Tunneled Structures as Model Cation Hosts for Energy Storage

Sep 27, 2017 - CONSPECTUS: Future advances in energy storage systems rely on identification of appropriate target materials and deliberate synthesis o...
0 downloads 3 Views 6MB Size
Article Cite This: Acc. Chem. Res. 2018, 51, 575−582

pubs.acs.org/accounts

Investigation of α‑MnO2 Tunneled Structures as Model Cation Hosts for Energy Storage Published as part of the Accounts of Chemical Research special issue “Energy Storage: Complexities Among Materials and Interfaces at Multiple Length Scales”. Lisa M. Housel,†,¶ Lei Wang,†,¶ Alyson Abraham,†,¶ Jianping Huang,† Genesis D. Renderos,† Calvin D. Quilty,† Alexander B. Brady,‡ Amy C. Marschilok,*,†,‡,§ Kenneth J. Takeuchi,*,†,‡ and Esther S. Takeuchi*,†,‡,§ †

Department of Chemistry, Stony Brook University, Stony Brook, New York 11794, United States Department of Materials Science and Chemical Engineering, Stony Brook University, Stony Brook, New York 11794, United States § Energy and Photon Sciences Directorate, Brookhaven National Laboratory, Upton, New York 11973, United States ‡

CONSPECTUS: Future advances in energy storage systems rely on identification of appropriate target materials and deliberate synthesis of the target materials with control of their physiochemical properties in order to disentangle the contributions of distinct properties to the functional electrochemistry. This goal demands systematic inquiry using model materials that provide the opportunity for significant synthetic versatility and control. Ideally, a material family that enables direct manipulation of characteristics including composition, defects, and crystallite size while remaining within the defined structural framework would be necessary. Accomplishing this through direct synthetic methods is desirable to minimize the complicating effects of secondary processing. The structural motif most frequently used for insertion type electrodes is based on layered type structures where ion diffusion in two dimensions can be envisioned. However, lattice expansion and contraction associated with the ion movement and electron transfer as a result of repeated charge and discharge cycling can result in structural degradation and amorphization with accompanying loss of capacity. In contrast, tunnel type structures embody a more rigid framework where the inherent structural design can accommodate the presence of cations and often multiple cations. Of specific interest are manganese oxides as they can exhibit a tunneled structure, termed α-MnO2, and are an important class of nanomaterial in the fields of catalysis, adsorption−separation, and ion-exchange. The α-MnO2 structure has one-dimensional 2 × 2 tunnels formed by corner and edge sharing manganese octahedral [MnO6] units and can be readily substituted in the central tunnel by a variety of cations of varying size. Importantly, α-MnO2 materials possess a rich chemistry with significant synthetic versatility allowing deliberate synthetic control of structure, composition, crystallite size, and defect content. This Account considers the investigation of α-MnO2 tunnel type structures and their electrochemistry. Examination of the reported findings on this material family demonstrates that multiple physiochemical properties influence the electrochemistry. The retention of the parent structure during charge and discharge cycling, the material composition including the identity and content of the central cation, the surface condition including oxygen vacancies, and crystallite size have all been demonstrated to impact electrochemical function. The selection of the α-MnO2 family of materials as a model system and the ability to control the variables associated with the structural family affirm that full investigation of the mechanisms related to active materials in an electrochemical system demands concerted efforts in synthetic material property control and multimodal characterization, combined with theory and modeling. This then enables more complete understanding of the factors that must be controlled to achieve consistent and desirable outcomes.



INTRODUCTION This Account examines the viability of a tunnel type structure for the robust insertion and deinsertion of cations when used in © 2018 American Chemical Society

Received: September 27, 2017 Published: February 19, 2018 575

DOI: 10.1021/acs.accounts.7b00478 Acc. Chem. Res. 2018, 51, 575−582

Article

Accounts of Chemical Research energy storage systems. Layered structures are often used where lattice expansion and contraction associated with the ion movement can result in structural degradation with accompanying loss of capacity.1 In contrast, tunnel type structures contain a more rigid framework where the inherent structure can accommodate the presence of cations. Tunnel structured manganese oxides have been investigated in the fields of catalysis, adsorption− separation, and ion exchange,2,3 and the tunnels can be substituted by various cations, including K+, Ag+, Na+, and Mg2+.4,5 The α-MnO2 structure has one-dimensional 2 × 2 (0.46 × 0.46 nm2) tunnels formed by corner and edge sharing manganese octahedral [MnO6] units adopting an I4/m tetragonal crystal structure. Further, manganese centers are mixed valence (3+ and 4+) where charge neutrality may be maintained by cations within the tunnel structure. This Account describes investigations of tunnel structured α-MnO2 (Mn8O16) materials providing a comprehensive view of their electrochemistry. It will be clear that the mechanisms related to active materials in an electrochemical system demands concerted efforts in synthetic material property control and multimodal characterization, combined with theory and modeling as multiple physiochemical properties play significant roles.6 α-MnO2 (Mn8O16) type materials often contain cations with 1+ or 2+ charge located within the manganese oxide tunnels where the Mn cations possess mixed 3+ and 4+ oxidation states (Figure 1).7 These materials are conceptually an attractive class

Table 1. Lattice Parameters from Single Crystal Diffraction for MxMn8O16 material

a = b (Å)

c (Å)

ref

Mn8O16 K1.33Mn8O16 Ag1.8Mn8O16

9.815 9.866 9.725

2.847 2.872 2.885

8 9 5

relative to Mn8O16, it does not explain the observed lattice contraction of Ag1.8Mn8O16. The contraction has been attributed to covalent-bonding behavior on the part of the silver atom,11 not accounted for in the hard-sphere ionic model.10 Synthesis of the Mn8O16 family of materials is worthy of consideration as carefully designed synthetic approaches can result in variation and control of composition, crystallite size, average Mn oxidation state, and surface oxygen vacancies. Potassium containing cryptomelane type KxMn8O16 materials have been prepared by a variety of methods including hydrothermal, sol−gel, and reflux methods.12−14 The synthesis of pure silver hollandite proved to be more elusive. In 1984, the first synthesis of silver hollandite (Ag1.8Mn8O16) was reported using a high temperature, high pressure solid-state approach.5 This was followed by an ionexchange procedure; however, potassium (K+) and silver (Ag0, Ag+) were in the final product.15 Hydrothermal approaches showed an impurity phase of pyrolusite (β-MnO2).16,17 A reflux-based synthesis16 was used for preparation of pure silver hollandite (AgxMn8O16, 1.0 ≤ x ≤ 1.8) with variation of silver content (x) and concomitant increase in crystallite size with increases of x.18 The compositional ranges achieved via reflux11,18−21 and hydrothermal19 synthesis of silver hollandite, and reflux22,23 and hydrothermal24−28 synthesis of potassium cryptomelane are summarized in Figure 2. Interestingly, while the compositional

Figure 1. General structure of hollandite, with 2 × 2 tunnels and central cation.

of secondary battery cathode materials due to the opportunity for (de)insertion of ions through the 1D structure combined with the high oxidation state of electroactive Mn. The examples presented here will focus on MxMn8O16 (M = no metal cation, Ag+, or K+). Notably, when the central cation (M) is potassium, K+, it is electrochemically inert; however, when M = Ag+, the cationic center is potentially electrochemically active.



Figure 2. Compositional ranges via reflux and hydrothermal synthesis for MxMn8O16 (M = Ag, K), with M/Mn ratio determined via optical emission or atomic absorption spectroscopy.

MATERIALS STRUCTURE AND SYNTHESIS Only one single-crystal structure analysis of each has been published for Mn8O16,8 KxMn8O16,9 and AgxMn8O16,5 and all structures share the I4/m space group. Notably, the equatorial lattice parameters (a = b) expand when potassium is added to the tunnel but contract when silver is added (Table 1). A parametrized geometric model was created to calculate theoretical lattice parameters for hollandite structured AxB8O16, with added terms to account for the structural cation (B) valence and size, and the tunnel cation (A) size.10 While this model describes the lattice expansion observed for several hollandite materials and is consistent with the observed lattice expansion of K1.33Mn8O16

ranges show some overlap, the compositional range for potassium cryptomelane is more limited (typically 1.0 per 8 Mn.



LITHIUM BASED ELECTROCHEMISTRY As a cathode material, α-MnO2 has been reported to deliver > 200 mAh/g29 but has also demonstrated significant capacity fade. The incorporation of cations in the tunnel such as Li2O, NH4+, K+, and Rb+ has been reported to improve capacity retention.12,24,30−33 576

DOI: 10.1021/acs.accounts.7b00478 Acc. Chem. Res. 2018, 51, 575−582

Article

Accounts of Chemical Research

Detailed characterization using local (TEM, electron diffraction, EELS) and bulk (synchrotron based X-ray diffraction, thermogravimetric analysis) techniques indicated a greater quantity of oxygen vacancies for the low-Ag materials, accompanied by lower average Mn valence (∼Mn3.5+) relative to their high-Ag counterparts (∼Mn3.7+).11 Notably, EELS data also indicated a lower Mn oxidation state and higher concentration of oxygen vacancies on the surface relative to the interior of the nanorods (Figure 3E). The 7-fold increase in delivered capacity was attributed to oxygen vacancies and MnO6 octahedra distortion for the small crystallite size, low silver content material, which facilitate Li+ diffusion (Figure 3F). This paper revealed that differences in physiochemical properties including crystallite size and oxygen vacancies play a significant role in influencing the electrochemistry where smaller crystallite size and higher oxygen vacancies are beneficial. The impact of potassium ion concentration on the electrochemistry of KxMn8O16 has recently been examined by several authors. One study used nitric acid treatment to lower the potassium content of K2Mn8O16 (described as K0.25MnO2).33 In that report, higher K+ content materials showed better rate capability with improved capacity delivery at higher discharge currents. At 5 C, Li/K0.25MnO2 delivered 62% and Li/K0.25−xMnO2 delivered 54% compared to their 0.1 C capacity. The study attributed the improved capacity delivery to higher potassium content; however, concomitant with lowering the K+ amount, the 4 day acid treatment changed the manganese average oxidation state and likely decreased the surface defects of the treated product. The report initially appeared in contrast to another report, which showed that the KxMn8O16 samples with lower K+ content (x = 0.00, 0.32) led to higher capacity and rate capability compared to higher K+ content (x = 0.51, 0.70, 0.75).24 However,

Specifically, previous experimental results indicated that tunnel stabilization by Li2O in α-MnO2 improves the cycle stability of the material.30,34 Further, the Li2O-doped or parent α-MnO2 powder was treated with NH3 gas resulting in α-[0.143Li2O· 0.007NH3]·MnO2 (Li2O/NH3 treated) or α-[0.20NH3]·MnO2 (NH3-treated). The ammonia treated α-MnO2 (0.20NH3· MnO2) maintained a reversible capacity of 167 mAh/g from cycle 5 to cycle 20. Lithia-stabilized α-MnO2 (0.143Li2O·MnO2) showed a capacity of 173 mAh/g with little capacity fade from cycle 5 to cycle 20. In contrast, untreated α-MnO2 showed more significant capacity fade from 218 mAh/g to 136 mAh/g after 10 cycles. It is important to note that the samples that initially delivered higher capacity showed more capacity fade. As will be noted below, the level of lithiation impacts structural stability of α-MnO2 with associated capacity fade and was likely a contributing factor in these experiments. This observation did motivate further studies into the role of the cation type and level as are described below. In the case of silver cations in the tunnel, composition and crystallite size were found to be related, where samples with higher silver content had larger crystallite size. For example, samples of AgxMn8O16, where x = 1.55 (high-Ag) and 1.25 (low-Ag) had corresponding crystallite sizes of 26 and 12 nm, respectively (Figure 3A,B).19 The cells assembled using the smaller crystallite size (low-Ag) material consistently displayed loaded voltages several hundred millivolts higher than the cells with the larger crystallite (high-Ag) material (Figure 3C).19 Further, while the theoretical capacities of the Ag1.20Mn8O16 (260 mAh/g) and Ag1.60Mn8O16 (247 mAh/g) differ by ∼5% based on composition, discharge tests (Figure 3D) indicated that Ag1.20Mn8O16 delivered 160 mAh/g while Ag1.60Mn8O16 delivered 23 mAh/g.

Figure 3. TEM images of high-Ag (A) and low-Ag (B) silver hollandite nanorods. (C) GITT discharge for lithium anode test cells with high-Ag sample (A) and low-Ag sample (B). (D) Initial discharge capacities of AgxM8O16, where x is 1.60 and 1.20. (E) EELS from low-Ag and high-Ag samples, as well as edge and interior of a nanorod of high-Ag sample. (F) Perspective view of Ag2Mn8O16 along [001] direction (top), showing the breakdown of the MnO6 octahedron wall due to a removal of an oxygen marked by orange circle (bottom). 577

DOI: 10.1021/acs.accounts.7b00478 Acc. Chem. Res. 2018, 51, 575−582

Article

Accounts of Chemical Research

Figure 4. (A) TEM experiment setup. (B, C) Snapshots of the lithiation process, showing three types of Li-transport pathways for three Ag1.63Mn8O16 rods (I−III), acquired after (B) 3 and (C) 320 s of lithiation. (D, E) EDPs acquired at γ (D) and β (E) regions. (F) EELS of Li−K edge comparing ex situ (0e−6e) and in situ lithiated samples. (G) Snap shot from phase-field simulation illustrating evolution of the β1 + β2 phase and the RF. (H, I) DFT predicted stable structures for high (H) Li concentration phases, showing disrupted tunnel walls, and low (I) Li concentration phases, showing intact tunnel structures.

of Ag1.63Mn8O16 particles and their phase evolution.37 Li+ ion diffusion along the c-axis was observed along with the first direct experimental observation of lateral Li+ transfer between rods along the ab plane (Figure 4B,C). Upon lithiation, the diameter of rod I expanded as the reaction front (RF) progressed (Figure 4B) in two regimes: the “β phase” and then the “γ phase”. As the rod expanded radially, the RF progressed toward unlithiated adjacent rods (II, III) where expansion of the adjacent rod III was observed, suggesting lithium ions diffuse both longitudinally along the c axis and laterally along the ab plane (Figure 4C). The operando TEM studies suggested a multistep lithiation process with a reaction front (RF) separated into different regions. Density functional theory (DFT) provided additional insight into the observed dynamics using coarse-grained thermodynamic simulations. Distinct areas were observed around the RF: an area of higher lithium concentration behind the reaction front (γ), and an area with low lithium concentration in front of the RF (β1 + β2). The γ regime (Figure 4D) evidenced the presence of both Ag0 and arc-shaped hollandite diffraction spots, while the β regime (Figure 4E) showed a tetragonal structure with crystal grains aligned along the c axis. Comparison of the data with chemically lithiated samples (0e−6e) using EELS measurements (Figure 4F) affirmed a greater Li concentration in the γ regime than the β regime. Coarse-grained thermodynamic simulations resulted in a phase field model that captured the formation of polyphase material (β1, β2) in the β regime at lower levels of lithiation and the γ regime at higher levels of lithiation (Figure 4G). Density functional theory (DFT) demonstrated expulsion of

this study controlled the cation tunnel content (K+) in KxMn8O16 synthetically while keeping other properties, including surface area, crystallite size, surface defects, and morphology similar across the series of materials. The results from the acid treatment study motivated study of Mn8O16 with no central cation and the effect of acid treatment. It was observed that the acid treatment decreased the specific capacity even in the absence of K+ cation.24 Deconvolution of the influence of K+ content and acid treatment revealed that the decrease in the observed delivered capacity resulted from the acid treatment and the associated changes rather than lower K+ content. In 2015, a direct mechanistic study using in situ transmission electron microscopy probed the lithiation of K0.25Mn8O16.35 While the lithiation was conducted in the absence of electrolyte, the method allows phase evolution to be directly visualized. Atomic resolution imaging showed 1 × 1 and 2 × 2 tunnels along the c directions of the wire, with K+ residing in the 2 × 2 tunnels. A tetragonal−orthorhombic−tetragonal (TOT) transition during lithiation of the K+-stabilized α-MnO2 was observed resulting from the asynchronous expansion of the α-MnO2 tetragonal unit cell along the a and b lattice directions. Density functional theory (DFT) calculations showed a Li+ occupancy sequence at Wyckoff 8h sites inside the 2 × 2 tunnels, leading to the asynchronous expansion and symmetry degradation of the host lattice as well as tunnel instability upon lithiation. More recently, a scanning/transmission electron microscopy (S/TEM) approach developed at Brookhaven National Laboratory36 (Figure 4A) was utilized to visualize operando lithiation 578

DOI: 10.1021/acs.accounts.7b00478 Acc. Chem. Res. 2018, 51, 575−582

Article

Accounts of Chemical Research Ag and loss of the tunnel structure at high levels of lithiation (>6Li, 6e equivalent) (Figure 4I) with retention of the tunnel framework at low lithiation (1Li, 1e equivalent) regime (Figure 4H). This study featured new insights including the possibility of lateral (a, b) ion diffusion and the observation of Ag metal formation at high levels of lithiation. A recent study elegantly modified the existing operando TEM experimental technique in order to probe lithium ion diffusion in the lateral a,b direction as well as along the c axis of the tunnels.38 The results verified that lithium diffusion down the tunnel in the c axis of potassium containing α-MnO2 was the fastest diffusion path. Additionally, the full lithiation process was probed exploring the α-MnO2 conversion reaction. At the early lithiation stage, Li insertion occurred and showed a- and b-lattice expansions but with little to no expansion along the longitudinal direction (c-lattice). Subsequently, the conversion reaction started, leading to the formation of Mn metal and Li2O phases through a MnO intermediate phase. The evolution of the MnO intermediate phase and the development of the MnO and Li2O phases with preferred orientation along the [1̅10]c direction was demonstrated for the first time. This study affirmed the a, b lateral lithium ion diffusion but also demonstrated that it was kinetically slower than diffusion along the c axis. It is thus important to consider the fundamental thermodynamic properties of the α-MnO2 family of compounds. Density functional theory (DFT) computation on α-MnO2 investigated the thermodynamics of Li ion and Na ion insertion.39,40 The 8h sites near the tunnel edges of α-MnO2 were considered as the preferred insertion sites for Li ions when x in LixMnO2 is ≤1. From MnO2 to Li0.25MnO2, lattices a and b increase by