Two-Dimensional Phosphorene-Derived Protective Layers on a

May 1, 2018 - Lithium-oxygen (Li-O2) batteries are desirable for electric vehicles ... Li plating is thermodynamically unfavorable on lithium phosphid...
0 downloads 0 Views 9MB Size
Download PDF
www.acsnano.org

Two-Dimensional Phosphorene-Derived Protective Layers on a Lithium Metal Anode for Lithium-Oxygen Batteries Youngjin Kim,† Dongho Koo,† Seongmin Ha,† Sung Chul Jung,§ Taeeun Yim,∥ Hanseul Kim,† Seung Kyo Oh,† Dong-Min Kim,† Aram Choi,† Yongku Kang,⊥ Kyoung Han Ryu,# Minchul Jang,¶ Young-Kyu Han,*,‡ Seung M. Oh,† and Kyu Tae Lee*,† †

School of Chemical and Biological Engineering, Institute of Chemical Processes, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea ‡ Department of Energy and Materials Engineering, Dongguk University-Seoul, Seoul 04620, Republic of Korea § Department of Physics, Pukyong National University, 45 Yongso-ro, Nam-Gu, Busan 48513, Republic of Korea ∥ Department of Chemistry, Incheon National University, 119 Academy-ro, Songdo-dong, Yeonsu-gu, Incheon 22012, Republic of Korea ⊥ Advanced Materials Division, Korea Research Institute of Chemical Technology, Yuseong, Daejeon 34114, Republic of Korea # Environment and Energy Research Team, Division of Automotive Research and Development, Hyundai Motor Company, 37 Cheoldobangmulgwan-ro, Uiwang, Gyeonggi-do 16082, Republic of Korea ¶ Future Technology Research Center, CRD, LG Chem, Ltd., 188 Munji-ro, Yuseong-gu, Daejeon 34122, Republic of Korea S Supporting Information *

ABSTRACT: Lithium-oxygen (Li-O2) batteries are desirable for electric vehicles because of their high energy density. Li dendrite growth and severe electrolyte decomposition on Li metal are, however, challenging issues for the practical application of these batteries. In this connection, an electrochemically active two-dimensional phosphorene-derived lithium phosphide is introduced as a Li metal protective layer, where the nanosized protective layer on Li metal suppresses electrolyte decomposition and Li dendrite growth. This suppression is attributed to thermodynamic properties of the electrochemically active lithium phosphide protective layer. The electrolyte decomposition is suppressed on the protective layer because the redox potential of lithium phosphide layer is higher than that of electrolyte decomposition. Li plating is thermodynamically unfavorable on lithium phosphide layers, which hinders Li dendrite growth during cycling. As a result, the nanosized lithium phosphide protective layer improves the cycle performance of Li symmetric cells and Li-O2 batteries with various electrolytes including lithium bis(trifluoromethanesulfonyl)imide in N,Ndimethylacetamide. A variety of ex situ analyses and theoretical calculations support these behaviors of the phosphorenederived lithium phosphide protective layer. KEYWORDS: phosphorene, lithium phosphide, lithium metal, protective layers, lithium-oxygen batteries The theoretical energy density (about 3505 W h kg−1) of Li-O2 batteries is much higher than that (about 387 W h kg−1) of LIBs; this is because Li-O2 batteries store energy based on the conversion chemistry of oxygen instead of the intercalation

A

great deal of attention has been given to improving lithium-ion batteries (LIBs) to extend the driving range of electric vehicles.1−7 However, even with LIBs reaching their theoretical energy density limit, it is not possible for these batteries to power battery electric vehicles (BEVs), which can drive up to 500 km on a single charge. Therefore, new battery systems are being sought as an alternative to LIBs, and Li-O2 batteries are considered a promising candidate.8−13 © 2018 American Chemical Society

Received: January 14, 2018 Accepted: May 1, 2018 Published: May 1, 2018 4419

DOI: 10.1021/acsnano.8b00348 ACS Nano 2018, 12, 4419−4430

Article

Cite This: ACS Nano 2018, 12, 4419−4430

Article

ACS Nano

Li dendrite formation. For this purpose, the first step is to find an electrochemically active protective layer that has a higher redox potential than the lowest unoccupied molecular orbital (LUMO) level of the electrolyte solvent (Figure 1). As

chemistry, as used in current LIBs.14−17 Li-O2 batteries comprise an oxygen cathode, an electrolyte, and a Li metal anode. Although remarkable advancements have been achieved for Li-O2 batteries through recent pioneering works on oxygen cathodes and electrolytes,18−28 there has not been enough focus on the challenging issues for Li metal anodes. Affordable electrolytes for use in Li-O2 batteries are restricted because of the reactivity of superoxide radicals and Li metal. For example, while organic carbonates are very common and stable solvents for LIBs, they have been excluded from use in Li-O2 batteries. This is because organic carbonates chemically react with the superoxide radicals obtained from the oxygen cathode, leading to their irreversible decomposition to lithium alkyl carbonates such as Li2CO3.29−31 In this regard, amidebased organic solvents, such as N,N-dimethylacetamide (DMA), have been considered as promising electrolyte solvents because they are more chemically and electrochemically stable against nucleophilic attacks of superoxide than organic solvents such as carbonates, ethers, and sulfoxides. However, unfortunately, DMA is known to be decomposed upon contact with Li metal, which is accompanied by corrosion of the Li metal anode.32,33 Therefore, Li metal protection is necessary for operating Li-O2 batteries with various electrolytes.34−38 Many efforts have been devoted to improving the Li metal stability using stable solid-electrolyte-interphase (SEI) layers formed by electrolyte decomposition, resulting in enhanced electrochemical performance.39−47 However, SEI layers are not stable enough to inhibit the deformation of the layers during cycling, resulting in an inhomogeneous local current and the formation of Li dendrites.48−51 In addition, although the SEI layers are not destroyed, the large current density enables Li metal plating on the SEI layers, leading to additional electrolyte decomposition caused by the exposure of new Li metal surfaces. For example, in previous work, the improved stability of LiNO3-derived SEI layers resulted in improved cycle retention for Li-O 2 batteries when compared with lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in DMA. Nevertheless, the LiNO3-derived SEI layers were not sufficiently stable, and they eventually deformed during cycling. This deformation caused repetitive exposure of new Li dendrites to the electrolyte, accompanied by the continuous consumption of LiNO3 during cycling; this resulted in depletion of the electrolyte and abrupt failure of the Li-O2 batteries.33,52 This indicates that there are at least a few requirements for a Li metal protective layer in Li-O2 batteries. Electrolytes and Li plating should be thermodynamically stable and unfavorable on the protective layer surface, respectively. The mechanical strength of the protective layer should be sufficiently strong to suppress its deformation due to Li dendrite growth.53 In this paper, taking these requirements into consideration, a twodimensional phosphorene-derived lithium phosphide on Li metal is introduced as a Li metal protective layer. The nanosized lithium phosphide protective layer exhibited the excellent electrochemical performance, with Li symmetric cells with various electrolytes showing stable cycle performance over 500 cycles, Li-O2 batteries showing no capacity fading over 50 cycles even with LiTFSI-dissolved in DMA, and suppressed Li dendrite growth on the lithium phosphide protective layer.

Figure 1. Schematic diagram of the electrochemically active protective layer concept.

electrolytes are electrochemically decomposed when the redox potential of the electrode surface is lower than the LUMO level of the solvent,54−58 electrolyte decomposition can be suppressed if the redox potential of the protective layer is higher than the LUMO level of the electrolyte solvent (less than about 0.8 V vs Li/Li+). One appropriate candidate for this limited redox potential range is black phosphorus, which is electrochemically active with Li. Black phosphorus is spontaneously transformed into lithium phosphide (Li3P) on Li metal, acting as a protective layer. The redox potential of black phosphorus converting into Li3P is about 0.9 V vs Li/Li+, which is considered to be higher than the LUMO level of the electrolyte solvent.59,60 However, as another form of black phosphorus, phosphorene is more practical for use as a protective layer than bulk black phosphorus as the two materials have similar redox potentials but different morphologies.61−63 Phosphorene is considered as a few layers of black phosphorus and obtained through the exfoliation of two-dimensional layered black phosphorus.64−66 Phosphorene, with its thickness of tens of nanometers, can form thin protective layers on Li metal through conventional coating processes; in contrast, it is not possible for black phosphorus to be coated on Li in the form of a thin film because the thickness of the bulk particles can reach a few millimeters. The second step is to consider whether the plating of Li on the surface of Li3P is energetically unfavorable, in the sense that Li dendrite growth on a Li3P protective layer can be hindered even at large current densities. For the consideration of the thermodynamics of Li plating on Li3P, we first calculated the

RESULTS AND DISCUSSION Theoretical Consideration for Li Protection. We propose a concept of a mechanically strong protective layer that thermodynamically inhibits electrolyte decomposition and 4420

DOI: 10.1021/acsnano.8b00348 ACS Nano 2018, 12, 4419−4430

Article

ACS Nano

energies from Li(111), Li(110), and Li(001) surfaces (Figure 2a). This result implies that the Li3P layer on Li metal is stably maintained during cycling because the removal of Li from the Li3P surface is energetically much more difficult than it is from the Li metal surface. Therefore, the Li addition and removal energy calculations suggest that the Li dendrite growth on the Li3P is thermodynamically hindered without structural collapse during cycling. The thermodynamically unfavorable addition and removal of Li on the Li3P surface can be explained in terms of electrostatic interactions. Bader charge analyses of the atomic layers of the Li3P(001) surface show that the first-layer Li, second-layer Li, and second-layer P atoms (Figure 2c) have charge states of +0.85, +0.82, and −2.50 e per atom, respectively. The P atoms deprive the first-layer Li atoms of their electrons to share them with Li adsorbates, thereby weakening Li adsorption on Li3P(001). The removal of a single Li atom from Li3P(001) is also difficult because the Li atom is tightly bound through attractive electrostatic interaction between the positively charged Li and negatively charged P atoms on the Li3P surface. In addition to this thermodynamic aspect, the electrically insulating nature of Li3P may contribute to inhibiting Li dendrite growth by blocking electron transport, as evidenced by the calculated density of states of Li metal and Li3P (Figure 2d). At the Fermi level (EF), the electron density of Li3P is much lower than that of Li metal, indicating that the number of electrons contributing to the electric conductivity in Li3P is considerably smaller than that in Li metal. Finally, we need to consider whether the mechanical strength of Li3P is great enough to suppress the deformation of the Li3P protective layer due to Li dendrite growth. It is known that protective layers are easily deformed because of Li dendrite growth with a high shear modulus, indicating that a protective layer should be sufficiently strong to endure the pressure originating from Li dendrite growth. The shear modulus of Li3P obtained from DFT calculations is 34.47 GPa,67 which is 10 times higher than that of Li dendrites (3.4 GPa).53 This result indicates that even in terms of mechanical strength, Li3P is a promising candidate for Li protection. Synthesis of a Phosphorene-Derived Protective Layer. A few layers of phosphorene sheets were obtained via facile exfoliation of black phosphorus in a carbonate-based solvent (ethylene carbonate: diethyl carbonate [EC:DEC] 1:1 volume ratio) using ultrasonication below 30 °C for 30 h. Figure 3a,b shows a high-resolution transmission electron microscope (HRTEM) image of exfoliated phosphorene and its corresponding fast Fourier transform (FFT) image, respectively. Phosphorene was several hundred nm in size. The FFT image reveals that the phosphorene layers retained the crystallinity of black phosphorus during exfoliation.64,65 Phosphorene-coated Li metal was obtained by spin-coating a solution of phosphorene in EC:DEC onto a Li metal foil in an Ar-filled glovebox, followed by drying. As shown in the X-ray photoelectron spectroscopy (XPS) spectra of P 2p for the phosphorenecoated Li metal at various etching times (Figure 3c), the phosphorene layers just above the Li metal surface chemically reacted to form Li3P.68 In contrast, the uppermost phosphorene layers remained unreacted because the reaction kinetics that forms Li3P is slow at room temperature. The spontaneous formation of Li3P is supported by ab initio molecular dynamics (AIMD) simulations, which revealed that P atoms on Li metal tend to mix with Li atoms to form Li3P with a large energy gain of 2.36 eV per P atom (Figure S1). We also

addition energy of a single Li atom on the Li3P surface using density functional theory (DFT) calculations. The (001) surface of crystalline Li3P was considered as the surface model because of its high stability and definite adsorption sites. In contrast to other surfaces, the (001) surface can have a top layer consisting of only Li atoms. Our ab initio molecular dynamics (AIMD) simulations of Li3P layer on Li metal show that only Li atoms at the surface are exposed to vacuum (Figure S1b), implying that the formation of Li top layer is energetically favored over that of P or mixed Li-P top layers. The calculated Li addition energy on the Li3P(001) surface was 0.70 eV, which is 50% lower than the average value (1.40 eV) of Li addition energies on Li(111), Li(110), and Li(001) surfaces (Figure 2a−

Figure 2. Thermodynamics of Li plating on lithium phosphide (Li3P). (a) Li addition and removal energies on Li metal and Li3P(001) surfaces. Structures of (b) Li(001) and (c) Li3P(001) surfaces. The Li addition and removal energies are defined as Eadd = −(Efinal − Esurf − ELi) and Erem = Efinal − Esurf + ELi, respectively. Efinal, Esurf, and ELi are the energy of Li-added or Li-removed surface, the energy of pristine surface, and the energy of an isolated Li atom, respectively. The Li addition and removal energies are the energy gain due to the adsorption of a Li atom on a pristine surface and the energy required to remove a Li atom from the pristine surface, respectively. (d) Density of states of bcc Li metal and amorphous Li3P. EF represents the Fermi level.

c). This result indicates that the energy gain due to Li adsorption on Li3P is much smaller than that on Li metal. Thus, Li plating on Li3P is thermodynamically unfavorable compared with that on Li metal. We also calculated the removal energy of a single Li atom from the Li3P(001) surface to estimate the structural robustness of the surface. The calculated Li removal energy for the Li3P(001) surface was 3.06 eV, which is 79% greater than the average value (1.71 eV) of the Li removal 4421

DOI: 10.1021/acsnano.8b00348 ACS Nano 2018, 12, 4419−4430

Article

ACS Nano

Figure 3. Characterization of the phosphorene-derived lithium phosphide (Li3P) protective layer. (a) HR-TEM and (b) FFT images of exfoliated phosphorenes. XPS spectra of P 2p for the phosphorene-coated Li metal at various etching times: (c) before and (d) after dropping an electrolyte solution. (e) Dynamic secondary ion mass spectrometry (SIMS) depth profiles of the phosphorene-coated Li metal. TOF-SIMS mapping images of Li and P for (f) pristine and (g) cycled phosphorene-coated Li metal electrodes.

under dynamic SIMS conditions (Figure 3e). The intensity of phosphorus abruptly decreased after sputtering with Cs+ ion for 50 min, indicating that the thickness of the Li3P layer was approximately