Atomistic Origins of High Rate Capability and Capacity of N-Doped

Jan 30, 2014 - Graduate School of System and Information Engineering, University of ... been found to possess high rate capability and capacity for li...
0 downloads 0 Views 4MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Atomistic Origins of High Rate Capability and Capacity of N‑Doped Graphene for Lithium Storage Xi Wang,*,† Qunhong Weng,† Xizheng Liu,‡ Xuebin Wang,† Dai-Ming Tang,*,† Wei Tian,*,† Chao Zhang,† Wei Yi,† Dequan Liu,† Yoshio Bando,† and Dmitri Golberg*,† †

International Center for Young Scientists (ICYS), World Premier International (WPI) Center for Materials Nanoarchitectonics (MANA), National Institute for Materials Science (NIMS), Namiki 1-1, Tsukuba, Ibaraki 305-0044, Japan ‡ Graduate School of System and Information Engineering, University of Tsukuba, Tennoudai 1-1-1, Tsukuba, 305-8573, Japan S Supporting Information *

ABSTRACT: Distinct from pure graphene, N-doped graphene (GN) has been found to possess high rate capability and capacity for lithium storage. However, there has still been a lack of direct experimental evidence and fundamental understanding of the storage mechanisms at the atomic scale, which may shed a new light on the reasons of the ultrafast lithium storage property and high capacity for GN. Here we report on the atomistic insights of the GN energy storage as revealed by in situ transmission electron microscopy (TEM). The lithiation process on edges and basal planes is directly visualized, the pyrrolic N “hole” defect and the perturbed solid-electrolyte-interface configurations are observed, and charge transfer states for three N-existing forms are also investigated. In situ high-resolution TEM experiments together with theoretical calculations provide a solid evidence that enlarged edge {0002} spacings and surface hole defects result in improved surface capacitive effects and thus high rate capability and the high capacity are owing to short-distance orderings at the edges during discharging and numerous surface defects; the phenomena cannot be understood previously by standard electron or X-ray diffraction analyses. KEYWORDS: N-doped graphene, lithium storage, atomistic origins, high rate capability, high capacity

R

experimental evidence and fundamental understanding of the lithium storage mechanisms for GN at the atomic scale. Revealing these for GN may shed light on the reasons of its ultrafast lithium storage feature and high capacity found in the experiments. Also, there still remain other issues, such as the contribution of surface capacitive charge to the whole energy, the differences between the material edges, and its surface (for both GN and G) in storing mechanism. To answer these questions, fabrication of advanced GN anodes and building a GN-based nanobattery are critical. In this study, we first created a binder-free N-doped graphene paper anode via a simple solution method, which exhibited both high capacity and ultrafast lithium storage property. Then we constructed a GNbased nano-LIBs device and analyzed its atomistic lithium storage mechanism. Finally, combined high-resolution transmission electron microscopy (HRTEM) experiments and theoretical structure calculations allow us to find the reasons for high capacity and rate capability to disclose the different storing mechanisms (for both GN and G, and also their edges and basal planes) and to investigate the defect configurations in GN and effects of three N-substituted configurations on lithium

ecently, advanced energy storage devices including lithium ion batteries (LIBs) have become one of the most important topics,1−6 which have attracted much attention in the scientific and industrial fields.7−9 There are increasing interests in developing high-power and high-energy anode materials for the next generation high-performance rechargeable LIBs.10−12 Among them, N-doped graphenes (GN) are expected to be a good candidate material, because GN has the intrinsically superior electrical conductivity for fast electron transport, high surface area, open and flexible porous structures available for numerous lithium storage sites, and short Li+ diffusion distances.13−18 For example, Sun et al. synthesized a series of GN nanosheet anodes13,14 with excellent cycle life and high lithium storage capability. Moreover, Cheng’s group reported the synthesis of GN by using NH3 thermal treatment of pristine graphene and found that it had exhibited ultrahigh capacity and rate performance.15 Ajayan and co-workers also reported GN film anodes with the enhanced electrochemical performance.16 On the other hand, direct visualization of the lithiation/ delithiation process can provide important insights into graphene-based LIBs and guide the development of advanced LIBs for powering future electrical vehicles and devices.19−22 For example, Liu and Huang et al.20 created a pure graphene (G) based battery and in situ observed lithiation/delithiation processes within it. Despite a big success achieved in both experiments and theoretical simulations on GN,23−27 there is no direct © XXXX American Chemical Society

Received: October 15, 2013 Revised: January 27, 2014

A

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 1. (a) High-resolution N1s XPS spectra of GN. N1, N2, and N3 represent graphitic N, pyrrolic N, and pyridinic N, respectively. (b) Raman spectra of G and GN at 1200−1800 cm−1. (c) A photograph showing the flexibility of a GN paper. (d) SEM image of the cross section of a piece of GN thin film. Scale bar: 10 μm. (e) N-elemental mapping of a GN nanosheet, the inset shows its HAADF-STEM image.

Figure 2. (a) Cycle performance and Coulombic efficiency and (b) galvanostatic charge/discharge profile of a GN electrode at C/5 between 3.0 and 0.05 V versus Li+/Li. (c) Comparative rate capability of GN and G, (d) Nyquist plots of GN and G.

B

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 3. (a) Typical cyclic voltammetric responses for GN and G at the sweep rate of 2 mV/s. (b) The corresponding comparison of the total stored charge distribution for GN and G, respectively.

and improved faradic capacitance of the surface and edge sites after N-doping and enhanced capacity by the enlarged interlayer distances (as discussed later). The electrochemical performance of the GN was then investigated at a current rate of C/5 (Figure 2). Note that the reversible capacity and cycle performance of the doped graphene are greatly improved in comparison with the pristine graphene. The initial Coulombic efficiency increases from 55% for G to 73% for GN, as shown in Figure 2a. It suggests that the nitrogen doping can, to some extent, improve the reversibility of lithium adsorption/desorption. During the first discharging, lithium will trap in porous electrode including GN and graphene but not release back, which also contributes to decreased efficiency. In addition, the nitrogen doping may suppress the electrolyte decomposition. The GN electrodes exhibit ultrahigh discharge capacity of 1284 mAh/g in the first cycle, which are much higher than 1080 mAh/g of the pristine G; this also demonstrates that GN has more energy storage sites than G. From the second cycle, illustrated in Figure 2b, GN exhibits better reversibility, nearly 100% Coulombic efficiency. Consequently, the reversible capacity retention is increased from 36% for the pristine graphene to 56% for GN after 20 cycles (Figure 2a). More significantly, GN electrode exhibits ultrafast lithium storage properties, as illustrated in Figure 2c. At a very high current rate of 50C, corresponding to a charge time of ∼72 s, the reversible capacity can still reach 432 mAh/g for GN, this value is much higher than the theoretical capacity of commercial graphite (372 mAh/g). These results are far superior to those of the pristine graphene (∼50 mAh/g at 50C) and other anode materials, such as graphite,33 porous carbon monoliths,34 graphitized carbon nanobeads,35 and carbon nanofibers,36 which clearly demonstrate that GN is a promising high-energy and high-power electrode material for LIBs. The ultrafast energy storage can be ascribed by both good electrical conductivity and the surface capacitive effects due to N-doping. This assumption can be verified by the electrochemical impedance spectroscopy (EIS) measurements. The Nyquist plots obtained were modeled and interpreted with the help of an appropriate electric equivalent circuit (Supporting Information Figure S2). It is revealed that the GN electrode shows much lower electrolyte resistances and charge transfer resistances than those of the pristine graphene. We also calculated the Li-ion diffusion coefficient for G and GN analyzing the low-frequency Warburg contribution in the EIS spectra.37 The value of GN (5.6 × 10−8 cm2 s−1) is obviously

adsorption; these studies were not possible before using traditional routes of electron and/or X-ray diffractions. GN paper was fabricated by a modified route previously used for making G-SnO2 paper.18 Therefore, certain amounts of nitrogen atoms must substitute carbon in G due to four N atoms per molecular in TCNQ−. As shown in X-ray photoemission spectroscopy (XPS) (Figure 1a), three fitted peaks in the N1s XPS spectra are assigned to pyridinic N (399.1 eV), pyrrolic N (400.2 eV), and graphitic or quaternary N (401.7 eV), respectively.28 The doping level is 3.9 atom %. The dominated types are pyrrolic N (N2), as clearly revealed by the inset of Figure 1a. Note that N2 (also N3) dopant can be located at the edge or surface “hole” defect sites in GN (inset, Figure 1a). Moreover, the fitted C1s core-level spectra (Figure S1, Supporting Information) also indicated the existence of sp2 CN and sp3 C−N. Figure 1b shows the Raman spectra of the pristine graphene and GN, in which there are two prominent peaks of D and G bands, typical characteristics of the chemically derived graphene. As inferred from Tuinstra-Koenig (TK) equation, the intensity ratio of D band to G band (ID/IG) is usually used to estimate the disordering degree of graphene.29,30 The relative high ID/IG value (1.05) for our GN suggests a more disordered phase compared with that of pristine G (0.58), which may be caused by the defects in graphene due to nitrogen doping. In addition, a downshift of the G band in the Raman spectra of GN was observed, which was caused by the inhomogeneous C−C and C−N (CN) bond distances.31,32 The as-prepared products look like a piece of flexible paper (Figure 1c). Scanning electron microscopy (SEM) analysis revealed that the film had a relatively uniform cross section (Figure 1d) and a disordered structure. This indicated that the majority of GN sheets did not restack back to graphite, despite of substantial heat-induced compression.12 Elemental mapping and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images (Figure 1e and inset) reveal that nitrogen heteroatoms are homogeneously distributed over the whole graphene sheet. Importantly, we note that N atoms are inclined to be at the edges and some active sites on surface. This generally enables GN to form a flexible interconnected conducting network with a porous structure, through N2 or N3 sites, as shown in the inset of Figure 1a. In theory, it would lead to more energy storage sites. Therefore, the above-mentioned features can simultaneously bring the following advantages to lithium storage: ultrafast lithium storage feature due to increased electrical conductivity C

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 4. (a) Schematic illustration of the in situ TEM electrochemical experiment setup. A GN nanosheet is glued to an Au rod, and Li metal on the Au probe makes the other electrode. The ion liquid electrolyte is dropped on the Li metal for Li+ transport. A negative bias is applied to the GN terminal to initiate lithiation. (b) TEM image of the nano-LIB. TEM images of a pristine GN (c) and lithiated GN (d); the insets show the corresponding schematic structures. Scale bar: 5 nm. (e) Well-defined Li−K edge on the electron energy loss spectrum (EELS) of GN at the lithiated state. (f) EELS elemental map showing Li distribution after the lithiation.

larger than G (2.6 × 10−8 cm2 s−1). This indicates that the unique structures of GN introduced by N-doing can improve the diffusivity of Li ions in the electrode. That is, the ultrafast energy storage of N-doped graphene can be attributed to both good electrical conductivity and the surface capacitive effects due to N-doping. In general, the total stored energy for an electrode material should be divided into two portions: the diffusion controlled insertion process, and the surface capacitive effects that include faradaic contribution caused by a charge transfer process on surface atoms (pseudocapacitance) and nonfaradaic contribution caused by a double-layer effect.37−39 In regular cases of Listorage materials, the latter is too negligible to be considered. Herein, we have to take into account the surface capacitive contribution to the current response because of bending and disordering features and many introduced defects to GN. To quantitatively distinguish it from the total storing energy, the voltammetric sweep rate dependence approach is applied according to the following equation i(V) ν1/2

= k1ν1/2 + k 2

As depicted in Figure 3b, the stored charge can be divided into insertion capacity and capacitive charge storage. For a sweep rate of 2 mV/s, GN features much higher insertion capacities than G. It suggests that introducing N dopants significantly enhances the insertion capacity by providing more active sites. In addition, it should also be mentioned that GN exhibits both the highest values of total stored charge. This is likely because the looser packing edges and many defects on the surfaces remain to be accessible by the electrolyte. Moreover, some other valuable information can be obtained. For example, the presence of a peak at 0.5−0.6 V for both GN and G is assigned to the formation of a solid electrolyte interphase (SEI) film.15 It is noted that there is a slight increase in potential for GN compared with pure G, as illustrated in Figure 3a. This red shift indicates that the reaction between lithium and GN is faster than that of Li/G, namely, the energy barriers for diffusion and reaction of Li/GN are lower than those of Li/G. It in turn proves that GN shown doping can effectively suppress the side reactions of G with electrolyte to form the SEI film. Furthermore, two relatively sharp reduction/oxidation peaks are found in CV curves of GN; these indicate the Li+ intercalation and deintercalation into/from the N-doped graphene. It is a typical graphite-like behavior. This may be ascribed to the formation of relatively compact and robust paper-like structures after high-temperature annealing treatment. In turn, this feature enables high mass or volumetric energy density for GN. In addition, the irreversible capacity from both the SEI formation/surface passivation and incomplete delithiation/extraction occurred in the lithiation/ delithiation process; one cause is that lithium will trap in porous electrodes including GN and graphene but cannot release back in the first cycle. Compared with graphene, GN showed a negative effect on the formation of SEI film, leading to the improved reversible insertion/deinsertion of Li ions. Overall, N-doping did contribute to better electrochemical activity and reversibility. These electrochemical improvements for GN are generally believed to be caused by N-doping. However, there is still a challenge to understand the storage mechanism at the atomic scale. Thus, we in situ TEM constructed a nanobattery protopype based on a GN nanosheet. Figure 4a presents a schematic drawing of this nanoscaled battery that consists of a

(1)

Where i represents the current; ν is the sweep rate; k1ν and k2ν1/2 correspond to the current contributions from surface capacitive effects and diffusion controlled insertion processes, respectively. Cyclic voltammetry (CV) is used to analyze the charge storage behavior. Figure 3a shows typical CV for GN and G at a sweep rate of 2 mV/s. There are two sharp reduction/oxidation peaks at ∼0.13 and 0.33 V found in it (also Supporting Information Figure S3),40,41 which are attributed to the lithium insertion/extraction processes into/from carbon. The electrochemical behavior of GN is clearly different from that of the pure graphene. By contrast, a broad peak for G occurs between 0.6 and 1.25 V. This process can be expressed by a scheme xC (graphene) + Li+ + e− → LiCx

(2)

Where x represents the mole fraction of inserted lithium. The area under the CV curves stands for the total amount of stored charge which originates from both faradaic and nonfaradaic processes.37 D

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

4f) suggests that LiC6 is formed due to a small peak at ∼57−62 eV. There are two significant plasmon peaks found, which is mainly attributed to the formation of Li2O in the SEI layer. However, the chemical composition of SEI is complex, and it depends on many factors (solvents, additives, and reaction conditions). In our case, SEI should consist of the lithiated organic compound (in addition to Li2O) due to the use of liquid ion electrolyte. It is noted that a broad shoulder has appeared on the C−K edge (Figure S6, Supporting Information) after 330 eV; this may be ascribed to the organic species formed in the SEI film, which is caused by the reaction between Li and electrolyte.45 This differs from the in situ observation on G17 in which just Li2O solid electrolyte was used; in contrast, the organic electrolyte was used in our work. In fact, utilizing organic electrolyte is the most commonly used procedure in real cell electrochemical tests. The C−K edge also exhibits a blunt π* peak followed by the σ* band, thus it can be deduced that the GN consists of a graphitic network with the typical sp2-bonds and is rather poorly crystalline.28 This is consistent with the Raman characterization (Figure 1b). Now we discuss the lithiation on the basal plane of GN. Very recently, researchers have used spectroscopic techniques or scanning tunneling microscopy (STM) to confirm the nitrogenintroduced interfaces or point defects on the basal plane of graphene. Meyer et al.46 analyzed a charge transfer at the singleatom level for nitrogen-substitution point defects in graphene by atomically resolved HRTEM. However, the atomic scale analysis of lithiation on the basal surfaces of the N-doped graphenes has not been experimentally detected. We analyze the configuration changes in basal plane for GN by using in situ TEM. Also, we demonstrate that the SEI configurations on GN surfaces are perturbed by defects compared to the case of the pure graphene. Figure 5a shows the HRTEM image of the GN surface at the initial stage. There are two kinds of configurations: ordered and disordered domains. From the large-magnification image of the ordered area (Figure 5b), concave-convex appearance and even dislocation sites (marked by green circle) are seen. Note that the defects in this case are quite few. In addition, a well-defined honeycomb lattice is discernible in GN. Observation of the disordered area (Figure 5c) shows that more atom-level Nsubstitution hole defects or disordered sites (missing atoms) appear on the surface. Most of these hole sites form valleys that may be attributed to the formation of a continuous channel consisting of more than three N2 or N3 defect sites. Of course, disordered domains may be a big cracked “pit” resulting from many connected holes. This also agrees with the domination of N2 and N3 states among N-doing types. Therefore, we experimentally identify the surface defect configurations. In theory, these disordered and defects sites are more active and thus enable storing more energy. However, it is very hard to directly recognize the point defect sites due to TEM resolution limitations, because the C−N bond length in these nonperiodic defects configuration is nearly identical to C−C bond length in graphene (with a difference less than 2 pm). Figure 5d−f shows the HRTEM images of the directly lithiated GN. There are crystalline and amorphous SEI layers on GN surface and their main composition has been identified as Li2O (Figure 4). In comparison, only crystalline SEI was observed in the previous report (ref 20). It is indicated that Liinsertion reaction takes place in the basal plane of GN leading to different forms, such as LiCyNx23 and LiC6. It is unlikely for pure graphene in which only one form, LiC6, was detected and

single GN nanosheet, an ionic liquid (IL) electrolyte, and a piece of metallic lithium.42 Figure 4b demonstrates a pristine GN nanosheet with the length of 600 nm. Note that many corrugations and scrolling are apparent on GN due to the substitution of C with N atoms. The edge of GN is clearly shown in Figure 4c, where an enlarged (0002) interplanar distance of 0.36 nm for GN is measured compared with that of 0.34 nm for G.20 This means that a much more loosened space between nanosheets exists in GN than in graphene, resulting in much more energy storage sites. From the atomistic point of view, there are two different structural regions in GN nanosheets. The first one is the basal plane, consisting of two-dimensional conjugated sp2 carbon atoms. The second one is the edge, few-atoms-thick defective graphitic line of carbon atoms having dandling bonds. Therefore, the electrochemical behavior of graphitic basal plane is believed to be different from that of its edge. However, this assumption is still controversial according to previous studies.43,44 Herein, we thus designed the lithiation processes for both basal plane and edge of GN using in situ TEM technology. First, when a potential of −3 V is applied between the GN and the Li metal electrodes, lithiation process takes place along the GN edge. The lithiation is initiated from the edge to the center. Figure 4d shows the edge morphology of the lithiated GN after 10 min of reaction. An SEI layer is formed on the surface of GN, whose main composition is identified to be Li2O, as revealed by the electron diffraction patterns (EDPs) of lithiated GN (Figure S4, Supporting Information). Also, as shown in Supporting Information Figure S5, the lattice spacing of 0.27 nm in the SEI layer can well be attributed to (111) plane of the Li2O nanocrystals. The averaged-spacing of the few-layer GN (0002) plane was increased from 3.6 to 4.2 Ǻ after lithiation (Figure 4d); this corresponds to a 16.6% expansion induced by lithium intercalation into the graphene layers. Also, more disordered structures are formed, and the original long-distance orderings have transformed into many short-rang ones, resulting in much more porous architecture and thus more storage sites (Figure S5, Supporting Information). This is quite different from the previously reported G materials in which pristine crystalline orientation was well maintained with a smaller d0002 enlargement (3.6 Å, 7.2% increase) after the lithiation.20 The inset illustration shown in Figure 4d depicts the structure of the lithiated GN, indicating that more defects and loose spaces between GN layers have been formed for ultrafast lithium storage. In addition, the transformation from long-distance orderings into short ones would largely affect the initial reversibility; that is why the Coulombic efficiency of GN is lowered for the first cycle. On the other hand, just as in the case of the amorphous or poorly crystalline silicon anodes that can store more energy than their crystalline forms, the as-made short-distance ordered GN structures should benefit from increasing active sites for storing energy. Note that the thinner SEI layers (∼2 nm thickness) are also apparent for GN when compared with G (∼5 nm thickness). This can be explained in line with the fact that doping can effectively suppress the side reactions of G with electrolyte to form the SEI film. This result is also inconsistent with the Coulombic efficiency of GN. The inhomogeneousn Li distribution is observed from spatially resolved Li EELS maps (Figure 4e). It indicates that the edge and some actives sites of the surface are preferred for Li insertion. The corresponding Li−K edge spectrum (Figure E

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

pure graphene. Finally, on the basis of comparisons of pristine GN’s surface with that of the lithiated GN, high concentration of nonperiodic defects is apparent, for example, bending and “wrinkles” are observed in Figure 5f. It further indicates that more effective charge transfer occurs due to the Li-insertion into these N-substituted sites. Especially, N2 and N3 induced defects are responsible for this. In order to better understand the differences between GN and G (and, also, between the edge and basal planes) with respect to the storage mechanisms, theoretical calculations were performed. Two structural regions (Scheme 1) were taken into Scheme 1. Illustrations of G and GN Nanosheets for LiInsertion Viewed from the Edge and Basal Plan Directions

consideration. First, there exists big discrepancy for G and GN in both structural geometries and bonding configurations. For example, G exhibits a uniform interlayer spacing distance and flat surface (Figure S7, Supporting Information). By contrast, the structure of GN shows the enlarged average (0002) distance, as well as many holes, and points defects owing to Nsubstitution. Obviously, these enable GN simultaneously to realize both high rate performance and high capacity. For example, N doping not only increases the electrical conductivity but also electrochemical activity of GN. In addition, in order to compare the conductivity of GN and the pure graphene papers, the standard Four Probe Method was used here. It is found that the conductivity of GN paper (1.0 × 105 S/m) at room temperature is dramatically improved, three times larger than that of the pure graphene paper. This deduction has also been confirmed by EIS tests (Figure 2d). In addition, the disordered configurations in both edge and surface regions (such as corrugations and scrolling) can improve faradic capacitance contribution in the total storing energy. This is also testified by the CV responses and analysis mentioned above (Figure 3). These two factors can lead to the GN ultrafast lithium storage. Also, the defects on surface and enlarged spacings distances caused by N3-GN and N2-GN not only provide more active sites for Li insertion, but also bring much larger Li adsorption energies at the vacancy sites. This leads to a low energy barrier for Li to penetrate into the defects. These factors will result in high storage capacity for GN. Furthermore, paperlike GN has other advantageous characteristics for the Li storage, such as peculiar 3D framework,47 high surface area,48 open porous structure,49 and mechanical flexibility.50 Above all, GN is

Figure 5. HRTEM images of surface structural evolution of GN during the discharging process: (a−c) GN, (d−f) lithiated GN. Panels b,c show the enlarged (FFT inverse) images of the corresponding framed parts marked in panel a; similar cases are also shown in panels e,f (FFT inverse image). The green circles represent some dislocations in GN (b), some hole defect sites (c), and some dislocations in the SEI layer (e).

no remarkable SEI changes had been found. In addition, two types of SEI films may be correlated with the ordered and disordered areas that originally exist in GN. In addition, many defects are found in the crystalline Li2O-dominated SEI film, such as stacking faults and dislocations, as illustrated in Figure 5e,f (marked by a green circle). It suggests that Li-intercalation reaction in basal plane of the GN is heterogeneous. Three different N defects may be responsible for this phenomenon, because they are much more active and thus lead to an easier Li incorporation. According to DFT calculation,23 N-doped graphene shows more charge transfer, larger average adsorption energy for Li, and thus higher capacity, when compared with F

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

calculations. This material is available free of charge via the Internet at http://pubs.acs.org.

favorable for fast electron and ion transfer resulting in superior rate capability, high capacity, and long cycle life. Then we study the effect of three N-substituted configurations on lithium adsorption in GN. As shown in Scheme 1b, there are three main differences for three types of defects. (1) The optimal geometry on lithium adsorption is different. For example, the most stable site of the graphitic N (N1-GN) is on top of a C atom, while Li atom is energetically more favorable to be adsorbed on the center of the defect for the pyrrolic and pyridinic defect structures (N2-GN and N3-GN).23 In contrast, for Li adsorption in the pure G, Li prefers to occupy the hollow sites of the hexagonal ring, namely, creating only one formLiC6. This is correlated with the binding energy based on firstprinciple density functional theory (DFT) calculations.23,51 As seen from Table S1 (Supporting Information), obviously N3GN and N2-GN exhibit much larger values than G, N1-GN has the smallest one. As a result, the N2-GN and N3-GN (here mainly N2) defects provide a deficiency to gain electron from the Li atom, whereas N1-GN shows a negative effect on Li adsorption. (2) Similarly, the charge transfer from Li is different. Li donates 0.82 |e| charge to GN for N2-GN, much larger than 0.08 |e| to N1-GN.23 This further demonstrates that pyrrolic N sites show higher efficiency to enhance the interaction between Li atoms and GN surface, leading to ultrafast lithium storage. (3) The pyrrolic N adsorbs Li much easier than graphitic N and pure G due to its larger average adsorption energy for N2-GN. Therefore, the ultrahigh capacity of 1198 mAh/g natural for N2-GN system is much larger than for G and N1-GN ones.23 Above all, after nitrogen doping the number of active binding sites for Li adsorption in the present GN increased, especially for a N2-GN configuration. In summary, we revealed the origin of the ultrafast lithium storage features and high capacity for N-doped graphene through fabrication of an advanced GN anode and building a prototype GN-based nanobattery device. The discharging process on edge and basal planes is directly visualized and the resultant disordered surface and nonperiodic hole defects in GN and nonuniform SEI layer in lithiated GN are also found. The complex structural changes taking place during a lithiation process enable more effective charge transfer. The calculations were helpful for the interpretation of experimental results, such as defects-discrepancy metrics for energy storage. We provide compelling evidence that enlarged edge {0002} spacings and surface hole defects have resulted in the improved surface capacitive effects. Thus high rate capability and high capacity are owing to short-distance orderings formed on edges and profound surface defects during discharging. We believe that high power and impressive energy capabilities of a GN paper electrode and in situ atomistic TEM study of its storage mechanism paired with theoretical calculations will open up an opportunity to develop optimized high-performance electrochemical storage devices.

■

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: *E-mail: *E-mail: *E-mail:

(X.W.) [email protected]. (D.-M.T.) [email protected]. (W.T.) [email protected]. (D.G.) [email protected].

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS

■

REFERENCES

This work was supported by the International Center for Young Scientists (ICYS), World Premier International (WPI) Research Center on Materials Nanoarchitectonics (MANA), MEXT, Japan.

(1) Armand, M.; Tarascon, J. M. Nature 2008, 451, 652−657. (2) Gao, W.; Singh, N.; Song, L.; Liu, Z.; Reddy, A. L. M.; Ci, L.; Vajtai, R.; Zhang, Q.; Wei, B. Q.; Ajayan, P. M. Nat. Nanotechnol. 2011, 6, 496−500. (3) Chen, Z.; Ren, W.; Gao, L.; Liu, B.; Pei, S.; Cheng, H. -M. Nat. Mater. 2011, 10, 424−428. (4) Miller, J. R.; Outlaw, R. A.; Holloway, B. C. Science 2010, 329, 1637−1639. (5) Ghaffari, M.; Zhou, Y.; Xu, H.; Lin, M.; Kim, T.; Ruoff, R. S. Adv. Mater. 2013, 25, 4879−4885. (6) Hsu, P. -C.; Wang, S.; Narasimhan, V. K.; Kong, D.; Lee, H. R.; Cui, Y. Nat. Commun. 2013, 4, 2522. (7) Wu, H.; Ruan, Z.; Hsu, P. -C.; Wang, S.; Yu, Z.; Carney, T. J.; Hu, L.; Fan, S.; Cui, Y. Nat. Nanotech. 2013, 8, 421−425. (8) Dunn, B.; Kamath, H.; Tarascon, J. M. Science 2011, 334, 928− 935. (9) Pasta, M.; Wessells, C. D.; Huggins, R. A.; Cui, Y. Nat. Commun. 2012, 3, 1149. (10) Yao, Y.; Huo, K.; Hu, L.; Liu, N.; Cha, J. J.; McDowell, M. T.; Chu, P. K.; Cui, Y. ACS Nano 2011, 5, 8346−8351. (11) Han, H.; Song, T.; Bae, J. -Y.; Nazar, L. F.; Kim, H.; Paik, U. Energy Environ. Sci. 2011, 4, 4532−4536. (12) Jeong, H. M.; Lee, J. W.; Shin, W. H.; Choi, Y. J.; Shin, H. J.; Kang, J. K.; Choi, J. W. Nano Lett. 2011, 11, 2472−2477. (13) Li, X.; Geng, D.; Zhang, Y.; Meng, X.; Li, R.; Sun, X. Electrochem. Commun. 2011, 13, 822−825. (14) Nethravathi, C.; Rajamathi, C. R.; Rajamathi, M.; Gautam, U. K.; Wang, X.; Golberg, D.; Bando, Y. ACS Appl. Mater. Interfaces 2013, 5, 2708−2714. (15) Wu, Z. -S.; Ren, W.; Xu, L.; Li, F.; Cheng, H. -M. ACS Nano 2011, 5, 5463−5471. (16) Reddy, A. L. M.; Srivastava, A.; Gowda, S. R.; Gullapalli, H.; Dubey, M.; Ajayan, P. M. ACS Nano 2010, 4, 6337−6342. (17) Yang, X.; Cheng, C.; Wang, Y.; Qiu, L.; Li, D. Science 2013, 341, 534−537. (18) Wang, X.; Cao, X.; Bourgeois, L.; Guan, H.; Chen, S.; Zhong, Y.; Tang, D. -M.; Li, H.; Zhai, T.; Li, L.; Bando, Y.; Golberg, D. Adv. Funct. Mater. 2012, 22, 2682−2690. (19) Huang, J. Y.; Li, Z.; Wang, C. M.; Sullivan, J. P.; Xu, W.; Zhang, L. Q.; Mao, S. X.; Hudak, N. S.; Liu, X. H.; Subramanian, A.; Fan, H.; Qi, L.; Kushima, A.; Li, J. Science 2010, 330, 1515−1520.

ASSOCIATED CONTENT

S Supporting Information *

High-resolution C1s XPS spectra of GN. The cyclic voltammetric responses of GN and G at the different sweep rates. SAED pattern of the lithiated GN. HRTEM images in the lithiation process of GN. The C−K EELS spectrum of GN at the lithiated state. HRTEM image of pure G, side view, and the corresponding structural scheme constructed by Material Studio software. Comparison of the theoretical capacity of graphite, G and the different types of GN according to DFT G

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

(20) Liu, X. H.; Wang, J. W.; Liu, Y.; Zheng, H.; Kushima, A.; Huang, S.; Zhu, T.; Mao, S. X.; Li, J.; Zhang, S.; Lu, W.; Tour, J. M.; Huang, J. Y. Carbon 2012, 50, 3836−3844. (21) Liu, X. H.; Huang, S.; Fan, F.; Huang, X.; Liu, Y.; Krylyuk, S.; Yoo, J.; Dayeh, S. A.; Davydov, A. V.; Mao, S. X.; Picraux, S. T.; Zhang, S.; Li, J.; Zhu, T.; Huang, J. Y. Nat. Nanotechnol. 2012, 7, 749−756. (22) Wang, J. W.; Narayanan, S.; Huang, J. Y.; Zhang, Z.; Zhu, T.; Mao, S. X. Nat. Commun. 2013, 4, 2340. (23) Ma, C.; Shao, X.; Cao, D. J. Mater. Chem. 2012, 22, 8911−8915. (24) Jin, Z.; Yao, J.; Kittrell, C.; Tour, J. M. ACS Nano 2011, 5, 4112−4117. (25) Panchakarla, L. S.; Subrahmanyam, K. S.; Saha, S. K.; Govindaraj, A.; Krishnamurthy, H. R.; Waghmare, U. V.; Rao, C. N. R. Adv. Mater. 2009, 21, 4726−4730. (26) Wang, X.; Tian, W.; Liu, D.; Zhi, C.; Bando, Y.; Golberg, D. Nano Energy 2013, 2, 257−267. (27) Giannis, M.; George, F. J. Chem. Phys. 2006, 125, 204707. (28) Wang, H.; Maiyalagan, T.; Wang, X. ACS Catal. 2012, 2, 781− 794. (29) Huang, X.; Yin, Z.; Wu, S.; Qi, X.; He, Q.; Zhang, Q.; Yan, Q.; Boey, F.; Zhang, H. Small 2011, 7, 1876−1902. (30) Wei, D.; Liu, Y. Adv. Mater. 2010, 22, 3225−3241. (31) Deng, D.; Pan, X.; Yu, L.; Cui, Y.; Jiang, Y.; Qi, J.; Li, W. X.; Fu, Q.; Ma, X.; Xue, Q.; Sun, G.; Bao, X. Chem. Mater. 2011, 23, 1188− 1193. (32) Li, X.; Wang, H.; Robinson, J. T.; Sanchez, H.; Diankov, G.; Dai, H. J. Am. Chem. Soc. 2009, 131, 15939−15944. (33) Buqa, H.; Goers, D.; Holzapfel, M.; Spahr, M. E.; Novak, P. J. Electrochem. Soc. 2005, 152, A474−A481. (34) Hu, Y. S.; Adelhelm, P.; Smarsly, B. M.; Hore, S.; Antonietti, M.; Maier, J. Adv. Funct. Mater. 2007, 17, 1873−1878. (35) Wang, H. Y.; Abe, T.; Maruyama, S.; Iriyama, Y.; Ogumi, Z.; Yoshikawa, K. Adv. Mater. 2005, 17, 2857−2860. (36) Subramanian, V.; Zhu, H. W.; Wei, B. Q. J. Phys. Chem. B 2006, 110, 7178−7183. (37) Jiang, Z.; Pei, B.; Manthiram, A. J. Mater. Chem. A 2013, 1, 7775−7781. (38) Brezesinski, T.; Wang, J.; Polleux, J.; Dunn, B.; Tolbert, S. H. J. Am. Chem. Soc. 2009, 131, 1802−1809. (39) Wang, J.; Polleux, J.; Lim, J.; Dunn, B. J. Phys. Chem. C 2007, 111, 14925−14931. (40) Lian, P.; Zhu, X.; Liang, S.; Li, Z.; Yang, W.; Wang, H.; Lian, P.; Zhu, X.; Liang, S.; Li, Z.; Yang, W.; Wang, H. Electrochim. Acta 2010, 55, 3909−3914. (41) Liu, W.; Kuo, S.; Lin, C.; Chiu, Y.; Su, C.; Wu, H.; Hsieh, C. Open Mater. Sci. J. 2011, 5, 236−241. (42) Wang, X.; Tang, D. -M.; Li, H.; Yi, W.; Zhai, T.; Bando, Y.; Golberg, D. Chem. Commun. 2012, 48, 4812−4814. (43) Yuan, W.; Zhou, Y.; Li, Y.; Li, C.; Peng, H.; Zhang, J.; Liu, Z.; Dai, L.; Shi, G. Sci. Rep. 2013, 3, 2248. (44) Banerjee, S. ACS Nano 2012, 7, 834−843. (45) Wang, F.; Graetz, J.; Moreno, M. S.; Ma, C.; Wu, L.; Volkov, V.; Zhu, Y. ACS Nano 2011, 5, 1190−1197. (46) Meyer, J. C.; Kurasch, S.; Park, H. J.; Skakalova, V.; Künzel, D.; Groß, A.; Chuvilin, A.; Algara-Siller, G.; Roth, S.; Iwasaki, T.; Starke, U.; Smet, J. H.; Kaiser, U. Nat. Mater. 2011, 10, 209−215. (47) Guan, H.; Wang, X.; Li, H.; Zhi, C.; Zhai, T.; Bando, Y.; Golberg, D. Chem. Commun. 2012, 48, 4878−4880. (48) Wang, X.; Guan, H.; Chen, S.; Li, H.; Bando, Y.; Golberg, D. Chem. Commun. 2011, 47, 12280−12282. (49) Wang, X.; Tian, W.; Zhi, C.; Zhai, T.; Bando, Y.; Golberg, D. J. Mater. Chem. 2012, 22, 23310−23326. (50) Guan, H.; Wang, X.; Chen, S.; Bando, Y.; Golberg, D. Chem. Commun. 2011, 47, 12098−12100. (51) Sun, Y.; Zhao, L.; Lu, X.; Gu, L.; Hu, Y. -S.; Li, H.; Armand, M.; Ikuhara, Y.; Chen, L.; Huang, X. Nat. Commun. 2013, 4, 1870.

H

dx.doi.org/10.1021/nl4038592 | Nano Lett. XXXX, XXX, XXX−XXX