Real-Time XRD Studies of Li–O2 Electrochemical Reaction in

Oct 16, 2012 - Herein we report a new analytical method based on in situ. XRD to evaluate the efficiency of Li−O2 electrochemical reaction in terms ...
11 downloads 4 Views 2MB Size
Download PDF
Letter pubs.acs.org/JPCL

Real-Time XRD Studies of Li−O2 Electrochemical Reaction in Nonaqueous Lithium−Oxygen Battery Hyunseob Lim,†,‡ Eda Yilmaz,† and Hye Ryung Byon*,† †

Byon Initiative Research Unit, RIKEN Advanced Science Institute, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan Department of Chemistry and Division of Advanced Materials Science, Pohang University of Science and Technology (POSTECH), San 31, Hyoja-Dong, Nam-Gu, Pohang 790-784, South Korea

‡

S Supporting Information *

ABSTRACT: Understanding of electrochemical process in rechargeable Li− O2 battery has suffered from lack of proper analytical tool, especially related to the identification of chemical species and number of electrons involved in the discharge/recharge process. Here we present a simple and straightforward analytical method for simultaneously attaining chemical and quantified information of Li2O2 (discharge product) and byproducts using in situ XRD measurement. By real-time monitoring of solid-state Li2O2 peak area, the accurate efficiency of Li2O2 formation and the number of electrons can be evaluated during full discharge. Furthermore, by observation of sequential area change of Li2O2 peak during recharge, we found nonlinearity of Li2O2 decomposition rate for the first time in ether-based electrolyte. SECTION: Energy Conversion and Storage; Energy and Charge Transport

A

chemical species has been achieved by ex situ X-ray diffraction (XRD), Raman, FT-IR, time-of-flight secondary ion mass spectrometry (ToF-SIMS), high-resolution transmission electron microscopy (HRTEM), or nuclear magnetic resonance (NMR) spectroscopy,10,11,19,20 and the number of electron has been achieved by differential electrochemical mass spectroscopy (DEMS).12,13,21 Although DEMS gives an insight into quantified efficiency of rechargeability by real-time monitoring, the targets are limited to gas species evolved during recharge.12−14,22 Namely, the analysis of quantity and the number of electrons by direct measurement of the solid-state Li2O2 and byproducts has not been reported. Therefore, it is highly demanded to establish a facile method to provide the information about the chemical species and the corresponding number of electrons involved in this process, which is essential to identify the obstacles to achieve high performance, thereby improving Li−O2 battery system from capacity fading. Herein we report a new analytical method based on in situ XRD to evaluate the efficiency of Li−O2 electrochemical reaction in terms of simultaneous identifications of discharge product and the number of electron involved in full discharge/ recharge cycle. This research approach is quite distinguishable from the conventional XRD techniques neither for lithium ion battery by in situ measurement to observe a phase transition of host cathode23,24 nor for Li−O2 battery by ex situ measurement

rechargeable lithium−oxygen (Li−O2) battery has a theoretical energy density higher than that of lithium-ion battery, which gives a potential to be developed for electrical vehicles requiring high battery capacity.1 However, critical scientific and technical challenges, especially understanding of electrochemical process on discharge/recharge cycle, still need to be resolved.2−15 The discharge/recharge process of Li−O2 battery employs the following reversible electrochemical reaction via a twoelectron process (eq 1)16,17 or possibly sequential one-electron process.18 O2 + 2Li+ + 2e− ↔ Li 2O2 (s)

E o = 2.96 V vs. Li/Li+ (1)

The performance of rechargeable Li−O2 battery is mainly governed by the yield of Li2O2 formed and decomposed in the presence of organic electrolyte. Therefore, it is critical to determine the accurate quantity of Li2O2 formed, that is, efficiency, and decomposed on discharge and recharge, respectively. Unfortunately, the precise evaluation of the Li2O2 quantity is frequently hindered by the byproducts mainly produced by the reaction of superoxide radical (O2•−), an intermediate of oxygen reduction reaction (O2 + e− → O2•−) for the Li2O2 (or LiO2) formation,17,18 with organic electrolytes.10,11 Two important issues acquiring an analogy of Li2O2 are the identification of chemical species involved in discharge/ recharge process and the accurate determination of the number of electrons involved with the formation/decomposition of Li2O2 as well as byproducts. Up to date, the identification of © 2012 American Chemical Society

Received: September 18, 2012 Accepted: October 16, 2012 Published: October 16, 2012 3210

dx.doi.org/10.1021/jz301453t | J. Phys. Chem. Lett. 2012, 3, 3210−3215

The Journal of Physical Chemistry Letters

Letter

to address discharge product.5,6,8,11,12,15 The benefit and novelty to use in situ XRD for Li−O2 battery, especially with comparison to DEMS that is the only real-time measurement tool to investigate Li−O2 battery mechanism, is that it is capable of being a simple, easy, and more importantly accurate evaluation by direct measurement of solid-state Li2O2, working for deep depth of discharge/recharge and cycles, and providing easy access and calculation without delicate and high-cost instrument and professional technique. In this study, we present the acquiring of a dominant presence of Li2O2 with different formation/decomposition rate by the monitoring of real-time Li2O2 peak-area change for the first time using in situ XRD. We explored Super P/binder cathode with ether-based electrolyte (triethylene glycol dimethyl ether (triglyme) and dimethoxyethane (DME)) as the first demonstration. In situ XRD measurements were performed using a homemade Li−O2 XRD cell. Li−O2 XRD cell structure was designed as the general Li−O2 battery cell (Scheme 1 and

Figure 1. Battery performance and XRD patterns of Super P/nafion cathode in 0.5 M Li salt of triglyme. (a) Discharge/recharge profiles in 2 to 4.75 V versus Li/Li+ at 0.1 mA cm−2geometry (158 mA g−1carbon) of current rate. (b) XRD patterns of i) as-prepared, ii) 1st discharged, iii) 1st recharged, and iv) 2nd discharged cathode, corresponding to the profile curves indicated by the colored circles in panel a.

(black line in Figure 1b), which is originated from the polymer membrane. The real-time XRD patterns upon the first discharge/ recharge are shown in Figure 2. Each XRD pattern was

Scheme 1. Schematic Views of (a) Li−O2 XRD Cell Structure for a Real-Time Collection of XRD Pattern and (b) the Cell Components Assembled and Posed on the Sample Stage (The Yellow Dashed Box in Panel a)

Figure S1 in the Supporting Information), which allowed the cathode to expose both a sufficient amount of O2 gas and X-ray beam path, thus removing the worries for the reliability of electrochemical tests during in situ XRD measurement. Super P/nafion (mass ratio of 6/4) pasted on a trilayer polymer membrane (Celgard C480, polypropylene (PP)/polyethylene (PE)/polypropylene (PP)) was assembled with a glassy-fiber separator, lithium metal, and 0.5 M LiTFSI (LiN(SO2CF3)2) in triglyme in an Ar-filled glovebox. The Li−O2 XRD cell filled with O2 gas through an O2 tank was loaded in an XRD instrument. The discharge/recharge profiles were recorded at 0.1 mA cm−2geometry (158 mA g−1carbon) of current rate at room temperature. (See more details in the Experimental Methods.) Figure 1a,b shows discharge/recharge profiles and a snapshot of XRD patterns corresponding to as-prepared and after full discharged/recharged cathode, respectively. The first discharge delivers ∼2934 mAh g−1electrode of capacity (green curve in Figure 1a), and a fully discharged cathode reveals a sharp XRD pattern of Li2O2 (green line in Figure 1b, 2θ ≈ 33.0° for (100), 35.1° for (101), and 40.7° for (102) plane). These Li2O2 peaks completely disappear after recharge (orange line in Figure 1b). The specific capacity of the following discharge is dramatically decreased to ∼1100 mAh g−1electrode (blue curve in Figure 1a). The Li2O2 pattern regenerated has much lower peak intensities and slightly wider widths (blue line in Figure 1b). No XRD peak of Li2O, LiOH, and Li2CO3 could be detected in cathode during Li−O2 battery cell tests. Note that PE peaks are observed in the XRD pattern from the as-prepared cathode

Figure 2. In situ XRD patterns on the first discharge/recharge. (a,b) Collections of XRD patterns in 30−45° of 2θ on (a) discharge and (b) recharge processes.

recorded in the 2θ region of 30−45° for every 30 min at 0.5 degree min−1 of scan rate with 50 s of retention interval between the measurements. The intensity changes of the peaks corresponding to Li2O2 (100), (101), and (102) planes exhibit distinctly different patterns: The peak intensities increase linearly upon discharge (Figure 2a) whereas they decrease nonlinearly upon recharge as they show a gradual decrease at the beginning of recharge but a rapid decrease after a certain point of the middle of recharge process (Figure 2b). We also calculated the area of Li2O2 (101) peak that shows the highest signal-to-noise ratio among the Li2O2 peaks by fitting with a Lorentzian function, then correlated with the discharge/ recharge time (Figure 3a; See more details in the Supporting Information). It is clearly seen that the Li2O2 peak area 3211

dx.doi.org/10.1021/jz301453t | J. Phys. Chem. Lett. 2012, 3, 3210−3215

The Journal of Physical Chemistry Letters

Letter

Figure 4. Time-dependent crystallite dimension of Li2O2 corresponding to full width at half-maximum (FWHM) of Li2O2 (101) XRD peaks on the first and second discharges. The green (blue) empty and filled circles indicate the crystallite domain size and fwhm of Li2O2 on the first (second) discharge, respectively.

recharge, the peak-area value of the Li2O2 becomes almost zero, implying the decomposition of entire Li2O2 crystal produced on discharge. The number of electrons (n), also expressed as the rate, involved during the Li2O2 formation/decomposition is estimated by following eq 2.

Figure 3. Efficiency of Li−O2 electrochemical reaction on the first discharge/recharge evaluated from Li2O2 peak areas of in situ XRD patterns. (a) Time-dependent Li2O2 (101) peak area. The red dashed line indicates the ideal rate of Li2O2 (101) formation/decomposition. (b) Values of number of electron used to form/decompose a Li2O2 (n e−/Li2O2) calculated from every five points of Li2O2 (101) peak area in panel a. The red dashed line guides for the ideal two-electron process of Li−O2. The green filled circles and orange empty squares indicate Li2O2 (101) formation and decomposition, respectively.

n × (slope of time‐dependent Li2O2(101)peak area) = 2 × (slope of time‐dependent ideal Li2O2 (101) peak area)

(2)

The number of electrons used for the formation of a Li2O2 (n e−/(+)Li2O2) was ∼3.12 e−/(+)Li2O2 on discharge (green filled circles in Figure 3b) calculated from every five point of Li2O2 (101) peak area in Figure 3a. Considering that ideally two electrons are involved in the formation of Li2O2 during discharge (red dashed line in Figure 3b), the surplus electron (1.12 e−/(+)Li2O2) should be employed for other electrochemical reactions, most likely electrolyte degradation,11,12 which causes the limited quantity of Li2O2 (∼64.1%). One remarkable feature observed from the peak-area change during recharge is that the decrease rate shows nonlinearity, which has not been observed before (orange empty squares in Figure 3a). The nonlinear change of peak area on recharging is attributed to the two different Li2O2 decomposition rates that are signified by the numbers of electrons involved in the process: (1) a slow decomposition process with dramatically decreasing number of electrons (8 to 2 e−/(−)Li2O2) on the first 15 h and (2) a rapid process involving almost ideal number of electron (∼2 e−/(−)Li2O2) after 15 h (Figure 3b). The large surplus electrons at the beginning seem to be consumed for the oxidation of byproducts formed during the discharge process like lithium alkyl carbonates10,11 and for the oxidation of cathode materials and electrolyte,12 which can be confirmed by 1 H NMR analysis (Figure S4 in the Supporting Information) and linear voltammograms (Figures S5 and S6 in the Supporting Information for a proposed mechanism). The reason for the sluggish decomposition of Li2O2 and huge overpotential accordingly is probably attributed to the low electronic conductivity of the Super P/binder electrode. In particular, at the starting point of recharge, the difficulty of electron transportation by a presence of Li2O2 that is either

increases linearly for the whole discharge process (0−31.5 h, green filled circles), whereas it decreases with two different rates (slopes) on the recharge process (0−15 h and 15−31.5 h, orange empty squares), which are consistent with the peakintensity behavior of Li2O2 (101) as well as the peak-area pattern of (100) (Figure S2 in the Supporting Information). In addition to the increase in the Li2O2 peak area on discharge, the crystallite dimension of Li2O2 indicating the coherently diffracted domain size was gradually decreased (the green empty circles), as shown in Figure 4, which was estimated by Scherrer equation from the increased full width at halfmaximum (FWHM, the green filled circles). It implies that the forming Li2O2 domain size becomes smaller and the surface coverage on the electrode is getting larger with poor crystallinity. The Li2O2 domain size eventually approached ∼25 nm at the end of discharge, which was comparable to the ones (15−20 nm) attained from ex situ XRD in the previous reports.12,15 The efficiency of Li2O2 formation on the first discharge, which is determined from the peak-area ratio of the ideal Li2O2 and the Li2O2 formed after full discharge, turns out to be ∼64.1% (Figure 3a). Note that the ideal peak-area value was obtained from a separate reference cathode composed of 9 wt % of commercially available Li2O2 (∼92.3% of purity measured by a redox titration) with Super P/nafion (see more details in the Supporting Information and Figure S3), and the ideal Li2O2 (101) peak area correlated with the discharge/recharge time is indicated in Figure 3a (red dashed line). Upon the successive 3212

dx.doi.org/10.1021/jz301453t | J. Phys. Chem. Lett. 2012, 3, 3210−3215

The Journal of Physical Chemistry Letters

Letter

efficiency, whereas no XRD peak of Li2O2 was observed.12 However, we also note that the XRD quantification method can allow for error if the amorphous Li2O2 discharge product exists. Therefore, the comparison of the quantitative analysis from XRD and DEMS systems is greatly helpful to investigate the accurate efficiency of Li−O2 electrochemical reaction. We propose that the carbon-based and catalyst-free Li−O2 cells have a typically comparable efficiency in 57−63% for the formation of Li2O2 using DME on the first cycle, which is estimated from the number of electrons by XRD measurement on discharge (∼3.5 e−/(+)Li2O2, Figure S7 in the Supporting Information) and by DEMS on recharge (∼3.2 e−/O2)12 under the assumption of 100% decomposition of Li2O2 to Li+ ion and O2 gas evolved with the same electrolyte of DME, although different carbon material, cell design, and depth of discharge should be further carefully considered. In summary, we demonstrate that in situ XRD-based analysis is reliable to evaluate quantitatively the efficiency of Li−O2 electrochemical reaction. The XRD pattern on cathode of Li− O2 battery cell showed dominant Li2O2 and no evidence of other inorganic crystal in ether-based electrolyte. With chemical identification of discharge product, the real-time acquisition of Li2O2 XRD pattern allowed us to estimate increasing and decreasing Li2O2 peak-area change, which revealed the rates of formation and decomposition of solid-state Li2O2 during discharge and recharge, respectively. The number of electrons estimated from time-dependent Li2O2 peak-area change showed the constant rate of Li2O2 formation but nonlinear rate of decomposition probably due to screening of Li2O2 decomposition by the oxidation of byproducts at the beginning of recharge. The quantity of Li2O2 was limited to 55−65% after the first full discharge and further reduced after the second discharge (∼35%) even though Li2O2 formed was completely decomposed after the first recharge. We believe that this simple and straightforward method provides an opportunity to investigate about the chemical species and the rate of Li−O2 electrochemical reaction on discharge/recharge cycle for various designs of Li−O2 battery systems.

clogged in electrode void or covered on the surface of carbon particle/binder chain suppresses the oxidation of Li2O2. Upon the second discharge, the capacity and efficiency of Li2O2 formation are greatly reduced (Figures 1a and 5). The

Figure 5. In situ XRD patterns and efficiency on the second discharge process. (a) Collection of XRD patterns and (b) time-dependent Li2O2 (101) peak area. The red dashed line indicates the ideal rate of Li2O2 (101) formation.

efficiency of Li2O2 (101) formation is ∼34.6% (Figure 5b), and the number of electrons for the Li2O2 formation is ∼5.77 e−/ (+)Li2O2. The distribution of crystallite dimension of Li2O2 (101) is 20−22 nm (the blue empty circles in Figure 4), which is smaller than the one measured after the first discharge (the green empty circles in Figure 4). The reason for capacity fading, low Li−O2 reaction efficiency, and poor Li2O2 crystallinity on the second discharge is not clear. A plausible hypothesis is that the byproducts produced on the first cycle are accumulated in both cathode and Li anode, resulting in deterioration of Li2O2 formation and fading Li−O2 cell performance.14 Other ether and carbonate-based electrolytes were also examined. The efficiency (∼57.1%) in 0.5 M LiTFSI of DME is slightly lower than that in triglyme on the first discharge (Figure S7 in the Supporting Information). The number of electrons for the Li2O2 formation is ∼3.50 e−/(+)Li2O2. The recharge slope also exhibits nonlinearity with two slopes like the case of triglyme, and the estimated number of electrons is 5 to 2 e−/(−)Li2O2 for the first 12 h and ∼2 e−/(−)Li2O2 after 12 h. The Li2O2 peak completely disappears after full recharge. The Li−O2 XRD cell prepared with carbonate electrolyte, propylene carbonate (PC)/dimethyl carbonate (DMC) (1/1 volume ratio) with 0.5 M of LiTFSI, does not present any clear XRD pattern for either Li2O2 or Li2CO3 after the first full discharge (Figure S8 in the Supporting Information). The efficiency of Li2O2 (55−65%) shows quite a large discrepancy to the value measured by DEMS (∼98%, 2.05 e−/ O2) in ether-based electrolytes. Because the efficiency value by DEMS comprehensively covers all of the byproducts possibly formed including Li2O2 during discharge,12 whereas in situ XRD-based analysis is calculated based on the amount of Li2O2 only via increase in solid-state Li2O2 peak area, such a large difference in the efficiency value might be attributed to the encountered total number of byproducts. It means that we should be careful to estimate the efficiency from the number of electrons that is obtained from either a cathode mass gained or pressure decay as much as O2 gas consumed without any evidence of Li2O2 product because we cannot rule out the consumption of O2 gas for parasitic reactions. For example, the number of electrons of Li−O2 cell with carbonate-based electrolyte exhibited 2.3 to 2.7 e−/O2, estimated to 74−87% of

■

EXPERIMENTAL METHODS Preparation of Cathode. The cathode was constructed by mixing Super P carbon and nafion (mass ratio 6/4) in isopropanol using a planetary mixer (Thinky) at 2000 rpm for 10 min, which was pasted on a trilayer of polymer membrane (PP/PE/ PP) and dried in vacuum at 60 °C overnight. The typical mass (Super P + nafion) and diameter of cathode were 1.2 mgelectrode and 12 mm, respectively. The Li2O2 electrode was prepared by mixing 1/6/4 mass ratio of Li2O2 powder/Super P/nafion with the same procedure. Assembly of Electrode on the Li−O2 XRD Battery Cell. A homemade Li−O2 battery cell for in situ XRD measurement (Li−O2 XRD cell) is composed of a top body and bottom stage part (Figure S1(a) in the Supporting Information). The top body part consists of two windows through which an X-ray beam passes. The two windows were sealed by polyimide (Kapton) films, and leak tests were examined in vacuum, ensuring a tight seal. An open planar spring was suspended from the edges of top body inside. On the bottom stage part, a thin stainless-steel plate, lithium metal, glassy-fiber separator, and the cathode were sequentially assembled with total 200 μL of ether-based (triglyme (