Letter pubs.acs.org/NanoLett
Crystal Orientation Tuning of LiFePO4 Nanoplates for High Rate Lithium Battery Cathode Materials Li Wang,† Xiangming He,*,†,∥ Wenting Sun,† Jianlong Wang,†,⊥ Yadong Li,‡ and Shoushan Fan§ †
Institute of Nuclear and New Energy Technology, Tsinghua University, Beijing 100084, P. R. China Department of Chemistry, Tsinghua University, Beijing 100084, P. R. China § Department of Physics, Tsinghua-Foxconn Nanotechnology Research Center, Tsinghua University, Beijing 100084, P. R. China ∥ State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing 100084, P. R. China ⊥ Beijing Key Laboratory of Fine Ceramics, Tsinghua University, Beijing 100084, P. R. China ‡
S Supporting Information *
ABSTRACT: We report the crystal orientation tuning of LiFePO4 nanoplates for high rate lithium battery cathode materials. Olivine LiFePO4 nanoplates can be easily prepared by glycol-based solvothermal process, and the largest crystallographic facet of the LiFePO4 nanoplates, as well as so-caused electrochemical performances, can be tuned by the mixing procedure of starting materials. LiFePO4 nanoplates with crystal orientation along the ac facet and bc facet present similar reversible capacities of around 160 mAh g−1 at 0.1, 0.5, and 1 C-rates but quite different ones at high C-rates. The former delivers 156 mAh g−1 and 148 mAh g−1 at 5 C-rate and 10 C-rate, respectively, while the latter delivers 132 mAh g−1 and only 28 mAh g−1 at 5 C-rate and 10 C-rate, respectively, demonstrating that the crystal orientation plays important role for the performance of LiFePO4 nanoplates. This paves a facile way to prepare high performance LiFePO4 nanoplate cathode material for lithium ion batteries. KEYWORDS: Cathode materials, LiFePO4, crystal orientation, solvothermal, intermixing of cations, Li-ion battery ince the first report by Padhi et al. in 1997, olivine lithium iron phosphate LiFePO4 has been considered as one of the most promising cathode candidates of lithium ion batteries for applications in electric vehicles and large-scale energy storage, due to its inherent merits such as high theoretical capacity (170 mAh g−1), long cyclability, high safety, low toxicity, and possibly low cost. The main obstacle for applicable electrochemical performances is its low electronic conductivity and lithium ion diffusivity.1 Progressive efforts, including carbon coating on particle surfaces,2,3 aliovalent cation substitution,1,4 particle size minimization,5 and customizing particle morphologies,6−8 have been made to overcome this obstacle. Among these approaches, nanosized electrodes have been intensively investigated for high-power density applications,2,5,6 as nanomaterials provide both a shorter diffusing path and a larger implantation surface area for charge carriers. In addition, according to the high anisotropy of olivine LiFePO4, customizing particle morphologies is of evidently high importance for material properties and electrochemical performances. The theoretical studies reported by Ceder’s group9,10 and Islam’s group11,12 confirmed that lithium ions hopping along the b-axis is preferable in the olivine structure, because of the low activation barrier along this direction. Chen et al.13 reported that the existence of particles with large ac faces can therefore improve the rate capability. From this standpoint, plates forming LiFePO4 nanomaterials with crystal growth orientations along the ac plane and thicknesses as low as possible may
S
© XXXX American Chemical Society
be very ideal to achieve high rate capabilities. In fact, various morphologies which indicate different crystal orientations of LiFePO4 particles, including nanorods, nanoflakes, and nanoplates, can be prepared by different methods and do show dramatically different electrochemical performance.14,15 In this sense, the efficacy of LiFePO4 nanomaterials largely depends on preparation and processing methods. Even though solid-state reactions are universally recognized as a useful methodology to prepare LiFePO4, the drawbacks associated with this method are unavoidable high-energy consumption and uncontrollable growth of the grains. Solution routes are more desirable to controlling nucleation and crystal growth than solid-state routes, and so the morphology of the products can be fine-tuned easily. Among them, the solvothermal process is the most promising, and it has been proven to be a predominant synthetic approach to prepare LiFePO4. The phase purity, intrinsic defect concentration, grain size, and morphology could be reasonably controlled by solvothermal conditions.8,15,16 It is worth noticing that samples prepared by solvothermal synthesis mostly grew along the bc plane, while only a few presented a large ac plane.8,15,17,18 Then, experimental results accounting for the differences arised from Received: July 27, 2012 Revised: September 16, 2012
A
dx.doi.org/10.1021/nl3027839 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
clearly for S1, indicative of possible preferred orientation of S1. The preferred orientation can be also deduced from the intensity ratio of I(200)/I(020). Kanumara et al.6 suggested that if the ratio of I(020)/I(200) is greater than that of the standard, platelet-type structures with preferential direction along the ac facet would be seen. In fact, the I(020)/I(200) ratio of S1 and JCPDS-81-1173 are 4.2 and 2.1, respectively, implying ac-plane plate morphology of S1. Based on similar reasoning, that the I(020)/I(200) ratio of S2 is 1.5 implies the bc-plane plate morphology of S2. The percentages of Fe•Li antisites for S1 and S2 samples are 0.38% and 0.87%, respectively, showing very low intermixing of cations in the prepared LiFePO4 nanoplates, as shown in Figure 1. The particle size, morphology, and crystal orientation of S1 and S2 are confirmed by SEM and TEM, as shown in Figure 2. Both S1 and S2 are well-dispersive and neat, indicative of high crystallinity and purity. Besides, they both present plate-form morphology, though S1 looks rounder than S2. Despite that the prominent surface area of S2 seems larger than that of S1, S1 and S2 show quite similar thicknesses. As labeled in Figure 2b and e, the average thicknesses of both S1 and S2 are around 30 nm. Further identification on crystal orientation, however, demonstrates that S1and S2 have totally different growth morphologies. That is, the exposure facet is mostly {010} for S1, while crystal growth orientations along the bc plane are prominent for S2. As reported by Islam et al., the growth morphology tends to be terminated by {010}, {100}, and {101} faces when taking into account the kinetic factors influencing crystal growth by using the attachment energy of each crystal plane, rather than its equilibrium surface energy.12 In this sense, the crystal orientations presented by S1 and S2 are both reasonable, when considering of the growth kinetics. As shown in Scheme 1, the two mixing sequences for S1 and S2 result in different final mixtures, which are in fact the different solvothermal precursors. The mixing procedure for S1 produces Li3PO4 as the main precipitation (Figure ES1, SI), and much free Fe2+ in the solution remains. However, in the cases for S2, Fe2+ mainly exists as Fe3(PO4)2 precipitation (Figure ES1), and the cation in the solution is mostly Li+. Related information can be tracked from the evolution of intermediates with time (Figure ES1). Though further investigations are needed to get a quantitative understanding, the above result indicates that two main differences during LiFePO4 formation and crystal growth are introduced by a simple change in feeding sequence. The first one lies in the solid part of the precursor. It is widely accepted that the mechanism for LiFePO4 crystal formation from solvothermal is “dissolution-nucleation-crystallization”.7 The difference in dissolution between Fe3(PO4)2 and Li3PO4 may lead to different nucleation speeds. It is worth noting that the appearance of olivine LiFePO4 in S2 is about 3 h earlier than that in S1 (Figure ES1), implying a fast nucleation in S2. That S2 presents a prominent surface along the bc-plane is reasonable. The second difference is the main cation in the liquid part of the precursor. S1 forms in the solution with a considerable amount of Fe2+, while S2 forms in the solution of Li+ salts. It is known that cations with a high charge number generally show strong adsorption on crystal nucleus, and they may act as a template to affect the growth speed of different facets. On the basis of above discussion, S1 and S2 present different growth kinetics due to different precursors. Here EG may contribute for the distinct crystal orientations due to its suitable viscosity.17,19
crystal orientation are still needed, and easy control techniques are still great challenge for LiFePO4 preparation. In this Letter, LiFePO4 nanoplates, prepared with the same starting materials under the same solvothermal process, show different crystal orientations and electrochemical performances. The mechanism for the different crystal growth is attempted. It provides an effective and reasonable way to prepare nanoLiFePO4 with excellent rate performance. In the typical route, FeSO4·7H2O, H3PO4, and LiOH·H2O were used as starting materials in a molar ratio of 1:1.5:2.7 and ethylene glycol (EG) was applied as solvent. For the synthesis of LiFePO4 nanoplates with the {010} face prominent, namely, S1, H3PO4 was slowly introduced to the LiOH solution under stirring. When a white suspension formed through the neutralization reaction, FeSO4 solution was added into the suspension. For the synthesis of LiFePO4 nanoplates with the {100} face prominent, namely, S2, the feeding sequence was changed. In this case, H3PO4 was slowly introduced to the FeSO4 solution under stirring. Then LiOH solution was added into the mixture, resulting in a sticky and dark green suspension. Then the same procedures were followed: after stirring for 30 min, the mixtures were transferred into a sealed autoclave, heated to 180 °C for 10 h, and then cooled down to room temperature. The obtained gray-green precipitates were washed with deionized water and ethanol. To achieve carbon coating, LiFePO4 nanoplates were mixed with 20 wt % of sucrose and then carbonized at 650 °C for 3 h in Ar atmosphere. After carbon coating, S1 and S2 become S1/C and S2/C, respectively. More details are in the Supporting Information (SI). Figure 1 shows the X-ray diffraction (XRD) patterns of S1 and S2, respectively. All of the reflection lines in the XRD
Figure 1. XRD spectra of S1 and S1, as well as Rietveld refinement of Pnma. S1 and S2 are LiFePO4 prepared under the same solvothermal conditions but with a different feeding sequence. The black line represents the observed pattern, the red cross corresponds to the calculated diffraction pattern, and the blue line is the difference pattern. FeLi% represents the percentage of Fe•Li antisite defects.
patterns can be indexed to an orthorhombic space group, Pnma (JCPDS card No. 81-1173), indicative of perfect crystalline structure of olivine LiFePO4. No characteristic lines of impurity phases are observed, and all lattice parameters coincide well with the standard, indicating the high purity of the samples. The Rietveld refinements using the Pnma phase provide acceptable fitting based on the Rwp and Rb fitting factors. However, a large mismatch of the (020) line can be observed B
dx.doi.org/10.1021/nl3027839 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 2. SEM images and TEM images (SEAD inside) of (a−c) S1 and (d−e) S2. The exposure facet is {010} for S1, while crystal growth orientations along the bc plane are prominent for S2.
Scheme 1. Reactions and Products during the Mixing Process for S1 and S2, Respectivelya
a
The solvothermal precursors are totally different for S1 and S2. For S1, the main solid is Li3PO4, and most of the Fe2+ ions remain in the liquid. However for S2, most of the Fe2+ ions exist as Fe3(PO4)2 precipitation, and about 2/3 of the total Li+ ions are in the liquid.
Figure 3. FTIR spectra of S1 and S1. According to Xue’s report,20 the symmetric stretching P−O vibration peak of the PO4 tetrahedron in defect-free LiFePO4 locates at 957 cm−1. The infrared absorption bands around 950 cm−1 imply that the defect concentrations of the Fe•Li antisite in both S1 and S2 are very low, indicative of the perfect crystal structure.
The Li ion migration in LiFePO4 is not only determined by particle size and diffusion pathway but also by point defect concentration. As reported, LiFePO4 prepared by a hydrothermal method at low temperatures innately forms Fe•Li antisite defects, which barricade Li+ migration. Then the antisite defects in S1 and S2 are characterized by Fourier transform infrared spectroscopy (FTIR; Figure 3). In particular, the broad bands with a shoulder above 900 cm−1 originate from the P−O symmetric and asymmetric stretching vibrations of olivine phosphate groups, respectively. A little shift between the two samples around 950 cm−1 (951 cm−1 for S2 and 946 cm−1 for S1) was obviously detected, while all the other peaks remain almost unchanged. According to Xue’s report,20 the symmetric stretching P−O vibration peak of the PO4 tetrahedron in defect-free LiFePO4 locates at 957 cm−1, while they are usually around 991 cm−1 or even absent for the LiFePO4 crystals prepared by the hydrothermal method.20,21 Then the infrared absorption bands around 950 cm−1 imply that the defect
concentrations of the Fe•Li antisite in both S1 and S2 are very low, indicative of the perfect crystal structure produced by the solvothermal method. This also further confirms the results of low intermixing of cations from XRD analysis. The electrochemical performances, including capacity and rate capability, of carbon-coated S1 and S2, namely, S1/C and S2/C, are evaluated (Figure 4). S1/C presents a high discharge capacity of about 164 mAh g−1, 160 mAh g−1, and 158 mAh g−1 at 0.1 C, 0.5 C, and 1 C, respectively, while S2/C presents 161 mAh g−1, 157 mAh g−1, and 155 mAh g−1 at 0.1 C, 0.5 C, and 1 C, respectively. The low Fe•Li defect concentration may contribute to these high capacities at low C-rates below 1 C for S1/C and S2/C. Though S1/C and S2/C present comparable capacities at low C-rates, their potentials at the middle of discharge capacity reveal that S1/C is superior to S2/ C
dx.doi.org/10.1021/nl3027839 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 5. Discharge capacity and Coloumbic efficiency vs cycle number plots of S1/C and S2/C. The loading density of the electrode is 17 mg cm−2. The voltage window is 2.0−4.2 V. The electrode formulation is active material (80 wt %), carbon (10 wt %), and binder (10 wt %). The poor reversibility of lithium metal anode may be responsible for the volatility in Coloumbic efficiency at a high C-rate.
For an electrochemical cell to deliver capacity at high rate, all parts of the Li+-electron path between the anode and the cathode active material have to be capable of sustaining this rate. The results in Figure 4 are obtained when testing a standard cathode preparation using 10 wt % carbon black and 10 wt % binder. Carbon black is added to facilitate electron transport from the active materials to the current collector. As this preparation has been optimized for materials that have substantially lower rates (
