Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Mn(II) deposition on anodes and its effects on capacity fade in spinel lithium manganate–carbon systems

Mn(II) deposition on anodes and its effects on capacity fade in spinel lithium manganate–carbon... ARTICLE Received 9 Jan 2013 | Accepted 13 Aug 2013 | Published 30 Sep 2013 DOI: 10.1038/ncomms3437 Mn(II) deposition on anodes and its effects on capacity fade in spinel lithium manganate–carbon systems 1,2, 2, 2 3 2 4 Chun Zhan *, Jun Lu *, A. Jeremy Kropf , Tianpin Wu , Andrew N. Jansen , Yang-Kook Sun , 1 2 Xinping Qiu & Khalil Amine Dissolution and migration of manganese from cathode lead to severe capacity fading of lithium manganate–carbon cells. Overcoming this major problem requires a better under- standing of the mechanisms of manganese dissolution, migration and deposition. Here we apply a variety of advanced analytical methods to study lithium manganate cathodes that are cycled with different anodes. We show that the oxidation state of manganese deposited on the anodes is þ 2, which differs from the results reported earlier. Our results also indicate that a metathesis reaction between Mn(II) and some species on the solid–electrolyte interphase takes place during the deposition of Mn(II) on the anodes, rather than a reduction reaction that leads to the formation of metallic Mn, as speculated in earlier studies. The concentration of Mn deposited on the anode gradually increases with cycles; this trend is well correlated with the anodes rising impedance and capacity fading of the cell. 1 2 Key Laboratory of Organic Optoelectronics and Molecular Engineering, Department of Chemistry, Tsinghua University, Beijing 100084, China. Chemical Sciences and Engineering Division, Argonne National Laboratory, Argonne, IL 60439, USA. X-ray Science Division, Argonne National Laboratory, Argonne, IL 60439, USA. Department of Energy Engineering, Hanyang University, Seoul 133 791, South Korea. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to X.Q. (email: [email protected]) or to K.A. (email: [email protected]). NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 1 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 15,16,23 pinel LiMn O (LMO) has been considered as one of the electrode surface . Both aspects were believed to directly 2 4 most attractive cathode materials for rechargeable lithium- contribute to the capacity fading of the cell. Sion batteries in many respects, such as low cost, thermal However, as characterization tools are not yet able to detect the 1–3 safety and excellent rate capability . The electrochemical extremely low concentration of manganese in the electrolyte and, reactions in these rechargeable batteries are based on the in particular, on the anode, there is not yet much detailed evidence reversible insertion and extraction of lithium ions between to pinpoint the exact nature of the reaction at the anode surface. active materials of the LMO cathode and carbon anode . Along Thus, there has been considerable speculation about the reactions with these reversible reactions, however, dissolution and at the anode surface. For instance, it has been speculated that Mn migration of manganese species from cathode to anode often was reduced and deposited on the graphite anode as metallic Mn, occur and are generally attributed to HF acid generated by as stated earlier, which consequently catalysed the decomposition fluorine from fluorinated anion (PF ) and protons from water of the electrolyte and/or modified the SEI to produce a high- 5 23 impurities. This irreversible process is believed to be associated impedance layer on the surface of the carbon anode . This with the severe capacity fading of the cells, in particular, when the conclusion was mainly based on the finding that the potential of cells are operated at elevated temperatures (455 C). In fact, the the lithium-intercalated graphite anode (o0.3 V versus Li/Li )is migration of transition metals from other cathode materials has much lower than the standard redox potential of Mn(II)/Mn also been observed—for instance, LiFePO (ref. 6) and Li Ni (1.87 V versus Li/Li ). As a result, Mn(II) may be 4 x 1/ 7 16 Co Mn O (NCM) —which likely leads to capacity fading of electrochemically reduced by the low anode potential or 3 1/3 1/3 2 the batteries, especially during the aging process. Overcoming this chemically reduced by lithiated graphite . Although energy- major problem requires a better understanding of mechanisms of dispersive spectroscopy , dynamic secondary ion mass 24 25 manganese dissolution, migration and deposition (DMD) on the spectrometry and atomic absorption spectrometry (AAS) anode. have proven the deposition of manganese species on anodes, In the past, researchers have proposed a mechanism to conclusive identification of the oxidation state of Mn deposited on interpret the DMD process for manganese-based cathodes, which anodes remains a big challenge, which needs to be addressed in can be briefly described in the following three steps: a order to fully understand the reaction mechanism at the anode disproportionation reaction of Mn(III) in LiMn O takes place surface and, as a consequence, the capacity fading of the batteries. 2 4 at the interface of the cathode and electrolyte, as shown in Here we apply several advanced characterization techniques to 3,5,8–14 equation (1) conclusively identify the oxidation state of manganese deposited on the anode surface. We are able to clearly demonstrate by using 3þ 2þ 4þ 2Mn ! Mn þ Mn : ð1Þ X-ray absorption spectroscopy (XAS) along with XPS that, surprisingly, the oxidation state of manganese deposited on the 4þ Consequently, Mn remains on the cathode surface because of anodes is þ 2, which differs from most results reported earlier. 2þ its insolubility in the electrolyte, whereas Mn dissolves into the The concentration of Mn deposited on the anode is found to 2þ electrolyte; the dissolved Mn species diffuse and migrate from increase gradually with increased cycles; moreover, this trend is 2þ the cathode to the anode. The Mn species are finally reduced well correlated with the anodes’ rising impedance. This work to metallic Mn on the anode surface. The third step is based on helps us to better understand the correlation between the the speculation in literatures that the lithiated anode has low manganese DMD process and capacity fading in Mn-based 15–19 chemical potential and high chemical activity . Although lithium-ion batteries and sheds some light on how to improve the 2þ 3þ there are a few studies that reported Mn /Mn being cell performance. deposited on the graphite anodes on the basis of X-ray photoelectron spectroscopy (XPS) measurements, these authors claimed that Mn was deposited as metal first (by reduction) and Results 20–21 then oxidized to Mn(II) , similar to the third step described Electrochemical performance and manganese deposition.To above. Nevertheless, as a consequence of the manganese DMD determine the amount of manganese deposited on the anode, a 3þ process, the loss of active material (Mn ) and the formation of Li/LMO half-cell was tested at a C/2 rate cycled between 3.5 and inactive spinel on the cathode surface because of the 4.3 V at room temperature. Figure 1a shows the normalized disproportionation reaction increases the impedance of the capacity versus cycle number for the Li/LMO cell. The data 11,22 cathode . The solid–electrolyte interphase (SEI) at the anode clearly indicate that the cell capacity is continuously fading to is also poisoned because of the deposition of Mn (as metal) on the about 85% of initial value after 100 cycles in agreement ab Capacity of LiMn O /Li cells 2 4 1.0 Concentration of Mn on Li anode 0.9 95 300 0.8 LiMn O versus dilithiated LiFePO 2 4 4 LiMn O versus Li Ti O 0.7 2 4 4 5 12 85 100 LiMn O versus graphite 2 4 LiMn O versus Li 2 4 80 0 0.6 0 25 50 75 100 0 20406080 100 Cycle number Cycle number Figure 1 | Cycle performance and ICP-AAS. (a) Cycle performance of Li/LiMn O half-cell (black curve) and concentration of Mn deposited on Li 2 4 anodes harvested after different charge–discharge cycles (red curve). (b) Cycle performance of coin cells with LiMn O cathodes versus anodes with 2 4 different Li-intercalation potentials. 2 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. –1 Capacity (mAh g ) Mn concentration on Li anode (p.p.m.) Normalized capacity NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ARTICLE 26 15–17,19–21 with previously reported data for similar materials . The was reduced to metallic Mn state . Moreover, our concentrations of Mn deposited on the Li anodes after different XANES data clearly suggest that the Mn oxidation state deposited cycles were measured using ICP-AAS (inductively coupled on the anodes does not depend on their chemical potential and plasma-atomic absorption spectrometry), and the results are chemical activity during the discharge/charge process. In addition also presented in Fig. 1a. A significant amount of Mn (B85 to the edge position, the fluorescence intensity of XANES can be p.p.m.) is already deposited on the anode after the formation used to determine the relative concentration of the tested element cycles, and it gradually increases with the cycling. The con- in the compound. The amount of Mn deposited on the selected centration of Mn reaches up to about 400 p.p.m. after 100 cycles anodes was calculated based on the Mn concentration on the Li of charge and discharge. As reference, a fully assembled Li/LMO anode from the ICP-AAS analysis and the fluorescence intensities cell was stored under the same condition as that used when the in Fig. 2a. After the 100 cycle test, about 320 p.p.m. of Mn is other cells were cycled. No Mn is detected on the un-cycled Li deposited on the mesocarbon microbead (MCMB) anode, anode, indicating that the DMD process does not occur in un- 260 p.p.m. on the D-LFP anode and 300 p.p.m. on the LTO anode cycled cell. As the ICP-AAS analysis provides accurate results for compared with 400 p.p.m. on the Li anode as seen in Fig. 1a. Mn concentration deposited on the anode (because the lithium Surprisingly, a large amount of Mn is deposited on the LTO and metal can be completely dissolved during the ICP-AAS mea- D-LFP anodes after 100 cycles. Figure 2b shows the XANES data surement), we adopt these data as baseline to obtain Mn con- for MCMB anodes collected after different cycles. The fluores- centration on the other anodes using the fluorescence intensity of cence intensity of the line at about 6,545 eV (1s-4p transition) X-ray absorption near edge spectra (XANES), as presented later. increases systematically with increasing cycle number, indicating 2þ Nevertheless, the half-cell results clearly suggest a correlation that Mn species continue to deposit on the anode surface between manganese deposition on the anode and capacity fading during the cycling tests. in LMO-based cells. Following the main edge, the X-ray absorption spectra show Figure 1b compares the cycle performance of cells with typical oscillations because of a scattering of the outgoing electron LiMn O cathodes versus different anodes with Li-insertion wave at the nearest-neighbour atoms. After a standard data 2 4 potentials ranging from 0 to 3.5 V (versus Li/Li ). The curves treatment (background reduction, normalization, conversion to k show that the LiMn O /Li Ti O (LTO) cell has almost no and m(0) fit), these waves yield the w(k) function and the w(k) k 2 4 4 5 12 capacity fade after 100 cycles, and the LiMn O /delithiated function. The later function and its Fourier transform (FT) 2 4 LiFePO (D-LFP) cell loses about 5% capacity after 100 cycles. contain valuable information about the nearest-neighbour dis- The curves also indicate that LiMn O exhibits much poorer tances, the coordination number and the coordination geometry. 2 4 cycling performance when cycled against graphite, losing 415% Figure 3a shows the FT of the w(k) k functions of all capacity after 100 cycles. Characterization of Mn deposition on investigated samples in R-space, where R is the Fourier conjugate those anodes could help to better understand the difference in to k, uncorrected for phase shifts, related to the distance to the capacity-fading performance. nearest-neighbour shells. The positions of the peaks in the Fourier-transformed w(k) k functions depend on the real radius of the scattering shells around the X-ray-absorbing atoms (that is, Oxidation state of Mn deposited on anodes. The Mn oxidation Mn). The intensities of these peaks depend on the scattering state on different anodes collected after 100 cycles was deter- amplitude, the coordination number and the Debye–Waller factor mined from XANES spectra recorded at room temperature. (s , the mean square disorder in the path lengths). Using feff 28 29 (version 9) and ifeffit we have been able to fit the region of the XANES reference spectra were recorded for Mn foil, MnO and 1 2 Mn O standards, as shown in Fig. 2a, whereas a more extensive FT from R¼ 1–2.3 Å (k¼ 2.75–10 Å , k weighting, modified 2 3 Hanning windows with width 1 Å ). A single Mn-O-scattering collection of standard Mn spectra has been made available by Manceau et al. The edge position of XANES is typically used to path suffices for three of the samples (D-LPF-100, LTO-100 and MCMB-100). For Li-100 a single scattering path no longer determine the oxidation states of the element in the compound. On the basis of the XANES results shown in Fig. 2a and the suffices, corresponding to the less distinct white line in the XANES standards available from Manceau et al. , it is clear that the spectrum (Fig. 2a). For the three samples, the Mn-O bond lengths oxidation state of Mn deposited on all of the anodes used in this are all within 2.15 0.03 Å. This bond length is slightly shorter study is predominately þ 2. This result is surprising because it than that for MnO (manganosite: R ¼ 2.222 Å) but is Mn-O has been speculated in most of the earlier publications that Mn generally consistent with the range of Mn(II)-O bond lengths . ab 1.8 0.030 Mn-metal MCMB-100 MnO 1.6 MCMB-50 Mn O 0.025 2 3 MCMB-25 1.4 D-LFP-100 MCMB-initial Li-100 1.2 0.020 MCMB-100 1.0 LTO-100 0.015 0.8 0.6 0.010 0.4 0.005 0.2 0.0 0.000 6,520 6,530 6,540 6,550 6,560 6,570 6,520 6,540 6,560 6,580 6,600 Photon energy (eV) Photon energy (eV) Figure 2 | X-ray absorption near edge spectra. (a) Normalized XANES of Mn K-edge for commercial MnO, Mn O , Mn foil and different anode samples 2 3 collected after 100 cycles, and (b) unnormalized XANES of Mn K-edge for MCMB anode samples collected after different cycles. NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 3 & 2013 Macmillan Publishers Limited. All rights reserved. Magnitude (a.u.) Magnitude (a.u.) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ab 2.0 2p 3/2 Mn 2p Li-100 2p 1/2 LTO-100 High B. E. 1.5 MCMB-100 LTO-100 D-LFP-100 1.0 D-LFP-100 0.5 MCMB-100 0.0 012 345678 660 655 650 645 640 635 R (Å) Binding energy (eV) Figure 3 | Extended X-ray absorption fine structure spectroscopy and XPS. (a) Fourier transforms of the absorption spectra (w(k) k function) for different anode samples collected after 100 cycles. (b) The Mn 2p XPS spectra of the selected anodes collected after 100 cycles. To further study the anode surface, XPS measurements were re-oxidized to Mn(II) or Mn(III) subsequently by the electrolyte. performed on different anodes collected after 100 cycles. Unlike In contrast, in our work, we found that the Mn deposition can the XANES technique, XPS is more surface-sensitive (measured occur on anodes with potential as high as 3.5 V (D-LFP anode) depth is about 1–3 nm); therefore, it is an ideal tool to analyse the versus Li/Li , and hence we believe that the Mn deposition 31–32 SEI on the anodes . The Mn 2p, Li 1s, C 1s, O 1s and F 1s was not a reduction reaction driven by the low potential of the XPS spectra were recorded and fitted. Figure 3b presents the Mn anodes but an ion-exchange reaction (metathesis reaction) driven 2p core level XPS spectra of the MCMB, LTO and D-LFP anodes by the solubility of the species involved. This is further confirmed after 100 cycles. The analysis of Mn 2p XPS spectra of all the by the observation of Mn(II) on delithiated carbon electrode 2þ samples shows that Mn is deposited on the anodes as Mn , (with SEI formed on the surface) after simply immersing it in the which is consistent with the XAS results discussed earlier. In electrolyte containing Mn(II) dissolved from LiMn O powder 2 4 addition, two more interesting features are noted from the Mn 2p (Supplementary Fig. S2). In addition, the SEI layer starts to form XPS spectra. First, the Mn 2p peak slightly shifts to higher on the carbon anode during the first charge. Therefore, if Mn(II) 3/2 binding energy from MCMB to D-LFP to LTO, indicating that in the electrolyte would be reduced on the carbon surface, it has Mn is bonded to a more electron-negative element in LTO than to diffuse through the SEI layer to reach the carbon surface to be in MCMB. In other words, the Mn-containing species that reduced by lithiated graphite particle, which is kinetically deposited on selective anodes are different, although they have unfavourable. Therefore, metathesis reaction of Mn(II) with the same oxidation state (þ 2). Second, according to the peak species in the SEI layer, which directly leads to the formation of amplitudes, the MCMB anodes have less Mn deposited, whereas Mn(II) on the anodes, most likely is the reaction that takes place LTO has much more Mn deposited on the surface. As the SEI in on the anode surface during the Mn deposition, rather than a LTO and MCMB are different, it is conceivable to expect that the reduction reaction that leads to the formation of metallic Mn, as 15–22 products resulting from the reaction between these SEIs and reported in most earlier studies . At this point, we are unable 2þ dissolved Mn in the electrolyte are different, as confirmed to pinpoint the exact metathesis reactions that occur on the using XPS. Analysis of Li 1s, C 1s, O 1s and F 1s XPS spectra of all anodes owing to the lack of knowledge of the composition of the samples did indicate that the SEI formed on various anodes is the SEI formed on the various anodes; however, we are able to different (Supplementary Fig. S1). propose the reaction mechanism based on the impedance results From the XAS and XPS data presented above, we conclude that presented later. Nevertheless, it is clear from the XAS and XPS the oxidation state of Mn deposited on anodes is predominately results that different Mn(II)-containing species deposit on the þ 2, which is independent of the anode potentials. As far as we anodes during the Mn DMD process, which certainly depends on know, this is the first time that solid evidence has been provided the nature of the SEI on the anodes. for the identification of the Mn oxidation state during the Mn DMD process in LMO-based cells. It should be mentioned that 2þ there are a few studies that reported the observation of Mn / Correlation between Mn deposition and capacity fading. Two 3þ Mn deposited on the carbon anode, especially in the papers of questions remain: What is the correlation between the quantity of 20 21 Komaba et al. and Ochida et al. However, they did not Mn deposited on anodes and the rate of capacity fading, especially 2þ provide as conclusive results as we presented in this work on the in the graphite anode? How does the Mn act in the SEI of the oxidation state of Mn deposition in a real LMO/graphite cell, and graphite anode? To address these issues, we first employed a simple especially on the correlation between the Mn deposition and calculation to estimate the capacity fading contributed solely from capacity fade of the cell. First, significant amount (4100 p.p.m.) the loss of active materials in the cathode because of the DMD 2þ of Mn was pre-added to the electrolyte in their studies either process. Both the amount of Mn deposited on the anodes (based using Mn(ClO ) or introduced by anodic dissolution of Mn on ICP-AAS and XAS data) and that dissolved in the electrolytes 4 2 metal, which does not represent the real situation in the LMO/C (based on literature data) were accounted for in the calculation. 2þ cells in which the Mn concentration in the electrolyte The calculated capacity fading in the MCMB-LMO cell resulting 3þ increases gradually from zero because of the dissolution of Mn from the loss of active Mn (o1%) is significantly smaller from the LMO cathode. For instance, in the first discharge/charge than the overall capacity fading determined experimentally from 2þ cycle, the Mn concentration in the electrolyte is lower than the 100-cycle data (15%). For the LTO-LMO and D-LFP-LMO 8 2þ 5 p.p.m. . Second, they believe that Mn was deposited as metal cells, much less overall capacity fading is found, even with first by reduction of the low-potential lithiated carbon and then almost the same amount of Mn deposited on the anodes. 4 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. 2 –2 FT [k x χ(k )] (Å ) Relative intensity (a.u.) NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ARTICLE –12 100 Graphite anode stored in the pure electrolyte Storage time 6 h 12 h –8 24 h 48 h –4 0 5 10 15 20 25 Surface-modified LiMn O versus graphite 2 4 Z′ (Ω) Bare LiMn O versus graphite 2 4 –12 2+ Graphite anode stored in the electrolyte with Mn Storage time 0 20 40 60 80 100 6 h Cycle number 12 h –8 24 h Figure 5 | Cycle performance. Cycle performance of coin cells with 48 h graphite anodes versus LMO cathodes (bare LMO and surface-modified LMO with TiO ) at C/2 and 55 C. –4 2 of cells with graphite anode versus bare LMO and surface- 0 5 10 15 20 25 modified LMO cathodes. The curves indicate that bare LMO Z′ (Ω) exhibits much poorer cycling performance when cycled against Figure 4 | The AC impedance. The impedance of graphite anodes stored in graphite at 55 C, losing 450% capacity after 100 cycles, 2þ 2þ an electrolyte (a) without Mn and (b) with Mn . whereas only about 20% capacity loss is observed for the surface-modified LMO. The AC impedance rise of the graphite anode and bare LMO This finding suggests that the capacity fading in the MCMB-LMO cathode is shown in Fig. 6a,b, respectively. The cells were 3þ cell is not directly caused by the loss of the active Mn . A pos- discharged to 50% state of charge (SOC) before each impedance sible reason for the severe capacity fading in the MCMB-LMO cell testing. The impedance rise of the bare LMO cathode slows 2þ is that Mn in the electrolyte reacts with the SEI on the graphite down after 25 cycles (Fig.6b), whereas that of the graphite anode, which modifies the nature of the SEI. As a result, the anode increases significantly and contributes to almost all the 2þ impedance of the graphite anode could change as the Mn increase in the impedance of the full cell after 50 cycles (Fig. 6a). concentration increases in the SEI during the cycle test. Corroborating the capacity versus cycle number diagram for the 2þ To understand the correlation between the reaction of Mn MCMB-LMO cell shown in Fig. 5 and these impedance results, it in the electrolyte with the SEI and the impedance change of the appears that the impedance rise of the cathode is not responsible graphite anode in the LiMn O -based cells, we first compared the for the capacity loss. The trend of impedance rise from the anode 2 4 AC impedance rise of graphite anodes (with SEI formed) stored matches that of the capacity loss during the cycling test, 2þ in electrolyte with or without Mn by using a three-electrode indicating that the anode impedance is the key factor that causes configuration. Graphite anodes were first cycled against Li metal the capacity fading of the full cell. Figure 6c shows the AC for one cycle to form an SEI layer and then fully discharged to impedance rise of the graphite anode cycling against surface- 5 mV versus Li/Li . The fully discharged graphite anodes were modified LMO at different cycles, which clearly indicates that the disassembled and transferred to a three-electrode cell (with Li impedance of the graphite anode grows much slower than that of metal as counter electrodes and Li Sn micro-reference electrode) the graphite anode in the bare LMO cell (Fig. 6a). Similar results filled with either a pure electrolyte or an electrolyte with 20 p.p.m. were also reported for MCMB-Li [Ni Mn Co ] O cells . 1.1 1/3 1/3 1/3 0.9 2 2þ Mn . As shown in Fig. 4, the impedance of graphite anode does In addition, we also performed the same in situ AC impedance not change much when stored in the pure electrolyte but spectroscopy for the LTO anode after different times of cycling 2þ increases notably when stored in the electrolyte with Mn . This against bare LMO at C/2 rate (Supplementary Fig. S3). The 2þ result clearly indicates that Mn in the electrolyte reacts with impedance of the LTO anode grows much slower than that of the the SEI on the graphite anode and modifies the nature of the SEI graphite anode in the bare LMO cell (Fig. 6a), corresponding to accordingly, leading to the increase in the impedance of graphite much slower capacity fading than the LTO/LMO cell (Fig. 1b). anode. It is also worthwhile to note that the impedance of There is, thus, a strong correlation between impedance rise of the 2þ 2þ graphite anode in the electrolyte with Mn is stabilized after anode and the concentration of Mn in the SEI on the graphite 2þ 24 h, likely because of the depletion of Mn in the electrolyte. anode, as well as the capacity fading. In situ AC impedance spectroscopy in a three-electrode configuration was performed to probe the capacitive and resistive behaviour of the graphite anodes after different times of cycling Discussion against bare LMO at C/2 and 55 C. For comparison, similar As assumed in earlier reports, Mn deposition could change the measurement was also performed on the graphite anode cycling structure or composition of SEI on graphite anodes, resulting in against surface-modified LMO with TiO , which is expected and the capacity fading of the full cell . To elucidate the effect of Mn confirmed with the ICP-AAS measurement (Supplementary deposition on the electrochemical properties of the SEI layer, a Supplementary Table S1) to have much less Mn dissolved detailed theoretical analysis of the anode impedance was from the cathode during the cycling test. In other words, performed, as described in the Supplementary Note 1. The fitting the dissolution of Mn is considerably suppressed from the results (Supplementary Table S2) showed the increase of R C , SEI SEI surface-modified LMO. Figure 5 compares the cycle performance which indicates that the SEI layer on the graphite anode is not NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 5 & 2013 Macmillan Publishers Limited. All rights reserved. Z′′ (Ω) Z′′ (Ω) Normalized capacity (%) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ab –16 –35 –5 LiMn O cathode 2 4 Graphite anode –4 Before cycling –30 Before cycling 25th cycle –3 25th cycle –12 50th cycle 50th cycle –25 –2 75th cycle 75th cycle –1 100th cycle 100th cycle –20 0 –8 0 1234 5 Z’ (Ω) –15 –10 –4 –5 0 5 10 15 20 25 30 35 04 812 16 Z′ (Ω) Z′ (Ω) cd –25 Graphite versus surface modified LiMn O Graphite anodes versus 2 4 Bare LiMn O 2 4 –20 Surface-modified LiMn O Before cycling 2 4 25th cycle 50th cycle –15 75th cycle 100th cycle –10 –5 0 5 10 15 20 25 020 40 60 80 100 Z′ (Ω) Cycle number Figure 6 | The AC impedance. (a) The graphite anode and (b) bare LiMn O cathode after different times of cycling at C/2 and 55 C. (c)AC 2 4 impedance rise of the graphite anode cycling against surface-modified LMO. (d) Plots of R C of graphite anodes cycling against bare and surface- SEI SEI modified LMO at different cycles based on the fitting results in Supplementary Table S2. simply thickened but also undergoes change in its nature Ion exchange model behaviour during cycling of the cell. On the basis of the fitting results in Supplementary Table S2, the change of R C of SEI SEI graphite anodes cycling against both bare and surface-modified LMO cathodes is plotted in Fig. 6d. It is clearly seen that a SEI SEI dramatic increase in R C of graphite anode was observed for SEI SEI the cell containing bare LMO, whereas a much smaller increase (by a magnitude of 2) in R C of the graphite anode was SEI SEI observed for the cell containing surface-modified LMO. This is well correlated to their capacity fade, as shown in Fig. 5. Graphite Graphite According to the results presented above, we believe that the + + 2+ Mobile Li Immobile Li Solvent Mn capacity fading is caused by changes in the SEI on the graphite 2þ anode because of the Mn that continuously migrated from the Figure 7 | Ion-exchange model. Proposed ion-exchange model for the Mn cathode and thus the change of the impedance of the graphite deposition on the graphite anode during the Mn DMD process in the anode. graphite/LMO cell. Corroborating XANES/XPS and AC impedance results, we propose a metathesis reaction (ion exchange) mechanism for the speculated in earlier studies. The capacity fading is caused by Mn deposition on the carbon anode, as illustrated in Fig. 7. We changing of the SEI on the graphite anode because of the believe that the ion exchange during the Mn deposition is 2þ 2þ þ continuous reaction with the dissolved Mn from the cathode. between Mn and mobile Li in the SEI—for example, lithium alkyl carbonate (ROCO Li), which blocks the lithium diffusion path—and thus, leads to the increase in R C of graphite SEI SEI Methods anodes (Fig. 6d). This is also confirmed by the absence of MnF LiMn O synthesis and characterization. Stoichiometric LiMn O particles were 2 2 4 2 4 synthesized via a solid-state method with Li CO (Sigma-Aldrich, 99.0%) and and MnCO on the SEI, which would result from the ion 2 3 2þ þ chemical MnO (Chemetal, 99.5%), which were mixed and ground thoroughly in a exchange between Mn and the immobile Li in LiF and molar ratio of 1.05: 4. The mixture was then heated at 800 C for 16 h. The Li CO on the SEI otherwise. 2 3 structure of the as-prepared spinel LiMn O was confirmed using X-ray powder 2 4 In summary, we demonstrated in this study that the oxidation diffraction and scanning electron microscopy (Supplementary Supplementary Figs. S4–S6 and Supplementary Note S1). The surface-modified LiMn O powders with state of manganese deposited on the anodes is þ 2 on the basis of 2 4 TiO was prepared as follows: Tetrabutyl titanate (Sigma-Aldrich, 99%) was first XAS and XPS analyses. The results also suggest a metathesis dissolved in ethanol to make a 10% solution, and then the solution was slowly 2þ reaction between Mn and some species on the SEI during the dropped into acetic acid solution (which contained 10% acetic acid and 56% deposition of Mn on the anodes, rather than a reduction reaction ethanol) with agitation to obtain the sol, which was further diluted to 5% by adding that leads to the formation of metallic Mn otherwise, as ethanol to prevent the aggregation of the colloid. Then the LiMn O powders were 2 4 6 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. Z′′ (Ω) Z′′ (Ω) Z’’ (Ω) R C (mF Ω) Z′′ (Ω) SEI SEI NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ARTICLE put into the diluted sol, agitated for 20 min, followed by drying at 80 C and Analysis of impedance spectra. According to the model first reported by Thomas 38 39 calcination at 750 C for 4 h. The resulting molar ratio of TiO :LiMn O in the et al. and further improved by Barsoukov et al. , we used the equivalent circuit 2 2 4 surface-modified LiMn O is B5%. (Supplementary Fig. S7) for interpreting our anode impedance spectra, where R is 2 4 s the resistance of the electrolyte, separator and wires; R and C represent the ct dl charge transfer resistance and double-layer capacitance at the electron/ion con- duction boundary for the Li-insertion reactions; R and C represent the SEI SEI Electrochemistry tests. 2032-type coin cells were assembled with a cathode resistance and the geometric capacitance of the SEI layer; and Z is the Warburg material of 85% LiMn O , 10% carbon black and 5% polyvinylidene difluoride 2 4 diffusion impedance of the Li ion into the bulk particles via solid-phase diffusion. (PVDF) binder. The GenII electrolyte was composed of 1.2 M LiPF in ethylene As shown in Fig. 5a, the impedance of the graphite electrode typically includes carbonate: ethyl methyl carbonate at a 3:7 ratio by weight. Anodes with different two depressed semicircles in the high-frequency region and a line in the low- Li-insertion potentials were tested, including 0 V (Li metal), 0.02 V (92% MCMB frequency region, which can be fitted to two serially connected parallel R elements ct graphite and 8% PVDF), 1.55 V (80% Li Ti O , 10% carbon black and 10% PVDF) 4 5 12 and the Warburg element, respectively, in the equivalent circuit. The first and 3.5 V (85% LiFePO , 10% carbon black and 5% PVDF, fully charged before semicircle (in higher-frequency region, as shown in the insertion of Fig. 5a) being used as anode). If not otherwise specified, the cells were first cycled with a remains almost the same from the 25th to 100th cycle, whereas the second one (at constant current of C/10. After the formation cycles, the cells were cycled at a lower-frequency region) enlarges constantly with the cycle time. As R and C ct dl constant current of C/2. Specific voltage windows (depending on the anode) were represent the Li-intercalation reaction impedance, which is mainly determined by selected in order to charge and discharge the LiMn O cathodes between 3.5 and 2 4 intrinsic properties of the active material, electrode SOC and temperature, it is þ 2þ 4.3 V (versus Li/Li ). The electrolyte with Mn was prepared by soaking 0.1 g reasonable to assume that R and C tested here remain constant with the cycling ct dl LiMn O in 10 ml GenII electrolyte in a sealed container at 55 C for 14 days. 2 4 number. On the other hand, the SEI layer will get thicker with the cycling of the 2þ The concentration of Mn in electrolyte was about 20 p.p.m. from the ICP-AAS cell, thereby changing R and C . We, therefore, assigned the first semicircle to SEI SEI analysis. the R C element and the second one to the R C element. The fitting results ct dl SEI SEI The AC impedance of cycled anodes and cathode was determined by using are shown in Supplementary Table S1. The deviation of the spectra in the high- 35–37 reference electrode cells, as described in detail elsewhere . The test cell was frequency region (inset to Fig. 5a) can be attributed to electron conduction and assembled with flat, single-sided, 32-cm electrodes. A micro-reference electrode 40 Li-ion diffusion in the porous electrode . was prepared from a 25-mm-diameter copper wire coated with 1-mm-thick tin and 41 According to the SEI model , R and C can be calculated based on the SEI SEI covered with a polyurethane coating. The reference wire was shorted to the following equations: R ¼ rl/S and C ¼ eS/l, where r and e are the specific SEI SEI lithiated anode, which caused the transfer of some lithium from the anode to Sn to resistance and permittivity of the layer, respectively, and l and S are the real make a Li Sn electrode. The reference electrode cells were first charged at a rate of electrochemically active thickness and area, respectively. The increase in R and SEI C/10 and cycled between 3.0 and 4.3 V at C/2 before being constant-voltage- decrease in C might be related to the thickening of the SEI layer, whereas r and e SEI charged to 50% SOC. The AC impedance data for the anode and cathode were could change with cycling as well. Therefore, R C ¼ re is selected here as a SEI SEI collected with a Solartron 1470 E and 1451A cell test system using a 5-mV AC reliable value to indicate the intrinsic electrochemical properties of the SEI layer perturbation with frequency ranging from 1 MHz to 20 MHz. determined by its composition, although independent of its geometry. The fitting results showed the increase in R C , which indicates that the SEI layer on the SEI SEI anode is not simply thickened but also undergoes change in its nature behaviour during cycling of the cell. Characterization of Mn deposition. The disassembled anodes were rinsed with a mixture of ethylene carbonate/ethyl methyl carbonate (3:7 by weight) to remove the residual electrolyte remaining on the anode surface. To avoid any contamination to the surface, this procedure was carefully performed in an argon- References filled glove box. Inductively coupled plasma (ICP) was applied to analyse the 1. Thackeray, M. M., David, W. I. F., Bruce, P. G. & Goodenough, J. B. Lithium amount of Mn deposited on the Li metal anode, as Li can be thoroughly dissolved insertion into manganese spinels. Mater. Res. Bull. 18, 461–472 (1983). by acid and hence provide reliable Mn concentration. 2. Guyomard, D. & Tarascon, J. M. Li metal-free rechargeable LiMn O /carbon 2 4 The oxidation states of Mn deposited on anodes were characterized by using cells their understanding and optimization. J. Electrochem. Soc. 139, 937–948 X-ray absorption spectroscopy (XAS) and XPS. For the XAS measurements, the (1992). samples are protected by Kapton tape to prevent the contamination from the air. 3. Gummow, R. J., Dekock, A. & Thackeray, M. M. Improved capacity retention Mn K-edge XAS measurements were conducted on both the bending magnet in rechargeable 4V lithium/lithium manganese oxide (spinel) cells. Solid State (10BM) and insertion device (10ID) beamlines of the Materials Research Ionics 69, 59–67 (1994). Collaborative Access Team in the Advanced Photon Source at Argonne National 4. Lazzari, M. & Scrosati, B. Cyclable lithium organic electrolyte cell based on 2 Laboratory. Harmonic rejection was accomplished with an Rh-coated mirror. Anode spectra were measured on 10BM in the fluorescence mode using a four- intercalation electrodes. J. Electrochem. Soc. 127, 773–774 (1980). element silicon drift detector (SII NanoTechnology USA Inc., Vortex-ME4). A 5. Amatucci, G. G. et al. Materials’ effects on the elevated and room temperature third detector in the series simultaneously collected a metallic Mn reference performance of C/LiMn O Li-ion batteries. J. Power Sources 69, 11–25 (1997). 2 4 spectrum with each measurement for energy calibration. The peak of the first 6. Aurbach, D. et al. Review on electrode-electrolyte solution interactions, related derivative was set to 6,537.7 eV. to cathode materials for li-ion batteries. J. Power Sources 165, 491–499 (2007). Samples were also analysed with XPS using a Kratos Axis Ultra DLD surface 7. Zheng, H., Sun, Q., Liu, G., Song, X. & Battaglia, V. S. Correlation between analysis instrument. The base pressure of the analysis chamber during these dissolution behavior and electrochemical cycling performance for 10  9 experiments was 3  10 torr, with operating pressures around 1 10 torr. LiNi Co Mn O -based cells. J. Power Sources 207, 134–140 (2012). 1/3 1/3 1/3 2 Spectra were collected with a monochromatic Al Ka source (1,486.7 eV) and a 8. Jang, D. H., Shin, Y. J. & Oh, S. M. Dissolution of spinel oxides and capacity 300 700-m spot size. For survey spectra, the data were collected at a pass energy of losses in 4V Li/LiMn O cells. J. Electrochem. Soc. 143, 2204–2211 (1996). 2 4 160 eV (fixed analyser transmission mode), a step size of 1 eV and a dwell time of 9. Jang, D. H. & Oh, S. M. Electrolyte effects on spinel dissolution and cathodic 200 ms. High-resolution regional spectra were collected with a pass energy of 20 eV capacity losses in 4V Li/LiMn O rechargeable cells. J. Electrochem. Soc. 144, 2 4 (fixed analyser transmission mode), a step size of 0.1 eV and a dwell time of 300 ms. 3342–3348 (1997). For low signal-to-noise regions, multiple passes were made and the results were 10. Xia, Y., Zhou, Y. & Yoshio, M. Capacity fading on cycling of 4V Li/LiMn O 2 4 averaged together. cells. J. Electrochem. Soc. 144, 2593–2600 (1997). Prior to their introduction into the load-lock vacuum chamber of the XPS 11. Choa, J. & Thackeray, M. M. Structural changes of LiMn O spinel electrodes 2 4 instrument, all air-sensitive samples were loaded into an inert transfer module during electrochemical cycling. J. Electrochem. Soc. 146, 3577–3581 (1999). interfaced with the instrument. Samples were prepared for analysis in an Ar-filled 12. Gnanaraj, J. S., Pol, V. G., Gedanken, A. & Aurbach, D. Improving the high- glove box, with no more than 1 p.p.m. O and 1 p.p.m. H O. Nonconductive 2 2 temperature performance of LiMn O spinel electrodes by coating the active 2 4 samples showed evidence of differential charging, resulting in peak shifts and mass with mgo via a sonochemical method. Electrochem. Commun. 5, 940–945 broadening. Photoelectron peak positions were shifted back towards their true (2003). values, and their peak widths were minimized by flooding the samples with low- 13. Yu, L. H., Qiu, X. P., Xi, J. Y., Zhu, W. T. & Chen, L. Q. Enhanced high- energy electrons and ions from the charge neutralizer system on the instrument. potential and elevated-temperature cycling stability of LiMn O cathode by 2 4 Peak position was further corrected by referencing the C 1s peak position of TiO modification for Li-ion battery. Electrochim. Acta. 51, 6406–6411 (2006). adventitious carbon for a sample (284.8 eV, PHI Handbook of Photoelectron 14. Walz, K. A. et al. Elevated temperature cycling stability and electrochemical Spectroscopy) and by shifting all other peaks in the spectrum accordingly. impedance of LiMn O cathodes with nanoporous ZrO and TiO coatings. 2 4 2 2 Fitting was carried out by using the program CasaXPS. Peaks were fit as J. Power Sources 195, 4943–4951 (2010). asymmetric Gaussian/Lorentzians, with 0–30% Lorentzian character. The full 15. Blyr, A. et al. Self-discharge of LiMn O /C Li-ion cells in their discharged state. 2 4 width at half maximum of all subpeaks was constrained to 0.7–2 eV, as dictated by J. Electrochem. Soc. 145, 194–209 (1998). instrumental parameters, lifetime broadening factors and broadening because of 16. Komaba, S., Kumagai, N. & Kataoka, Y. Influence of manganese(II), cobalt(II), sample charging. With this native resolution set, peaks were added, and the best fit, and nickel(II) additives in electrolyte on performance of graphite anode for using a least-square fitting routine, was obtained while adhering to the constraints lithium-ion batteries. Electrochim. Acta. 47, 1229–1239 (2002). mentioned above. NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 7 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 17. Komaba, S., Itabashi, T., Kaplan, B., Groult, H. & Kumagai, N. Enhancement of 37. Dee, D. W., Jansen, A. N. & Abraham, D. P. Theoretical examination of Li-ion battery performance of graphite anode by sodium ion as an electrolyte reference electrodes for Li-ion cells. J. Power Sources 174, 1001–1006 (2007). additive. Electrochem. Commun. 5, 962–966 (2003). 38. Thomas, M. G. S. R., Bruce, P. G. & Goodenough, J. B. AC impedance analysis 18. Komaba, S. et al. Inorganic electrolyte additives to suppress the degradation of of polycrystalline insertion electrodes: application to Li CoO . J. Electrochem. 1-x 2 graphite anodes by dissolved Mn(II) for lithium-ion batteries. J. Power Sources Soc. 132, 1521–1528 (1985). 119-121, 378–382 (2003). 39. Barsoukov, E., Kim, J. H., Kim, J. H., Yoon, C. O. & Lee, H. Kinetics of lithium 19. Whittingham, M. S. Lithium batteries and cathode materials. Chem. Rev. 104, intercalation into carbon anodes: in situ impedance investigation of thickness 4271–4301 (2004). and potential dependence. Solid State Ionics 116, 249–261 (1999). 20. Komaba, S. et al. Impact of 2-Vinylpyridine as electrolyte additive on surface 40. Barsoukov, E. & Macdonald, J. R. Impedance Spectroscopy: Theory, Experiment, and electrochemistry of graphite for C/LiMn O Li-Ion Cells. J. Electrochem. and Applications 2nd edn (Blackwell Science Public, Osney Mead, 2005). 2 4 Soc. 152, A937–A946 (2005). 41. Peled, E., Golodnitsky, D., Ardel, G. & Eshkenazy, V. The SEI model - 21. Ochida, M. et al. Influence of manganese dissolution on the degradation of application to lithium polymer electrolyte batteries. Electrochim. Acta. 40, surface films on edge plane graphite negative-electrodes in Lithium-Ion 2197–2204 (1995). Batteries. J. Electrochem. Soc. 159, A961–A966 (2012). 22. Hirayama, M. et al. Dynamic structural changes at LiMn O /electrolyte 2 4 Acknowledgements interface during lithium battery reaction. J. Am. Chem. Soc. 132, 15268–15276 (2010). J.L. was supported by the Department of Energy (DOE) Office of Energy Efficiency and 23. Amine, K. et al. Improved lithium manganese oxide spinel/graphite Li-ion cells Renewable Energy (EERE) Postdoctoral Research Award under the EERE Vehicles for high-power applications. J. Power Sources 129, 14–19 (2004). Technology Program administered by the Oak Ridge Institute for Science and Education (ORISE) for the DOE managed by Oak Ridge Associated Universities (ORAU) under 24. Abraham, D. P., Spila, T., Furczon, M. M. & Sammann, E. Evidence of transition-metal accumulation on aged graphite anodes by SIMS. Electrochem. DOE contract number DE-AC05-06OR23100. K.A. and A.J.K. (X-ray absorption studies) Solid State 11, A226–A228 (2008). were supported by the Center for Electrical Energy Storage, an Energy Frontier Research 25. Tsunekawa, H. et al. Capacity fading of graphite electrodes due to the Center funded by the U.S. Department of Energy, Office of Science, Office of Basic deposition of manganese ions on them in Li-ion batteries. J. Electrochem. Soc. Energy Sciences. A.N.J. was supported by the Applied Battery Research for Transpor- 149, A1326–A1331 (2002). tation (ABR) Program from the U.S. DOE-EERE Office of Vehicle Technologies. Use of 26. Sun, Y.-K., Hong, K.-J. & Prakash, J. The effect of ZnO coating on the Advanced Photon Source, an Office of Science User Facility operated for DOE, Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under electrochemical cycling behavior of spinel LiMn O cathode materials at 2 4 elevated temperature. J. Electrochem. Soc. 150, A970–A972 (2003). Contract No. DE-AC02-06CH11357. MRCAT operations are supported by the DOE 27. Manceau, A., Marcus, M. A. & Grangeon, S. Determination of Mn valence and the MRCAT member institutions. Financial support from the 973 Program states in mixed-valent manganates by XANES spectroscopy. Am. Mineral. 97, (2009CB220105) of China, Beijing Natural Science Foundation (2120001), National 816–827 (2012). Natural Science Foundation of China (21273129) and Bosch (China) Ltd. is gratefully 28. Rehr, J. J. & Albers, R. C. Theoretical approaches to X-ray absorption fine acknowledged. This work was also supported by the Human Resources Development structure. Rev. Mod. Phys. 72, 621–654 (2000). of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean government, Ministry of Trade, Industry and Energy 29. Ravel, B. & Newville, M. Athena, artemis, hephaestus: data analysis for X-ray absorption spectroscopy using ifeffit. J. Synchrotron. Radiat. 12, 537–541 (No. 20124010203310) and by the National Research Foundation of Korea (NRF) (2005). grant funded by the Korea government (MEST) (No. 2009-0092780). 30. Pacalo, R. E. & Graham, E. K. Pressure and temperature-dependence of the elastic properties of synthetic MnO. Phys. Chem. Miner. 18, 69–80 (1991). 31. Peled, E. et al. Composition, depth profiles and lateral distribution of materials Author contributions in the SEI built on HOPG-TOF SIMS and XPS studies. J. Power Sources 97-98, J.L. designed the experiments; C.Z. synthesized the cathode materials, performed and analysed the electrochemical experiments, A.J.K., J.L. and T.W. performed and analysed 52–57 (2001). 32. Aurbach, D. et al. Recent studies on the correlation between surface chemistry, the X-ray absorption spectroscopy measurements; J.L. performed and analysed the XPS morphology, three-dimensional structures and performance of Li and Li-C experiments; A.N.J. fabricated the three-electrode cells for impedance measurements; X.Q. and K.A. supervised the project; J.L., C.Z. and K.A. wrote the paper. All of the intercalation anodes in several important electrolyte systems. J. Power Sources 81-82, 91–98 (1997). authors discussed the results and reviewed the manuscript. 33. Amine, K. et al. Mechanism of capacity fade of MCMB/Li (Ni Mn Co 1.1 1/3 1/3 1/ ) O cell at elevated temperature and additives to improve its cycle life. 3 0.9 2 Additional information J. Mater. Chem 21, 17754–17759 (2011). 34. Amatucci, G., Du Pasquier, A., Blyr, A., Zheng, T. & Tarascon, J. M. The Supplementary Information accompanies this paper on http://www.nature.com/ elevated temperature performance of the LiMn O /C system: failure and naturecommunications 2 4 solutions. Electrochim. Acta. 45, 255–271 (1999). Competing financial interests: The authors declare no competing financial interests. 35. Abraham, D. P., Poppen, S. D., Jansen, A. N., Liu, J. & Dees, D. W. Application of a lithium-tin reference electrode to determine electrode contributions to Reprints and permission information is available online at http://npg.nature.com/ impedance rise in high-power lithium-ion cells. Electrochim. Acta. 49, reprintsandpermissions/ 4763–4775 (2004). 36. Jansen, A. N., Dee, D. W., Abraham, D. P., Amine, K. & Henriksen, G. L. Low- How to cite this article: Zhan, C. et al. Mn(II) deposition on anodes and its effects on temperature study of lithium-ion cells using a Li Sn micro-reference electrode. capacity fade in spinel lithium manganate–carbon systems. Nat. Commun. 4:2437 J. Power Sources 174, 372–379 (2007). doi: 10.1038/ncomms3437 (2013). 8 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Communications Springer Journals

Mn(II) deposition on anodes and its effects on capacity fade in spinel lithium manganate–carbon systems

Download PDF
8 pages

Loading next page...
 
/lp/springer-journals/mn-ii-deposition-on-anodes-and-its-effects-on-capacity-fade-in-spinel-4ln7R4eVex

References (44)

Publisher
Springer Journals
Copyright
Copyright © 2013 by Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.
Subject
Science, Humanities and Social Sciences, multidisciplinary; Science, Humanities and Social Sciences, multidisciplinary; Science, multidisciplinary
eISSN
2041-1723
DOI
10.1038/ncomms3437
Publisher site
See Article on Publisher Site

Abstract

ARTICLE Received 9 Jan 2013 | Accepted 13 Aug 2013 | Published 30 Sep 2013 DOI: 10.1038/ncomms3437 Mn(II) deposition on anodes and its effects on capacity fade in spinel lithium manganate–carbon systems 1,2, 2, 2 3 2 4 Chun Zhan *, Jun Lu *, A. Jeremy Kropf , Tianpin Wu , Andrew N. Jansen , Yang-Kook Sun , 1 2 Xinping Qiu & Khalil Amine Dissolution and migration of manganese from cathode lead to severe capacity fading of lithium manganate–carbon cells. Overcoming this major problem requires a better under- standing of the mechanisms of manganese dissolution, migration and deposition. Here we apply a variety of advanced analytical methods to study lithium manganate cathodes that are cycled with different anodes. We show that the oxidation state of manganese deposited on the anodes is þ 2, which differs from the results reported earlier. Our results also indicate that a metathesis reaction between Mn(II) and some species on the solid–electrolyte interphase takes place during the deposition of Mn(II) on the anodes, rather than a reduction reaction that leads to the formation of metallic Mn, as speculated in earlier studies. The concentration of Mn deposited on the anode gradually increases with cycles; this trend is well correlated with the anodes rising impedance and capacity fading of the cell. 1 2 Key Laboratory of Organic Optoelectronics and Molecular Engineering, Department of Chemistry, Tsinghua University, Beijing 100084, China. Chemical Sciences and Engineering Division, Argonne National Laboratory, Argonne, IL 60439, USA. X-ray Science Division, Argonne National Laboratory, Argonne, IL 60439, USA. Department of Energy Engineering, Hanyang University, Seoul 133 791, South Korea. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to X.Q. (email: [email protected]) or to K.A. (email: [email protected]). NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 1 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 15,16,23 pinel LiMn O (LMO) has been considered as one of the electrode surface . Both aspects were believed to directly 2 4 most attractive cathode materials for rechargeable lithium- contribute to the capacity fading of the cell. Sion batteries in many respects, such as low cost, thermal However, as characterization tools are not yet able to detect the 1–3 safety and excellent rate capability . The electrochemical extremely low concentration of manganese in the electrolyte and, reactions in these rechargeable batteries are based on the in particular, on the anode, there is not yet much detailed evidence reversible insertion and extraction of lithium ions between to pinpoint the exact nature of the reaction at the anode surface. active materials of the LMO cathode and carbon anode . Along Thus, there has been considerable speculation about the reactions with these reversible reactions, however, dissolution and at the anode surface. For instance, it has been speculated that Mn migration of manganese species from cathode to anode often was reduced and deposited on the graphite anode as metallic Mn, occur and are generally attributed to HF acid generated by as stated earlier, which consequently catalysed the decomposition fluorine from fluorinated anion (PF ) and protons from water of the electrolyte and/or modified the SEI to produce a high- 5 23 impurities. This irreversible process is believed to be associated impedance layer on the surface of the carbon anode . This with the severe capacity fading of the cells, in particular, when the conclusion was mainly based on the finding that the potential of cells are operated at elevated temperatures (455 C). In fact, the the lithium-intercalated graphite anode (o0.3 V versus Li/Li )is migration of transition metals from other cathode materials has much lower than the standard redox potential of Mn(II)/Mn also been observed—for instance, LiFePO (ref. 6) and Li Ni (1.87 V versus Li/Li ). As a result, Mn(II) may be 4 x 1/ 7 16 Co Mn O (NCM) —which likely leads to capacity fading of electrochemically reduced by the low anode potential or 3 1/3 1/3 2 the batteries, especially during the aging process. Overcoming this chemically reduced by lithiated graphite . Although energy- major problem requires a better understanding of mechanisms of dispersive spectroscopy , dynamic secondary ion mass 24 25 manganese dissolution, migration and deposition (DMD) on the spectrometry and atomic absorption spectrometry (AAS) anode. have proven the deposition of manganese species on anodes, In the past, researchers have proposed a mechanism to conclusive identification of the oxidation state of Mn deposited on interpret the DMD process for manganese-based cathodes, which anodes remains a big challenge, which needs to be addressed in can be briefly described in the following three steps: a order to fully understand the reaction mechanism at the anode disproportionation reaction of Mn(III) in LiMn O takes place surface and, as a consequence, the capacity fading of the batteries. 2 4 at the interface of the cathode and electrolyte, as shown in Here we apply several advanced characterization techniques to 3,5,8–14 equation (1) conclusively identify the oxidation state of manganese deposited on the anode surface. We are able to clearly demonstrate by using 3þ 2þ 4þ 2Mn ! Mn þ Mn : ð1Þ X-ray absorption spectroscopy (XAS) along with XPS that, surprisingly, the oxidation state of manganese deposited on the 4þ Consequently, Mn remains on the cathode surface because of anodes is þ 2, which differs from most results reported earlier. 2þ its insolubility in the electrolyte, whereas Mn dissolves into the The concentration of Mn deposited on the anode is found to 2þ electrolyte; the dissolved Mn species diffuse and migrate from increase gradually with increased cycles; moreover, this trend is 2þ the cathode to the anode. The Mn species are finally reduced well correlated with the anodes’ rising impedance. This work to metallic Mn on the anode surface. The third step is based on helps us to better understand the correlation between the the speculation in literatures that the lithiated anode has low manganese DMD process and capacity fading in Mn-based 15–19 chemical potential and high chemical activity . Although lithium-ion batteries and sheds some light on how to improve the 2þ 3þ there are a few studies that reported Mn /Mn being cell performance. deposited on the graphite anodes on the basis of X-ray photoelectron spectroscopy (XPS) measurements, these authors claimed that Mn was deposited as metal first (by reduction) and Results 20–21 then oxidized to Mn(II) , similar to the third step described Electrochemical performance and manganese deposition.To above. Nevertheless, as a consequence of the manganese DMD determine the amount of manganese deposited on the anode, a 3þ process, the loss of active material (Mn ) and the formation of Li/LMO half-cell was tested at a C/2 rate cycled between 3.5 and inactive spinel on the cathode surface because of the 4.3 V at room temperature. Figure 1a shows the normalized disproportionation reaction increases the impedance of the capacity versus cycle number for the Li/LMO cell. The data 11,22 cathode . The solid–electrolyte interphase (SEI) at the anode clearly indicate that the cell capacity is continuously fading to is also poisoned because of the deposition of Mn (as metal) on the about 85% of initial value after 100 cycles in agreement ab Capacity of LiMn O /Li cells 2 4 1.0 Concentration of Mn on Li anode 0.9 95 300 0.8 LiMn O versus dilithiated LiFePO 2 4 4 LiMn O versus Li Ti O 0.7 2 4 4 5 12 85 100 LiMn O versus graphite 2 4 LiMn O versus Li 2 4 80 0 0.6 0 25 50 75 100 0 20406080 100 Cycle number Cycle number Figure 1 | Cycle performance and ICP-AAS. (a) Cycle performance of Li/LiMn O half-cell (black curve) and concentration of Mn deposited on Li 2 4 anodes harvested after different charge–discharge cycles (red curve). (b) Cycle performance of coin cells with LiMn O cathodes versus anodes with 2 4 different Li-intercalation potentials. 2 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. –1 Capacity (mAh g ) Mn concentration on Li anode (p.p.m.) Normalized capacity NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ARTICLE 26 15–17,19–21 with previously reported data for similar materials . The was reduced to metallic Mn state . Moreover, our concentrations of Mn deposited on the Li anodes after different XANES data clearly suggest that the Mn oxidation state deposited cycles were measured using ICP-AAS (inductively coupled on the anodes does not depend on their chemical potential and plasma-atomic absorption spectrometry), and the results are chemical activity during the discharge/charge process. In addition also presented in Fig. 1a. A significant amount of Mn (B85 to the edge position, the fluorescence intensity of XANES can be p.p.m.) is already deposited on the anode after the formation used to determine the relative concentration of the tested element cycles, and it gradually increases with the cycling. The con- in the compound. The amount of Mn deposited on the selected centration of Mn reaches up to about 400 p.p.m. after 100 cycles anodes was calculated based on the Mn concentration on the Li of charge and discharge. As reference, a fully assembled Li/LMO anode from the ICP-AAS analysis and the fluorescence intensities cell was stored under the same condition as that used when the in Fig. 2a. After the 100 cycle test, about 320 p.p.m. of Mn is other cells were cycled. No Mn is detected on the un-cycled Li deposited on the mesocarbon microbead (MCMB) anode, anode, indicating that the DMD process does not occur in un- 260 p.p.m. on the D-LFP anode and 300 p.p.m. on the LTO anode cycled cell. As the ICP-AAS analysis provides accurate results for compared with 400 p.p.m. on the Li anode as seen in Fig. 1a. Mn concentration deposited on the anode (because the lithium Surprisingly, a large amount of Mn is deposited on the LTO and metal can be completely dissolved during the ICP-AAS mea- D-LFP anodes after 100 cycles. Figure 2b shows the XANES data surement), we adopt these data as baseline to obtain Mn con- for MCMB anodes collected after different cycles. The fluores- centration on the other anodes using the fluorescence intensity of cence intensity of the line at about 6,545 eV (1s-4p transition) X-ray absorption near edge spectra (XANES), as presented later. increases systematically with increasing cycle number, indicating 2þ Nevertheless, the half-cell results clearly suggest a correlation that Mn species continue to deposit on the anode surface between manganese deposition on the anode and capacity fading during the cycling tests. in LMO-based cells. Following the main edge, the X-ray absorption spectra show Figure 1b compares the cycle performance of cells with typical oscillations because of a scattering of the outgoing electron LiMn O cathodes versus different anodes with Li-insertion wave at the nearest-neighbour atoms. After a standard data 2 4 potentials ranging from 0 to 3.5 V (versus Li/Li ). The curves treatment (background reduction, normalization, conversion to k show that the LiMn O /Li Ti O (LTO) cell has almost no and m(0) fit), these waves yield the w(k) function and the w(k) k 2 4 4 5 12 capacity fade after 100 cycles, and the LiMn O /delithiated function. The later function and its Fourier transform (FT) 2 4 LiFePO (D-LFP) cell loses about 5% capacity after 100 cycles. contain valuable information about the nearest-neighbour dis- The curves also indicate that LiMn O exhibits much poorer tances, the coordination number and the coordination geometry. 2 4 cycling performance when cycled against graphite, losing 415% Figure 3a shows the FT of the w(k) k functions of all capacity after 100 cycles. Characterization of Mn deposition on investigated samples in R-space, where R is the Fourier conjugate those anodes could help to better understand the difference in to k, uncorrected for phase shifts, related to the distance to the capacity-fading performance. nearest-neighbour shells. The positions of the peaks in the Fourier-transformed w(k) k functions depend on the real radius of the scattering shells around the X-ray-absorbing atoms (that is, Oxidation state of Mn deposited on anodes. The Mn oxidation Mn). The intensities of these peaks depend on the scattering state on different anodes collected after 100 cycles was deter- amplitude, the coordination number and the Debye–Waller factor mined from XANES spectra recorded at room temperature. (s , the mean square disorder in the path lengths). Using feff 28 29 (version 9) and ifeffit we have been able to fit the region of the XANES reference spectra were recorded for Mn foil, MnO and 1 2 Mn O standards, as shown in Fig. 2a, whereas a more extensive FT from R¼ 1–2.3 Å (k¼ 2.75–10 Å , k weighting, modified 2 3 Hanning windows with width 1 Å ). A single Mn-O-scattering collection of standard Mn spectra has been made available by Manceau et al. The edge position of XANES is typically used to path suffices for three of the samples (D-LPF-100, LTO-100 and MCMB-100). For Li-100 a single scattering path no longer determine the oxidation states of the element in the compound. On the basis of the XANES results shown in Fig. 2a and the suffices, corresponding to the less distinct white line in the XANES standards available from Manceau et al. , it is clear that the spectrum (Fig. 2a). For the three samples, the Mn-O bond lengths oxidation state of Mn deposited on all of the anodes used in this are all within 2.15 0.03 Å. This bond length is slightly shorter study is predominately þ 2. This result is surprising because it than that for MnO (manganosite: R ¼ 2.222 Å) but is Mn-O has been speculated in most of the earlier publications that Mn generally consistent with the range of Mn(II)-O bond lengths . ab 1.8 0.030 Mn-metal MCMB-100 MnO 1.6 MCMB-50 Mn O 0.025 2 3 MCMB-25 1.4 D-LFP-100 MCMB-initial Li-100 1.2 0.020 MCMB-100 1.0 LTO-100 0.015 0.8 0.6 0.010 0.4 0.005 0.2 0.0 0.000 6,520 6,530 6,540 6,550 6,560 6,570 6,520 6,540 6,560 6,580 6,600 Photon energy (eV) Photon energy (eV) Figure 2 | X-ray absorption near edge spectra. (a) Normalized XANES of Mn K-edge for commercial MnO, Mn O , Mn foil and different anode samples 2 3 collected after 100 cycles, and (b) unnormalized XANES of Mn K-edge for MCMB anode samples collected after different cycles. NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 3 & 2013 Macmillan Publishers Limited. All rights reserved. Magnitude (a.u.) Magnitude (a.u.) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ab 2.0 2p 3/2 Mn 2p Li-100 2p 1/2 LTO-100 High B. E. 1.5 MCMB-100 LTO-100 D-LFP-100 1.0 D-LFP-100 0.5 MCMB-100 0.0 012 345678 660 655 650 645 640 635 R (Å) Binding energy (eV) Figure 3 | Extended X-ray absorption fine structure spectroscopy and XPS. (a) Fourier transforms of the absorption spectra (w(k) k function) for different anode samples collected after 100 cycles. (b) The Mn 2p XPS spectra of the selected anodes collected after 100 cycles. To further study the anode surface, XPS measurements were re-oxidized to Mn(II) or Mn(III) subsequently by the electrolyte. performed on different anodes collected after 100 cycles. Unlike In contrast, in our work, we found that the Mn deposition can the XANES technique, XPS is more surface-sensitive (measured occur on anodes with potential as high as 3.5 V (D-LFP anode) depth is about 1–3 nm); therefore, it is an ideal tool to analyse the versus Li/Li , and hence we believe that the Mn deposition 31–32 SEI on the anodes . The Mn 2p, Li 1s, C 1s, O 1s and F 1s was not a reduction reaction driven by the low potential of the XPS spectra were recorded and fitted. Figure 3b presents the Mn anodes but an ion-exchange reaction (metathesis reaction) driven 2p core level XPS spectra of the MCMB, LTO and D-LFP anodes by the solubility of the species involved. This is further confirmed after 100 cycles. The analysis of Mn 2p XPS spectra of all the by the observation of Mn(II) on delithiated carbon electrode 2þ samples shows that Mn is deposited on the anodes as Mn , (with SEI formed on the surface) after simply immersing it in the which is consistent with the XAS results discussed earlier. In electrolyte containing Mn(II) dissolved from LiMn O powder 2 4 addition, two more interesting features are noted from the Mn 2p (Supplementary Fig. S2). In addition, the SEI layer starts to form XPS spectra. First, the Mn 2p peak slightly shifts to higher on the carbon anode during the first charge. Therefore, if Mn(II) 3/2 binding energy from MCMB to D-LFP to LTO, indicating that in the electrolyte would be reduced on the carbon surface, it has Mn is bonded to a more electron-negative element in LTO than to diffuse through the SEI layer to reach the carbon surface to be in MCMB. In other words, the Mn-containing species that reduced by lithiated graphite particle, which is kinetically deposited on selective anodes are different, although they have unfavourable. Therefore, metathesis reaction of Mn(II) with the same oxidation state (þ 2). Second, according to the peak species in the SEI layer, which directly leads to the formation of amplitudes, the MCMB anodes have less Mn deposited, whereas Mn(II) on the anodes, most likely is the reaction that takes place LTO has much more Mn deposited on the surface. As the SEI in on the anode surface during the Mn deposition, rather than a LTO and MCMB are different, it is conceivable to expect that the reduction reaction that leads to the formation of metallic Mn, as 15–22 products resulting from the reaction between these SEIs and reported in most earlier studies . At this point, we are unable 2þ dissolved Mn in the electrolyte are different, as confirmed to pinpoint the exact metathesis reactions that occur on the using XPS. Analysis of Li 1s, C 1s, O 1s and F 1s XPS spectra of all anodes owing to the lack of knowledge of the composition of the samples did indicate that the SEI formed on various anodes is the SEI formed on the various anodes; however, we are able to different (Supplementary Fig. S1). propose the reaction mechanism based on the impedance results From the XAS and XPS data presented above, we conclude that presented later. Nevertheless, it is clear from the XAS and XPS the oxidation state of Mn deposited on anodes is predominately results that different Mn(II)-containing species deposit on the þ 2, which is independent of the anode potentials. As far as we anodes during the Mn DMD process, which certainly depends on know, this is the first time that solid evidence has been provided the nature of the SEI on the anodes. for the identification of the Mn oxidation state during the Mn DMD process in LMO-based cells. It should be mentioned that 2þ there are a few studies that reported the observation of Mn / Correlation between Mn deposition and capacity fading. Two 3þ Mn deposited on the carbon anode, especially in the papers of questions remain: What is the correlation between the quantity of 20 21 Komaba et al. and Ochida et al. However, they did not Mn deposited on anodes and the rate of capacity fading, especially 2þ provide as conclusive results as we presented in this work on the in the graphite anode? How does the Mn act in the SEI of the oxidation state of Mn deposition in a real LMO/graphite cell, and graphite anode? To address these issues, we first employed a simple especially on the correlation between the Mn deposition and calculation to estimate the capacity fading contributed solely from capacity fade of the cell. First, significant amount (4100 p.p.m.) the loss of active materials in the cathode because of the DMD 2þ of Mn was pre-added to the electrolyte in their studies either process. Both the amount of Mn deposited on the anodes (based using Mn(ClO ) or introduced by anodic dissolution of Mn on ICP-AAS and XAS data) and that dissolved in the electrolytes 4 2 metal, which does not represent the real situation in the LMO/C (based on literature data) were accounted for in the calculation. 2þ cells in which the Mn concentration in the electrolyte The calculated capacity fading in the MCMB-LMO cell resulting 3þ increases gradually from zero because of the dissolution of Mn from the loss of active Mn (o1%) is significantly smaller from the LMO cathode. For instance, in the first discharge/charge than the overall capacity fading determined experimentally from 2þ cycle, the Mn concentration in the electrolyte is lower than the 100-cycle data (15%). For the LTO-LMO and D-LFP-LMO 8 2þ 5 p.p.m. . Second, they believe that Mn was deposited as metal cells, much less overall capacity fading is found, even with first by reduction of the low-potential lithiated carbon and then almost the same amount of Mn deposited on the anodes. 4 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. 2 –2 FT [k x χ(k )] (Å ) Relative intensity (a.u.) NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ARTICLE –12 100 Graphite anode stored in the pure electrolyte Storage time 6 h 12 h –8 24 h 48 h –4 0 5 10 15 20 25 Surface-modified LiMn O versus graphite 2 4 Z′ (Ω) Bare LiMn O versus graphite 2 4 –12 2+ Graphite anode stored in the electrolyte with Mn Storage time 0 20 40 60 80 100 6 h Cycle number 12 h –8 24 h Figure 5 | Cycle performance. Cycle performance of coin cells with 48 h graphite anodes versus LMO cathodes (bare LMO and surface-modified LMO with TiO ) at C/2 and 55 C. –4 2 of cells with graphite anode versus bare LMO and surface- 0 5 10 15 20 25 modified LMO cathodes. The curves indicate that bare LMO Z′ (Ω) exhibits much poorer cycling performance when cycled against Figure 4 | The AC impedance. The impedance of graphite anodes stored in graphite at 55 C, losing 450% capacity after 100 cycles, 2þ 2þ an electrolyte (a) without Mn and (b) with Mn . whereas only about 20% capacity loss is observed for the surface-modified LMO. The AC impedance rise of the graphite anode and bare LMO This finding suggests that the capacity fading in the MCMB-LMO cathode is shown in Fig. 6a,b, respectively. The cells were 3þ cell is not directly caused by the loss of the active Mn . A pos- discharged to 50% state of charge (SOC) before each impedance sible reason for the severe capacity fading in the MCMB-LMO cell testing. The impedance rise of the bare LMO cathode slows 2þ is that Mn in the electrolyte reacts with the SEI on the graphite down after 25 cycles (Fig.6b), whereas that of the graphite anode, which modifies the nature of the SEI. As a result, the anode increases significantly and contributes to almost all the 2þ impedance of the graphite anode could change as the Mn increase in the impedance of the full cell after 50 cycles (Fig. 6a). concentration increases in the SEI during the cycle test. Corroborating the capacity versus cycle number diagram for the 2þ To understand the correlation between the reaction of Mn MCMB-LMO cell shown in Fig. 5 and these impedance results, it in the electrolyte with the SEI and the impedance change of the appears that the impedance rise of the cathode is not responsible graphite anode in the LiMn O -based cells, we first compared the for the capacity loss. The trend of impedance rise from the anode 2 4 AC impedance rise of graphite anodes (with SEI formed) stored matches that of the capacity loss during the cycling test, 2þ in electrolyte with or without Mn by using a three-electrode indicating that the anode impedance is the key factor that causes configuration. Graphite anodes were first cycled against Li metal the capacity fading of the full cell. Figure 6c shows the AC for one cycle to form an SEI layer and then fully discharged to impedance rise of the graphite anode cycling against surface- 5 mV versus Li/Li . The fully discharged graphite anodes were modified LMO at different cycles, which clearly indicates that the disassembled and transferred to a three-electrode cell (with Li impedance of the graphite anode grows much slower than that of metal as counter electrodes and Li Sn micro-reference electrode) the graphite anode in the bare LMO cell (Fig. 6a). Similar results filled with either a pure electrolyte or an electrolyte with 20 p.p.m. were also reported for MCMB-Li [Ni Mn Co ] O cells . 1.1 1/3 1/3 1/3 0.9 2 2þ Mn . As shown in Fig. 4, the impedance of graphite anode does In addition, we also performed the same in situ AC impedance not change much when stored in the pure electrolyte but spectroscopy for the LTO anode after different times of cycling 2þ increases notably when stored in the electrolyte with Mn . This against bare LMO at C/2 rate (Supplementary Fig. S3). The 2þ result clearly indicates that Mn in the electrolyte reacts with impedance of the LTO anode grows much slower than that of the the SEI on the graphite anode and modifies the nature of the SEI graphite anode in the bare LMO cell (Fig. 6a), corresponding to accordingly, leading to the increase in the impedance of graphite much slower capacity fading than the LTO/LMO cell (Fig. 1b). anode. It is also worthwhile to note that the impedance of There is, thus, a strong correlation between impedance rise of the 2þ 2þ graphite anode in the electrolyte with Mn is stabilized after anode and the concentration of Mn in the SEI on the graphite 2þ 24 h, likely because of the depletion of Mn in the electrolyte. anode, as well as the capacity fading. In situ AC impedance spectroscopy in a three-electrode configuration was performed to probe the capacitive and resistive behaviour of the graphite anodes after different times of cycling Discussion against bare LMO at C/2 and 55 C. For comparison, similar As assumed in earlier reports, Mn deposition could change the measurement was also performed on the graphite anode cycling structure or composition of SEI on graphite anodes, resulting in against surface-modified LMO with TiO , which is expected and the capacity fading of the full cell . To elucidate the effect of Mn confirmed with the ICP-AAS measurement (Supplementary deposition on the electrochemical properties of the SEI layer, a Supplementary Table S1) to have much less Mn dissolved detailed theoretical analysis of the anode impedance was from the cathode during the cycling test. In other words, performed, as described in the Supplementary Note 1. The fitting the dissolution of Mn is considerably suppressed from the results (Supplementary Table S2) showed the increase of R C , SEI SEI surface-modified LMO. Figure 5 compares the cycle performance which indicates that the SEI layer on the graphite anode is not NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 5 & 2013 Macmillan Publishers Limited. All rights reserved. Z′′ (Ω) Z′′ (Ω) Normalized capacity (%) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ab –16 –35 –5 LiMn O cathode 2 4 Graphite anode –4 Before cycling –30 Before cycling 25th cycle –3 25th cycle –12 50th cycle 50th cycle –25 –2 75th cycle 75th cycle –1 100th cycle 100th cycle –20 0 –8 0 1234 5 Z’ (Ω) –15 –10 –4 –5 0 5 10 15 20 25 30 35 04 812 16 Z′ (Ω) Z′ (Ω) cd –25 Graphite versus surface modified LiMn O Graphite anodes versus 2 4 Bare LiMn O 2 4 –20 Surface-modified LiMn O Before cycling 2 4 25th cycle 50th cycle –15 75th cycle 100th cycle –10 –5 0 5 10 15 20 25 020 40 60 80 100 Z′ (Ω) Cycle number Figure 6 | The AC impedance. (a) The graphite anode and (b) bare LiMn O cathode after different times of cycling at C/2 and 55 C. (c)AC 2 4 impedance rise of the graphite anode cycling against surface-modified LMO. (d) Plots of R C of graphite anodes cycling against bare and surface- SEI SEI modified LMO at different cycles based on the fitting results in Supplementary Table S2. simply thickened but also undergoes change in its nature Ion exchange model behaviour during cycling of the cell. On the basis of the fitting results in Supplementary Table S2, the change of R C of SEI SEI graphite anodes cycling against both bare and surface-modified LMO cathodes is plotted in Fig. 6d. It is clearly seen that a SEI SEI dramatic increase in R C of graphite anode was observed for SEI SEI the cell containing bare LMO, whereas a much smaller increase (by a magnitude of 2) in R C of the graphite anode was SEI SEI observed for the cell containing surface-modified LMO. This is well correlated to their capacity fade, as shown in Fig. 5. Graphite Graphite According to the results presented above, we believe that the + + 2+ Mobile Li Immobile Li Solvent Mn capacity fading is caused by changes in the SEI on the graphite 2þ anode because of the Mn that continuously migrated from the Figure 7 | Ion-exchange model. Proposed ion-exchange model for the Mn cathode and thus the change of the impedance of the graphite deposition on the graphite anode during the Mn DMD process in the anode. graphite/LMO cell. Corroborating XANES/XPS and AC impedance results, we propose a metathesis reaction (ion exchange) mechanism for the speculated in earlier studies. The capacity fading is caused by Mn deposition on the carbon anode, as illustrated in Fig. 7. We changing of the SEI on the graphite anode because of the believe that the ion exchange during the Mn deposition is 2þ 2þ þ continuous reaction with the dissolved Mn from the cathode. between Mn and mobile Li in the SEI—for example, lithium alkyl carbonate (ROCO Li), which blocks the lithium diffusion path—and thus, leads to the increase in R C of graphite SEI SEI Methods anodes (Fig. 6d). This is also confirmed by the absence of MnF LiMn O synthesis and characterization. Stoichiometric LiMn O particles were 2 2 4 2 4 synthesized via a solid-state method with Li CO (Sigma-Aldrich, 99.0%) and and MnCO on the SEI, which would result from the ion 2 3 2þ þ chemical MnO (Chemetal, 99.5%), which were mixed and ground thoroughly in a exchange between Mn and the immobile Li in LiF and molar ratio of 1.05: 4. The mixture was then heated at 800 C for 16 h. The Li CO on the SEI otherwise. 2 3 structure of the as-prepared spinel LiMn O was confirmed using X-ray powder 2 4 In summary, we demonstrated in this study that the oxidation diffraction and scanning electron microscopy (Supplementary Supplementary Figs. S4–S6 and Supplementary Note S1). The surface-modified LiMn O powders with state of manganese deposited on the anodes is þ 2 on the basis of 2 4 TiO was prepared as follows: Tetrabutyl titanate (Sigma-Aldrich, 99%) was first XAS and XPS analyses. The results also suggest a metathesis dissolved in ethanol to make a 10% solution, and then the solution was slowly 2þ reaction between Mn and some species on the SEI during the dropped into acetic acid solution (which contained 10% acetic acid and 56% deposition of Mn on the anodes, rather than a reduction reaction ethanol) with agitation to obtain the sol, which was further diluted to 5% by adding that leads to the formation of metallic Mn otherwise, as ethanol to prevent the aggregation of the colloid. Then the LiMn O powders were 2 4 6 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. Z′′ (Ω) Z′′ (Ω) Z’’ (Ω) R C (mF Ω) Z′′ (Ω) SEI SEI NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 ARTICLE put into the diluted sol, agitated for 20 min, followed by drying at 80 C and Analysis of impedance spectra. According to the model first reported by Thomas 38 39 calcination at 750 C for 4 h. The resulting molar ratio of TiO :LiMn O in the et al. and further improved by Barsoukov et al. , we used the equivalent circuit 2 2 4 surface-modified LiMn O is B5%. (Supplementary Fig. S7) for interpreting our anode impedance spectra, where R is 2 4 s the resistance of the electrolyte, separator and wires; R and C represent the ct dl charge transfer resistance and double-layer capacitance at the electron/ion con- duction boundary for the Li-insertion reactions; R and C represent the SEI SEI Electrochemistry tests. 2032-type coin cells were assembled with a cathode resistance and the geometric capacitance of the SEI layer; and Z is the Warburg material of 85% LiMn O , 10% carbon black and 5% polyvinylidene difluoride 2 4 diffusion impedance of the Li ion into the bulk particles via solid-phase diffusion. (PVDF) binder. The GenII electrolyte was composed of 1.2 M LiPF in ethylene As shown in Fig. 5a, the impedance of the graphite electrode typically includes carbonate: ethyl methyl carbonate at a 3:7 ratio by weight. Anodes with different two depressed semicircles in the high-frequency region and a line in the low- Li-insertion potentials were tested, including 0 V (Li metal), 0.02 V (92% MCMB frequency region, which can be fitted to two serially connected parallel R elements ct graphite and 8% PVDF), 1.55 V (80% Li Ti O , 10% carbon black and 10% PVDF) 4 5 12 and the Warburg element, respectively, in the equivalent circuit. The first and 3.5 V (85% LiFePO , 10% carbon black and 5% PVDF, fully charged before semicircle (in higher-frequency region, as shown in the insertion of Fig. 5a) being used as anode). If not otherwise specified, the cells were first cycled with a remains almost the same from the 25th to 100th cycle, whereas the second one (at constant current of C/10. After the formation cycles, the cells were cycled at a lower-frequency region) enlarges constantly with the cycle time. As R and C ct dl constant current of C/2. Specific voltage windows (depending on the anode) were represent the Li-intercalation reaction impedance, which is mainly determined by selected in order to charge and discharge the LiMn O cathodes between 3.5 and 2 4 intrinsic properties of the active material, electrode SOC and temperature, it is þ 2þ 4.3 V (versus Li/Li ). The electrolyte with Mn was prepared by soaking 0.1 g reasonable to assume that R and C tested here remain constant with the cycling ct dl LiMn O in 10 ml GenII electrolyte in a sealed container at 55 C for 14 days. 2 4 number. On the other hand, the SEI layer will get thicker with the cycling of the 2þ The concentration of Mn in electrolyte was about 20 p.p.m. from the ICP-AAS cell, thereby changing R and C . We, therefore, assigned the first semicircle to SEI SEI analysis. the R C element and the second one to the R C element. The fitting results ct dl SEI SEI The AC impedance of cycled anodes and cathode was determined by using are shown in Supplementary Table S1. The deviation of the spectra in the high- 35–37 reference electrode cells, as described in detail elsewhere . The test cell was frequency region (inset to Fig. 5a) can be attributed to electron conduction and assembled with flat, single-sided, 32-cm electrodes. A micro-reference electrode 40 Li-ion diffusion in the porous electrode . was prepared from a 25-mm-diameter copper wire coated with 1-mm-thick tin and 41 According to the SEI model , R and C can be calculated based on the SEI SEI covered with a polyurethane coating. The reference wire was shorted to the following equations: R ¼ rl/S and C ¼ eS/l, where r and e are the specific SEI SEI lithiated anode, which caused the transfer of some lithium from the anode to Sn to resistance and permittivity of the layer, respectively, and l and S are the real make a Li Sn electrode. The reference electrode cells were first charged at a rate of electrochemically active thickness and area, respectively. The increase in R and SEI C/10 and cycled between 3.0 and 4.3 V at C/2 before being constant-voltage- decrease in C might be related to the thickening of the SEI layer, whereas r and e SEI charged to 50% SOC. The AC impedance data for the anode and cathode were could change with cycling as well. Therefore, R C ¼ re is selected here as a SEI SEI collected with a Solartron 1470 E and 1451A cell test system using a 5-mV AC reliable value to indicate the intrinsic electrochemical properties of the SEI layer perturbation with frequency ranging from 1 MHz to 20 MHz. determined by its composition, although independent of its geometry. The fitting results showed the increase in R C , which indicates that the SEI layer on the SEI SEI anode is not simply thickened but also undergoes change in its nature behaviour during cycling of the cell. Characterization of Mn deposition. The disassembled anodes were rinsed with a mixture of ethylene carbonate/ethyl methyl carbonate (3:7 by weight) to remove the residual electrolyte remaining on the anode surface. To avoid any contamination to the surface, this procedure was carefully performed in an argon- References filled glove box. Inductively coupled plasma (ICP) was applied to analyse the 1. Thackeray, M. M., David, W. I. F., Bruce, P. G. & Goodenough, J. B. Lithium amount of Mn deposited on the Li metal anode, as Li can be thoroughly dissolved insertion into manganese spinels. Mater. Res. Bull. 18, 461–472 (1983). by acid and hence provide reliable Mn concentration. 2. Guyomard, D. & Tarascon, J. M. Li metal-free rechargeable LiMn O /carbon 2 4 The oxidation states of Mn deposited on anodes were characterized by using cells their understanding and optimization. J. Electrochem. Soc. 139, 937–948 X-ray absorption spectroscopy (XAS) and XPS. For the XAS measurements, the (1992). samples are protected by Kapton tape to prevent the contamination from the air. 3. Gummow, R. J., Dekock, A. & Thackeray, M. M. Improved capacity retention Mn K-edge XAS measurements were conducted on both the bending magnet in rechargeable 4V lithium/lithium manganese oxide (spinel) cells. Solid State (10BM) and insertion device (10ID) beamlines of the Materials Research Ionics 69, 59–67 (1994). Collaborative Access Team in the Advanced Photon Source at Argonne National 4. Lazzari, M. & Scrosati, B. Cyclable lithium organic electrolyte cell based on 2 Laboratory. Harmonic rejection was accomplished with an Rh-coated mirror. Anode spectra were measured on 10BM in the fluorescence mode using a four- intercalation electrodes. J. Electrochem. Soc. 127, 773–774 (1980). element silicon drift detector (SII NanoTechnology USA Inc., Vortex-ME4). A 5. Amatucci, G. G. et al. Materials’ effects on the elevated and room temperature third detector in the series simultaneously collected a metallic Mn reference performance of C/LiMn O Li-ion batteries. J. Power Sources 69, 11–25 (1997). 2 4 spectrum with each measurement for energy calibration. The peak of the first 6. Aurbach, D. et al. Review on electrode-electrolyte solution interactions, related derivative was set to 6,537.7 eV. to cathode materials for li-ion batteries. J. Power Sources 165, 491–499 (2007). Samples were also analysed with XPS using a Kratos Axis Ultra DLD surface 7. Zheng, H., Sun, Q., Liu, G., Song, X. & Battaglia, V. S. Correlation between analysis instrument. The base pressure of the analysis chamber during these dissolution behavior and electrochemical cycling performance for 10  9 experiments was 3  10 torr, with operating pressures around 1 10 torr. LiNi Co Mn O -based cells. J. Power Sources 207, 134–140 (2012). 1/3 1/3 1/3 2 Spectra were collected with a monochromatic Al Ka source (1,486.7 eV) and a 8. Jang, D. H., Shin, Y. J. & Oh, S. M. Dissolution of spinel oxides and capacity 300 700-m spot size. For survey spectra, the data were collected at a pass energy of losses in 4V Li/LiMn O cells. J. Electrochem. Soc. 143, 2204–2211 (1996). 2 4 160 eV (fixed analyser transmission mode), a step size of 1 eV and a dwell time of 9. Jang, D. H. & Oh, S. M. Electrolyte effects on spinel dissolution and cathodic 200 ms. High-resolution regional spectra were collected with a pass energy of 20 eV capacity losses in 4V Li/LiMn O rechargeable cells. J. Electrochem. Soc. 144, 2 4 (fixed analyser transmission mode), a step size of 0.1 eV and a dwell time of 300 ms. 3342–3348 (1997). For low signal-to-noise regions, multiple passes were made and the results were 10. Xia, Y., Zhou, Y. & Yoshio, M. Capacity fading on cycling of 4V Li/LiMn O 2 4 averaged together. cells. J. Electrochem. Soc. 144, 2593–2600 (1997). Prior to their introduction into the load-lock vacuum chamber of the XPS 11. Choa, J. & Thackeray, M. M. Structural changes of LiMn O spinel electrodes 2 4 instrument, all air-sensitive samples were loaded into an inert transfer module during electrochemical cycling. J. Electrochem. Soc. 146, 3577–3581 (1999). interfaced with the instrument. Samples were prepared for analysis in an Ar-filled 12. Gnanaraj, J. S., Pol, V. G., Gedanken, A. & Aurbach, D. Improving the high- glove box, with no more than 1 p.p.m. O and 1 p.p.m. H O. Nonconductive 2 2 temperature performance of LiMn O spinel electrodes by coating the active 2 4 samples showed evidence of differential charging, resulting in peak shifts and mass with mgo via a sonochemical method. Electrochem. Commun. 5, 940–945 broadening. Photoelectron peak positions were shifted back towards their true (2003). values, and their peak widths were minimized by flooding the samples with low- 13. Yu, L. H., Qiu, X. P., Xi, J. Y., Zhu, W. T. & Chen, L. Q. Enhanced high- energy electrons and ions from the charge neutralizer system on the instrument. potential and elevated-temperature cycling stability of LiMn O cathode by 2 4 Peak position was further corrected by referencing the C 1s peak position of TiO modification for Li-ion battery. Electrochim. Acta. 51, 6406–6411 (2006). adventitious carbon for a sample (284.8 eV, PHI Handbook of Photoelectron 14. Walz, K. A. et al. Elevated temperature cycling stability and electrochemical Spectroscopy) and by shifting all other peaks in the spectrum accordingly. impedance of LiMn O cathodes with nanoporous ZrO and TiO coatings. 2 4 2 2 Fitting was carried out by using the program CasaXPS. Peaks were fit as J. Power Sources 195, 4943–4951 (2010). asymmetric Gaussian/Lorentzians, with 0–30% Lorentzian character. The full 15. Blyr, A. et al. Self-discharge of LiMn O /C Li-ion cells in their discharged state. 2 4 width at half maximum of all subpeaks was constrained to 0.7–2 eV, as dictated by J. Electrochem. Soc. 145, 194–209 (1998). instrumental parameters, lifetime broadening factors and broadening because of 16. Komaba, S., Kumagai, N. & Kataoka, Y. Influence of manganese(II), cobalt(II), sample charging. With this native resolution set, peaks were added, and the best fit, and nickel(II) additives in electrolyte on performance of graphite anode for using a least-square fitting routine, was obtained while adhering to the constraints lithium-ion batteries. Electrochim. Acta. 47, 1229–1239 (2002). mentioned above. NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications 7 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3437 17. Komaba, S., Itabashi, T., Kaplan, B., Groult, H. & Kumagai, N. Enhancement of 37. Dee, D. W., Jansen, A. N. & Abraham, D. P. Theoretical examination of Li-ion battery performance of graphite anode by sodium ion as an electrolyte reference electrodes for Li-ion cells. J. Power Sources 174, 1001–1006 (2007). additive. Electrochem. Commun. 5, 962–966 (2003). 38. Thomas, M. G. S. R., Bruce, P. G. & Goodenough, J. B. AC impedance analysis 18. Komaba, S. et al. Inorganic electrolyte additives to suppress the degradation of of polycrystalline insertion electrodes: application to Li CoO . J. Electrochem. 1-x 2 graphite anodes by dissolved Mn(II) for lithium-ion batteries. J. Power Sources Soc. 132, 1521–1528 (1985). 119-121, 378–382 (2003). 39. Barsoukov, E., Kim, J. H., Kim, J. H., Yoon, C. O. & Lee, H. Kinetics of lithium 19. Whittingham, M. S. Lithium batteries and cathode materials. Chem. Rev. 104, intercalation into carbon anodes: in situ impedance investigation of thickness 4271–4301 (2004). and potential dependence. Solid State Ionics 116, 249–261 (1999). 20. Komaba, S. et al. Impact of 2-Vinylpyridine as electrolyte additive on surface 40. Barsoukov, E. & Macdonald, J. R. Impedance Spectroscopy: Theory, Experiment, and electrochemistry of graphite for C/LiMn O Li-Ion Cells. J. Electrochem. and Applications 2nd edn (Blackwell Science Public, Osney Mead, 2005). 2 4 Soc. 152, A937–A946 (2005). 41. Peled, E., Golodnitsky, D., Ardel, G. & Eshkenazy, V. The SEI model - 21. Ochida, M. et al. Influence of manganese dissolution on the degradation of application to lithium polymer electrolyte batteries. Electrochim. Acta. 40, surface films on edge plane graphite negative-electrodes in Lithium-Ion 2197–2204 (1995). Batteries. J. Electrochem. Soc. 159, A961–A966 (2012). 22. Hirayama, M. et al. Dynamic structural changes at LiMn O /electrolyte 2 4 Acknowledgements interface during lithium battery reaction. J. Am. Chem. Soc. 132, 15268–15276 (2010). J.L. was supported by the Department of Energy (DOE) Office of Energy Efficiency and 23. Amine, K. et al. Improved lithium manganese oxide spinel/graphite Li-ion cells Renewable Energy (EERE) Postdoctoral Research Award under the EERE Vehicles for high-power applications. J. Power Sources 129, 14–19 (2004). Technology Program administered by the Oak Ridge Institute for Science and Education (ORISE) for the DOE managed by Oak Ridge Associated Universities (ORAU) under 24. Abraham, D. P., Spila, T., Furczon, M. M. & Sammann, E. Evidence of transition-metal accumulation on aged graphite anodes by SIMS. Electrochem. DOE contract number DE-AC05-06OR23100. K.A. and A.J.K. (X-ray absorption studies) Solid State 11, A226–A228 (2008). were supported by the Center for Electrical Energy Storage, an Energy Frontier Research 25. Tsunekawa, H. et al. Capacity fading of graphite electrodes due to the Center funded by the U.S. Department of Energy, Office of Science, Office of Basic deposition of manganese ions on them in Li-ion batteries. J. Electrochem. Soc. Energy Sciences. A.N.J. was supported by the Applied Battery Research for Transpor- 149, A1326–A1331 (2002). tation (ABR) Program from the U.S. DOE-EERE Office of Vehicle Technologies. Use of 26. Sun, Y.-K., Hong, K.-J. & Prakash, J. The effect of ZnO coating on the Advanced Photon Source, an Office of Science User Facility operated for DOE, Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under electrochemical cycling behavior of spinel LiMn O cathode materials at 2 4 elevated temperature. J. Electrochem. Soc. 150, A970–A972 (2003). Contract No. DE-AC02-06CH11357. MRCAT operations are supported by the DOE 27. Manceau, A., Marcus, M. A. & Grangeon, S. Determination of Mn valence and the MRCAT member institutions. Financial support from the 973 Program states in mixed-valent manganates by XANES spectroscopy. Am. Mineral. 97, (2009CB220105) of China, Beijing Natural Science Foundation (2120001), National 816–827 (2012). Natural Science Foundation of China (21273129) and Bosch (China) Ltd. is gratefully 28. Rehr, J. J. & Albers, R. C. Theoretical approaches to X-ray absorption fine acknowledged. This work was also supported by the Human Resources Development structure. Rev. Mod. Phys. 72, 621–654 (2000). of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korean government, Ministry of Trade, Industry and Energy 29. Ravel, B. & Newville, M. Athena, artemis, hephaestus: data analysis for X-ray absorption spectroscopy using ifeffit. J. Synchrotron. Radiat. 12, 537–541 (No. 20124010203310) and by the National Research Foundation of Korea (NRF) (2005). grant funded by the Korea government (MEST) (No. 2009-0092780). 30. Pacalo, R. E. & Graham, E. K. Pressure and temperature-dependence of the elastic properties of synthetic MnO. Phys. Chem. Miner. 18, 69–80 (1991). 31. Peled, E. et al. Composition, depth profiles and lateral distribution of materials Author contributions in the SEI built on HOPG-TOF SIMS and XPS studies. J. Power Sources 97-98, J.L. designed the experiments; C.Z. synthesized the cathode materials, performed and analysed the electrochemical experiments, A.J.K., J.L. and T.W. performed and analysed 52–57 (2001). 32. Aurbach, D. et al. Recent studies on the correlation between surface chemistry, the X-ray absorption spectroscopy measurements; J.L. performed and analysed the XPS morphology, three-dimensional structures and performance of Li and Li-C experiments; A.N.J. fabricated the three-electrode cells for impedance measurements; X.Q. and K.A. supervised the project; J.L., C.Z. and K.A. wrote the paper. All of the intercalation anodes in several important electrolyte systems. J. Power Sources 81-82, 91–98 (1997). authors discussed the results and reviewed the manuscript. 33. Amine, K. et al. Mechanism of capacity fade of MCMB/Li (Ni Mn Co 1.1 1/3 1/3 1/ ) O cell at elevated temperature and additives to improve its cycle life. 3 0.9 2 Additional information J. Mater. Chem 21, 17754–17759 (2011). 34. Amatucci, G., Du Pasquier, A., Blyr, A., Zheng, T. & Tarascon, J. M. The Supplementary Information accompanies this paper on http://www.nature.com/ elevated temperature performance of the LiMn O /C system: failure and naturecommunications 2 4 solutions. Electrochim. Acta. 45, 255–271 (1999). Competing financial interests: The authors declare no competing financial interests. 35. Abraham, D. P., Poppen, S. D., Jansen, A. N., Liu, J. & Dees, D. W. Application of a lithium-tin reference electrode to determine electrode contributions to Reprints and permission information is available online at http://npg.nature.com/ impedance rise in high-power lithium-ion cells. Electrochim. Acta. 49, reprintsandpermissions/ 4763–4775 (2004). 36. Jansen, A. N., Dee, D. W., Abraham, D. P., Amine, K. & Henriksen, G. L. Low- How to cite this article: Zhan, C. et al. Mn(II) deposition on anodes and its effects on temperature study of lithium-ion cells using a Li Sn micro-reference electrode. capacity fade in spinel lithium manganate–carbon systems. Nat. Commun. 4:2437 J. Power Sources 174, 372–379 (2007). doi: 10.1038/ncomms3437 (2013). 8 NATURE COMMUNICATIONS | 4:2437 | DOI: 10.1038/ncomms3437 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved.

Journal

Nature Communications – Springer Journals

Published: Sep 30, 2013

There are no references for this article.