Introduction

Zinc has a rich history as an anode for static,1, 2 redox,3, 4 semi-solid5, 6 and flow assisted cells,7 and remains attractive for secondary applications despite the current popularity of lithium ion batteries due to its abundance, safety and scalability. Historically, it has been limited by morphological uncertainty, which is dependent on local current density and pH.8 Two typically undesirable regimes are ramification or the formation of dendrites that lead to electrical-short-circuits9 and precipitation of ZnO10, 11 that can irreversibly passivate a surface. In conventional static zinc systems (composites, pastes and powders), these mechanisms lead to terminal failure, and in flow and flow-assisted cells they can be reversed only at the cost of additional system overhead.

In this work, we present a method of synthesizing zinc foam electrochemically, in situ. The formation of zinc foam in situ may enable new possibilities of designing structural electrodes because of the freedom of shape and the advantage of replacing depleted or contaminated electrolyte for long-life usage. The zinc foam proposed in this paper is formed by a three-dimensional network of dendrites that is electrochemically active and electronically conductive, at the nanoscale. We find that the nanoporous zinc structure presents a modified activation energy for electrochemical cycling. Conductive paths throughout the internal structure result in a more uniform current distribution, reducing non-uniform concentration gradients and anode polarization losses.12 We exploit the dendrites, which are normally a limitation against rechargeable zinc electrodes, by initially operating at states far from equilibrium (cathodic overpotentials from ~200 to ~600 mV) and thereafter cycling toward equilibrium. This presents a rethinking of the process of recharging batteries. The dendritic network, upon recharging a battery at standard operating conditions, may densify the architecture over cycles and suppress initiation of dendrites if the three-dimensional core is kept intact. Recent work has indicated that zinc structures with submicron features can undergo 50 charge–discharge cycles without failure.13 Here, we describe a simple mechanism for making such structures, with the option to alter the surface kinetics as a function of process regime.

We investigated the electrodeposition of zinc with hydrogen at different overpotentials by characterizing its electrochemical and morphological properties. We studied the kinetics of the produced material via electrochemical techniques such as chronocoulometry and Butler-Volmer modeling. We characterized the morphology of zinc by scanning and transmission electron microscopy methods. We determined the particle size distribution and specific surface area of zinc foam through X-ray diffraction, differential scanning calorimetry and BET (Brunauer-Emmet-Teller) techniques. Finally, we investigated the reversibility of Zn0/Zn2+ redox chemistry and discharge-rate capability of the material for battery applications. The hyper-dendritic zinc electrode showed significant capacity retention over 100 cycles, and due to the electrochemical synthesis procedure, it can be reformed in situ if needed.

Materials and methods

Experiments carried out in alkaline conditions where 8.9 M KOH pellets (Sigma-Aldrich, St Louis, MO, USA) and 0.61 M ZnO (Sigma-Aldrich) were dissolved in deionized (DI) water and stirred overnight. The zinc foam was prepared electrochemically from flooded undivided cells containing 10 ml of electrolyte in a three-electrode setup that was controlled by a potentiostat (Keithley Model 2401 Low Voltage SourceMeter Instrument, Keithley Instruments Inc., Cleveland, OH, USA). The setup consisted of a 0.5-mm diameter platinum wire (Sigma-Aldrich) working electrode, an Hg/HgO reference electrode (Koslow Scientific Company, Englewood, NJ, USA) and a nickel mesh (Dexmet Corporation, Wallingford, CT, USA) that served as a counter electrode. Electrodeposited zinc on a platinum wire was controlled with a defined charge passed of 100 coulombs (C), using a controlled-potential coulometry method and a surface area of 0.260 cm2. For studies of the rechargeability of the zinc foam, the working electrode was prepared with a charge passed of 3000 C, and a conventional nickel electrode with excess capacity served as the counter electrode. The platinum electrode was initially activated and cleaned by dipping it in nitric acid (Sigma-Aldrich) for 1 h at 60 °C, and then rinsed with DI water. The nickel electrode was prepared by mixing a slurry consisting of 50 wt.% NiOOH (Sigma-Aldrich), 40 wt.% graphite (Timcal Timrex KS6, Imerys Graphite & Carbon, Bodio, Switzerland) and 10 wt.% PTFE (Sigma-Aldrich), and the resulting paste onto a flat surface using a rolling pin. The rolled electrode was then dried in a vacuum oven at 125 °C for 2 h. The assembly was compressed in a uniaxial press using 4 tons of force on a nickel mesh current collector. The full cell was sandwiched between two layers of non-woven cellulose membrane (Freudenberg), and topped up with either 10 ml electrolyte for flooded cells or 1 ml electrolyte for non-flooded cells. The uncompensated resistance, Ru, was evaluated for each setup in this work through a current interrupt method. The Ru values obtained were under 4 Ω for all methods and hence neglected, due to the absence of any significant effect on the measured potential. For in situ studies, the zinc was kept in solution. For ex situ studies, the sample was rinsed three times in 8.9 M KOH to ensure the dissolution of ZnO, neutralized with DI water five times, and then dried under vacuum for 1 h at 110 °C. The samples were prepared by grinding with mortar, pestle and small amounts of acetone until a fine powder was obtained. Purum bulk zinc powder (Sigma-Aldrich) was used as a standard comparison. The Tafel studies were carried out using a 0.5-mm thick zinc foil with a surface area of 0.260 cm2 (Sigma-Aldrich) cleaned in 10 wt.% sulfuric acid (Sigma-Aldrich) and rinsed with DI water.

Scanning electron microscopy (SEM) images and elemental composition quantifications were assessed by an FEI Quanta 200 FEG Environmental SEM (FEI, Hillsboro, OR, USA) connected with an integrated electron diffraction spectroscopy system. High-resolution transmission electron microscopy (TEM) images were taken by a Philips CM200 FEG-TEM operating at an accelerating voltage of 200 kV. The samples were attached onto a lacey carbon support film on a gold 200 mesh grid (Electron Microscopy Sciences, Hatfield, PA, USA). An OptixCam OCS-1.3 optical microscope (The Microscopic Store, LLC, Roanoke, VA, USA) was used to image the zinc foam growth on a nickel mesh current collector. BET surface area measurements were carried out using a Micromeritics ASAP 2010 apparatus (Micromeritics Instrument Corporation, Norcross, GA, USA) using nitrogen as an adsorbent. The samples were prepared by heating overnight at 120 °C while flowing N2 gas. The N2 adsorption was measured from a relative pressure range of 0.01–1 at 77.3 K. Differential scanning calorimetry (DSC) measurements were performed using a Perkin-Elmer Pyris 1 DSC (PerkinElmer Inc., Waltham, MA, USA) with a process scheme of rapid isothermal treatment to 340 °C in 5 min and then heated at rate of 1 °C/min up to 440 °C at ambient pressure in argon atmosphere. High resolution X-ray diffraction patterns were taken at room temperature with a Bruker D8 Discover X-ray Diffractometer (Bruker Corporation, Billerica, MA, USA) and in 0.2° intervals with 1 s settling time using a Cu-Kα source operating at 40 kV and 40 mA.

Results and discussion

Zinc metal is soluble in strong alkaline solutions and equilibrates with zincate ions, Zn(OH)42−, at pH values >12. The charge and discharge reaction of zinc electrodes, the precipitation of ZnO at supersaturated concentration of zincate ions and hydrogen evolution are shown in equations (1), (2) and (3), respectively:14, 15

Zinc has an equilibrium potential that is about 0.4 V lower than the equilibrium potential of hydrogen evolution at pH 14 thus having a direct influence on the efficiency of electrodeposited zinc.

Electrodeposited zinc foam on platinum at 100 C charge passed and different potentials is included with the Supplementary Information. The metallic foam structure was achieved at potentials applied above −1.6 V and a substantial amount of hydrogen evolved with zinc foam was observed at −2.0 V. Figure S1 in the Supplementary Information shows that the measured current increased more rapidly at higher applied potentials, despite being well beyond the limiting current for zinc in this electrolyte. The increase in current over time was most likely due to increased surface: while reaction (3) is competitive with reaction (1), we found in all described experiments coulombic efficiencies of 84.9±3.1%, indicating a similar fraction of current formed H2(g) at each test.

Electrodeposited zinc has been reported as mossy, epitaxial layers, boulder-like and dendritic morphologies at low to high overpotentials.7, 9, 16, 17, 18 The morphology of electrodeposited zinc is dependent on the intermediate product zincate. Precipitation of zincate at the surface interface generally initiates dendritic morphologies at limiting current densities due to non-uniform concentration gradients. Effort on controlling mass transport by convection was carried out by Ito et al.7 where they quantified the change in morphology of deposited zinc. In this work, we mainly operated at limiting current densities with no convection control at high overpotentials. Hexagonal crystals organized into fernlike dendrites with particles sized between 2 and 10 μm are shown in Figures 1a–d. At high cathodic overpotentials (>~500 mV, assuming an equilibrium potential of zinc at −1.38 V), oriented zinc pillars were formed at nanoscale, seen in Figures 1h and j, with a mean particle size of 73.05±8.95 nm measured from 49 randomly selected zinc particles. At higher overpotentials, transport rate of zincate ions, Zn(OH)42− will be limited at the surface-liquid interface, causing unpredictable growth of primary and secondary dendrites.

Figure 1
figure 1figure 1

SEM images of electrodeposited zinc in alkaline solution at (a and b) −1.6 V, (c and d) −1.7 V, (e and f) −1.8 V, (g and h) −1.9 V and (i and j) −2.0 V vs Hg/HgO.

Full size image

The elemental composition of the samples in Figure 1 was analyzed with EDS. A small amount of the oxide was formed: 3.2, 3.8, 5.9, 6.5 and 8.2 wt.%, in increasingly negative potentials, respectively. The amount of potassium was below 0.6 wt.% for all measurements thus negligible. This supports our expectation that the majority of the deposited foam was metallic zinc. The hyper-dendritic nanoporous zinc foam prepared at −2.0 V was further investigated under a TEM. The TEM image taken of hyper-dendritic nanoporous zinc foam is seen in Figure 2a. A schematic representation of our interpretation of zinc dendrite formation, based on the diffusion-limited aggregation model19, 20, 21 is seen in Figure 2b. Our hypothesis is that branch growth on secondary dendrites at the edge of the primary dendritic spines led to the formation of nanoporous zinc. The growth of these branches may be the cause of three-dimensional threadlike networks consisting of zinc particles. López and Choi12 presented a similar result of stabilized dendrite growth forming motif crystals that were interconnected in a matrix with dendritic backbones. A mean particle size of 72.15±18.05 nm was assessed from the TEM image based on 72 randomly selected zinc nanoparticles. Typical ‘large’ zinc hexagonal crystals were no longer evident due to the initiation of numerous branches resulting in spherical boundaries, but rather we see small defective zinc hexagonal aggregates.

Figure 2
figure 2

(a) TEM image taken at nanoscale of electrochemical deposited hyper-dendritic porous zinc foam formed at −2.0 V vs Hg/HgO. (b) Scheme of zinc dendrite formation based on the diffusion-limited aggregation model.

Full size image

Of particular note is the way in which the zinc dendritic foam grows. The zinc foam grows in an isotropic manner at the micron scale as shown in the time-lapse movie of zinc foam growth on nickel mesh in Figure 3. At first this is counterintuitive as one expects an ‘electrochemical lighting rod’ under these conditions, but upon inspection these fast directional growths are occurring, just at the nanometer scale. In concert, the exploitation of the small directional growth from edge facets seems to result in a micron scale morphology of randomly oriented particles. This growth then continues, apparently creating a conformal ‘hyper-dendritic’ foam.

Figure 3
figure 3

Time-lapse microscopy movie of hyper-dendritic zinc foam growth on a nickel mesh current collector at −2.0 V vs Hg/HgO in a chronological order, from top left to bottom right.

Full size image

The kinetics of electrodeposited zinc at varying potentials were evaluated in operando, in comparison with a standard bulk zinc sample, by first depositing zinc on platinum at a constant potential with a charge passed of 100 C to ensure a thick layer of deposited metal for all samples. The samples were kept in solution at an open-circuit voltage for 1 h to relax the surface. Afterwards, polarization curves were swept linearly from −80 to 80 mV in overpotentials at a sweeping rate of 1 mV s−1. The kinetic parameters of zinc were calculated through the Butler-Volmer equation at electron-transferred limited regions. The Butler-Volmer relationship is seen in (4) where I0 is the exchange current, η the activation overpotential, αc the cathodic charge transfer coefficient. Figure 4 shows the measured Tafel plots and Butler-Volmer fits at different potentials. Calculated kinetic parameters are given for the different potentials in Table 1.

Figure 4
figure 4

Tafel plot and Butler-Volmer fit of electrodeposited zinc at different potentials and standard bulk zinc. The lines show the experimental data and the markers show the calculated fit.

Full size image
Table 1 Measured kinetic parameters for electrodeposited zinc from Butler-Volmer fit to Tafel slope at different potentials
Full size table

The literature typically concerns kinetic parameters of electrodeposited zinc at low cathodic overpotentials (30–100 mV) in alkaline solutions.22, 23, 24, 25, 26, 27, 28, 29 Cathodic Tafel slope values of ~120 mV dec−1 and cathodic charge transfer coefficient of ~0.55 are suggested. The kinetic results evaluated from the Butler-Volmer fit at potentials of electrodeposited zinc at −1.6 V and the standard bulk zinc is in agreement with the literature.22, 23, 24, 25, 26, 27, 28, 29 At applied potentials above −1.6 V, more rapid kinetics were evident: we observed an increased exchange current of 36.8±4.3 mA at −2.0 V in comparison with 12.9±0.5 mA for the standard. In addition to the evidence of more rapid kinetics, the anodic and cathodic Tafel slopes were calculated at 212.9±5.1 and 190.1±4.7 mV dec−1 at −2.0 V in comparison with 125.6±7.5 and 107.3±1.5 mV dec−1 for the standard. The kinetic transition led to a lower activation energy at potentials of −1.8 to −2.0 V and may have occurred because the nucleation of zinc particles on the edge (branch formation) was more probable than the continuous growth of existing primary dendrites. The anodic reaction, the oxidation of Zn0 to Zn(OH)42−, was also dominant. The anodic and cathodic charge transfer coefficients of 0.69±0.06 and 0.31±0.06 at −2.0 V, compared with 0.45±0.01 and 0.55±0.01 for the standard, indicated the oxidation reaction being dominant. A lower equilibrium potential was measured for the samples prepared above the potentials of −1.8 V (E0=−1.38 V) in comparison with the ones prepared under potentials of −1.7 V (E0=−1.395 V) and the bulk sample (E0=−1.41 V). Osório et al.30 presented the results of corrosion rates as a function of secondary arm spacing for zinc alloys. They concluded that a shorter secondary dendrite arm spacing resulted in an increased amount of hydrogen evolved at equilibrium states for zinc alloys. A more rapid corrosion rate with shorter dendrite arm spacing achieved at increasingly negative potentials was evident for the anode material in this work as well.

The N2 adsorption–desorption isotherms of electrodeposited hyper-dendritic nanoporous zinc foam at −2.0 V were measured. A specific surface area of 12.16±0.62 m2 g−1 was obtained from the pressure range of 0.05<P/P0<0.3. In contrast to bulk zinc, which has a specific surface area of 0.5–1 m2 g−1, the specific surface area of zinc foam was 10–25 times larger. Assuming the dendrites are composed of solid, spherical particles with smooth surface in narrow size distribution and a theoretical density of 7.14 g cm−3, a mean particle size diameter of 69.10 nm was calculated from (5) where ρ is the theoretical density in g/cm3 and SBET is the specific surface area in m2 g−1. This is in agreement with what was observed in the TEM.

DSC profiles of prepared hyper-dendritic zinc foam and conventional bulk zinc are seen in Figure 5, presenting a melting-point depression of the zinc foam because of reduction in particle size. The thermograms showed a dip in heat flow, which indicated an endothermic phase transition. A decrease in heat of fusion of the zinc foam, 6.56 kJ mol−1, in contrast to the bulk zinc, 7.26 kJ mol−1, was achieved. A melting point for the zinc foam was seen at 417.42 °C and for the bulk zinc at 420.24 °C. A mean particle diameter size of 85.30 nm was calculated based on Gibbs-Thomson thermodynamic model of melting point depression for nanoparticles seen in:

Figure 5
figure 5

Heat flow vs temperature from DSC measurements of nanoporous zinc foam produced from −2.0 V vs Hg/HgO (solid line) in comparison with bulk zinc (dashed line). The onset indicated the melting point of both metals.

Full size image

where γsl is the solid–liquid interfacial energy in J m−2, Hf is the bulk heat of fusion in J g−1, ρs is the density of the solid in g m−3, TMB is the bulk melting temperature in K and TM(d) is the nanoparticle melting temperature in K. All physical constants were taken from the literature and the melting points were measured from the onset of the endothermic phase transition.31, 32

X-ray diffraction patterns of prepared zinc foam at −2.0 V and bulk zinc are presented in Figure 6. The patterns showed highly crystalline zinc with well-defined peaks for zinc corresponding to the lattice planes {002}, {100}, {101} and {012}. No corresponding peaks of ZnO or Zn(OH)2 were found, indicating that pure hexagonal zinc crystals were formed. An increased relative intensity of the peak at 36.29° for the zinc foam (B) compared with the bulk zinc (A) indicates that zinc was preferentially deposited in highly oriented {002} direction. X-ray peak broadening analysis was carried out through the Scherrer equation, seen in (7), which estimates the mean crystallite size from the full width at half maximum of a peak:

Figure 6
figure 6

X-ray diffraction patterns (intensity vs 2θ) of zinc foam and bulk zinc.

Full size image

where d is the mean crystallite size in nm, K is a dimensionless shape factor (0.94), λ is the diffraction wavelength in nm, β is the full width at half maximum in radians and θ is the Bragg angle in radians. Assuming Gaussian peak profiles and spherical crystals with cubic symmetry, the mean crystallite sizes were calculated to be 35.28, 36.90, 37.10 and 38.61 nm for the peaks of 36.29°, 38.93°, 43.16° and 54.28°, respectively.

Discharge profiles as a function of rate of prepared zinc foam on platinum at −2.0 V and at ambient temperature with charge passed of 3000 C is seen in Figure 7. Flat discharge-profile plateaus at −1.375 to −1.1 V and specific capacities of 719.2±12.1 mAh g−1 corresponding to a consistent coulombic efficiency of 87.7±1.5% at discharge rates from 2C to C/10 were measured. The discharge profiles of zinc foam showed no significant decrease in discharge capacity at 100% depth-of-discharge (DOD) and at high rates. The decrease in the voltage plateau at high discharge rates was directly correlated to higher ohmic resistance. The high utilization of the hyper-dendritic nanoporous zinc foam showed great rate capability and was comparable to results of conventional zinc anodes in a sponge form-factor.13, 33

Figure 7
figure 7

Discharge capacity profiles as a function of rate of the hyper-dendritic nanoporous zinc foam initially prepared in situ with charge passed of 3000 C. The discharge rates used are 2C, C, C/2 and C/10.

Full size image

Conventional zinc anodes usually consist of active material, additives to suppress hydrogen and control the morphological uniformity of zinc deposition, and polymeric binders to keep zinc particles intact.1, 23, 27, 34 In this work, the zinc foam was prepared, in situ, by reducing zincate ions on a current collector with no other necessary compounds. The structural foam provided mechanical strength with an interconnected zinc current pathway throughout the core when cycled. For proof-of-concept of the hyper-dendritic zinc foam reversibility in both flooded and non-flooded systems, the zinc foam was charge–discharge cycled against a conventional nickel electrode in a full cell. At the working electrode, the reversible reaction seen in (1) took place upon charging and discharging. The active material was initially prepared by electrodeposition of zinc on nickel mesh at −2.0 V for the zinc foam and −1.4 V for the bulk or ‘equilibrium’ zinc, both with a charge passed of 3000 C controlled through chronocoulometry. The three-dimensional dendritic structure of nanoporous zinc on nickel mesh was confirmed under a SEM, seen in Figure 8d, and Wang et al.16 results support that zinc deposits on face-centered cubic nickel substrates with no structural misfits. The capacity of the nickel counter electrode was roughly five times higher than the working electrode to ensure no capacity limitations. The prototype cells were cycled at rate of C/5 corresponding to a current of 166.67 mA, and to 40% DOD which is equivalent to a theoretical specific capacity of 328 mAh/g, to ensure that the core dendritic structure was intact. A 15-min rest between each charge and discharge step and a voltage cutoff during discharge at −1.35 V vs Hg/HgO or 1.6 V vs Ni(OH)2/NiOOH was used. Figure 8a shows the summarized cycling data of the specified electrodeposited zinc up to 100 cycles, and Figures 8b and c present the charge-discharging curves for the flooded and non-flooded hyper-dendritic zinc cells. The flooded cell was less constrained than the non-flooded cells, therefore the solubility of zincate never reached a saturation point. Ex-situ SEM images taken at the cycle number points of 0, 40 and 100 from the non-flooded cell are seen in Figures 8d–f, respectively. The dendritic zinc foam was cycling toward equilibrium and the electrodes after 40 cycles exhibited no dendrites at macroscale. Instead, a more densified texture was evident where primary or secondary dendrites were no longer evident but rather the well-reported ‘ramified’ morphology. Additionally, upon cycling no precipitation of visible ZnO was formed. The hypothesis of a more dense structure over cycles of the hyper-dendritic foam was supported both by the microscopic evidence of morphology change presented, and by the shift in the open circuit potential measured at the start of the charging. Similar to the trend in Figure 4, the open circuit potential was lower for the foam architecture, thus the charge–discharge trend over cycles indicated a densification of the foam to bulk structures. This indicates that if a cell starts with metallic zinc that was formed further from equilibrium than the average cycling protocol demands, the zinc will restructure to form a more dense morphology, filling ‘inwards’, before creating ramified structures, typically misstated as dendrites. The discharge voltage profile for the flooded cell showed promising stability with less than ~1% potential fade per cycle, a specific discharge capacity of 303.1±24.9 mAh g−1, and low resistive losses across the cell over time (0.4–0.8 Ω). At the 100th cycle, a specific discharge capacity of 282.3 mAh g−1 was achieved corresponding to 86.1% of the initial capacity. For the non-flooded cells, the hyper-dendritic zinc was compared with equilibrium processed zinc (−1.4 V vs Hg/HgO) (Supplementary Figure S2), and presented superior capacity retention over 100 cycles.

Figure 8
figure 8

(a) Cell summary of flooded and non-flooded zinc electrodes cycled against conventional nickel electrodes over 100 cycles. (b) Galvanostatic charge–discharge cycling profiles of hyper-dendritic nanoporous zinc foam in a flooded cell at a rate of C/5 to a capacity limit of 40% DOD. (c) Galvanostatic charge–discharge cycling profiles of hyper-dendritic nanoporous zinc foam in a non-flooded cell at a rate of C/5 to a capacity limit of 40% DOD. SEM images taken from the non-flooded zinc cell at 0 (d), 40 (e) and 100 cycles (f).

Full size image

Conclusion

We have introduced a method of synthesizing inexpensive three-dimensional zinc anodes in the form-factor of hyper-dendritic nanoporous foam, in situ. The zinc foam exhibits stable primary anode performance at high coulombic efficiency (87.7±1.5%) and secondary anode performance beyond 100 cycles at 40% DOD. The superior performance is attributed to the three-dimensional dendritic matrix at nanoscale formed by branch growth on secondary dendrites. The branch formation resulted in zinc nanoparticles with a size distribution of 54.1–96.0 nm. The achieved nanoparticles contributed to: more rapid kinetics in comparison with conventional bulk zinc of the redox chemistry, 10–25 times larger specific surface area than bulk zinc, a 2.8 °C decrease in melting point in comparison with bulk zinc and highly oriented zinc deposition. Most interesting is that operating mesoscale isotropic three-dimensional network of dendrites far from equilibrium seems to densify the structure at standard battery operating conditions, as evidenced by the morphological change, and the shift in open circuit potential from hyper-dendritic zinc foam as a function of cycle number. This is in contrast to most literature work with zinc which starts flat or packed and seeks to emulate the ‘equilibrium condition’. If hyper-dendritic structures can be recreated locally in operando, then a battery with a zinc anode capable of indefinite cycle life may be possible. Perhaps, if dendritic behavior can be maximized and controlled locally, then it may deter system limiting short circuits globally, as the system will move toward more dense structures as opposed to more branched structures.