Design and structures of hosts for Se
Figure 2a depicts the design rationale of the Se host with abundant continuous porosity and electrochemical catalytic functionality of double polar sites. The intrinsically porous structure of Prussian blue analogues enables a facile guests’ ions infiltration and ion transportation, while the highly tunable transition metal species pledges the incorporation of electroactive sites for effective battery chemistries. Such a structure can significantly promote the redox kinetics and greatly mitigate the shuttle effects under high Se loading, benefiting from the improved electrocatalytic and enhanced confinement effect. In comparison, the binding strength and conversion efficiency of frequently used porous carbon host towards pristine Se and reactive Sen are unsatisfactory.
The typical construction of Se-in-(Cu/Ni/Co)[Co(CN)6] is obtained by extracting primary K ions from K(Cu/Ni/Co)[Co(CN)6] frameworks and impregnating Se molecules into their pores. It should be mentioned that the (Cu/Ni/Co)[Co(CN)6] exhibits good thermostability. Its structure is completely maintained even after heating treatment at 300 oC under N2 atmosphere (Figure S1). The Se molecules are successfully locked in (Cu/Ni/Co)[Co(CN)6] pores in a quantitative manner via melt diffusion at 260 oC, as reflected in the drastic decrease in N2 uptakes of these Se-in-(Cu/Ni/Co)[Co(CN)6] samples in contrast to that of the pristine (Cu/Ni/Co)[Co(CN)6] (Figure S2). Meanwhile, the crystallinities of these (Cu/Ni/Co)[Co(CN)6] are well-maintained as evidenced by the sharp peaks and systematic variation in the peak intensity of (Cu/Ni/Co)[Co(CN)6] confirms the confinement of Se species within (Cu/Ni/Co)[Co(CN)6] pores (Figure 2b). The pinpoint management of Se content is conducted by veering from the feeding ratio, and the concrete loads is determined by thermalgravimetric analysis (TGA). In consideration of (Cu/Ni/Co)[Co(CN)6] structure with high porosity and strong chemisorption characteristic to guest ions (Table S1), the content of Se in Cu[Co(CN)6] framework can reach to 71.3% by weight, higher than that of Ni[Co(CN)6] (42.1%), Co[Co(CN)6] (38.8%) frameworks and porous carbon (25.7%) (Figure 2c, Figure S3). The high Se loads with Cu[Co(CN)6] framework can be attributed to that the synergetic effect of metallic Cu and Co species together with its effectively porous structure. The morphologies of (Cu/Ni/Co)[Co(CN)6] are approximate cubic with a size of ~500 nm, as revealed by scanning electron microscope (SEM) (Figure 2d-f). Encouragingly, the original morphology of (Cu/Ni/Co)[Co(CN)6] frameworks is well preserved even under these high Se loading levels, without any observable Se species on the crystal surface, indicating the consummate confinement of Se molecules into (Cu/Ni/Co)[Co(CN)6] matrixes (Figure 2g-i).
Electrochemical kinetics and process of the electrocatalytic Se redox reaction
To experimentally in-depth study the fundamental electrocatalytic behavior of the hosts on Sen oxidation, the linear sweep voltammetry (LSV) of (Cu/Ni/Co)[Co(CN)6] hosts and referenced porous carbon is shown in Figure 3a. Among them, Cu[Co(CN)6] gives the highest current density response, validating its enhanced redox kinetics of Sen conversion. The Tafel slope (η) determined from LSV curves is utilized to characterize the reaction kinetics and catalytic activity of Sen electrocatalyst. The Tafel curves witness the smallest η of Cu[Co(CN)6] (188.06 mV·dec−1) compared with Ni[Co(CN)6] (459.25 mV·dec−1), Co[Co(CN)6] (598.42 mV·dec−1), and porous carbon (629.82 mV·dec−1), indicating the best kinetics of Cu[Co(CN)6] for Sen oxidation process. From a computational perspective, the ab initio molecular dynamics (AIMD) simulation on (Cu/Ni/Co)[Co(CN)6] surface with one adsorbed ZnSe are performed to visualize the Zn-Se decomposition process. Compared with Co[Co(CN)6], the radial distribution function (RDF) curves of Zn-Se on Cu[Co(CN)6] and Ni[Co(CN)6] exhibit much wider Zn-Se bond length distributions ranging from 2.2 to 3.2 Å (Figure 3b), while Cu[Co(CN)6] is more capable of elongate Zn-Se bond than Ni[Co(CN)6] with larger average Zn-Se bond length, validating the highest tendency of Zn-Se bond breakage on Cu[Co(CN)6]. Furthermore, the decomposition energy and barrier are calculated employing the climbing-image nudged elastic band method to examine the dezincation reaction kinetics of ZnSe in (Cu/Ni/Co)[Co(CN)6] frameworks. The calculated energy barriers of ZnSe transformation on Cu[Co(CN)6] framework (1.83 eV) is smaller than that of Ni[Co(CN)6] (2.81 eV) and Co[Co(CN)6] (3.43 eV) frameworks (Figure 3c). We also plot the partial density of states (PDOS) curves before and after ZnSe adsorption in Figure S4. Hybridization of transition (Cu/Co/Ni) 3d and Se 4p orbitals with the energy ranging from -1.0 to 0.5 eV is observed, leading to changes in DOS around the Fermi level after ZnSe adsorption. For Se 4p orbital, the tendency of its downshift can be concluded for Cu[Co(CN)6], manifesting a high level of ZnSe activation. The above results suggest that Cu and Co metal species serve as active sites to enhance the phase transformation of ZnSe to Sen facilitate the whole reaction in Zn||Se batteries.
To further understand the origin of the improved catalytic activity and kinetics, the LSV and electrochemical impedance spectra (EIS) measurements are conducted following the protocols well developed by oxygen reduction reaction (ORR) community22. Before the LSV tests, the (Cu/Ni/Co)[Co(CN)6]-based electrodes are activated by cyclic voltammetry (CV) for 50 cycles at scan rate of 50 mV⋅s−1 in the non-Faradaic range to reach a stable electrochemical active surface area. Figure 3d exhibits the Se reduction polarization curves of (Cu/Ni/Co)[Co(CN)6] electrocatalysts and referred porous carbon deposited on a glassy carbon electrode. The Se reduction LSV curves show similar features to those of the ORR, including an onset potential, diffusion-limited current density (Jd) and half-wave potential (E1/2) to evaluate electrocatalytic effects. The E1/2 and Jd for the Cu[Co(CN)6] is 1.11 V and 5.53 mA⋅cm−2, respectively, which is considerably higher than those of Ni[Co(CN)6] (1.05 V, 4.35 mA⋅cm−2), Co[Co(CN)6] (1.04 V, 3.63 mA⋅cm−2) and porous carbon (0.96 V, 2.12 mA⋅cm−2), revealing a considerable lower overpotential and larger reduction current for the Cu[Co(CN)6] (Figure 3e). The reaction kinetics and catalytic activity of (Cu/Ni/Co)[Co(CN)6] hosts and compared carbon are characterized by η determined from the LSV curves. Notably, the Cu[Co(CN)6] electrocatalyst shows the smallest η of 95.23 mV⋅dec−1, compared with 112.19, 115.76 and 184.21 mV⋅dec−1 for Ni[Co(CN)6], Co[Co(CN)6] and porous carbon, respectively, indicating considerable accelerated reaction kinetics and higher electrocatalytic activity of Cu[Co(CN)6] framework (Figure 3f).
To further probe the origin of the enhanced electrocatalytic activity and kinetics of Cu[Co(CN)6]-catalyzed Se reduction, the EIS measurements are carried out at the onset potential to examine the charge transfer resistance. Charge transfer is a necessary step, in which ions and electrons are transferred to the active centers to participate in the electrochemical reactions. The charge transfer kinetics at the electrocatalyst-adsorbate interface therefore represents the primary factor determining the electrocatalytic Se reduction kinetics. The EIS curves show that the Cu[Co(CN)6] electrocatalysts shows the smallest charge transfer resistance (28.9 Ω⋅cm−2) during the Se reduction in comparison with those of Ni[Co(CN)6] (36.9 Ω⋅cm−2), Co[Co(CN)6] (38.7 Ω⋅cm−2) and porous carbon (63.4 Ω⋅cm−2), indicating its superior charge transfer kinetics (Figure 3g). We further extend these EIS measurements and determine the temperature dependence of charge transfer resistance at the onset potential. The Ea was extracted by using the Arrhenius equation. The logarithmic values of the reciprocal of the charge transfer resistance keeps to a linear relationship with the inverse of the absolute temperature following the Arrhenius relation, we determine Ea to be 1.63 kJ⋅mol−1 for Cu[Co(CN)6], 2.85 kJ⋅mol−1 for Ni[Co(CN)6], 2.97 kJ⋅mol−1 for Co[Co(CN)6] and 4.23 kJ⋅mol−1 for porous carbon (Figure 3h, i), respectively. The lowest Ea corresponds to the fastest kinetics of Cu[Co(CN)6] electrocatalyst for Se reduction.
Coin-cell performances of the cathode materials
As revealed by the above characterizations, Se have been successfully confined in (Cu/Ni/Co)[Co(CN)6] frameworks and they can electro-catalyze the Se redox reaction. The propelled Se/Se2- conversion kinetics of three different Se-in-(Cu/Ni/Co)[Co(CN)6] cathodes and Se-in-porous carbon for Zn batteries are investigated by CV and galvanostatic charge/discharge (GCD) tests with an aqueous gel polymer electrolyte. Compared with aqueous electrolyte of 4 M Zn(OTf)2, the aqueous gel polymer electrolyte of 4 M Zn(OTf)2/poly(ethylene oxide) (PEO) extends the voltage stability window of the aqueous electrolyte to 2.64 V, which can completely cover the operating voltage range of 0.1-2.1 V for the Zn||Se batteries (Figure S5). Meanwhile, the aqueous gel polymer electrolyte maintains high ionic conductivity of 0.12 S×cm-1 (Figure S6). The CV curves of batteries with different cathodes shown in Figure 4a delivers two pairs of redox peaks, which are ascribed to the Se4+/Se0 and Se0/Se2- redox reactions. The Se-in-Cu[Co(CN)6] cathode exhibits the lowest cathodic and the highest anodic peak potential of 1.12 V and 1.82 V, respectively, compared with Se-in-Ni[Co(CN)6] (1.9 V, 1.0 V), Se-in-Co[Co(CN)6] (1.89 V, 1.01 V) and Se-in-porous carbon (1.93 V, 0.97 V). Meanwhile, the discharged capacity for Se-in-Cu[Co(CN)6] cathode is 664.7 mAh×g-1 at current density of 0.2 A×g-1, significantly higher than that of Se-in-Ni[Co(CN)6] (599.6 mAh×g-1), Se-in-Co[Co(CN)6] (528.9 mAh×g-1) and Se-in-porous carbon (409.4 mAh×g-1) (Figure 4b). On the other hand, benefiting from effectively electrocatalytic effects of Cu[Co(CN)6] framework, the Se-in-Cu[Co(CN)6] cathode exhibits the lowest voltage hysteresis of 0.54 V, compared that of Se-in-Ni[Co(CN)6] (0.69), Se-in-Cu[Co(CN)6] (0.70 V) and Se-in-porous carbon (1.07 V). The above results reveal the highest electrocatalytic capability and the fastest redox kinetics of Se-in-Cu[Co(CN)6] cathode.
The rate capability tests further certify that the Se-in-Cu[Co(CN)6] cathode delivers the highest specific capacity among the four samples under all testing rates (0.2 A×g-1 to 10 A×g-1) (Figure 4c). Moreover, the Zn battery with Se-in-Cu[Co(CN)6] cathode shows obvious discharge plateau and high specific capacity of 430.6 mAh×g-1 even at a high current density of 10 A×g-1, which also confirms the fast Se4+/Se2- redox reaction in Cu[Co(CN)6] framework (Figure 4d).
Meanwhile, the Se-in-Cu[Co(CN)6] cathode shows a higher capacity retention rate (98.6%) and Coulombic efficiency (~100%) after cycling at 0.2 A×g-1 for 100 cycles, compared with Se-in-Ni[Co(CN)6] (75.3 % and 96 %, respectively), Se-in-Co[Co(CN)6] (60.3 % and 94 %) and Se-in-porous carbon (25.9 % and 88 %) (Figure 4e). These findings demonstrate that the (Cu/Ni/Co)[Co(CN)6] framework can effectively immobilize the dissolved Sen eliminate the capacity decay caused by the shuttle effect. The Se-in-Cu[Co(CN)6] cathode also shows the outstanding long-term cyclability with 90.6 % initial capacity retained after 6000 cycles at current density of 5 A×g-1 (Figure 4f). The above results reveals that the enhanced performance is not only a physical increment of polar (Cu/Ni/Co)[Co(CN)6], but more of an optimal ‘Sen confinement-catalysis’ process by the double polar sites.
Phase transformation and interfacial chemistry
To securitize the Zn2+ storage mechanism in the Se-in-Cu[Co(CN)6] cathode, we analyzed the Se conversion reactions and the concomitant phases evolution by in-situ XRD and Raman tests at second GCD cycle at of 0.2 A×g-1 (Figure 5a). As shown in Figure 5b, three peaks located at 29.6o, 30.5o and 34.2o can be indexed to the Sen crystal phase, which appears at the depth of discharge of 1.7 V. Meanwhile, the ZnSe phase is also detected at discharge process. While, the intensity of characteristic peaks representing SeO2 at 23.4o gradually decreases with discharging process and almost completely disappear at full discharge state. Similarly, the vibrational mode with frequency of 231.5 cm-1 corresponding to Sen, appears at discharge process in Raman spectra, which presents the formation of Sen phase. In comparison, the intensity of vibrational mode representing SeO2 at 250.4 cm-1 gradually disappears during discharge process (Figure 5c)23. The weak vibrational modes with frequency of 251 cm-1 is assigned to the ZnSe, indicating the formation of ZnSe during the discharging process.
We also employed ex-situ X-ray photoelectron spectroscopy (XPS) to study the redox reactions of the Se cathode at different discharged states of 1.0 V, 0.1 V, and following charge states of 1.5V, 2.1 V. When the battery is discharged to 0.1 V, a pair of peaks of Se located at 54.6 eV (3d5/2) and 55.5 eV (3d3/2) confirms the existence of Se2- (Figure 5d). According to above XRD and Raman analysis, the present of Se2- detected is formation of ZnSe. Subsequently, the peaks moved back to 55.3 eV (3d5/2) and 56.1 eV (3d3/2) for Se 3d when charging to 1.5 V, which is defined as Se0 (Figure 5e). Upon further charge to 2.1 V, a sharp peak located at 59.2 eV appears in the Se 3d spectrum, manifesting Se0 is oxidized to Se4+ (Figure 5f). When discharged to 1.0 V, a pair of peaks located at 55.4 eV (3d5/2) and 56.3 eV (3d3/2) are derived from Se0, which indicates Se4+ is reduced to Se0 (Figure 5g). Based on the above-mentioned results, this Zn||Se battery achieves a Se-based reversible 6-electron transfer reaction of Se2- Se0 Se4+.
The time-of-flight secondary ion mass spectrometry (ToF-SIMS) on the cycled Zn metal are performed to directly characterize Zn metal corrosion. The results show that, with Se-in-porous carbon cathode, many Se agglomerates are identified on the surface of cycled Zn metal anodes in the cells (Figure 5h), demonstrating a severe shuttle effect of Sen and Zn metal corrosion. The intensities of the Se element on the surface of the cycled Zn metal anodes became weaker in the case of the Se-in-Ni[Co(CN)6] and Se-in-Co[Co(CN)6] cathode (Figure 5i, j), but were still visible, demonstrating that the Ni[Co(CN)6] and Co[Co(CN)6] frameworks cannot completely eliminate the shuttle effect and Zn metal corrosion. By sharp contrast, only a very weak Se signal that is assigned to the formation of a Se-containing solid-electrolyte interphase layer can be observed on the surface of the cycled Zn metal anode employing the Cu[Co(CN)6] host (Figure 5k). The suppressed shuttle effect is attributed to the highly effective Se conversion and strong chemical affinity to pristine Sen, reactive Sen and ZnSe (Table S1).
Validation in Zn||Se pouch cells with ultra-high mass loading
In light of the excellent coin-cell performance of the Se-in-Cu[Co(CN)6] cathode, we fabricated a pouch-type cell with a loading mass of 4.45 mg·cm-2 and total 100 mg Se loading in a single-piece cathode with size of 5×4.5 cm. The as-prepared pouch cell is cycled at high current density of 5 A·g-1. The cell exhibited a specific capacity of over 500 mA h·g-1 with a capacity retention rate of 90.4 % and high Coulombic efficiency of ~100 % for 450 cycles (Figure 6a). Pouch cells with loading mass up to 8.9 mg·cm-2 and total 200 mg Se cathode using Se-in-Cu[Co(CN)6], cathode and Se-in-porous carbon compartment coupled with Zn metal anode, are thus developed and tested for comparison. As shown in Figure 6b, the Se-in-Cu[Co(CN)6], cathode-based pouch cell still demonstrates very stable cycle life with 91.8 % initial capacity retained and Coulombic efficiency close to 100 % within 500 cycles. By contrast, the Se-in-porous carbon-based pouch-type cell experiences continuous capacity fading (with only 18.7 % capacity retained after 480 cycles) and fluctuant Coulombic efficiency. We also assembled an A-h-level Zn||Se-in-Cu[Co(CN)6] pouch-type battery consisting of four whole sets of anode-electrolyte-cathode stacks, to validate our concept in high Se loading operation (the inset of Figure 6c). It is mentioned that the Zn metal is only 200% excess (corresponding to a theoretically negative/positive capacity ratio of ~2.4), and the electrolyte/Se ratio is controlled at 5 ml×mg-1. We increase the total Se loading of the pouch cell to 2.4 g with a double-side coating (corresponding to loading mass of ~12.3 mg×cm-2). As shown in Figure 6c, the as-assembled cell delivers ~1350 mA·h g-1 at 0.5 A·g-1, manifesting a high Se utilization (~83.3 %) even under high Se loads and with lean electrolyte. As a result, our A-h-level Zn||Se pouch cell attains a high specific energy density of 675 Wh·kgSe-1. It also exhibits high Coulombic efficiency (> 98 %) and stable cycling performance for 400 cycles (89.4 % capacity retention), supporting the effectiveness of the double polar sites in immobilizing Sen and eliminating Zn metal corrosion. Meanwhile, based on delivered capacity of Se-in-Cu[Co(CN)6] cathode with different mass loading, the delivered capacity linearly scale with increase of areal mass loading from 2 to 12.3 mg·cm-2. The specific capacity almost maintains with increased Se loadings (Figure 6d).
We further compare the electrochemical performance of the Zn batteries based on different cathode materials to reveal the advantages of the Zn||Se system. Remarkably, the Zn||Se batteries deliver a highest discharge capacity among reported Zn batteries (Figure 6e), which benefits from the effective conversion reaction mechanism of Se0 and Se2-, and high Se utilization. Meanwhile, our delivered Zn||Se batteries delivered an ultra-high energy density, 728.9 Wh·kg-1(Se) and maximum power density of 7078.8 W·kg-1(Se), which is greatly higher than that of other reported cathodes in rechargeable Zn batteries (Figure. 6f)24-34.