with the six characteristic plateaus attributed to consecutive multiphase reversible reactions (figures 4(c) and (d)) [47,48].These observations indicate that the electrochemical signature for different C-rates were not affected when the direct ink writing 3D printing process was employed to manufacture the 3D-Na 0.44 MnO 2 electrode.
The planar (100% infill) 3D-Na 0.44 MnO 2 electrode was observed by SEM to evaluate the printing quality of the direct ink writing process (figures 5(a)-(d)).This non-conventional manufacturing method produces electrodes with a smooth surface, while also keeping a minimum of microporosity.As expected, the extrusion 3D printing process did not alter the distribution of the carbon-binder domain and active material, similarly to conventional tape casted electrodes.

3D printing of GPEs
Among the different VPP printing parameters that can be tuned to improve the print quality are the initial exposure time, the layer exposure time, the UV light intensity, and the layer thickness.The most relevant parameter for the printing of composite resins containing NaClO 4 -based electrolyte is the exposure time, since it dictates how long the UV light remains for each single layer.Too long exposure may result in a loss of intricate details and printing inaccuracy.On the other hand, a too short exposure may under-cure the composite resin.In this work it was found that a short exposure time (5 s) is adequate for the GPE 1:0 and 1:2 (table 2), but it is not enough for the resins containing a higher amount of electrolyte to allow adequate handleability.The ideal exposure time for composite resins 1:4 and 1:5 were 80 s and 100 s, respectively.Applying a too long exposure time can result in the local heating of the GPE resin that may promote volatilization of the electrolyte solvents (EC and PC) affecting the resin composition, as well as early photopolymerization resulting in an overcuring phenomenon and printing defects.In order to prevent these detrimental effects, the composite resin feedstock was stored and printed at a temperature not exceeding 10 • C, so that the local temperature during the printing step did not exceed 20 • C.Besides selecting a GPE with the highest possible ionic conductivity, the adequate printability of the composite resin feedstock is an equally important characteristic that must be considered to allow shape-conformability.The GPE resin composition 1:5 was thus discarded, since it is at the edge of the printability window (figure 3(a)), and the following electrochemical tests were focused exclusively on the GPE 1:4.
The printing accuracy of the 3D printed GPE 1:4 (3D-GPE 1:4) was analyzed (table 3) to assess the reliability of the printing process to produce high-quality specimens with precise dimensions as established in the computer-aid designs.For this study, a digital file consisting of a 100 µm thick planar disc was used with a view to estimate the first layer printing resolution limitation.Printing accuracy has been established as the theoretical thickness of the disc within the computer-aid model divided by the thickness of the VPP 3D printed GPE.
The printing accuracy was observed to vary according to the overall exposure time, which refers to both the initial and the basic exposure time.The printing accuracy values for all overall exposure times were always lower than 100%, indicating that the measured dimensions of the printed GPEs were greater than those of the computer-aid models.By lowering the overall exposure time, the printing accuracy was improved but at the expense of the 3D printed GPE thin disc handleability.Indeed, only 3D printed GPE discs with thickness ⩾400 µm were smoothly detached from the build plate without breaking thanks to their flexibility.Therefore, a higher exposure time was chosen at the detriment of printing accuracy.In future, besides tuning the printing parameters, the chemical composition of the resin may also be altered with the  introduction of a photoabsorber to regulate the exposure, and improve the accuracy while also ensuring adequate handleability [49,50].
As shown in figure 6, 3D printing of numerous complex designs was demonstrated with the composition resin:electrolyte 1:4 v/v.Planar designs were investigated, including a planar full disc (figures 6(a) and (e)), a mandala (figures 6(b) and (f)), and a spiral (figures 6(c) and (g)).The thickness of the 3D printed planar full disc was kept between 400 and 500 µm in order to avoid any short circuits during electrochemical characterization in half-cell batteries.In future studies, the ability of VPP 3D printing to manufacture shape-conformable GPEs could be leveraged to favor adhesion and interfacial compatibility between electrodes and electrolyte.More complex designs with smaller features can also be printed, such as the face-centered cubic volume lattice shown on figures 6(d) and (h).This design exhibits struts with thickness of about 250 µm.Note that these designs did not leverage the highest resolution of the printer in the Z-axis, that can be as low as 10 µm, so that handleability is not compromised.
The SEM examination to the surface of the spiral 3D-GPE 1:4 (figures 6(c) and (g)) revealed a clear pattern with the intended different printed heights (figure 7(a)).Since the solvents contained within this sample had to be vacuumed prior to SEM observation, well-dispersed crystals of NaClO 4 can be observed in figure 7(b).A higher magnification to the surface revealed an interesting surface arrangement of the polymer chains in the form of crosslinked networks that cover every NaClO 4 crystal (figures 7(c) and (d)).A similar arrangement has been also observed by SEM by Subba Reddy et al [39], in a PVC/NaClO 4 polymerized mixture in 80:20 wt.% ratio, who stated that the organization of the polymer strongly depends on the amount of salt within the composite, rather than the polymerization method.Chen et al [23] also observed a similar arrangement in a PEGDA/LiClO 4 -based electrolyte polymerized mixture in 20:80 vol.% ratio, and mentioned that the sub-micron scale channels formed during photopolymerization can help to enhance the ion transport by reducing interfacial scattering in an analogous manner to conventional porous separators.

Electrochemical testing of 3D printed GPEs
The VPP 3D printed GPEs were tested electrochemically by galvanostatic and potentiostatic means, using both conventional tape casted and 3D printed electrodes (manufactured by direct ink writing) containing Na 0.44 MnO 2 as active material.The ionic conductivity of the 3D-GPE 1:4 was found to be 3.3 ± 0.6 × 10 −3 S•cm −1 , relatively close to the values obtained for tape casted Ar-GPE 1:4 (3.6 ± 0.7 × 10 −3 S•cm −1 ) and air-GPE 1:4 (4.8 ± 0.7 × 10 −3 S•cm −1 ) (table 4).Despite the tape casted air-GPE and the printed 3D-GPE being prepared under a similar air atmosphere, both present slightly different ionic conductivity values.The effect of ambient water exposure is more significant on the tape casted film as it was spread out and punched, while the printed item was immersed with the resin feedstock during the VPP printing step until being transferred to an air-tight container.
The cycling stability of 3D-GPE 1:4 was evaluated with galvanostatic polarization tests within symmetric cells Na 0 /3D-GPE 1:4/Na 0 (figure 8(a)).For comparison purposes, the behavior of tape casted Ar-GPE, serving here as reference, is also displayed (figure 8(b)).The symmetric cell containing the 3D-GPE 1:4 shows a slightly erratic behavior (figure 8(a)), in comparison with the cell containing the Ar-GPE 1:4 (figure 8(b)), most probably due to the water uptake during the 3D printing process.Nonetheless, no short circuit was observed for the duration of the test.After 110 h (55 cycles), the overpotential stabilizes at 3 mV and does not exceed 7 mV in further cycling (figure 8(a)).
On the other hand, the symmetric cell containing the Ar-GPE 1:4 (figure 8(b)) shows clean voltage profiles with low overpotential of only 6 mV that lasted for at least 280 h (140 cycles).The zoom to the selected middle area shows that the interface resistance did not go up with each charge/discharge.Comparably, Castillo et al [51] showed that a non-printed GPE made of a resin and a liquid electrolyte for Li-ion batteries generated an overpotential of 70 mV within similar conditions.Overall, these results highlight the necessity to control the water uptake during the printing process.In future, this could be prevented by placing the printers in a dry room.The water uptake was also visible on the electrochemical stability test of 3D-GPE 1:4, where a new process took place below 1.5 V vs. Na 0 /Na + (figure 8(c)).Indeed, the water redox reaction theoretically occurs at −1.23 V vs. NHE (1.48 vs. Na 0 /Na + ) or at lower potentials depending on the pH of the medium [52].Although not quantifiable from this graph, the current involved in this process did not exceed 0.5 µA.Moreover, the 3D-GPE 1:4 presents an upper-limit operational window of 4.8 V vs. Na 0 /Na + , equivalent to what is observed for the tape casted Ar-GPE 1:4 reference.
Half-cell batteries Na 0.44 MnO 2 /3D (or Ar)-GPE 1:4/Na 0 were assembled to evaluate the impact of water on the cycling performance (figure 8(d)).Here, the Na 0.44 MnO 2 working electrode was tape casted, while the GPE was either 3D printed (figure 1-configuration #3) or tape casted (figure 1-configuration #1, shown for comparison purposes).Initially, the performance of the 3D-GPE 1:4 is similar to that of the Ar-GPE 1:4 at the low rates C/20 and C/10 for the first ten cycles, however, the specific capacity dramatically drops at the high rate of 1 C to only 10 mAh•g −1 .The capacity is effectively recovered to just below 100 mAh•g −1 when coming back to C/20 for the last five cycles.The capacity drop at high C-rate may be due to several reasons including (i) differences in polymer arrangement as a result of the manufacturing process, (ii) cross-linking degree differences caused by the exposure either to the UV 3D printer projector or the UV lamp, and (iii) local temperature during photopolymerization.

Multiprocess SIBs additive manufacturing
The last part of this work was focused on combining the direct ink writing 3D printed Na 0.44 MnO 2 electrode and the VPP 3D printed GPE 1:4 into the same battery cell (figure 1-configuration #4), towards the manufacturing of a 3D printed battery thanks to two additive manufacturing processes.Four designs of 3D printed 3D-Na 0.44 MnO 2 electrodes with different path spacings were cycled versus Na 0 using the 3D-GPE 1:4 (figure 9).Design #1, Design #2 and Design #3 exhibit paths spacing of 0.6 mm, 0.9 mm, and 1.2 mm, respectively.As a reference, a full electrode disc was also printed.Optical images of the different electrode designs and their electrochemical performance are shown in figure 9.
The full disc presents an initial discharge capacity just below 110 mAh•g −1 , similar to what was observed previously in figure 4(b) when combining a 3D printed Na 0.44 MnO 2 electrode and a tape casted Ar-GPE.Design #1, #2 and #3 exhibit initial higher specific discharge capacities with values of 118, 122 and 128 mAh•g −1 , respectively.The specific capacity data has been normalized by the active material weight, hence it is possible to compare one to each other.For all the C-rate tested in figure 9, increasing the path spacing in the electrode design resulted in enhanced specific capacity.Design #3 with the highest path spacing exhibits the best electrochemical results with capacity values of 126 mAh•g −1 , 120 mAh•g −1 , 86 mAh•g −1 , and 119 mAh•g −1 after five cycles at C/20, C/10, 1 C, and back to C/20, respectively.This experiment proves that it is possible to assemble two 3D printed components manufactured with different additive manufacturing processes (direct ink writing and VPP) to build a functional half-cell battery.Specific capacity values can be optimized by modifying the path spacing and printed pattern.

Conclusion
In this work, GPEs with a resin:electrolyte 1:4 v/v ratio were successfully 3D printed and electrochemically tested by various methods.The 3D printed GPE 1:4 presented an ionic conductivity of 3.3 × 10 −3 S•cm −1 at room temperature, a very similar value to the one from an Ar-GPE 1:4, and comparable to literature in the field.Inside a Na 0.44 MnO 2 /Na half-cell, the 3D printed GPE 1:4 delivered equivalent results to a cell using an Ar-GPE 1:4 at the C-rates of C/20 and C/10, meaning that the 3D printing process kept the regular performance of tape casted electrodes, while also expanding the electrochemical stability window up to 4.8 V vs. Na 0 /Na + .In future studies, the developed GPE composite resins may also be employed with other 3D printer systems that use UV light, such as direct ink writing printers equipped with a UV-curing option, or also the two-photon polymerization (2PP) processes.The latter option is particularly interesting for battery applications since 2PP allows printing with resolution of only tens of nanometers.In addition, the versatility of these composite resins makes them candidates for multi-material VPP printing [53,54], which will allow future one-shot batteries 3D printing [15].On the other hand, a composite ink loaded with Na 0.44 MnO 2 and conductive additives was developed to 3D print positive electrodes via direct ink writing.After demonstrating the functionality of the independent 3D printed components in SIBs, the two 3D printed components, manufactured with different additive manufacturing processes (direct ink writing and VPP), were combined together into the same cell to build a functional half-cell battery.When employing the 3D printed GPE 1:4 inside half-cells 3D-Na 0.44 MnO 2 /Na 0 , a tendency was observed as a function of the printed pattern.Indeed, by increasing the path spacing, the performances in terms of specific capacities at the different C-rates were improved.So far, only planar 3D printed GPEs were tested inside half-cells, but the authors anticipate that the complex GPE designs that are otherwise not attainable by conventional tape casting methods will further favor adhesion and interfacial compatibility with both electrodes.This work appears as the first step towards the manufacturing of a complete shape-conformable battery thanks to different additive manufacturing processes.

Figure 1 .
Figure 1.Coin cell scheme detailing the four configurations of the working electrode and GPE that are discussed throughout the paper.Note that the illustrations of the 3D printed electrodes and GPEs in the blue boxes are for demonstration purposes only, and the actual 3D printed items shapes are described in the results and discussion section.

Figure 2 .
Figure 2. (a) Viscosity measurements of a GPE composite resin with resin:electrolyte v/v 1:0 (black data) and 1:1 (orange data) ratios; (b) FTIR spectra taken in transmission mode of the Ar-tape casted GPE 1:0 and 1:1.Green arrows correspond to peak coincidences with the NaClO4 spectrum and blue to peak coincidences with the EC and PC spectra.

Figure 3 .
Figure 3. (a) Ionic conductivity data of different resin:electrolyte v/v ratios tape casted inside of an Ar-filled glovebox (Ar-GPE) or under a regular atmosphere (air-GPE).The inset corresponds to the equivalent circuit that was used for data fitting.Printability zone was defined according to experimental observations.(b) Cathodic LSV experiments from 0.01 to 5 V vs. Na 0 /Na + using only Ar-GPEs.The inset is a zoom to the region 2.5-5.0V vs. Na 0 /Na + .

Figure 4 .
Figure 4. Half-cell battery performance using a combination of an Ar-GPEs 1:4 and a: (a) tape casted Na0.44MnO2 electrode (configuration #1), or a (b) 3D printed Na0.44MnO2 electrode (configuration #2).As a reference, tape casted Na0.44MnO2/Na 0 using liquid electrolyte is added for comparison in (a).(c) and (d) Voltage-current curves for the first and sixth cycle of cells shown in (a) and (b).

Table 2 .
Exposure time as a function of GPE resin:electrolyte ratio.

Table 3 .
Printing accuracy data of the GPE 1:4 resin.