A Photolithographable Electrolyte for Deeply Rechargeable Zn Microbatteries in On‐Chip Devices

Zn batteries show promise for microscale applications due to their compatibility with air fabrication but face challenges like dendrite growth and chemical corrosion, especially at the microscale. Despite previous attempts in electrolyte engineering, achieving successful patterning of electrolyte microscale devices has remained challenging. Here, successful patterning using photolithography is enabled by incorporating caffeine into a UV‐crosslinked polyacrylamide hydrogel electrolyte. Caffeine passivates the Zn anode, preventing chemical corrosion, while its coordination with Zn2+ ions forms a Zn2+‐conducting complex that transforms into ZnCO3 and 2ZnCO3·3Zn(OH)2 over cycling. The resulting Zn‐rich interphase product significantly enhances Zn reversibility. In on‐chip microbatteries, the resulting solid‐electrolyte interphase allows the Zn||MnO2 full cell to cycle for over 700 cycles with an 80% depth of discharge. Integrating the photolithographable electrolyte into multilayer microfabrication creates a microbattery with a 3D Swiss‐roll structure that occupies a footprint of 0.136 mm2. This tiny microbattery retains 75% of its capacity (350 µAh cm−2) for 200 cycles at a remarkable 90% depth of discharge. The findings offer a promising solution for enhancing the performance of Zn microbatteries, particularly for on‐chip microscale devices, and have significant implications for the advancement of autonomous microscale devices.


Introduction
Rechargeable Zn batteries are emerging as compelling alternatives to lithium-ion batteries due to their high safety and costeffectiveness.While Zn batteries are primarily used for large-scale energy storage, [1] they also offer significant engineering benefits in microscale applications due to its safety and adaptability to new materials. [2,3]n contrast to high energy density lithiumion batteries, which necessitate an inert atmosphere during assembly and face challenges in accommodating multiple deposition, lithography, and etching iterations, Zn batteries prove highly adaptable to a range of microfabrication processes, which stems from the inherent stability of Zn metal in both air and aqueous electrolytes.However, challenges such as dendrite growth and chemical corrosion continue to affect the reversibility of Zn batteries. [4]Although some studies have demonstrated reversibility over extended periods, [5] the practical performance of Zn batteries in real devices remains uncertain due to their limited depth of discharge (DOD).This limitation particularly hampers the utilization of the Zn anode, especially in microscale devices. [6]n particular, interdigitated microbatteries exhibit poor cycling stability when the DOD exceeds 30% (Figure 1a; the DOD calculation is elaborated in the Supplementary text, and Figures S1  and S2, Supporting Information show the detailed cycling performance at different DODs).A low DOD corresponds to a diminished energy density attained within constrained material loads, particularly in batteries utilized for microdevice applications.To strive for an enhancement in DOD and subsequently, energy density, the implementation of asymmetric electrode layouts emerges as a viable solution.Notably, opting for a solitary Zn microstrip electrode instead of seven microstrips yields a noteworthy 40% surge in areal energy density.However, it is important to acknowledge that this design exhibits a relatively limited lifespan of less than 100 rechargeable cycles due to its high DOD of ≈80% (Figure 1a).In order to facilitate the self-sufficient operation of microscale devices, ranging from a few millimeters to micrometers in size, through the integration of Zn microbatteries, the achievement of a high areal energy density while maintaining high reversibility becomes a critical endeavor.[7] To enhance the Zn reversibility at high DODs, various techniques have been developed including the use of polymers and exotic metals to modify the surface and the creation of an interface layer via electrolyte engineering.[8] However, for these strategies to work in microscale devices, the crux of the matter lies in their amenability to patterning.Photolithography is a pivotal tool in the production of microscale electronic devices, enabling parallel patterning of materials.[9] In the standard process, photolithography entails the exposure of a photosensitive material, such as photoresists, to UV light, thereby generating a patterned layer that serves as a mask for subsequent etching or deposition steps, enabling the desired patterning of solid materials like current collectors and electrode materials (Figure 1b).Nevertheless, when it comes to integrating Zn microbatteries on a chip, the major challenge lies in the intricate task of patterning electrolytes, as solid-state electrolytes for aqueous Zn batteries are currently unavailable.[10] One potential solution involves the use of hydrogels that can be crosslinked under UV light to generate patterned electrolytes.However, the successful implementation of this approach is accompanied by a multitude of technical con-siderations, including the need to devise strategies that enhance Zn reversibility, address the complexities associated with the fabrication process, ensure stability during multilayer microfabrication steps, achieve high-resolution patterning, and maintain uniformity throughout.
Here, we develop a photolithographable electrolyte (Figure 1c) based on a UV-crosslinked polyacrylamide (PAM) hydrogel consisting of ZnSO 4 and caffeine.We find that caffeine can be spontaneously adsorbed to the Zn surface, thus passivating the Zn anode and preventing Zn chemical corrosion.The coordination complex formed between Zn 2+ ions and caffeine molecules is Zn 2+ -conductive, which gradually transforms into Zn 2+ -conductive solids-ZnCO 3 and 2ZnCO 3 •3Zn(OH) 2 .This Zn-rich layer brings about a significant enhancement in Zn reversibility, with the Zn foil anode demonstrating remarkable reversibility even at a 90% DOD.The high efficiency of the caffeine additive in improving Zn reversibility was further evidenced in microscale devices.The caffeine additive inhibits the self-discharge of the on-chip Zn||MnO 2 microbattery and renders a cycling stability of > 700 cycles at a DOD of 80%.More importantly, our research delves deeper into the realm of microfabrication by integrating the caffeine-containing photolithographable electrolyte into multilayer stacks, thereby enabling the creation of a 3D microbattery with a Swiss-roll architecture achieved through an on-chip self-assembly process.This Swiss-roll microbattery delivers an areal capacity of 350 μAh cm −2 with a small footprint of 0.136 mm 2 , and maintains an impressive 75% of its capacity over 200 cycles, even under an exceptionally high DOD of 90%.

Adsorption of Caffeine on Zn Surface
Recent work on highly reversible Zn anodes has pursued the in situ formation of a Zn 2+ conducting interface layer in aqueous electrolytes by fluorine-containing additives such as trifluoromethanesulfonate (OTF − ) anions; however, fluorine often prevents monomers (e.g., acrylamide) from crosslinking under UV light.Given that water-soluble organic additives with carboxyl, [11] amino, [12] imidazole, [13] and amine functional groups, [5c,14] coordinate with Zn 2+ ions, forming coordination complexes as the INTERFACE LAYER, we posit that caffeine with an imidazole ring in the aqueous electrolyte (2 m ZnSO 4 ) could facilitate the formation of a Zn complex interphase with cycling, thereby improving Zn reversibility while also enabling photo-induced polymerization of acrylamide.As shown in Figure 1c, patterned squares (200 by 200 μm 2 ) of the polyacrylamide hydrogel electrolyte containing 20 mm caffeine are successfully produced.Figure 2a describes the binding energies of caffeine molecules on prototypical Zn surfaces through density functional theory (DFT) calculations.A strong electron transfer between both (002) and (100) facets of Zn and caffeine molecules is observed, indicating that caffeine can be adsorbed onto the Zn surface.The spontaneous adsorption of caffeine molecules is further experimentally evidenced by the 13 mV increase in the time-dependent open circuit potential (Figure S3a, Supporting Information), [15] as well as negative values of Gibbs free energy (△G ads ) of caffeine adsorption (Figure 2b, calculations of △G ads by different methods are described in Supplementary text).We fit △G ads from the Tafel plot of the Zn foil in the electrolyte with 20 mm caffeine delivering −19.29 kJ mol −1 .Additionally, △G ads derived from cyclic voltammetry (CV) curves and the electrochemical impedance (EIS) spectrum (Figure S3b,c, Supporting Information) are −17.39 and −25.83 kJ mol −1 , respectively.The adsorption of caffeine is found to inhibit unwanted Zn corrosion, [16] as seen by the increase in corrosion potential from −0.94 to −0.91 V and the decrease in corrosion current from 43 to 1.24 μA cm −2 after adding 20 mm caffeine (Figure 2c).Notably, the hydrogen evolution reaction accompanying Zn corrosion is also suppressed as the onset potential decreases from −1.07 to −1.14 V (Figure S4, Supporting Information).Such a good corrosion inhibition performance is attributed to the high caffeine coverage (94%) of the Zn surface (see Supplementary text and Table S1, Supporting Information for a detailed calculation and discussion about the inhibition efficiency).
As shown in Figure 2d, the exchange current density in the electrolyte with caffeine is 15.83 mA cm −2 , more than three times higher than in the caffeine-free electrolyte (4.24 mA cm −2 ).The large exchange current density suggests a considerable concentration of Zn 2+ at the surface, resulting in a discernible concentration gradient. [17]This effect is attributed to the coordination of Zn 2+ with caffeine molecules, which cover over 90% of the Zn anode surface and coordinate with Zn 2+ ions in the electrolyte, increasing the interface Zn 2+ concentration as shown in Figure 2e.However, in the bulk solution, the large cluster consisting of Zn 2+ and caffeine causes a dramatic drop in Zn 2+ concentration.It is noteworthy that the adsorbed caffeine regulates the Zn deposition process.Under an overpotential of −150 mV, the deposition current decreases in the caffeine-containing electrolyte, while in the caffeine-free electrolyte the current continuously increases (Figure S5, Supporting Information).The continuous increase in deposition current is attributed to the rampant Zn 2+ diffusion that facilitates the formation of Zn tips and therefore dendrites. [18]In contrast, constrained diffusion through the Zn caffeine complex allows for the uniform distribution of Zn deposits. [19]We first examined the impact of caffeine additive concentrations (0, 10, 20, 30, and 40 mm, Figure S6, Supporting Information).The addition of caffeine improves the cycling stability of Zn anodes.Nevertheless, at the elevated concentration of 40 mm, the cell showed soft short circuits at 728 h, underscoring that a thick interface layer also increases the risk of short circuits.Beyond 40 mm, observable precipitates of caffeine became evident.For instance, at 60 mm caffeine, dissolution in the electrolyte proved challenging due to the strong coordination between caffeine and Zn 2+ (Figure S7, Supporting Information).As it is crucial to reduce the quantity of additives in the electrolyte to minimize any compromise in energy density, [20] our findings indicate that the optimal concentration of caffeine in 2 m ZnSO 4 electrolyte is 20 mm.In addition, as shown in Figure S8, Supporting Information, the ionic conductivity increased from 26 to 36 mS cm −1 after adding 20 mm caffeine to the electrolyte.The Nyquist plots measured in symmetric cells without/with 20 mm caffeine (Figure S9, Supporting Information) also reveal a decrease in charge transfer resistance.The increase in the stripping and plating overpotential is attributed to the caffeine adsorbed on the Zn surface, acting as a regulator for Zn depositions that prevent random growth.

Highly Reversible Zn Anode
We investigate the caffeine effect on the long-term cycling stability of the Zn anode in a coin cell (Figure 3a).In the caffeine-containing electrolyte, the Zn||Ti cell exhibits >99% plating/stripping efficiency (Coulombic efficiency, CE) in the initial 12 cycles, while the CE slowly increases to 95% without caffeine (Figure 3b).The high initial CE is attributed to the dehydration of Zn 2+ ions by hydrophilic-ionic interactions with caffeine, which prevents water from approaching the electrode surface. [21]This finding is also in line with the excellent anticorrosion property of caffeine (Figure 2c).The Zn plating overpotential increases by 50 mV (Figure S10, Supporting Information), which is likely caused by a large nucleation overpotential (Figure S11, Supporting Information).The increased nucleation overpotential is often correlated to a regulated Zn growth and hence improved cycling stability. [22]Caffeine results in an average CE of 99.8% over 1000 cycles at a current of 2 mA cm −2 and a capacity of 2 mAh cm −2 (Figure 3b).In contrast, the Zn||Ti cell without caffeine fails after 265 cycles, with an average CE of 96%.A Zn||Zn cell with caffeine demonstrates excellent Zn reversibility at 2 mA cm −2 with a capacity of 2 mAh cm −2 (9% DOD) for over 800 cycles, while in the caffeine-free electrolyte, it fails after 30 cycles (Figure 3c).5a] When the DOD is increased to 45%, Zn reversibility in the caffeine-containing electrolyte still reaches more than 180 cycles.On the contrary, the Zn anode was shortened after 30 cycles in the caffeine-free electrolyte (Figure 3d).At 90% DOD, the Zn anode in the caffeine-free electrolyte is irreversible.By contrast, the caffeine additive maintains good Zn reversibility (Figure S12, Supporting Information).Caffeine also improves the rate capability of the Zn anode (Figure 3e).The overpotential at 4 mA cm −2 of the Zn||Zn cell with caffeine is 74 mV smaller than for the caffeine-free electrolyte.Moreover, caffeine prevents soft short-circuit, as seen by the significantly reduced Zn stripping/plating overpotential in the cell without caffeine when the current density returns to 0.5 from 4 mA cm −2 .We used the common interdigitated structure to scale down the electrode size to the micrometer scale (Figure 3f).The Zn||Zn cell (interdigitated structure) coated by the photopatternable electrolyte containing caffeine can be cycled over 500 times at a large current (Figure 3g).In comparison, the Zn||Zn cell only survives for 85 cycles with obvious fluctuation in galvanostatic cycling profiles.We also observed the short circuit at 120 μA without caffeine (Figure 3h), which is caused by the irregular growth of Zn producing large Zn plates/dendrites that short microelectrodes (Figure S13, Supporting Information).The addition of caffeine refines the Zn growth and thereby avoids the risk of short-circuit at high current densities.

Gradient Zn 2+ -Conducting Solid ElectrolyteInterface (SEI)
We recorded cross-sectional scanning electron microscopy (SEM) images of Zn anodes cycled in pure ZnSO4 electrolyte and 20 mm caffeine-contained electrolyte as shown in Figure S14, Supporting Information.The Zn foil cycled in pure ZnSO4 electrolyte shows a rough surface consisting of irregular and agglomerated Zn clusters, indicating that Zn is pulverized over cycling.By contrast, a compact morphology is maintained on the surface cycled in the electrolyte containing 20 mm caffeine.Moreover, a clear interface layer (indicated by red lines) is observed on the Zn foil, which is attributed to the Zn-caffeine complex.The topview SEM image (Figure 4a) shows horn structures on the cycled Zn surface, which are completely different from hexagonal deposits of most cycled interface products such as Zn crystals [4c,23] and Zn basic salt [11,24] (Figure S15, Supporting Information).Thus, the atypical structures indicate a new interface product.We collected the interface product on the Zn anode cycled for 20 times.The Raman spectrum of the interface product (Figure S16, Supporting Information) suggests a complex consisting of Zn and caffeine molecules.The negative peak shifts at 7.78 and 3.0 ppm in the 1 H nuclear magnetic resonance (NMR) spectrum (Figure 4b) indicate that the product is a coordination complex.
A new signal at 8.13 ppm (H8) confirms the coordination of caffeine with Zn 2+ ions, which forms a Zn complex deposit on the electrolyte/electrode interface.The Zn caffeine complex was predicted to be Zn 2+ -conductive by DFT calculations as shown in Figure S17, Supporting Information.Ab-initio molecular dynamics (AIMD) calculations were performed to explore the pathways of the Zn 2+ ion transport in the Zn caffeine complex.As shown in Figure 4c and Figure S17, Supporting Information, Zn 2+ ions are able to move across the coordination complex within restricted transport pathways, which also contributes to the increased stripping/plating overpotential.
The X-ray diffraction (XRD) result of the Zn anode after 20 cycles displays additional components such as ZnCO 3 and 2ZnCO 3 •3Zn(OH) 2 (Figure S18, Supporting Information).The absence of the peaks associated with basic Zn sulfate reveals that the in situ formed interphase can effectively prevent the side reaction (i.e., hydrogen evolution) that produces basic Zn sulfate.To clarify the gradient interphase, we characterized the Zn anode by XPS after etching by 20 and 150 nm.Peaks at 1022.8 eV (Figure 4d) display coordinated Zn 2+ on the surface, which is ascribed to the Zn caffeine complex.Peaks at 1022.2 eV in the Zn 2p 2/3 spectrum and 532.1 eV in O 1s spectrum (Figure 4e) are assigned to Zn-O, stemming from the interaction between Zn 2+ and C═O in caffeine.The 13 C NMR spectrum (Figure S19, Supporting Information) also reveals that Zn 2+ ions induce the cleavage of C═O over cycling.The Zn 2+ -caffeine complex compound covered Zn anode shows an increase in the Zn transference number, indicating a selectivity of Zn 2+ ion transport through such compound (Figure S20, Supporting Information).A small peak for CO 3 2− in the C1s spectrum (Figure 4f) implies that caffeine molecules are likely to be decomposed.As such, the surface layer mainly comprises the coordination complex and small amounts of carbonate produced by the decomposition of caffeine.
At deeper levels, the composition of CO 3 2− significantly increases from 10.6% to 16.4% at 20 nm depth and 44.8% at 150 nm depth, along with the shift to carbonate characteristics with a higher binding energy (Figure 4f).Given the existence of ZnCO 3 and 2ZnCO 3 •3Zn(OH) 2 in the XRD result (Figure S18, Supporting Information), we suggest a gradient interface layer.The composition under the surface transforms to inorganic carbonates and hydroxides (Figure 4g).At a depth of 20 nm, the coordination complex is still present, as the C─C bond is clear in the C 1s spectrum (Figure 4f, 20 nm).However, in 150 nm depth, the C─C bond from caffeine is difficult to be detected and is covered by the C─O bond from inorganic compounds (Figure 4f, 150 nm).In Figure 4f, when the etch depth was 0 and 20 nm, the peak position at 284.8 eV was assigned to the C─C bond.However, when the etching depth increases to 150 nm, the zinc carbonate dominates (increased CO 3 2− intensity), which shifts the C 1s peak to a higher binding energy. [25]The Zn 2p 3/2 and O 1s spectra still show the possible presence of the coordination product because coordinated Zn 2+ and C═O bond are found (Figure 4d-f, 150 nm).As such, a complicated mixture of the coordination complex and ZnCO 3 and 2ZnCO 3 •3Zn(OH) 2 is formed near the metallic Zn surface, where the inorganic compounds are the main composition (Figure 4g).Notably, the inorganic compounds, ZnCO 3 and 2ZnCO 3 •3Zn(OH) 2 are Zn 2+ -conducting and able to improve the reversibility in the long-term operation as presumed in previous reports. [26]All the compositions in the gradient interface layer are highly conductive for Zn 2+ ion transportation, thus generating a Zn-rich interface layer.As shown in Figure S21, Supporting Information, the Zn ratio increases from 18.8% to 43.1% for the Zn anode cycled in the caffeine-containing electrolyte.By contrast, the Zn ratio is less than 10% for the Zn cycled in the caffeine-free electrolyte.The Zn-rich interface layer can facilitate the solid-tosolid conversion of Zn 0/2+ and therefore further improves the Zn reversibility. [27]

Full Cell Performance
We began with investigating the effect of caffeine on the full cell performance in Zn||MnO 2 coin cells.The CV results show typ-ical redox peaks of the Zn||MnO 2 battery (Figure S22, Supporting Information): a merged oxidation peak at 1.6 V and two distinct reduction peaks at 1.4 V and 1.2 V, which is consistent with charge/discharge plateaus in the galvanostatic charge/discharge profiles at 1C (Figure 5a).The initial capacity of the Zn||MnO 2 battery with caffeine (251 mAh g −1 ) is higher than in the caffeinefree electrolyte (172 mAh g −1 ), which is likely due to the improved initial CE of the Zn anode.Moreover, the Zn||MnO 2 battery capacity with caffeine at current densities from 1 C to 5 C is always higher than without caffeine (Figure 5b).When the current density is reduced to 1C, a reversible capacity of 253 mAh g −1 is attained by the caffeine-containing cell, which is almost twice the capacity without caffeine (133 mAh g −1 ). Figure 5c compares the stability of the Zn||MnO 2 battery cycled at a current density of 2 C. Caffeine renders a reversible capacity retention of 95% across 500 cycles while only 35% capacity is retained without caffeine.Furthermore, the CE remains at 99% over cycling for the Zn||MnO 2 with caffeine.Without caffeine, an abrupt drop and fluctuation in CE after 300 cycles indicates irreversibility.A high reversibility of the Zn||MnO 2 battery in the caffeine-containing electrolyte is also observed at 0.5 C and 10 C (Figure S23, Supporting Information).For small-scale and on-chip performance, we fabricated onchip interdigitated Zn||MnO 2 batteries with 7 mm long and 0.5 mm wide electrode strips.The clear hydrogel electrolyte (Figure S24, Supporting Information) was patterned on the electrodes by photolithography.Our lithographable electrolytes, distinguished by their superior resolution, allow for the realization of diverse shapes-from simple rectangles to intricate designs such as a panda (Figure S25, Supporting Information).The DOD was set to about 80% as only one electrode strip was assigned for the Zn anode.The first indicator for Zn anode stability is selfdischarge.The Zn||MnO 2 battery is almost fully discharged by itself after a 24-h rest in a caffeine-free electrolyte (Figure 5d).In contrast, caffeine suppresses the self-discharge as a typical discharge plateau above 1.3 V is observed and 71% capacity is retained (Figure 5e).Moreover, the capacity of the Zn||MnO 2 battery decreases dramatically in the first 50 cycles, while caffeine stabilizes the capacity of the Zn||MnO 2 battery (Figure 5f).The gradual increase in capacity over cycling of the Zn||MnO 2 battery is attributed to the redeposition of manganese oxide from Mn 2+ ions in the electrolyte. [28]The rate performance shows a similar trend to that of the coin cells (Figure S26, Supporting Information).

Multilayer Fabrication Compatibility and Swiss-Roll Microbattery Performance
Furthermore, the suitability of caffeine-containing hydrogel electrolytes for on-chip device fabrication can be evaluated based on their compatibility with photolithography and multilayer processes.To verify this, we patterned a multilayer stack consisting of a sacrificial layer and five functional layers (Figure 6a).As illustrated in Figure 6b, a polyimide film was layered on the sacrificial layer (Figure S27, Supporting Information).As polyimide is a Zn 2+ -conductive polymer, [29] a 40-nm-thick SiO 2 layer was deposited on the polyimide to prevent current leakage across interdigitated electrodes.Subsequently, interdigitated electrodes are prepared, including the current collector (15 nm Cr and 60 nm Au) and electrode materials (2 microstrips of Zn and 6 microstrips of MnO 2 ).The interdigitated electrodes were coated by caffeine-containing hydrogel with different thicknesses, ranging from 1300 to 480 nm (Figure S28, Supporting Information).The ability to pattern the hydrogel into a square (200 by 200 μm 2 , Figure 6a) confirmed its photolithographic processability.By adding an additional hydrogel layer with lower swelling capacity, we achieved the self-roll-up of the multilayer stack, resulting in a micro-Swiss-roll Zn||MnO 2 battery with a length of 850 μm and a diameter of 160 μm (Figure 6c).It is important to note that the PAM layer also undergoes swelling in aqueous solutions, introducing additional strain within the layer stack and consequently impacting the radius of the Swissroll.As depicted in Figure S28, Supporting Information, a decrease in PAM thickness correlates with an increase in the radius of the Swiss-roll, transitioning from 110 to 175 μm. Figure 6d shows the parallel formation of hair-thin micro-Swiss-rolls.The implementation of this on-chip self-assembly technique to shape a 3D structure demonstrates that the caffeine-containing hydrogel electrolyte can be used for the processing of complex microstructures.CV curves (Figure S29, Supporting Information) show typical redox peaks of the Zn||MnO 2 battery.The micro-Swiss-roll Zn||MnO 2 battery attains a capacity of about 470 nAh on a footprint of 0.136 mm 2 , resulting in a footprint capacity of 350 μAh cm −2 (Figure 6e).At 90% DOD, the micro-Swissroll battery with caffeine shows a capacity retention of about 75% over 200 charge/discharge cycles at a current of 1 μA (Figure 6f).In contrast, the capacity of the micro-Swiss-roll battery in the caffeine-free electrolyte decreases to less than 20% in 50 cycles.We also employed focused ion beam etching to cut the Swiss-roll microbattery (Figure S30, Supporting Information).The crosssection reveals that the thin Zn microanode maintained its compact structure throughout the cycling process, in contrast to the loose structure formed by randomly distributed Zn crystals in the thin Zn microanode cycled in the caffeine-free electrolyte, as depicted in Figure S15, Supporting Information.In addition, as depicted in Figure S31, Supporting Information, the cycling life of the Swiss-roll microbattery with a 480 nm PAM layer (8000 rpm) is similar to its counterpart with a 670 nm PAM layer (6000 rpm), as detailed in our manuscript: both demonstrate over 200 cycles at a current of 1 μA.However, when the PAM layer exceeded a thickness comparable to that of the Zn anode (>1000 nm, 4000 rpm), the reversibility of the microbatteries was significantly affected as evidenced by the cycling profiles with notable fluctuations.This phenomenon is attributed to the increased polarization during the charging process (Figure S32a, Supporting Infor-mation), arising from the increased charge transfer resistance of a thick PAM layer (Figure S32b, Supporting Information).

Conclusion
In summary, we enhance the performance of Zn microbatteries by introducing caffeine into the widely employed ZnSO 4 electrolyte.By incorporating this water-soluble organic compound, we effectively combat corrosion and hinder dendrite growth on the Zn anode.Caffeine acts as a passivating agent for the Zn anode, interacting with Zn 2+ ions to forge a Znrich interface layer that facilitates the solid-to-solid conversion of Zn 0/2+ with remarkable reversibility.The efficacy of our approach is exemplified by coin-cell tests, where the Zn||Ti cell displays a Coulombic efficiency of 99.8% over the course of 1000 cycles.Moreover, the Zn||Zn cell consistently maintains its reversibility even at 90% DOD, thereby underscoring the potential of our microbattery design, even with limited Zn mass loading.Leveraging the power of photolithography, we successfully pattern the caffeine-containing electrolyte, enabling the fabrication of on-chip Zn||MnO2 microbatteries that exhibit enhanced reversibility at an impressive 80% DOD, with a remarkable cycling stability surpassing 700 cycles.Taking our advancements further, we seamlessly integrate the photolithographable electrolyte into a multilayer microfabrication process, yielding Swiss-roll microbatteries with a compact footprint area measuring 0.136 mm 2 .These Swiss-roll microbatteries retain 75% of their capacity over the course of 200 cycles subjected to a high DOD of 90%.This exceptional reversibility at elevated DOD levels presents an exciting opportunity for electrode downsizing, propelling the development of high-performance microbatteries and expanding the capabilities of microscale devices to new horizons.

Experimental Section
Half/Full Cell Fabrication: Zn foils with a thickness of 50 μm were used as Zn anodes in coin cells.MnO 2 electrodes were prepared according to a previous work.1. 2 m ZnSO 4 and 2 m ZnSO 4 with 20 mm caffeine were used as electrolyte in the coin cells.Glass fiber films were used as separators.To synthesize photolithographable electrolyte, 10 mL 2 m ZnSO 4 with/without 20 mm caffeine was heated up to 90 °C, followed by adding 1.5 g gelatin under stirring.After the mixture cooled down, the 3 g acrylamide was dissolved in the solution.45 mg N, N′methylenebis (acrylamide) and 45 mg 2-hydroxy-4′-(2-hydroxyethoxy)−2methylpropiophenone were subsequently added into the solution and stirred at room temperature overnight.
To fabricate interdigitated microbatteries, Cr/Au current collectors were deposited using nanoPVD (Moorfield Nanotechnology Ltd.) and patterned by a standard lift-off process using an AZ-5214E photoresist (Microchemicals GmbH).The Zn microanode was electrochemically deposited under −0.008A in the electrolyte of 1 m ZnSO 4 .All the Zn electrochemical depositions were conducted in the three-electrode cell, where a Pt net was used as the counter electrode and Ag/AgCl was used as the reference electrode.The MnO 2 microcathode was electrochemically deposited according to the previous work. [30]Briefly, the PEDOT buffer layer was electrodeposited in the mixture solution (1 mm 3,4-ethylenedioxythiophene (EDOT), 1 mm sodium dodecyl sulfate, and 1 m H 2 SO 4 ) at constant voltage for 30 s. MnO 2 was deposited in the electrolyte containing 0.01 m Na 2 SO 4 and Mn(CH 3 COO) 2 using CV scan between 0.3 and 0.6 V at a scan rate of 50 mV s −1 for 12 cycles followed by a constant voltage of 0.6 V for 90 s and this deposition process was repeated for six times.To change DOD, the interdigitated microbatteries were designed as asymmetric with different ratios between fingers of anode and cathode as shown in Figure S1, Supporting Information.
To fabricate micro-Swiss-roll battery, sacrificial, polyimide, and hydrogel layers were created as previously reported. [31]Briefly, the sacrificial layer solution was first spin-coated onto the O 2 plasma cleaned Si/SiO 2 substrates at 6000 rpm for 60 s.The samples were baked at 35 °C for 10 min for solvent removal and further exposed (365 nm, 15 mW cm −2 ) for 60 s through a glass/Cr photomask using a SUSS MJB4 mask aligner (Karl Suss KG-Gmbh).The patterned samples were developed in deionized water for 20 s and baked at 220 °C for 10 min to obtain a stable sacrificial layer.Second, polyimide (PI) precursor solution was spin-coated at 4000 rpm for 60 s and baked at 50 °C for 10 min.After exposed (365 nm, 15 mW cm −2 ) for 2 min through the glass/Cr photomask, the samples were then developed in a mixture of 1 part (v/v) of dimethylacetamide and 9 parts (v/v) of propylene carbonate until the unexposed layer was removed.Finally, the samples were rinsed in propylene glycol methyl ether acetate and baked at 220 °C for 30 min.SiO 2 with 35 nm thickness was deposited on the PI layer using plasma enhanced chemical vapor deposition chemical vapor deposition (CVD, SI 500 D, Sentech Instruments).Then the deposition of Cr/Au current collector was followed by lift-off process.The micro-anode was deposited under −0.002A (chrono potentiometry method) in the electrolyte of 1 m ZnSO 4 , while micro-cathode deposition was conducted using CVchronocoulometry method for three times.Subsequently, the patternable electrolyte was spin-coated at 6000 rpm for 60 s.After baking at 90 °C for 25 min, samples were exposed for 300 s through the glass/Cr photomask.The sample was developed in deionized water for 20 s.Finally, the solution for the swellable hydrogel layer was spin-coated at 6000 rpm for 60 s.After baking at 40 °C for 10 min, samples were exposed (365 nm, 15 mW cm −2 ) for 90 s through the glass/Cr photomask using the SUSS MJB4 mask aligner.The final layer was developed in 2-(2-methoxyethoxy) ethanol for 45 s and rinsed in propylene glycol methyl ether acetate to remove unexposed polymer.The samples were baked at 220 °C for 30 min.
The rolling process of planar devices was controlled by selectively etching the sacrificial layer in the solution of 0.5 m ethylenediaminetetraacetic acid disodium salt with a pH value of 8.After the rolling process, the microscale battery was taken out of the etching solution and dried at room temperature.
Material Characterization: XRD characterizations were performed on X'Pert PRO MPD, Philips (Co K radiation,  = 1.78910Å).XPS tests were performed on ESCALAB 250Xi (Thermo Fisher) spectrometer equipped with Al-K radiation (8.34 Å) as power source.SEM images were taken in the TESCAN.Raman spectra were applied to test the composition of the materials with a laser wavelength of 442 nm.Solid state NMR was run on a Bruker 850 MHz spectrometer equipped with a 3.2 mm MASEfree probe.All HC-CP spectra were recorded at a spinning frequency of 11 kHz and a temperature of 287 K.The proton to carbon CP transfer time of 2 ms and interscan delay (d1) of 4 s were kept constant between different measurements.During CP, a ramped pulse (100 to 90 percent) was applied to the proton channel.A total of 16 384 scans were acquired for each sample.
Electrochemical Measurement: Time-dependent open-circuit potentials, polarization curves, CV, EIS, linear scan voltammetry, and chronoamperograms were measured by an electrochemical workstation (MULTIAU-TOLAB/M101).The impedance measurement was carried out at the open circuit potential.Galvanostatic charge/discharge profiles were taken from the Biologic battery cycler (BCS-805).
Simulation Method: All simulations were carried out in the Vienna ab initio simulation package (VASP) based on the DFT scheme. [32]The exchange and correlation functional adopted the Perdew-Burke-Ernzerhof form under the generalized gradient approximation and the van der Waals correction adopted the zero damping DFT-D3 method of Grimme. [33]A high energy cutoff of 600 eV was performed.For the AIMD simulation, the Nosé thermostat [34] was used at the room temperature of 300 K. VASPKIT 1.4.0 was used for data post-processing. [35]

Figure 2 .
Figure 2. Adsorption of caffeine on Zn surface.a) Charge density distribution of caffeine on (002) and (100) face of metallic Zn. b) Gibbs free energy of caffeine adsorption derived from Tafel plots, CV curves, and EIS spectra.c) Tafel plots and d) exchange current density of the Zn anode in caffeinecontaining and caffeine-free electrolytes.e) Schematic of electrode-electrolyte interface with and without caffeine.

Figure 3 .
Figure 3. Zn reversibility.a) Schematic coin cell assembly for the measurements.b) Coulombic efficiency derived from plating/striping test at 2 mA cm −2 and 2 mAh cm −2 .The cut-off voltage of stripping is 1.5 V. Galvanostatic plating/stripping cycles at c) 2 mA cm −2 and 2 mAh cm −2 and d) 10 mA cm −2 and 10 mAh cm −2 (45% DOD), and e) rate performance of the Zn||Zn coin cell.f) Schematic interdigitated microcell.g) Galvanostatic plating/stripping cycles at 50% DOD and h) rate performance of the Zn||Zn interdigitated cell.

Figure 4 .
Figure 4. Zn 2+ -conducting SEI.a) SEM image of cycled Zn foil in the caffeine-containing electrolyte.Scale bar: 2 μm.b) 1 H NMR spectra of fresh and cycled caffeine.c) Simulated pathways of Zn 2+ ion in Zn-caffeine complex.d-f) XPS spectra of Zn anode cycled in the ZnSO 4 electrolyte with caffeine for 20 times at 10 mA cm −2 /10 mAh cm −2 The Zn anodes were etched by Ar + sputtering for 20 and 150 nm to provide the chemical information at different depth.g) Schematic of the gradient organic-inorganic SEI.

Figure 5 .
Figure 5. Full cell performances.a) Initial galvanostatic charge/discharge curves of Zn||MnO 2 coin cells in the electrolyte with and without caffeine.b) Rate performance and c) galvanostatic cycles at current density of 2 C in the coin cells with different electrolytes.Self-discharge of Zn||MnO 2 interdigitated microbatteries in the electrolyte d) without caffeine and e) with 20 mm caffeine.f) Galvanostatic charge/discharge cycles of interdigitated microbatteries at a current of 50 μA.

Figure 6 .
Figure 6.Multilayer fabrication toward self-assembled Swiss-roll microbattery.a) Optical image showing the parallel fabrication of planar devices with the photolithographable electrolyte, scale bar: 200 μm.b) Schematic of the multilayer stack to form the micro-Swiss-roll.c) Optical image of a micro-Swiss roll, scale bar: 100 μm.d) Optical image of a micro-Swiss-roll array on a chip, scale bar: 1 mm.e) Galvanostatic charge/discharge curve of the micro-Swiss-roll battery.f) Cycling performance of micro-Swiss-roll battery using the photolithographable electrolyte with and without caffeine.