High-quality mesoporous graphene particles as high-energy and fast-charging anodes for lithium-ion batteries

The application of graphene for electrochemical energy storage has received tremendous attention; however, challenges remain in synthesis and other aspects. Here we report the synthesis of high-quality, nitrogen-doped, mesoporous graphene particles through chemical vapor deposition with magnesium-oxide particles as the catalyst and template. Such particles possess excellent structural and electrochemical stability, electronic and ionic conductivity, enabling their use as high-performance anodes with high reversible capacity, outstanding rate performance (e.g., 1,138 mA h g−1 at 0.2 C or 440 mA h g−1 at 60 C with a mass loading of 1 mg cm−2), and excellent cycling stability (e.g., >99% capacity retention for 500 cycles at 2 C with a mass loading of 1 mg cm−2). Interestingly, thick electrodes could be fabricated with high areal capacity and current density (e.g., 6.1 mA h cm−2 at 0.9 mA cm−2), providing an intriguing class of materials for lithium-ion batteries with high energy and power performance.

1. Likewise, the author said "The mesoporous structure provides lithium-storage sites, effective ion-transport pathways, as well as space to accommodate the volume change", it seems that proofs are not enough and convincing just from simple rate and cycling performance. There is no information for reaction or diffusion kinetics analysis on their materials.

Response:
We thank the reviewer for this comment and agree that reaction or diffusion kinetics analysis on HNMG materials should be added. Regarding the roles of porous structure in mass transport and chemical reaction, it is a common understanding that porous structure does provide surface sites and transport pathways in the science community. In term of using voids (pores) to accommodate the volume change of electrode materials during the charging and discharging, it is also a common strategy used in the battery community. To further address this comment, EIS studies were conducted on electrodes based on our mesoporous graphene and non-porous graphite. As expected, significantly higher ion-diffusion coefficient and lower charge-transfer resistance are observed for the mesoporous graphene electrode, which is consistent with the excellent rate performance observed. We have provided this information in the revised manuscript.
Please see page 8, line 28-page 9, line 8 in the revised manuscript: "The outstanding rate performance of the HNMG electrodes can be attributed from their fast ion-electron transport. Fig. S8 compares the Nyquist plots of a HNMG electrode and a traditional graphite electrode, where the HNMG electrode shows a significantly lower charge-transfer resistance. Note that, during lithiation/delithiation of an electrode, the exchange current density is generally inversely proportional to the charge-transfer resistance. 53 The significantly lower charge-transfer resistance observed for the HNMG electrode indicates a more effective lithiation/delithiation process. Furthermore, the diffusion coefficient of lithium ions within the electrodes was estimated using the Warburg impedance model. 54 As expected, the 2. The materials was paired with Li foil for high rate cycling performance, for example, 40 C (calculated for 29.76 mA cm -2 @1 mg cm -2 ) with capacity of ~500 mA h g -1 (calculated for ~0.5 mA h cm -2 ) are cycled for 3000 cycles. We are wondering that why conventional Li foil can sustain the high current density of 30.0 mA cm -2 with the areal capacity of 0.5 mA h cm -2 for stable cycling over 3000 times in unmodified carbonate-based electrolyte? This ultrahigh current density under this areal capacity will induce dramatical dendrite growth phenomenon.

Response:
We sincerely appreciate the reviewer's important comments. As reviewer rightly pointed out, this ultrahigh current density under high areal capacity will induce dramatically dendrite growth phenomenon. In our work, the mas loading of the HNMG electrode in this test is 0.25 mg cm -2 (rather than 1 mg cm -2 ). Therefore, the areal capacity of such electrodes is 0.125 mA h cm -2 (rather than 0.5 mA h cm -2 ).
To demonstrate that such cells can be operated under such a current density, Li||Li cells, as well as Li||HNMG cells, were assembled with separators having a hole at the center to allow effective penetration of dendrites. Li plating/stripping was conducted at a high current density of 45 mA cm -2 . With a plating/striping capacity of 0.125 mA h cm -2 per cycle, the Li||Li cell showed increasing voltage hysteresis and was shorted after 6,000 cycles. For the Li||HNMG cell, the voltage hysteresis was significantly smaller and the voltage remained unchanged for more than 6,000 cycles (Fig 1). Furthermore, the Li||HNMG cell could be cycled at a current density of 45 mA cm -2 and a higher capacity of 0.25 mA h cm -2 for 3,000 cycles without shorting (Fig 2).
The mitigated growth of dendrite of the Li||HNMG cell is associated with the unique structure and composition of the HNMG electrodes. For example, the high surface area could reduce the local current density; while the nitrogen doping makes the HNMG lithiophilic with sufficient nucleation sites and low nucleation overpotential.  Please see page 16, line 26-line 28 in the revised manuscript: "HNMG electrodes with a mass loading of 0.25 mg cm -2 were used for high-rate cycling performance testing, where 4 M lithium bis(fluorosulfonyl) imide in 1,2-dimethoxyethane was used as the electrolyte." 3. Besides the BET measurements for mesoporous morphology, there are no other proofs about how the mesopores work to affect ion-transportation for high-rate ability?
Response: We thank the reviewer for this valuable suggestion. It is commonly known that, porous structure, particularly, pore size and tortuosity, can dramatically affect mass transport. To address this comment, EIS studies were conducted on a HNMG electrode and a graphite electrode. It was found that the HNMG electrode exhibits a significantly higher lithium-ion-diffusion coefficient, which is two to three orders of magnitude higher than that of the graphite electrode; meanwhile, the charge-transfer resistance of the HNMG electrode is significantly lower than that of the non-porous graphite electrode. This study confirms the mesoporous structure facilitate ion transport with reduced charge-transport resistance. We have provided this information in the revised manuscript.
Please see page 8, line 28-page 9, line 8 in the revised manuscript: "The outstanding rate performance of the HNMG electrodes can be attributed from their fast ion-electron transport. Fig. S8 compares the Nyquist plots of a HNMG electrode and a traditional graphite electrode, where the HNMG electrode shows a significantly lower charge-transfer resistance. Note that, during lithiation/delithiation of an electrode, the exchange current density is generally inversely proportional to the charge-transfer resistance. 53 The significantly lower charge-transfer resistance observed for the HNMG electrode indicates a more effective lithiation/delithiation process. Furthermore, the diffusion coefficient of lithium ions within the electrodes was estimated using the Warburg impedance model. 54 As expected, the HNMG electrode exhibits a significantly higher lithium-ion-diffusion coefficient, which is two to three orders of magnitude higher than that of the graphite electrode, further confirming the roles of porous structure in facilitating the ion transport in the electrode (See the Supplementary Information for details, Fig. S9-Fig. S12)." Please see page 7, line 6-page 8, line 8 in the revised supplementary information: "According to the Warburg impedance model, Z re = δ ω -1/2 , where Z re is the real part of the resistance, δ is the Warburg prefactor and ω is the angular frequency. 28 δ is related to the diffusion coefficient of lithium ions (D) by where V m is the molar volume of the lithiated HNMG or graphite, F is the Faraday constant, A is the electrode area, and dE oc /dx is the gradient of the coulometric titration curve. dE oc /dx can be obtained from a plot of the open-circuit potential (E oc ) vs. the molar fraction of lithium "x" in the HNMG or graphite at each charged state.
Assuming that graphite and graphene have a similar molar volume (Vm) and that both the electrode have a similar electrode area (A), and F is the Faraday constant, therefore, δ and dE oc /dx related to the diffusion coefficient of lithium ions (D) by Based on the δ measured (16 Ω/s 0.5 and 64 Ω/s 0.5 for the HNMG and graphite electrode, respectively), the ratio of δ between the HNMG and the graphite electrode is 1/4. Based on the results shown in Fig. S9-Fig. S12, the ratio of dE oc /dx for the HNMG electrode vs the graphite electrode varies in the range of 5 to 15. Accordingly, the ratio of the diffusion coefficient of lithium ion in the HNMG and graphite electrode, [ ], can be estimated with a number ranging from 400 to 3600. The diffusion coefficient of lithium ion in the HNMG electrode, based on the calculation above, is 2 to 3 orders of magnitude higher than that of the graphite electrode." Supplementary Figure 9. The relationship between the real part of the impedance spectra (Z re ) and ω -1/2 in the low-frequency region, where ω is the angular frequency in the low-frequency region, ω = 2πf.   Fig. 3G, the lithiation time of a HNMG structure is 3 min, corresponding to the high charge/discharge rate of 20 C. And no significant change of the structure and morphology can be observed. Therefore, we think that these experiments allow us to directly observe the mechanical robustness of HNMG electrode during battery operation.

Supplementary
Please see page 11, line 3-line 15 in the revised manuscript: "The in situ TEM investigation by applying a voltage bias of 3 V was conducted using a setup illustrated in Fig 3G. A HNMG particle dispersed on a Cu half-grid was connected to a Li metal electrode that was deposited on a tungsten (W) wire; while the native Li 2 O layer on the Li electrode was used as the electrolyte. Applying a negative voltage to the Cu grid initiated the lithiation process of the HNMG particle, while the delithiation was started by applying a reversed voltage bias. Fig 3G also shows the structure and morphology evolution of the HNMG particle during lithiation and delithiation (also see the Supplementary Movie 1). Before lithiation (0 min), the HNMG particle exhibits a porous structure similar to that observed in Figure 1. The morphology, structure, and dimension of the particle remain unchanged after lithiation (3 min) and delithiation (6 min), confirming the structural stability of the particle during lithiation and delithiation. Particularly, the unchanged dimension of the HNMG particles during the charging and discharging preserves the electron-and ion-conduction networks of the electrode, which is essential to ensure the cycling stability." 5. SEM images showed the cycled particle, we are wonder why the surface of cycled one is smooth than the original one. More importantly, what happened when such high current density and areal capacity applied on nitrogen-doped mesoporous graphene particles?
Response: We sincerely appreciate the reviewer's time and important comments. This is an excellent question. It is noteworthy that the surface of N-doped graphene is full of lithiophilic functional groups, offering uniform nucleation sites and small nucleation overpotential (Angew. Chem. Int. Ed. 56, 7764-7768, 2017). After long-term cycles, the surface of HNMG electrode in lithium intercalation state can be covered with uniform and smooth Li deposits. Therefore, the surface of cycled one is smoother than the original one. And we have provided this information in the revised manuscript.
Please see page 10, line 28-page 11, line 2 in the revised manuscript: " Fig. 3D and 3F further show the SEM images of a HNMG particle before and after the cycling process, which show a similar morphology and indicate an excellent structural stability of the particles. It is noteworthy that N-doped graphene is relatively lithiophilic, offering uniform nucleation sites with small nucleation overpotential. 55 Consistently, it was found that the HNMG electrode appears smoother after the cycling process possibly due to the formation of uniform deposition on the particles." "55. Zhang, R. et al. Lithiophilic sites in doped graphene guide uniform lithium nucleation for dendrite-free lithium metal anodes. Angew. Chem. Int. Ed. 56, 7764-7768 (2017)."

The title about the nitrogen-doped graphene particles, however, we nearly found no information about how doped nitrogen atoms functions and contributes on high current density situations.
Response: We thank the Reviewer for highlighting this important point. The roles of nitrogen doping on energy-storage performance of graphene have been well documented. We have provided this information in the revised manuscript.
Please see page 3, line 15-line 18 in the revised manuscript: "(4) Nitrogen doping improves electrode-electrolyte interactions, provides lithiophilic surface moieties, and offers uniform nucleation sites with small nucleation overpotential, which has advantages in high energy density with fast-charging capability."

Response to Reviewer #2:
Overall, this is an excellent piece of work. The authors demonstrated a simple template method to prepare high-quality graphene electrodes for fast-charging and high-energy lithium ion batteries. The high-quality, nitrogen-doped, mesoporous graphene particles own higher capacity and great rate performance. The stability is also pretty good. The authors also carried out extensive characterization such as in situ TEM to find why the performance is so great. Their explanation is sound. This work is recommended for publication in Nature Communications after minor revision. Some detailed comments are provided as follows: 1. Theoretical capacity of N-doped mesoporous graphene should be different from pure graphene, authors set 1C of HNMG as 744 mA g -1 , this should be more careful.

Response:
We sincerely appreciate the reviewer's time and extremely valuable comments. As pointed out, surface modification can enhance the Li storage capability of graphene materials. For example, Wang et al. [1] reported that nitrogen-doped graphene nanosheets exhibit a high reversible capacity up to 900 mA h g -1 . Ma et al. [2] investigated lithium storage of N-doped graphene using first-principle calculations. The calculation suggests the capacity is depended on the doping structure, which is 1262, 1198 and 1087 mA h g -1 for the pyridinic, pyrrolic and graphitic structures, respectively. In this context, C-rate is commonly used to reflect the electrochemical performance of such materials. [1,3] For consistency, C-rate is also adapted as a performance index in this work (1C = 744 mA g -1 ). [3] Wang, X. et al. Atomistic origins of high rate capability and capacity of N-doped graphene for lithium storage. Nano Lett. 14, 1164-1171 (2014). figure 3A, the coulombic efficiency under 40C or 60C should be added.

Response:
We thank the reviewer for this comment and agree that the coulombic efficiency under 40 C or 60 C should be added. We have provided this information in the revised manuscript.
Please see page 10, line 6-line 7 in the revised manuscript: "For example, after cycling at 40 C and 60 C for 3000 cycles, HNMG electrodes still show a capacity of 475 and 436 mA h g -1 with a capacity retention of 99.2 % and 99.1%, respectively (Fig. 3A). Meanwhile, HNMG electrodes exhibit the high coulombic efficiency (Fig. S13)." Supplementary Figure 13. Coulombic efficiency for HNMG electrodes with a mass loading of 0.25 mg cm -2 at rates of 40 C and 60 C for 3000 cycles. Please see page 10, line 19-line 21 in the revised manuscript: "Considering various carbon materials have been synthesized using hard-template methods, their electrochemical performance is compared with that of the HNMG particles (Table S3). Clearly, the HNMG particles show outstanding performance."  19679-19683 (2012)."

Response to Reviewer #3:
I found this work to be very exciting and believe it could be a game changer in the lithium-ion battery world. The capacity and rate performance of their HNMG is surprisingly high. If this material is successfully scaled up, it could eliminate the need for silicon-based anodes and perhaps even lithium metal anodes. This paper should be accepted with minor revisions/corrections.
1. My one major concern regards the lithium metal counter electrode used in this work. The authors show cycling data of 500 to 3,000 cycles with their HNMG electrodes versus lithium metal electrodes, often at very high C-rates (40C to 60C in Figure 3A). What is the material loading (mg/cm 2 or mA h/cm 2 ) of the HNMG electrodes in Figure 3A? Response: We sincerely appreciate the reviewer's important comments. As pointed out, a high current density could induce dendrite growth. The electrode used in this study has a mass loading of 0.25 mg cm -2 , which allowed us to conduct these studies. Please also refer to the Response for the Comment #2, Reviewer 1. The related information has been incorporated to the revised manuscript.
To demonstrate that such cells can be operated under such a current density, Li||Li cells, as well as Li||HNMG cells, were assembled with separators having a hole at the center to allow effective penetration of dendrites. Li plating/stripping was conducted at a high current density of 45 mA cm -2 . With a plating/striping capacity of 0.125 mA h cm -2 per cycle, the Li||Li cell showed increasing voltage hysteresis and was shorted after 6,000 cycles. For the Li||HNMG cell, the voltage hysteresis was significantly smaller and the voltage remained unchanged for more than 6,000 cycles (Fig 1). Furthermore, the Li||HNMG cell could be cycled at a current density of 45 mA cm -2 and a higher capacity of 0.25 mA h cm -2 for 3,000 cycles without shorting (Fig 2).
The mitigated growth of dendrite of the Li||HNMG cell is associated with the unique structure and composition of the HNMG electrodes. For example, the high surface area could reduce the local current density; while the nitrogen doping makes the HNMG lithiophilic with sufficient nucleation sites and low nucleation overpotential.  Please see page 16, line 26-line 28 in the revised manuscript: "HNMG electrodes with a mass loading of 0.25 mg cm -2 were used for high-rate cycling performance testing, where 4 M lithium bis(fluorosulfonyl) imide in 1,2-dimethoxyethane was used as the electrolyte."