- Email: [email protected]
Rapid aqueous synthesis of ordered mesoporous carbons: Investigation of synthesis variables and application as anode materials for Li-ion batteries
Microporous and Mesoporous Materials 195 (2014) 92–101
Contents lists available at ScienceDirect
Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso
Rapid aqueous synthesis of ordered mesoporous carbons: Investigation of synthesis variables and application as anode materials for Li-ion batteries Alexandre F. Léonard ⇑, Cedric J. Gommes, Marie-Laure Piedboeuf, Jean-Paul Pirard, Nathalie Job Laboratory of Chemical Engineering – Nanomaterials, Catalysis, Electrochemistry, Institute of Chemistry (B6a), University of Liège, B-4000 Liège, Belgium
a r t i c l e
i n f o
Article history: Received 28 January 2014 Received in revised form 10 March 2014 Accepted 14 April 2014 Available online 22 April 2014 Keywords: Ordered mesoporous carbons Soft-templating Hexamethylenetetramine Anodes Li-ion batteries
a b s t r a c t Ordered mesoporous carbons (OMC) were synthesized via a direct templating pathway by a synthesis route that features short duration, moderate temperature and aqueous media. Resorcinol was used as carbon precursor and hexamethylenetetramine as a source of formaldehyde and ammonia to respectively cross-link the framework and regulate the pH. The temperature of the heat treatment leading to the formation of the solid polymer was shown to have a strong influence on the structural and textural parameters. In particular, moderate temperatures led to the coexistence of differently-sized entangled hexagonal mesostructures, whereas the higher temperatures led to a sharp decrease in the mesopore volume. The performance of these materials as anode materials for Li-ion batteries has been investigated in detail. Although these OMC show reversible capacities similar to those reported for hard carbons, their long-term cycling remains very stable for over 100 cycles of charge/discharge. The optimization of the reported short preparation pathway offers new possibilities regarding the application of ordered mesoporous carbons in various fields, such as energy storage, sorption and heterogeneous catalysis. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Li-based batteries hold a bright future to power ever higher energy-demanding portable electronics, electric vehicles or for the storage of energy produced from renewable but often intermittent sources, like windmills for example [1–3]. Indeed, such batteries display unique advantages, like high volumetric and massic energy densities, combined with the absence of memory effect and a high open-circuit voltage [4–6]. In the quest of even more performant Li-based batteries that feature best security issues, many researches around the world deal with the development of new (solid) electrolytes as well as new electrode materials [7–9]. On the anode side, graphite is currently the most used insertion compound, but its theoretical capacity is limited to 372 mA h/g, i.e. 1 Li per 6 C atoms [10]. Considering the relationship existing between the total capacity of a battery and the capacity of anode materials, it appears that the best anode should bear a capacity of about 1200 mA h/g when used with the current available cathode materials, like LiCoO2 or LiFePO4 [11]. Such a value of capacity could be attained with so-called hard carbons, i.e. porous carbons bearing a turbostratic structure. Among these, ordered mesoporous ⇑ Corresponding author. Tel.: +32 4 366 4875. E-mail address: [email protected] (A.F. Léonard). http://dx.doi.org/10.1016/j.micromeso.2014.04.028 1387-1811/Ó 2014 Elsevier Inc. All rights reserved.
carbons (OMC) represent a particular class of materials because of their highly regular mesostructure. It has indeed been reported that such materials could exhibit specific capacities up to 1100 mA h/g with a very good cycling stability that has been attributed to the regular mesoporous structure favorable to the impregnation of the carbon by the electrolyte [12–13]. The most widespread procedure to prepare ordered mesoporous carbons consists in a nanocasting route, where a mesoporous silica acts as hard template that leads to its pure carbon inverted replica [14]. The method as such is however impossible to scaleup since it is costly – synthesis of mesoporous silica as sacrificial template – and harmful – use of HF as etching agent. A more elegant and convenient pathway is the direct synthesis of ordered mesoporous carbons in presence of a surfactant as structuredirecting agent in a similar fashion as used for ordered mesoporous silica [15–17]. Moreover, the obtained material is expected to be more robust since the framework is continuous in comparison to a stacking of carbon rods in the case of the replica [18]. Pioneering work has been reported during the last years in the direct synthesis of ordered mesoporous carbons, using mainly non-ionic surfactant micelles to control the growth of polymeric networks based on phenol-derivatives and formaldehyde. The most widespread is the Evaporation-Induced Self-Assembly (EISA) method [19–20]. Via this pathway, highly ordered mesoporous
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
carbon films – or monoliths – can be obtained upon evaporating the solvent at ambient temperatures. Interestingly, this route is also very versatile since it allows for the direct fabrication of nanocomposites like C–Al2O3 or C–TiO2 [21–22]. Also, it is possible to incorporate directly heteroatoms leading to W, Mo, Ti, B or N-doped OMC, avoiding the necessity of tedious post-treatments [23–25]. Nevertheless, it should be noted also that this route requires the evaporation of large volumes of organic solvents, especially if scaling-up is considered. Synthesis of OMC was also reported using a phase separation method, where the growing polymer-rich phase separates from the solvent [26–27]. Highly acidic conditions (HCl as a catalyst) are nevertheless required, making this route ill-adapted when poisoning by Cl ions has to be avoided and when synthesis equipment is not designed for corrosive environments. Hydrothermal autoclaving processes, commonly employed in the preparation of zeolites and mesoporous oxides, were reported first by Zhao and colleagues [28]. Ordered mesoporous carbons were synthesized successfully, but adapted equipment is necessary to work under high temperature and high pressure. Finally, the best pathway for large-scale production could very well be the diluted aqueous route reported by Zhao and colleagues [29–30]. OMC with hexagonal or cubic stacking of their channels were reported, depending on the non-ionic surfactant employed. Nevertheless, this synthesis appears to be very sensitive both to temperature and pH of the mixture and this narrow window of conditions could be problematic if large-scale preparations have to be realized. Moreover, very long durations (5–7 days) are required for the polymerization to be accomplished. In fact, compared to mesoporous silica, the cooperative selfassembly process between surfactant and framework-building blocks is more difficult in the synthesis of OMC since it requires the presence of two precursors [31–32]. In that way, the key of a successful preparation relies on the careful control of the phenol–formaldehyde polycondensation, making it not too fast in order to allow for the interaction of the growing oligomers with the hydrophilic heads of the surfactant molecules. This was very well-illustrated by Liu et al. who described the rapid aqueous synthesis of OMC by using resorcinol with hexamethylenetetramine instead of formaldehyde as cross-linking agent [33–34]. This precursor releases in situ OH ions and formaldehyde upon hydrolysis, allowing for the control of pH and polymer network growth. Remarkably, this pathway avoids the use of metal or halide ions, making the resulting carbon free of any poisoning species. As a general trend, whatever the synthesis route employed, the surfactant used is a block-copolymer polyethyleneoxide–polypropyleneoxide–polyethyeleneoxide (PEO–PPO–PEO) and the regular stacking of mesopores in OMC always belong to 3D bi-continuous cubic (Ia3d), 2D hexagonal (P6mm) or body-centered cubic (Im3m) space groups [15–16]. This symmetry is mainly controlled by the choice of the surfactant, i.e. the hydrophilic/hydrophobic balance (PEO/PPO) that dictates the curvature of the micelles. The hydrophilic volume is nevertheless changed upon interaction between PEO moieties of the surfactant and the hydrophilic resol oligomers. For that reason, the phenol/surfactant ratio will have an influence on the structure of the final carbon [18,25,35]. Increasing the hydrophobic volume for instance by adding a swelling agent like trimethylbenzene or decane, which will locate in the cores of the micelles, will favor the formation of structures with less curvature. On the other hand, the use of reverse copolymers PPO–PEO–PPO in conjunction with hexane as a cosolvent with the EISA method has led to the formation OMC with intergrown hexagonal and cubic pore structures, where one or the other phase was favored upon tuning the amount of added cosolvent [36]. When using ammonia in addition to hexamethylenetetramine in the aqueous synthesis pathway, it was also shown that the pore structure of OMC could be oriented from body-centered cubic
93
(Im3m) to hexagonal (P6mm) space groups upon addition of trimethylbenzene (TMB) [34]. The same authors reported that, in the absence of ammonia, the presence of TMB was necessary to obtain ordered mesostructures [33]. In addition, temperature is known to play a key role in the orientation of the structure of ordered mesoporous materials, mainly by influencing the micelle packing parameter g, defined by the volume and length of surfactant chains as well as the effective area of the hydrophilic head groups [37]. Additionally, temperature can also have an influence on the pore size distribution of the final mesoporous material [38–39]. For these reasons and also to solve some reproducibility issues, we decided to explore in more detail the HMT-based aqueous synthesis route by realizing a systematic study of the influence of thermal treatment temperature on the parameters of the final ordered mesoporous carbons. The use of these materials as anodes for Li-ion batteries was investigated in terms of their electrochemical performances, which were compared to those of OMC obtained via the hard-templating (nanocasting) route. 2. Experimental 2.1. Reagents Resorcinol (1,3 dihydroxybenzene, C6H4(OH)2, 99%) and electrolyte LP-71 (1 M LiPF6 in Ethylenecarbonate:Diethylcarbonate:Dimethylcarbonate 1:1:1, EC:DEC:DMC) were supplied by Merck chemicals. Block copolymer [poly(ethylene oxide)b-poly(propylene oxide)-poly(ethylene oxide)] surfactant Pluronic F127 (EO106PO70EO106, Mav = 12,600) as well as hexamethylenetetramine (HMT, C6H12N4) were purchased from Sigma–Aldrich. 1,3,5-trimethylbenzene (TMB, C9H12, 99%) was bought from Acros chemicals and N-methyl-2-pyrrolidone (NMP, C5H9NO, 99.0+%) from Alfa Aesar. 2.2. Synthesis of ordered mesoporous carbons In a typical synthesis, 6.0 g of Pluronic F127 were dissolved in 108 ml deionized water in an autoclavable glass bottle. After stirring during 3 h at ambient temperature, 1.2 g of trimethylbenzene, 3.3 g of resorcinol and 2.1 g of HMT were added sequentially, leaving 1 h stirring between each. After 1 h further stirring, the bottle was sealed and heated in an oil bath overnight, at a precisely chosen temperature ranging from 65 to 127 °C. Depending on the chosen temperature, a pink or orange aggregate was observed, surrounded by a dark brown supernatant. The latter was discarded and the aggregate washed with deionized water and i-propanol, followed by drying in air. Pink to orange carbon polymer powders were recovered at this stage. Finally, the surfactant was removed upon carbonization of the materials under nitrogen in a tubular furnace, with a heating ramp of 2 °C min1 up to 800 °C, with a step of 2 h. 2.3. Characterizations The structure of the materials was investigated by powder X-ray diffraction (XRD) on a Siemens D-5000 diffractometer (Cu K-a radiation, 40 kV, 40 mA) as well as by Transmission Electron Microscopy (Tecnai T10, 100 kV). Selected samples were also characterized by Small Angle X-ray Scattering (SAXS) in order to get a better insight into the mesostructural order. The SAXS data were measured on a XEUSS SAXS/WAXS system from Xenocs, equipped with a Molybdenum K-a X-ray source. The 2D scattering patterns were rotationally averaged using the ConeX software [40] and corrected for background intensity. The scattered intensity was finally plotted against wave-vector q = 4p/k sin(h/2), where k is
94
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
the wavelength and h is the scattering angle. The textural features of the polymers and carbons were assessed by nitrogen adsorption–desorption data acquired on a Micromeritics ASAP 2420 MP at 77 K. Prior to measurements, the samples were outgassed at 170 °C (polymers) or 270 °C (carbons). The specific surface area was calculated using the BET (Brunauer–Emmet–Teller) method with adsorption data taken in the relative pressure range 0.01– 0.10. The micropore volume was calculated by the Dubinin– Radushkevich equation and the total pore volume was determined from the amount of gas adsorbed at a relative pressure of 0.99. Indeed, since these materials are mesoporous, their nitrogen adsorption–desorption isotherm is of type IV, with a saturation plateau at high relative pressures, meaning that the volume adsorbed at saturation corresponds to the total pore volume. The pore size distributions were determined from the DFT method, provided with the software of the apparatus. Thermogravimetric measurements were performed on a Setaram TG-DSC-111 microbalance under He with heating at 3 °C min1 or under air with heating at 2 °C min1. The particle sizes were measured by laser diffraction using a Malvern Mastersizer 2000 particle size analyzer. Prior to measurements, the OMC was dispersed in distilled water with high-intensity ultrasonic irradiation during 30 s. 2.4. Electrochemical analysis Prior to measurements, the materials to be tested were prepared as electrodes. OMC (92 wt.%) and PVDF (binder, 8 wt.%) were mixed in N-methyl-2-pyrrolidone (NMP) to form a homogeneous slurry. The solid/NMP mass ratio was fixed at 1/2.5. This ink was then spread on a Cu foil with a bar coater, the opening of the knife being adjusted at 100 lm. After drying at ambient temperature during 3 h and at 60 °C overnight, 13-mm disk electrodes were punched from this coating. The thickness of the coatings after drying was measured by stylus profilometry (Veeco Dektak 150, stylus radius: 12.5 lm, force: 1.00 mg). In order to have access to the thickness of the coating, disk electrodes were fixed on a microscope slide and the carbon was scratched off from the copper at regular intervals to set the baseline. The thickness was calculated as an average value taken from three electrodes with two scans each. The electrochemical measurements were carried out in homemade Swagelok-type cells, where the tested material acts as cathode and a Li-metal disk as anode. A Celgard separator soaked with 100 ll of LP71 (1 M LiPF6 in EC:DEC:DMC 1:1:1) electrolyte was placed in-between. The cell assembly was done in an Ar-filled glove-box (MBraun). Charge–discharge curves were recorded at C/5 between 0.005 and 1.5 V (vs. Li+/Li) with a Biologic VMP3 multichannel potentiostat.
is usually recorded for amorphous hard carbons, indicating the absence of long-range graphitic ordering of the graphene planes [42]. Indeed, in the case of carbon materials obtained from resol precursors, graphitization can only be reached after a heat treatment at temperatures higher than 2000 °C. Another way is to induce catalytic graphitization upon addition of heteroatoms or transition metals. In this case, the doped OMC display a better graphitization degree, even after thermal treatments at temperatures lower than 1000 °C [23–25]. Fig. 1 shows the low-angle powder XRD patterns of the samples prepared at different temperatures and pyrolyzed under N2. As a general trend, at temperatures above 115 °C, only one single reflection can be detected at low angles, indicative of a disordered wormhole-like array of mesopores of homogeneous sizes. In this case, that single peak is an indication of the average pore-to-pore separation distance. It could be hypothesized that too high temperatures either weaken the H-bondings between the hydrophilic heads of the surfactant micelles and the growing resorcinol-based polymers, or that the cross-linking of the framework becomes too fast, thereby disturbing the hexagonal organization of the surfactant-polymer supramolecular assemblies. In the intermediate range of temperatures, in addition to a very sharp reflection at very low angles (2h = 0.84–0.90°), other less intense peaks can be observed at angles in the regions 2h = 1.45–1.59° as well as 2h = 1.68–1.84°. According to literature, the presence of these secondary reflections indicates the long-range ordering of the mesoporous channels in a hexagonal array [43]. Indeed, mesoporous materials displaying a hexagonal ordering of their channels exhibit a typical four-peak pattern with a strong reflection line at low angle (1 0 0) and weaker peaks at higher angles (1 1 0, 2 0 0 and 2 1 0) that can be indexed on a hexagonal unit cell. From this symmetry, the unit cell parameter a0 can be calculated as:
a0 ¼ 2d100 =31=2
ð1Þ
where d100 is the repetition distance calculated from the first diffraction peak. This unit cell parameter represents the sum of the pore diameter and the thickness of the pore wall. Table 1 gives the values of the repetition distances and unit cell parameters as a function of synthesis temperature. The experimentally observed positions of the peaks on the XRD patterns (and the calculated corresponding distances) are consistent with a hexagonal symme-
3. Results and discussion 3.1. Structural features of the prepared materials If the temperature of the thermal treatment was kept below 72 °C, i.e. 65 and 70 °C in our case, no solid was recovered after the filtration and washing steps. This could be due to the fact that HMT is not hydrolyzed at low temperature [41], so that the framework cannot be cross-linked sufficiently to form solid particles that can be separated from the supernatant solution. Beyond that temperature, pink to orange aggregates can be isolated and processed to powders upon filtration and washing steps. A slight color change of the dried powders could be noticed, with a progressive shift from pink to orange with the temperature increase (Fig. S1). Wide-angle powder XRD measurements only show two broad peaks at 2h angles of about 24° and 44°. These correspond to the (0 0 2) and (1 0 0) diffractions of amorphous carbon. Such a pattern
Fig. 1. Low angle powder XRD patterns of the OMC materials prepared at different temperatures.
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101 Table 1 Structural parameters from the low angle powder XRD data of the OMC prepared at different temperatures. T (°C)
d100 (nm)a
d110 (nm)a
d200 (nm)a
a0 (nm)b
77 87 89 90 95 100 105 110 115 120 121
10.3 10.0 10.0 9.8 10.0 10.5 10.0 9.8 9.6 10.3 10.8
6.0 6.0 5.7 5.8 5.7 6.0 5.7 6.0 5.5 –c –c
5.0 5.0 4.9 4.8 4.9 5.1 5.1 5.0 4.6 –c –c
11.9 11.6 11.6 11.3 11.6 12.1 11.6 11.3 11.1 –c –c
a dhkl distances calculated from the Bragg law on the positions of the 2h reflection lines: dhkl = k/2sin(2h/2). b Unit cell parameter calculated on the basis of a hexagonal unit cell: a0 = 2d100/31/2. c Not observed.
try. Obviously, the unit cell parameter, as calculated from low angle XRD data, remains constant whatever the synthesis temperature, with values comprised between 11.3 and 12.1 nm. These structural data are very close to those reported by Liu et al. [33] The structural regularity of the materials prepared at temperatures ranging from 72 to 115 °C is further confirmed by TEM observations (Fig. 2). The honeycomb-like array of mesopores can clearly be observed and the 6-fold symmetry is confirmed by the FFT-calculations of these images shown in insets. The longitudinal views also confirm that the channels run parallel to each other. Very interestingly, some particles display a hexagonal macroscopic shape, suggesting a very good long-range ordering (Fig. 2A, C, and D). In contrast, for the material prepared at higher temperatures, the order is lost: only wormhole-like mesopores can be evidenced and the macroscopic shape of the particles becomes more random (Fig. 2H). At first glance, the materials prepared at the intermediate temperatures could be described as highly ordered mesoporous carbons with a hexagonal stacking of their channels. Nevertheless, a closer observation of the low-angle XRD patterns shows that in some cases, the main reflection line (1 0 0) seems split. For that reason, we decided to perform SAXS measurements on selected samples (Fig. 3). From these measurements, it appears clearly that different structures are present in these materials, especially for the lower temperatures. The analysis of the positions of the peaks suggests the presence of two frameworks with a hexagonal symmetry in the stacking of mesopores, but with different sizes. This feature is most visible on the pattern corresponding to the sample prepared at 95 °C and is certainly also present for the samples prepared at 77 and 100 °C. Above this temperature, the distinction of frameworks becomes less defined, but some shoulders present on the first most intense peak could also suggest the presence of hexagonal arrays bearing different sizes. The repetition distances as well as the unit cell parameters calculated from the SAXS data are given in Table 2 and fit well with an indexation on hexagonal unit cells. Based on these remarkable observations, we decided to examine the TEM images of the samples in more detail in order to assess the homogeneity of the structure. Although the differences observed from the SAXS data could not be distinguished by TEM, the distances measured between the channels very well correspond to those from the diffraction and diffusion data. Nevertheless, as can be seen on Fig. 4, an additional, smaller level of organization could be evidenced, which is illustrated by the white bars on the pictures. The observation of the images suggests that, in some way, the channel walls split for the formation of new mesoporous channels within a continuous network. The continuity of the framework is clearly seen in Fig. 4A and also supported from Fig. 4C and D where the angles between the planes of small and
95
larger regular arrays is close to 60°, in accordance with the hexagonal symmetry existing in these materials. The presence of this smaller repetition distance is however not supported by XRD or SAXS data, suggesting that the largest part of the material is constituted of channels with thick pore walls. Moreover, such an inhomogeneity could not be evidenced for the materials prepared at temperatures higher than 110 °C. It should also be mentioned that the SAXS data of the corresponding non-pyrolyzed as-prepared samples display the same patterns, but with peaks located at shorter lengths of the scattering vector q. From this, it can be stated that the pyrolysis step does not affect the global structure of the framework, but only induces an isotropic shrinkage. The shrinkage extent was calculated based on the reduction of the unit cell parameter a0 and is comprised between 35% and 38%, whatever the preparation temperature of the materials. To summarize, from a structural point of view, highly ordered mesoporous carbons with a hexagonal stacking of their channels can be prepared in a temperature interval ranging from 72 to 115 °C. The unit cell parameters are very close, whatever the temperature, but the lowest temperatures seem to favor the formation of materials displaying several hexagonal frameworks entangled inside one single continuous network. Moreover, TEM observations suggest the presence of an additional array of mesoporous channels with thin pore walls. A tentative explanation could be the progressive coalescence of pore walls with the progression of the polycondensation of the carbon polymer framework. In this case, in a first stage of the reaction, the formation of surfactant-carbon precursor supramolecular assemblies occurs. During this step, resorcinol interacts with the hydrophilic heads of the surfactant, whereas the hydrophobic parts join to form hybrid micelles. These assemblies then progressively join to form a continuous carbon polymer network with the surfactant molecules embedded, leading to the development of porosity. In a second stage, with further advancement of the reaction and continuous release of formaldehyde from the HMT precursor, the polymerization of the carbon polymer continues and eventually expels surfactant molecules, inducing the disappearance of ‘‘channels’’ and increasing the fraction of the carbon skeleton, and thus leading to thicker pore walls. This phenomenon would be even more pronounced if the preparation temperature is increased, explaining why such entangled differently-sized hexagonal arrays could not be evidenced above 110 °C. 3.2. Thermal analysis The washed and dried polymer/surfactant composites were analyzed by thermogravimetry–differential scanning calorimetry in order to determine their behavior during the carbonization step (Fig. S2). All of the investigated samples behave the same, independently of the preparation temperature. Under helium flow (Fig. S2a), a first small mass loss (2%) is recorded below 100 °C, corresponding to the evaporation of remaining solvent molecules. In a second stage, a large weight loss of 50–55% occurs between 300 and 400 °C, with an endothermic signal at 383 °C. This loss mainly corresponds to the decomposition of the surfactant. When the temperature is further increased up to 700 °C, a continuous mass loss up to 80% in total is recorded, which corresponds to the carbonization of the material. TG-DSC measurements under air atmosphere in turn show two distinct mass losses with exothermic signals at 310 and 480 °C, corresponding respectively to the burning of the surfactant and the carbon framework (Fig. S2b). 3.3. Textural parameters of the OMC as a function of preparation temperature Nitrogen adsorption–desorption measurements were performed in order to assess the pore volume, specific surface area
96
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Fig. 2. TEM images of the OMC prepared at different temperatures: (a) 77 °C, (b) 87 °C, (c) 89 °C, (d) 95 °C, (e) 100 °C, (f) 105 °C, (g) 110 °C, (h) 120 °C. Insets show the Fast Fourier Transformations (FFT) of the corresponding pictures.
and pore size distribution. In the case of the dried samples, the isotherms do not show substantial nitrogen uptake and the specific surface area calculated by the BET equation is comprised between 2 and 10 m2/g. This indicates that the surfactant is still present within the porosity of the samples after the different washing steps, which is consistent with the data from the thermal analysis. After pyrolysis at 800 °C under inert atmosphere and decomposition of the surfactant, the materials display a much higher porosity. Nitrogen adsorption–desorption isotherms as well as the corresponding pore size distribution of representative pyrolyzed materials are represented in Fig. 5 and the textural properties as a function of preparation temperature of all the samples are given in Table 3. For all the materials prepared in the considered temperature range (72–127 °C), the isotherms are of type IV, with a high adsorbed volume in the very low relative pressure range [44]. The materials can thus be described as micro-mesoporous materials. The micropore volume, calculated by the Dubinin–Radushkevich equation, is very
close for all the samples, with values comprised between 0.18 and 0.24 cm3/g, which are values commonly observed for porous hard carbons. The specific surface area does not show any clear evolution with the synthesis temperature, with values ranging from 450 to 630 m2/g. The pore size distributions remain quite broad with mesopore diameters centered on about 3.5 nm and a shoulder at 2.7 nm, independently of the temperature, except for the OMC prepared at 77 °C that shows a peak at 4.3 nm. The main difference occurring refers to the change in mesopore volume. Indeed, for temperatures higher than 115 °C, the fraction of mesopores decreases sharply to volumes of about 0.05 cm3/g, whereas it is twice higher at the lower preparation temperatures, the highest mesopore volume being measured for the sample with the largest pore sizes (77 °C). The combination of data obtained from nitrogen adsorption– desorption and from the structural investigations by XRD and SAXS further confirms that the materials can be described as
97
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Table 2 Structural parameters from the SAXS data of the OMC prepared at different temperatures. T (°C)a
77 77⁄ 95 95⁄ 100⁄ 110⁄
d100 (nm)b
d110 (nm)b
d200 (nm)b
a0 (nm)c
12.2 9.7 11.0 9.0 9.4 8.8
6.7 5.6 6.5 5.2 5.4 5.2
5.9 4.8 5.6 4.6 4.7 4.5
14.1 11.2 12.7 10.4 10.8 10.2
a and ⁄ symbols referring to the SAXS patterns of the different hexagonal structures in Fig. 3. b dhkl distances calculated from the length of the scattering vector q: dhkl = 2p/q. c Unit cell parameter calculated on the basis of a hexagonal unit cell: a0 = 2d100/ 31/2.
data, let us consider a unit cell, with an arbitrary extension L in the direction parallel to the mesopores. The mass of carbon in the unit cell can be calculated as:
m¼
Fig. 3. SAXS patterns of the OMC materials prepared at different temperatures ( and ⁄ symbols refer to the different hexagonal structures).
hexagonally-stacked tubular mesoporous channels separated by microporous walls. The following considerations are illustrated in Fig. 6, where the structural model represents cylindrical mesopores with radius R positioned on the nodes of a 2D hexagonal lattice, with spacing a between neighboring mesopores. The space between mesopores, sometimes referred to as the ‘‘pore wall’’, is microporous. We shall make no assumption about the shape of the latter pores. Their overall porosity is e. In order to calculate the mesoporous and microporous specific volumes corresponding to that simple model and compare them to the characterization
"pffiffiffi # 3 2 a pR2 Lð1 eÞq 2
ð2Þ
where L times the bracket is the volume of the microporous wall, and q is the specific mass of dense carbon. Because one unit cell contains only one mesopore, with volume pR2L, the mesoporous specific volume is obtained simply as:
V meso pR2 i ¼ hpffiffiffi m 3=2a2 pR2 ð1 eÞq
ð3Þ
Similarly, the microporous volume in one unit cell is:
hpffiffiffi i 3=2a2 pR2 Le
ð4Þ
which leads to the following expression of the specific microporous volume:
V micro e ¼ m ð1 eÞq
ð5Þ
Fig. 4. Enlarged TEM images showing the different arrays of mesoporous channels of different sizes for OMC prepared at different temperatures: (a) 77 °C, (b) 87 °C, (c) 95 °C, (d) 100 °C.
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Differenal pore volume (cm³/g)
Volume adsorbed (cm³/g-STP)
98
Relave pressure p/p0
Pore size (nm)
Fig. 5. Nitrogen adsorption–desorption isotherms and the corresponding pore size distribution of the OMC prepared at different temperatures: (a) 77 °C, (b) 95 °C, (c) 100 °C, (d) 110 °C, (e) 120 °C.
Assuming a specific mass q = 2 g/cm3, which is an average value for dense amorphous carbon, the experimental value Vmicro = 0.22 cm3/g converts to a wall porosity e = 0.31 [45]. This value is similar to the 30% porosity of the wall in SBA-15 mesoporous silica synthesized with the same template [46–47]. Finally, using the value a = 10.8 nm obtained from SAXS, and R = 1.8 nm obtained from physisorption, the mesopore volume should be 0.08 cm3/g. This value is remarkably close to the 0.09 cm3/g measured directly from physisorption (Table 2). The present calculations therefore prove that the structural model of Fig. 6 is quantitatively accurate. In this case, although the global mesostructure is not fully homogeneous, the average pore wall thickness can be deduced from the data of XRD and nitrogen adsorption. Indeed, it
was reported that in the case of a hexagonal stacking of mesoporous channels, the unit cell parameter, a0, is the sum of pore diameter and pore wall thickness [48]. The calculated values in Table 3 show an average pore wall thickness comprised between 7.3 and 8.5 nm without any correlation with the preparation temperature. These values are quite high in comparison to the mesopore sizes but are consistent with those obtained in similar syntheses carried out in presence of ammonia [34]. Indeed, in the synthesis of ordered mesoporous silica, the surfactant is fully removed upon solvent extraction and calcination under air. In the present case, the carbonization process under nitrogen flow leads to a partial decomposition of the surfactant molecules, but a large part of it is converted into carbon that thickens the pore walls, leading to smaller-sized mesopores.
99
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101 Table 3 Textural parameters from the nitrogen adsorption–desorption data of the OMC prepared at different temperatures. T (°C) 65 70 72 77 87 89 90 95 100 105 110 115 120 127 a b c d e f g
SBETa (m2/g) g
– –g 460 624 560 480 441 560 567 520 607 497 567 450
Vmicrob (cm3/g) g
– –g 0.19 0.25 0.23 0.19 0.18 0.22 0.22 0.21 0.24 0.20 0.22 0.18
Vp, g
tot
– –g 0.27 0.38 0.33 0.28 0.27 0.34 0.31 0.26 0.32 0.24 0.28 0.23
c
(cm3/g)
Vmesod (cm3/g) g
– –g 0.08 0.13 0.10 0.09 0.09 0.12 0.09 0.05 0.08 0.04 0.06 0.05
De (nm) g
– –g 3.7 4.3 3.8 3.4 3.2 3.2 3.6 3.4 3.5 3.1 3.4 3.0
tf (nm) –g –g n.a. 8.2 7.3 7.8 7.9 8.4 8.5 8.2 7.8 8.0 n.a. n.a.
SBET: specific surface area calculated from the BET equation between 0.01 and 0.10 p/p0. Vmicro: micropore volume calculated by the Dubinin–Radushkevich equation. Vp, tot: total pore volume at p/p0 = 0.995. Vmeso: mesopore volume calculated by Vp, tot – Vmicro. D: pore diameter determined from the maximum of the pore size distribution. It should be noted that a shoulder centered on 2.7 nm is present in each case. t: pore wall thickness calculated from: t = a0D, a0 being the unit cell parameter (Table 1). No values available since no solid materials was obtained.
It should be further noted that the presence of differently-sized hexagonal frameworks distinguished by the SAXS measurements or from the TEM observations could not be related to any differences in mesopore size distributions resulting from sorption measurements. This suggests that the difference in unit cell parameters only results from the presence of mesopores with various pore wall thicknesses. 3.4. Electrochemical performance of the OMC
Fig. 6. Structural model of the ordered mesoporous carbon (OMC): the mesopores are cylindrical with radius R and positioned on the nodes of a 2D hexagonal lattice with spacing a between pores. The space between mesopores is microporous, with overall porosity e.
Fig. 7. Charge–discharge curves for the 1st, 2nd and 100th cycle of OMC-100 in a Li/C half-cell at a C/5 rate.
In order to evaluate the performance of the ordered mesoporous carbons to be used as anodes for Li-ion batteries, electrodes were prepared upon spreading the carbon materials onto Cu-foil as a current collector. OMC prepared at different temperatures (77, 100 and 110 °C) were used for evaluating the electrochemical performances (denoted as OMC-77, OMC-100 and OMC-110). The particle size distribution of the powders was measured by laser diffraction and the thickness of the final coatings was evaluated by profilometry (Fig. S3). As can be seen from Fig. S3a, the particle size distribution for OMC-100 is centered on 10 lm (a similar distribution is obtained for OMC-77 and OMC-110), which is a size suitable for the coating on current collectors. As described in the experimental part, for each sample, coatings were realized with the bar coater at 100 lm knife opening. The thickness of the final coatings after drying is about 25 lm (Fig. S3b). Nitrogen adsorption–desorption measurements were carried out on the electrodes after drying in order to assess the textural parameters. The isotherms show the type IV shape identical to that of the powder OMC, the pore size distribution is centered on 3.6 nm and the specific surface area is reduced to 70 m2/g. This decrease in specific surface area can be explained by the presence of the PVDF binder. Note also that the error on the weighed mass of carbon material is much larger in the case of electrodes than in the case of a raw materials used for nitrogen adsorption analysis. The electrochemical performances were determined in homemade Swagelok-type cells with metallic lithium as anode, the lithium insertion into the carbon material being referred to as discharge and the lithium extraction as charge. Fig. 7 illustrates the discharge (intercalation) and charge (deintercalation) curves recorded on OMC-100 for the first, second and 100th cycle. No definite stages can be observed on these curves, which is in agreement with the observations commonly made on non-graphitic hard
100
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Fig. 8. Cycle performance of OMC-77, OMC-100 and OMC-110. The cycling of the half-cell with OMC-77 was stopped after 70 cycles due to a technical problem.
carbons [49]. For the first cycle, the discharge and charge capacities are respectively 726 and 228 mA h/g, giving a coulombic efficiency of 31%. Such a high loss during the first cycle is very common for non-graphitic hard carbons and may be related to several factors, the most important being the formation of a Solid Electrolyte Interphase (SEI film). The latter is generally more pronounced when the specific surface area is high and turns a significant amount of cations into dead metallic lithium [50]. Another factor could be the irreversible entrapment of Li ions into the micropores of the carbons. In the subsequent cycles, the coulombic efficiency remains higher than 95%. This stabilization is best illustrated in Fig. 8, where the charge capacity is represented against the number of cycles, for all the investigated OMC. All of the investigated materials show comparable performances, independently of the synthesis temperature, and so of the pore size distribution, especially if OMC-77 is considered. It also seems that the structural differences observed (presence of two hexagonal frameworks from the SAXS data) do not have any influence on cycling behavior and capacity values. Table 4 summarizes the reversible and irreversible capacities of all the investigated samples. From a global point of view, the first cycle reversible capacities are very close (206–259 mA h/g) and the same is true for the irreversible capacities. After a few cycles, the latter become negligible and the reversible capacities remain very stable, with values around 180 mA h/g. It has to be noted that these values are in line with those commonly observed for some OMC and, more generally for hard carbons, but very low in comparison to those reported by Zhou et al. (>1000 mA h/g) [12]. Our materials show approximately the same mesopore sizes (3.6 vs. 3.9 nm), but only half the specific surface area and total pore volume. The main difference arises from the preparation pathway and the resulting mesostructure. Indeed, Zhou’s OMC were prepared via the nanocasting route (CMK-3 materials), the structure
Table 4 Reversible and irreversible capacities of the OMC materials prepared at different temperatures. Material
Qrev 1st Qirr 1st Qrev 10th (mA h/g) (mA h/g) (mA h/g)
Qirr 10th (mA h/g)
Qrev 100th (mA h/g)
Qirr 100th (mA h/g)
OMC-77
209 206 228 222 241 237
3 4 5 5 2 5
172* 174* 178 177 171 177
0.04* 0.7* 5 1 0.7 4
OMC-100 OMC-110
*
After the 70th cycle.
435 449 497 496 486 472
177 178 182 176 190 192
is thus an inverted replica of the starting ordered mesoporous silica. In our case, the OMC were prepared via a direct soft-templating route, implying the formation of the polymer carbon network around surfactant micelles, leading to materials with quite thick mesopore walls. As proposed by Li et al., the elimination of the silica in the hard-templating route could cause the formation of microcrystalline defects within the carbon walls, which are absent in the direct preparation pathway [13]. These defects could account for the very high capacity observed in nanocasting-derived OMC, although the mechanisms involved are not understood yet. It should also be mentioned that, in our case, no conducting carbon was added to the preparation of the slurry before coating, which could also account for smaller observed capacities. Morevover, there was no pressing step involved in the electrode preparation, which results in a lower tap density. This has to be kept in mind regarding the interpretation and comparison of data since, for car manufacturers, the volumetric capacity is more important than the gravimetric capacity. If we calculate the irreversible/reversible capacity ratio (Qirr/Qrev) for the first insertion–deinsertion cycle, a value of 2.0–2.2 is obtained. It is worth mentioning that this value is lower than that observed for other hard carbons, like carbon xerogels for instance. Indeed, our studies reveal that this value is generally comprised between 3.0 and 3.5 for carbon xerogels (unpublished results). Since the microporosity of OMC is quite close to that of carbon xerogels (0.22 cm3/g vs. 0.28 cm3/g), the lower irreversible losses should be attributed to other parameters, like for instance the structural regularity of the OMC. Regarding the long-term cycling performance, Fig. 8 and Table 4 clearly show that the OMC prepared in the present work feature excellent charge/discharge stability after the initial formation of the SEI layer. From cycles 10–100, the capacity retention exceeds 90%, and, after 100 cycles, capacities of over 75% that of the first cycle were measured. Such a good cycling behavior could be related to the uniform mesoporous structure that is favorable for impregnation by the electrolyte, where Li+ ions can move quickly to be intercalated in the carbon walls [13]. Also, the presence of this porosity can account for the absorption of volume changes caused by the intercalation/deintercalation of lithium ions within the carbon framework. In this case, one can expect an improved long-life behavior of the electrode by avoiding the loss of contact between the active material and the current collector. Note that the cycling of the half-cell with OMC-77 was stopped after 70 cycles due to a technical problem, but its behavior is obviously the same as for the other materials.
4. Conclusions Ordered mesoporous carbons were successfully prepared via a direct templating pathway that features important benefits such as short synthesis time, moderate temperature and an aqueous medium. To solve reproducibility issues, the effect of preparation temperature on the structural and textural parameters has been studied in detail. Highly ordered mesoporous carbons could be obtained at temperatures lower than 100 °C whereas the increase of synthesis temperature tends to decrease the overall mesopore volume, thereby reducing benefits such as good impregnation of electrolyte in the case of anode materials or, more generally, diffusion in the case of heterogeneous catalysis, where high mesoporosity is required. In Li-based half-cells, these OMC show reversible and irreversible capacities in line with those generally reported for high-specific surface area hard carbons, but most importantly, a very good long-term cycling behavior has been shown for over 100 charge–discharge cycles, which could be attributed to the very regular stacking of homogeneous mesoporous channels. The
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
optimization of the synthesis route reported in this work will certainly open-up a wide range of perspectives in the preparation of OMC-based composites for Li-based batteries and also in other domains such as heterogeneous catalysis for example. Acknowledgements The authors are grateful to Prof. Bart Goderis and Mrs. Maja Vanhalle (KULeuven) for measuring the SAXS patterns. This work was realized within the framework of the SOMABAT Project that is financially supported by the 7th Framework Programme of the European Commission (Grant Agreement No. NMP3-SL-2010266090). CJG and MLP thank the FNRS for their ‘‘Chercheur Qualifié’’ and ‘‘FRIA’’ fellowship grants respectively. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.micromeso. 2014.04.028. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
B. Scrosati, J. Garche, J. Power Sources 195 (2010) 2419–2430. P.G. Bruce, Solid State Ionics 179 (2008) 752–760. M. Armand, J.-M. Tarascon, Nature 451 (2008) 652–657. P.G. Bruce, B. Scrosati, J.M. Tarascon, Angew. Chem. Int. Ed. 47 (2008) 2930– 2946. J.M. Tarascon, M. Armand, Nature 414 (2001) 359–367. F. Cheng, Z. Tao, J. Liang, J. Chen, Chem. Mater. 20 (2008) 667–681. J. Li, C. Daniel, D. Wood, J. Power Sources 196 (2011) 2452–2460. T. Yamamoto, T. Hara, K. Segawa, K. Honda, H. Akashi, J. Power Sources 174 (2007) 1036–1040. C. Cai, Y. Wang, Materials 2 (2009) 1205–1238. S. Flandrois, B. Simon, Carbon 37 (1999) 165–180. M. Yoshio, T. Tsumura, N. Dimov, J. Power Sources 146 (2005) 10–14. H.S. Zhou, S.M. Zhu, M. Hibino, I. Honma, M. Ichihara, Adv. Mater. 15 (2003) 2107–2111. H.Q. Li, R.L. Liu, D.Y. Zhao, Y.Y. Xia, Carbon 45 (2007) 2628–2635. R. Ryoo, S.H. Joo, S. Jun, J. Phys. Chem. B 103 (1999) 7743–7746. T.Y. Ma, L. Liu, Z.Y. Yuan, Chem. Soc. Rev. 42 (2013) 3977–4003. I. Muylaert, A. Verberckmoes, J. De Decker, P. Van Der Voort, Adv. Coll. Interf. Sci. 175 (2012) 39–51. C.D. Liang, Z.J. Li, S. Dai, Angew. Chem. Int. Ed. 47 (2008) 3696–3717. Y. Meng, D. Gu, F.Q. Zhang, Y.F. Shi, L. Cheng, D. Feng, Z.X. Wu, Z.X. Chen, Y. Wang, A. Stein, D.Y. Zhao, Chem. Mater. 18 (2006) 4447–4464.
101
[19] C.D. Liang, K.L. Hong, G.A. Guiochon, J.W. Mays, S. Dai, Angew. Chem. Int. Ed. 43 (2004) 5785–5789. [20] S. Tanaka, N. Nishiyama, Y. Egashira, K. Ueyama, Chem. Commun. (2005) 2125–2127. [21] T. Wang, J. He, J. Zhou, X. Ding, J. Zhao, S. Wu, Y. Guo, Micropor. Mesopor. Mater. 134 (2010) 58–64. [22] R. Liu, Y. Ren, Y. Shi, F. Zhang, L. Zhang, B. Tu, D.Y. Zhao, Chem. Mater. 20 (2008) 1140–1146. [23] J. Tang, T. Wang, X. Sun, Y. Guo, H. Xue, H. Guo, M. Liu, X. Zhang, J. He, Micropor. Mesopor. Mater. 177 (2013) 105–112. [24] T. Wang, C.X. Zhang, X. Sun, Y. Guo, H. Guo, J. Tang, H. Xue, M. Liu, X. Zhang, L. Zhu, Q. Xie, J. He, J. Power Sources 212 (2012) 1–12. [25] J. Tang, T. Wang, X. Pan, X. Sun, X. Fan, Y. Guo, H. Xue, J. He, J. Phys. Chem. C 117 (2013) 16896–16906. [26] C.D. Liang, S. Dai, J. Am. Chem. Soc. 128 (2006) 5316–5317. [27] X.C. Zhao, A. Wang, J.W. Yan, G.Q. Sun, L.X. Sun, T. Zhang, Chem. Mater. 22 (2010) 5463–5473. [28] Y. Huang, H. Cai, D. Feng, D. Gu, Y. Deng, B. Tu, H. Wang, P.A. Webley, D.Y. Zhao, Chem. Commun. (2008) 2641–2643. [29] F.Q. Zhang, Y. Meng, D. Gu, Y. Yan, C.Z. Yu, B. Tu, D.Y. Zhao, J. Am. Chem. Soc. 127 (2005) 13508–13509. [30] F.Q. Zhang, Y. Meng, D. Gu, Y. Yan, Z.X. Chen, B. Tu, D.Y. Zhao, Chem. Mater. 18 (2006) 5279–5288. [31] A.H. Lu, B. Spliethoff, F. Schüth, Chem. Mater. 20 (2008) 5314–5319. [32] D. Gu, H. Bongard, Y. Meng, K. Miyasaka, O. Terasaki, F.Q. Zhang, Y. Deng, Z.X. Wu, D. Feng, Y. Fang, B. Tu, F. Schüth, D.Y. Zhao, Chem. Mater. 22 (2010) 4828– 4833. [33] D. Liu, J.H. Lei, L.P. Guo, K.J. Deng, Carbon 49 (2011) 2113–2119. [34] D. Liu, J.H. Lei, L.P. Guo, D. Qu, Y. Li, B.L. Su, Carbon 50 (2012) 476–487. [35] J.M. Kim, Y. Sakamoto, Y.K. Hwang, Y.U. Kwon, O. Terasaki, S.E. Park, G.D. Stucky, J. Phys. Chem. B 106 (2002) 2552–2558. [36] P. Li, Y. Song, Z. Tang, G. Yang, Q. Guo, L. Liu, J. Yang, J. Coll. Interf. Sci. 401 (2013) 161–163. [37] Q. Huo, D.I. Margolese, G.D. Stucky, Chem. Mater. 8 (1996) 1147–1160. [38] J.L. Blin, A. Léonard, B.L. Su, J. Phys. Chem. B 105 (2001) 6070–6079. [39] A. Galarneau, H. Cambon, F. Di Renzo, F. Fajula, Langmuir 17 (2001) 8328– 8335. [40] C.J. Gommes, B. Goderis, J. Appl. Crystallogr. 43 (2010) 352–355. [41] S. Yin, T. Sato, J. Mater. Chem. 15 (2005) 4584–4587. [42] J.R. Dahn, W. Xing, Y. Gao, Carbon 35 (1997) 825–830. [43] S. Inagaki, Y. Fukushima, K. Kuroda, J. Chem. Soc. Chem. Commun. (1993) 680– 682. [44] A.J. Lecloux, in: J.R. Anderson, M. Boudart (Eds.), Catalysis, Science and Technology, vol. II, Springer-Verlag, Berlin, 1981, pp. 171–230. [45] J. Robertson, Mat. Sci. Eng. R 37 (2002) 129–281. [46] C.J. Gommes, H. Friedrichs, M. Wolters, P.E. de Jongh, K.P. de Jong, Chem. Mater. 21 (2009) 1311–1317. [47] J. Prass, D. Mueter, P. Fratzl, O. Paris, Appl. Phys. Lett. 95 (2009) art n° 083121. [48] C.Y. Chen, S.Q. Xiao, M.E. Davis, Micropor. Mater. 4 (1995) 1–20. [49] Z. Hou, F. Zeng, B. He, W. Tao, C. Ge, Y. Kuang, J. Zeng, Mater. Lett. 65 (2011) 897–900. [50] K. Guerin, A. Fevrier-Bouvier, S. Flandrois, B. Simon, P. Biensan, Electrochim. Acta 45 (2000) 1607–1615.
Contents lists available at ScienceDirect
Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso
Rapid aqueous synthesis of ordered mesoporous carbons: Investigation of synthesis variables and application as anode materials for Li-ion batteries Alexandre F. Léonard ⇑, Cedric J. Gommes, Marie-Laure Piedboeuf, Jean-Paul Pirard, Nathalie Job Laboratory of Chemical Engineering – Nanomaterials, Catalysis, Electrochemistry, Institute of Chemistry (B6a), University of Liège, B-4000 Liège, Belgium
a r t i c l e
i n f o
Article history: Received 28 January 2014 Received in revised form 10 March 2014 Accepted 14 April 2014 Available online 22 April 2014 Keywords: Ordered mesoporous carbons Soft-templating Hexamethylenetetramine Anodes Li-ion batteries
a b s t r a c t Ordered mesoporous carbons (OMC) were synthesized via a direct templating pathway by a synthesis route that features short duration, moderate temperature and aqueous media. Resorcinol was used as carbon precursor and hexamethylenetetramine as a source of formaldehyde and ammonia to respectively cross-link the framework and regulate the pH. The temperature of the heat treatment leading to the formation of the solid polymer was shown to have a strong influence on the structural and textural parameters. In particular, moderate temperatures led to the coexistence of differently-sized entangled hexagonal mesostructures, whereas the higher temperatures led to a sharp decrease in the mesopore volume. The performance of these materials as anode materials for Li-ion batteries has been investigated in detail. Although these OMC show reversible capacities similar to those reported for hard carbons, their long-term cycling remains very stable for over 100 cycles of charge/discharge. The optimization of the reported short preparation pathway offers new possibilities regarding the application of ordered mesoporous carbons in various fields, such as energy storage, sorption and heterogeneous catalysis. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Li-based batteries hold a bright future to power ever higher energy-demanding portable electronics, electric vehicles or for the storage of energy produced from renewable but often intermittent sources, like windmills for example [1–3]. Indeed, such batteries display unique advantages, like high volumetric and massic energy densities, combined with the absence of memory effect and a high open-circuit voltage [4–6]. In the quest of even more performant Li-based batteries that feature best security issues, many researches around the world deal with the development of new (solid) electrolytes as well as new electrode materials [7–9]. On the anode side, graphite is currently the most used insertion compound, but its theoretical capacity is limited to 372 mA h/g, i.e. 1 Li per 6 C atoms [10]. Considering the relationship existing between the total capacity of a battery and the capacity of anode materials, it appears that the best anode should bear a capacity of about 1200 mA h/g when used with the current available cathode materials, like LiCoO2 or LiFePO4 [11]. Such a value of capacity could be attained with so-called hard carbons, i.e. porous carbons bearing a turbostratic structure. Among these, ordered mesoporous ⇑ Corresponding author. Tel.: +32 4 366 4875. E-mail address: [email protected] (A.F. Léonard). http://dx.doi.org/10.1016/j.micromeso.2014.04.028 1387-1811/Ó 2014 Elsevier Inc. All rights reserved.
carbons (OMC) represent a particular class of materials because of their highly regular mesostructure. It has indeed been reported that such materials could exhibit specific capacities up to 1100 mA h/g with a very good cycling stability that has been attributed to the regular mesoporous structure favorable to the impregnation of the carbon by the electrolyte [12–13]. The most widespread procedure to prepare ordered mesoporous carbons consists in a nanocasting route, where a mesoporous silica acts as hard template that leads to its pure carbon inverted replica [14]. The method as such is however impossible to scaleup since it is costly – synthesis of mesoporous silica as sacrificial template – and harmful – use of HF as etching agent. A more elegant and convenient pathway is the direct synthesis of ordered mesoporous carbons in presence of a surfactant as structuredirecting agent in a similar fashion as used for ordered mesoporous silica [15–17]. Moreover, the obtained material is expected to be more robust since the framework is continuous in comparison to a stacking of carbon rods in the case of the replica [18]. Pioneering work has been reported during the last years in the direct synthesis of ordered mesoporous carbons, using mainly non-ionic surfactant micelles to control the growth of polymeric networks based on phenol-derivatives and formaldehyde. The most widespread is the Evaporation-Induced Self-Assembly (EISA) method [19–20]. Via this pathway, highly ordered mesoporous
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
carbon films – or monoliths – can be obtained upon evaporating the solvent at ambient temperatures. Interestingly, this route is also very versatile since it allows for the direct fabrication of nanocomposites like C–Al2O3 or C–TiO2 [21–22]. Also, it is possible to incorporate directly heteroatoms leading to W, Mo, Ti, B or N-doped OMC, avoiding the necessity of tedious post-treatments [23–25]. Nevertheless, it should be noted also that this route requires the evaporation of large volumes of organic solvents, especially if scaling-up is considered. Synthesis of OMC was also reported using a phase separation method, where the growing polymer-rich phase separates from the solvent [26–27]. Highly acidic conditions (HCl as a catalyst) are nevertheless required, making this route ill-adapted when poisoning by Cl ions has to be avoided and when synthesis equipment is not designed for corrosive environments. Hydrothermal autoclaving processes, commonly employed in the preparation of zeolites and mesoporous oxides, were reported first by Zhao and colleagues [28]. Ordered mesoporous carbons were synthesized successfully, but adapted equipment is necessary to work under high temperature and high pressure. Finally, the best pathway for large-scale production could very well be the diluted aqueous route reported by Zhao and colleagues [29–30]. OMC with hexagonal or cubic stacking of their channels were reported, depending on the non-ionic surfactant employed. Nevertheless, this synthesis appears to be very sensitive both to temperature and pH of the mixture and this narrow window of conditions could be problematic if large-scale preparations have to be realized. Moreover, very long durations (5–7 days) are required for the polymerization to be accomplished. In fact, compared to mesoporous silica, the cooperative selfassembly process between surfactant and framework-building blocks is more difficult in the synthesis of OMC since it requires the presence of two precursors [31–32]. In that way, the key of a successful preparation relies on the careful control of the phenol–formaldehyde polycondensation, making it not too fast in order to allow for the interaction of the growing oligomers with the hydrophilic heads of the surfactant molecules. This was very well-illustrated by Liu et al. who described the rapid aqueous synthesis of OMC by using resorcinol with hexamethylenetetramine instead of formaldehyde as cross-linking agent [33–34]. This precursor releases in situ OH ions and formaldehyde upon hydrolysis, allowing for the control of pH and polymer network growth. Remarkably, this pathway avoids the use of metal or halide ions, making the resulting carbon free of any poisoning species. As a general trend, whatever the synthesis route employed, the surfactant used is a block-copolymer polyethyleneoxide–polypropyleneoxide–polyethyeleneoxide (PEO–PPO–PEO) and the regular stacking of mesopores in OMC always belong to 3D bi-continuous cubic (Ia3d), 2D hexagonal (P6mm) or body-centered cubic (Im3m) space groups [15–16]. This symmetry is mainly controlled by the choice of the surfactant, i.e. the hydrophilic/hydrophobic balance (PEO/PPO) that dictates the curvature of the micelles. The hydrophilic volume is nevertheless changed upon interaction between PEO moieties of the surfactant and the hydrophilic resol oligomers. For that reason, the phenol/surfactant ratio will have an influence on the structure of the final carbon [18,25,35]. Increasing the hydrophobic volume for instance by adding a swelling agent like trimethylbenzene or decane, which will locate in the cores of the micelles, will favor the formation of structures with less curvature. On the other hand, the use of reverse copolymers PPO–PEO–PPO in conjunction with hexane as a cosolvent with the EISA method has led to the formation OMC with intergrown hexagonal and cubic pore structures, where one or the other phase was favored upon tuning the amount of added cosolvent [36]. When using ammonia in addition to hexamethylenetetramine in the aqueous synthesis pathway, it was also shown that the pore structure of OMC could be oriented from body-centered cubic
93
(Im3m) to hexagonal (P6mm) space groups upon addition of trimethylbenzene (TMB) [34]. The same authors reported that, in the absence of ammonia, the presence of TMB was necessary to obtain ordered mesostructures [33]. In addition, temperature is known to play a key role in the orientation of the structure of ordered mesoporous materials, mainly by influencing the micelle packing parameter g, defined by the volume and length of surfactant chains as well as the effective area of the hydrophilic head groups [37]. Additionally, temperature can also have an influence on the pore size distribution of the final mesoporous material [38–39]. For these reasons and also to solve some reproducibility issues, we decided to explore in more detail the HMT-based aqueous synthesis route by realizing a systematic study of the influence of thermal treatment temperature on the parameters of the final ordered mesoporous carbons. The use of these materials as anodes for Li-ion batteries was investigated in terms of their electrochemical performances, which were compared to those of OMC obtained via the hard-templating (nanocasting) route. 2. Experimental 2.1. Reagents Resorcinol (1,3 dihydroxybenzene, C6H4(OH)2, 99%) and electrolyte LP-71 (1 M LiPF6 in Ethylenecarbonate:Diethylcarbonate:Dimethylcarbonate 1:1:1, EC:DEC:DMC) were supplied by Merck chemicals. Block copolymer [poly(ethylene oxide)b-poly(propylene oxide)-poly(ethylene oxide)] surfactant Pluronic F127 (EO106PO70EO106, Mav = 12,600) as well as hexamethylenetetramine (HMT, C6H12N4) were purchased from Sigma–Aldrich. 1,3,5-trimethylbenzene (TMB, C9H12, 99%) was bought from Acros chemicals and N-methyl-2-pyrrolidone (NMP, C5H9NO, 99.0+%) from Alfa Aesar. 2.2. Synthesis of ordered mesoporous carbons In a typical synthesis, 6.0 g of Pluronic F127 were dissolved in 108 ml deionized water in an autoclavable glass bottle. After stirring during 3 h at ambient temperature, 1.2 g of trimethylbenzene, 3.3 g of resorcinol and 2.1 g of HMT were added sequentially, leaving 1 h stirring between each. After 1 h further stirring, the bottle was sealed and heated in an oil bath overnight, at a precisely chosen temperature ranging from 65 to 127 °C. Depending on the chosen temperature, a pink or orange aggregate was observed, surrounded by a dark brown supernatant. The latter was discarded and the aggregate washed with deionized water and i-propanol, followed by drying in air. Pink to orange carbon polymer powders were recovered at this stage. Finally, the surfactant was removed upon carbonization of the materials under nitrogen in a tubular furnace, with a heating ramp of 2 °C min1 up to 800 °C, with a step of 2 h. 2.3. Characterizations The structure of the materials was investigated by powder X-ray diffraction (XRD) on a Siemens D-5000 diffractometer (Cu K-a radiation, 40 kV, 40 mA) as well as by Transmission Electron Microscopy (Tecnai T10, 100 kV). Selected samples were also characterized by Small Angle X-ray Scattering (SAXS) in order to get a better insight into the mesostructural order. The SAXS data were measured on a XEUSS SAXS/WAXS system from Xenocs, equipped with a Molybdenum K-a X-ray source. The 2D scattering patterns were rotationally averaged using the ConeX software [40] and corrected for background intensity. The scattered intensity was finally plotted against wave-vector q = 4p/k sin(h/2), where k is
94
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
the wavelength and h is the scattering angle. The textural features of the polymers and carbons were assessed by nitrogen adsorption–desorption data acquired on a Micromeritics ASAP 2420 MP at 77 K. Prior to measurements, the samples were outgassed at 170 °C (polymers) or 270 °C (carbons). The specific surface area was calculated using the BET (Brunauer–Emmet–Teller) method with adsorption data taken in the relative pressure range 0.01– 0.10. The micropore volume was calculated by the Dubinin– Radushkevich equation and the total pore volume was determined from the amount of gas adsorbed at a relative pressure of 0.99. Indeed, since these materials are mesoporous, their nitrogen adsorption–desorption isotherm is of type IV, with a saturation plateau at high relative pressures, meaning that the volume adsorbed at saturation corresponds to the total pore volume. The pore size distributions were determined from the DFT method, provided with the software of the apparatus. Thermogravimetric measurements were performed on a Setaram TG-DSC-111 microbalance under He with heating at 3 °C min1 or under air with heating at 2 °C min1. The particle sizes were measured by laser diffraction using a Malvern Mastersizer 2000 particle size analyzer. Prior to measurements, the OMC was dispersed in distilled water with high-intensity ultrasonic irradiation during 30 s. 2.4. Electrochemical analysis Prior to measurements, the materials to be tested were prepared as electrodes. OMC (92 wt.%) and PVDF (binder, 8 wt.%) were mixed in N-methyl-2-pyrrolidone (NMP) to form a homogeneous slurry. The solid/NMP mass ratio was fixed at 1/2.5. This ink was then spread on a Cu foil with a bar coater, the opening of the knife being adjusted at 100 lm. After drying at ambient temperature during 3 h and at 60 °C overnight, 13-mm disk electrodes were punched from this coating. The thickness of the coatings after drying was measured by stylus profilometry (Veeco Dektak 150, stylus radius: 12.5 lm, force: 1.00 mg). In order to have access to the thickness of the coating, disk electrodes were fixed on a microscope slide and the carbon was scratched off from the copper at regular intervals to set the baseline. The thickness was calculated as an average value taken from three electrodes with two scans each. The electrochemical measurements were carried out in homemade Swagelok-type cells, where the tested material acts as cathode and a Li-metal disk as anode. A Celgard separator soaked with 100 ll of LP71 (1 M LiPF6 in EC:DEC:DMC 1:1:1) electrolyte was placed in-between. The cell assembly was done in an Ar-filled glove-box (MBraun). Charge–discharge curves were recorded at C/5 between 0.005 and 1.5 V (vs. Li+/Li) with a Biologic VMP3 multichannel potentiostat.
is usually recorded for amorphous hard carbons, indicating the absence of long-range graphitic ordering of the graphene planes [42]. Indeed, in the case of carbon materials obtained from resol precursors, graphitization can only be reached after a heat treatment at temperatures higher than 2000 °C. Another way is to induce catalytic graphitization upon addition of heteroatoms or transition metals. In this case, the doped OMC display a better graphitization degree, even after thermal treatments at temperatures lower than 1000 °C [23–25]. Fig. 1 shows the low-angle powder XRD patterns of the samples prepared at different temperatures and pyrolyzed under N2. As a general trend, at temperatures above 115 °C, only one single reflection can be detected at low angles, indicative of a disordered wormhole-like array of mesopores of homogeneous sizes. In this case, that single peak is an indication of the average pore-to-pore separation distance. It could be hypothesized that too high temperatures either weaken the H-bondings between the hydrophilic heads of the surfactant micelles and the growing resorcinol-based polymers, or that the cross-linking of the framework becomes too fast, thereby disturbing the hexagonal organization of the surfactant-polymer supramolecular assemblies. In the intermediate range of temperatures, in addition to a very sharp reflection at very low angles (2h = 0.84–0.90°), other less intense peaks can be observed at angles in the regions 2h = 1.45–1.59° as well as 2h = 1.68–1.84°. According to literature, the presence of these secondary reflections indicates the long-range ordering of the mesoporous channels in a hexagonal array [43]. Indeed, mesoporous materials displaying a hexagonal ordering of their channels exhibit a typical four-peak pattern with a strong reflection line at low angle (1 0 0) and weaker peaks at higher angles (1 1 0, 2 0 0 and 2 1 0) that can be indexed on a hexagonal unit cell. From this symmetry, the unit cell parameter a0 can be calculated as:
a0 ¼ 2d100 =31=2
ð1Þ
where d100 is the repetition distance calculated from the first diffraction peak. This unit cell parameter represents the sum of the pore diameter and the thickness of the pore wall. Table 1 gives the values of the repetition distances and unit cell parameters as a function of synthesis temperature. The experimentally observed positions of the peaks on the XRD patterns (and the calculated corresponding distances) are consistent with a hexagonal symme-
3. Results and discussion 3.1. Structural features of the prepared materials If the temperature of the thermal treatment was kept below 72 °C, i.e. 65 and 70 °C in our case, no solid was recovered after the filtration and washing steps. This could be due to the fact that HMT is not hydrolyzed at low temperature [41], so that the framework cannot be cross-linked sufficiently to form solid particles that can be separated from the supernatant solution. Beyond that temperature, pink to orange aggregates can be isolated and processed to powders upon filtration and washing steps. A slight color change of the dried powders could be noticed, with a progressive shift from pink to orange with the temperature increase (Fig. S1). Wide-angle powder XRD measurements only show two broad peaks at 2h angles of about 24° and 44°. These correspond to the (0 0 2) and (1 0 0) diffractions of amorphous carbon. Such a pattern
Fig. 1. Low angle powder XRD patterns of the OMC materials prepared at different temperatures.
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101 Table 1 Structural parameters from the low angle powder XRD data of the OMC prepared at different temperatures. T (°C)
d100 (nm)a
d110 (nm)a
d200 (nm)a
a0 (nm)b
77 87 89 90 95 100 105 110 115 120 121
10.3 10.0 10.0 9.8 10.0 10.5 10.0 9.8 9.6 10.3 10.8
6.0 6.0 5.7 5.8 5.7 6.0 5.7 6.0 5.5 –c –c
5.0 5.0 4.9 4.8 4.9 5.1 5.1 5.0 4.6 –c –c
11.9 11.6 11.6 11.3 11.6 12.1 11.6 11.3 11.1 –c –c
a dhkl distances calculated from the Bragg law on the positions of the 2h reflection lines: dhkl = k/2sin(2h/2). b Unit cell parameter calculated on the basis of a hexagonal unit cell: a0 = 2d100/31/2. c Not observed.
try. Obviously, the unit cell parameter, as calculated from low angle XRD data, remains constant whatever the synthesis temperature, with values comprised between 11.3 and 12.1 nm. These structural data are very close to those reported by Liu et al. [33] The structural regularity of the materials prepared at temperatures ranging from 72 to 115 °C is further confirmed by TEM observations (Fig. 2). The honeycomb-like array of mesopores can clearly be observed and the 6-fold symmetry is confirmed by the FFT-calculations of these images shown in insets. The longitudinal views also confirm that the channels run parallel to each other. Very interestingly, some particles display a hexagonal macroscopic shape, suggesting a very good long-range ordering (Fig. 2A, C, and D). In contrast, for the material prepared at higher temperatures, the order is lost: only wormhole-like mesopores can be evidenced and the macroscopic shape of the particles becomes more random (Fig. 2H). At first glance, the materials prepared at the intermediate temperatures could be described as highly ordered mesoporous carbons with a hexagonal stacking of their channels. Nevertheless, a closer observation of the low-angle XRD patterns shows that in some cases, the main reflection line (1 0 0) seems split. For that reason, we decided to perform SAXS measurements on selected samples (Fig. 3). From these measurements, it appears clearly that different structures are present in these materials, especially for the lower temperatures. The analysis of the positions of the peaks suggests the presence of two frameworks with a hexagonal symmetry in the stacking of mesopores, but with different sizes. This feature is most visible on the pattern corresponding to the sample prepared at 95 °C and is certainly also present for the samples prepared at 77 and 100 °C. Above this temperature, the distinction of frameworks becomes less defined, but some shoulders present on the first most intense peak could also suggest the presence of hexagonal arrays bearing different sizes. The repetition distances as well as the unit cell parameters calculated from the SAXS data are given in Table 2 and fit well with an indexation on hexagonal unit cells. Based on these remarkable observations, we decided to examine the TEM images of the samples in more detail in order to assess the homogeneity of the structure. Although the differences observed from the SAXS data could not be distinguished by TEM, the distances measured between the channels very well correspond to those from the diffraction and diffusion data. Nevertheless, as can be seen on Fig. 4, an additional, smaller level of organization could be evidenced, which is illustrated by the white bars on the pictures. The observation of the images suggests that, in some way, the channel walls split for the formation of new mesoporous channels within a continuous network. The continuity of the framework is clearly seen in Fig. 4A and also supported from Fig. 4C and D where the angles between the planes of small and
95
larger regular arrays is close to 60°, in accordance with the hexagonal symmetry existing in these materials. The presence of this smaller repetition distance is however not supported by XRD or SAXS data, suggesting that the largest part of the material is constituted of channels with thick pore walls. Moreover, such an inhomogeneity could not be evidenced for the materials prepared at temperatures higher than 110 °C. It should also be mentioned that the SAXS data of the corresponding non-pyrolyzed as-prepared samples display the same patterns, but with peaks located at shorter lengths of the scattering vector q. From this, it can be stated that the pyrolysis step does not affect the global structure of the framework, but only induces an isotropic shrinkage. The shrinkage extent was calculated based on the reduction of the unit cell parameter a0 and is comprised between 35% and 38%, whatever the preparation temperature of the materials. To summarize, from a structural point of view, highly ordered mesoporous carbons with a hexagonal stacking of their channels can be prepared in a temperature interval ranging from 72 to 115 °C. The unit cell parameters are very close, whatever the temperature, but the lowest temperatures seem to favor the formation of materials displaying several hexagonal frameworks entangled inside one single continuous network. Moreover, TEM observations suggest the presence of an additional array of mesoporous channels with thin pore walls. A tentative explanation could be the progressive coalescence of pore walls with the progression of the polycondensation of the carbon polymer framework. In this case, in a first stage of the reaction, the formation of surfactant-carbon precursor supramolecular assemblies occurs. During this step, resorcinol interacts with the hydrophilic heads of the surfactant, whereas the hydrophobic parts join to form hybrid micelles. These assemblies then progressively join to form a continuous carbon polymer network with the surfactant molecules embedded, leading to the development of porosity. In a second stage, with further advancement of the reaction and continuous release of formaldehyde from the HMT precursor, the polymerization of the carbon polymer continues and eventually expels surfactant molecules, inducing the disappearance of ‘‘channels’’ and increasing the fraction of the carbon skeleton, and thus leading to thicker pore walls. This phenomenon would be even more pronounced if the preparation temperature is increased, explaining why such entangled differently-sized hexagonal arrays could not be evidenced above 110 °C. 3.2. Thermal analysis The washed and dried polymer/surfactant composites were analyzed by thermogravimetry–differential scanning calorimetry in order to determine their behavior during the carbonization step (Fig. S2). All of the investigated samples behave the same, independently of the preparation temperature. Under helium flow (Fig. S2a), a first small mass loss (2%) is recorded below 100 °C, corresponding to the evaporation of remaining solvent molecules. In a second stage, a large weight loss of 50–55% occurs between 300 and 400 °C, with an endothermic signal at 383 °C. This loss mainly corresponds to the decomposition of the surfactant. When the temperature is further increased up to 700 °C, a continuous mass loss up to 80% in total is recorded, which corresponds to the carbonization of the material. TG-DSC measurements under air atmosphere in turn show two distinct mass losses with exothermic signals at 310 and 480 °C, corresponding respectively to the burning of the surfactant and the carbon framework (Fig. S2b). 3.3. Textural parameters of the OMC as a function of preparation temperature Nitrogen adsorption–desorption measurements were performed in order to assess the pore volume, specific surface area
96
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Fig. 2. TEM images of the OMC prepared at different temperatures: (a) 77 °C, (b) 87 °C, (c) 89 °C, (d) 95 °C, (e) 100 °C, (f) 105 °C, (g) 110 °C, (h) 120 °C. Insets show the Fast Fourier Transformations (FFT) of the corresponding pictures.
and pore size distribution. In the case of the dried samples, the isotherms do not show substantial nitrogen uptake and the specific surface area calculated by the BET equation is comprised between 2 and 10 m2/g. This indicates that the surfactant is still present within the porosity of the samples after the different washing steps, which is consistent with the data from the thermal analysis. After pyrolysis at 800 °C under inert atmosphere and decomposition of the surfactant, the materials display a much higher porosity. Nitrogen adsorption–desorption isotherms as well as the corresponding pore size distribution of representative pyrolyzed materials are represented in Fig. 5 and the textural properties as a function of preparation temperature of all the samples are given in Table 3. For all the materials prepared in the considered temperature range (72–127 °C), the isotherms are of type IV, with a high adsorbed volume in the very low relative pressure range [44]. The materials can thus be described as micro-mesoporous materials. The micropore volume, calculated by the Dubinin–Radushkevich equation, is very
close for all the samples, with values comprised between 0.18 and 0.24 cm3/g, which are values commonly observed for porous hard carbons. The specific surface area does not show any clear evolution with the synthesis temperature, with values ranging from 450 to 630 m2/g. The pore size distributions remain quite broad with mesopore diameters centered on about 3.5 nm and a shoulder at 2.7 nm, independently of the temperature, except for the OMC prepared at 77 °C that shows a peak at 4.3 nm. The main difference occurring refers to the change in mesopore volume. Indeed, for temperatures higher than 115 °C, the fraction of mesopores decreases sharply to volumes of about 0.05 cm3/g, whereas it is twice higher at the lower preparation temperatures, the highest mesopore volume being measured for the sample with the largest pore sizes (77 °C). The combination of data obtained from nitrogen adsorption– desorption and from the structural investigations by XRD and SAXS further confirms that the materials can be described as
97
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Table 2 Structural parameters from the SAXS data of the OMC prepared at different temperatures. T (°C)a
77 77⁄ 95 95⁄ 100⁄ 110⁄
d100 (nm)b
d110 (nm)b
d200 (nm)b
a0 (nm)c
12.2 9.7 11.0 9.0 9.4 8.8
6.7 5.6 6.5 5.2 5.4 5.2
5.9 4.8 5.6 4.6 4.7 4.5
14.1 11.2 12.7 10.4 10.8 10.2
a and ⁄ symbols referring to the SAXS patterns of the different hexagonal structures in Fig. 3. b dhkl distances calculated from the length of the scattering vector q: dhkl = 2p/q. c Unit cell parameter calculated on the basis of a hexagonal unit cell: a0 = 2d100/ 31/2.
data, let us consider a unit cell, with an arbitrary extension L in the direction parallel to the mesopores. The mass of carbon in the unit cell can be calculated as:
m¼
Fig. 3. SAXS patterns of the OMC materials prepared at different temperatures ( and ⁄ symbols refer to the different hexagonal structures).
hexagonally-stacked tubular mesoporous channels separated by microporous walls. The following considerations are illustrated in Fig. 6, where the structural model represents cylindrical mesopores with radius R positioned on the nodes of a 2D hexagonal lattice, with spacing a between neighboring mesopores. The space between mesopores, sometimes referred to as the ‘‘pore wall’’, is microporous. We shall make no assumption about the shape of the latter pores. Their overall porosity is e. In order to calculate the mesoporous and microporous specific volumes corresponding to that simple model and compare them to the characterization
"pffiffiffi # 3 2 a pR2 Lð1 eÞq 2
ð2Þ
where L times the bracket is the volume of the microporous wall, and q is the specific mass of dense carbon. Because one unit cell contains only one mesopore, with volume pR2L, the mesoporous specific volume is obtained simply as:
V meso pR2 i ¼ hpffiffiffi m 3=2a2 pR2 ð1 eÞq
ð3Þ
Similarly, the microporous volume in one unit cell is:
hpffiffiffi i 3=2a2 pR2 Le
ð4Þ
which leads to the following expression of the specific microporous volume:
V micro e ¼ m ð1 eÞq
ð5Þ
Fig. 4. Enlarged TEM images showing the different arrays of mesoporous channels of different sizes for OMC prepared at different temperatures: (a) 77 °C, (b) 87 °C, (c) 95 °C, (d) 100 °C.
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Differenal pore volume (cm³/g)
Volume adsorbed (cm³/g-STP)
98
Relave pressure p/p0
Pore size (nm)
Fig. 5. Nitrogen adsorption–desorption isotherms and the corresponding pore size distribution of the OMC prepared at different temperatures: (a) 77 °C, (b) 95 °C, (c) 100 °C, (d) 110 °C, (e) 120 °C.
Assuming a specific mass q = 2 g/cm3, which is an average value for dense amorphous carbon, the experimental value Vmicro = 0.22 cm3/g converts to a wall porosity e = 0.31 [45]. This value is similar to the 30% porosity of the wall in SBA-15 mesoporous silica synthesized with the same template [46–47]. Finally, using the value a = 10.8 nm obtained from SAXS, and R = 1.8 nm obtained from physisorption, the mesopore volume should be 0.08 cm3/g. This value is remarkably close to the 0.09 cm3/g measured directly from physisorption (Table 2). The present calculations therefore prove that the structural model of Fig. 6 is quantitatively accurate. In this case, although the global mesostructure is not fully homogeneous, the average pore wall thickness can be deduced from the data of XRD and nitrogen adsorption. Indeed, it
was reported that in the case of a hexagonal stacking of mesoporous channels, the unit cell parameter, a0, is the sum of pore diameter and pore wall thickness [48]. The calculated values in Table 3 show an average pore wall thickness comprised between 7.3 and 8.5 nm without any correlation with the preparation temperature. These values are quite high in comparison to the mesopore sizes but are consistent with those obtained in similar syntheses carried out in presence of ammonia [34]. Indeed, in the synthesis of ordered mesoporous silica, the surfactant is fully removed upon solvent extraction and calcination under air. In the present case, the carbonization process under nitrogen flow leads to a partial decomposition of the surfactant molecules, but a large part of it is converted into carbon that thickens the pore walls, leading to smaller-sized mesopores.
99
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101 Table 3 Textural parameters from the nitrogen adsorption–desorption data of the OMC prepared at different temperatures. T (°C) 65 70 72 77 87 89 90 95 100 105 110 115 120 127 a b c d e f g
SBETa (m2/g) g
– –g 460 624 560 480 441 560 567 520 607 497 567 450
Vmicrob (cm3/g) g
– –g 0.19 0.25 0.23 0.19 0.18 0.22 0.22 0.21 0.24 0.20 0.22 0.18
Vp, g
tot
– –g 0.27 0.38 0.33 0.28 0.27 0.34 0.31 0.26 0.32 0.24 0.28 0.23
c
(cm3/g)
Vmesod (cm3/g) g
– –g 0.08 0.13 0.10 0.09 0.09 0.12 0.09 0.05 0.08 0.04 0.06 0.05
De (nm) g
– –g 3.7 4.3 3.8 3.4 3.2 3.2 3.6 3.4 3.5 3.1 3.4 3.0
tf (nm) –g –g n.a. 8.2 7.3 7.8 7.9 8.4 8.5 8.2 7.8 8.0 n.a. n.a.
SBET: specific surface area calculated from the BET equation between 0.01 and 0.10 p/p0. Vmicro: micropore volume calculated by the Dubinin–Radushkevich equation. Vp, tot: total pore volume at p/p0 = 0.995. Vmeso: mesopore volume calculated by Vp, tot – Vmicro. D: pore diameter determined from the maximum of the pore size distribution. It should be noted that a shoulder centered on 2.7 nm is present in each case. t: pore wall thickness calculated from: t = a0D, a0 being the unit cell parameter (Table 1). No values available since no solid materials was obtained.
It should be further noted that the presence of differently-sized hexagonal frameworks distinguished by the SAXS measurements or from the TEM observations could not be related to any differences in mesopore size distributions resulting from sorption measurements. This suggests that the difference in unit cell parameters only results from the presence of mesopores with various pore wall thicknesses. 3.4. Electrochemical performance of the OMC
Fig. 6. Structural model of the ordered mesoporous carbon (OMC): the mesopores are cylindrical with radius R and positioned on the nodes of a 2D hexagonal lattice with spacing a between pores. The space between mesopores is microporous, with overall porosity e.
Fig. 7. Charge–discharge curves for the 1st, 2nd and 100th cycle of OMC-100 in a Li/C half-cell at a C/5 rate.
In order to evaluate the performance of the ordered mesoporous carbons to be used as anodes for Li-ion batteries, electrodes were prepared upon spreading the carbon materials onto Cu-foil as a current collector. OMC prepared at different temperatures (77, 100 and 110 °C) were used for evaluating the electrochemical performances (denoted as OMC-77, OMC-100 and OMC-110). The particle size distribution of the powders was measured by laser diffraction and the thickness of the final coatings was evaluated by profilometry (Fig. S3). As can be seen from Fig. S3a, the particle size distribution for OMC-100 is centered on 10 lm (a similar distribution is obtained for OMC-77 and OMC-110), which is a size suitable for the coating on current collectors. As described in the experimental part, for each sample, coatings were realized with the bar coater at 100 lm knife opening. The thickness of the final coatings after drying is about 25 lm (Fig. S3b). Nitrogen adsorption–desorption measurements were carried out on the electrodes after drying in order to assess the textural parameters. The isotherms show the type IV shape identical to that of the powder OMC, the pore size distribution is centered on 3.6 nm and the specific surface area is reduced to 70 m2/g. This decrease in specific surface area can be explained by the presence of the PVDF binder. Note also that the error on the weighed mass of carbon material is much larger in the case of electrodes than in the case of a raw materials used for nitrogen adsorption analysis. The electrochemical performances were determined in homemade Swagelok-type cells with metallic lithium as anode, the lithium insertion into the carbon material being referred to as discharge and the lithium extraction as charge. Fig. 7 illustrates the discharge (intercalation) and charge (deintercalation) curves recorded on OMC-100 for the first, second and 100th cycle. No definite stages can be observed on these curves, which is in agreement with the observations commonly made on non-graphitic hard
100
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
Fig. 8. Cycle performance of OMC-77, OMC-100 and OMC-110. The cycling of the half-cell with OMC-77 was stopped after 70 cycles due to a technical problem.
carbons [49]. For the first cycle, the discharge and charge capacities are respectively 726 and 228 mA h/g, giving a coulombic efficiency of 31%. Such a high loss during the first cycle is very common for non-graphitic hard carbons and may be related to several factors, the most important being the formation of a Solid Electrolyte Interphase (SEI film). The latter is generally more pronounced when the specific surface area is high and turns a significant amount of cations into dead metallic lithium [50]. Another factor could be the irreversible entrapment of Li ions into the micropores of the carbons. In the subsequent cycles, the coulombic efficiency remains higher than 95%. This stabilization is best illustrated in Fig. 8, where the charge capacity is represented against the number of cycles, for all the investigated OMC. All of the investigated materials show comparable performances, independently of the synthesis temperature, and so of the pore size distribution, especially if OMC-77 is considered. It also seems that the structural differences observed (presence of two hexagonal frameworks from the SAXS data) do not have any influence on cycling behavior and capacity values. Table 4 summarizes the reversible and irreversible capacities of all the investigated samples. From a global point of view, the first cycle reversible capacities are very close (206–259 mA h/g) and the same is true for the irreversible capacities. After a few cycles, the latter become negligible and the reversible capacities remain very stable, with values around 180 mA h/g. It has to be noted that these values are in line with those commonly observed for some OMC and, more generally for hard carbons, but very low in comparison to those reported by Zhou et al. (>1000 mA h/g) [12]. Our materials show approximately the same mesopore sizes (3.6 vs. 3.9 nm), but only half the specific surface area and total pore volume. The main difference arises from the preparation pathway and the resulting mesostructure. Indeed, Zhou’s OMC were prepared via the nanocasting route (CMK-3 materials), the structure
Table 4 Reversible and irreversible capacities of the OMC materials prepared at different temperatures. Material
Qrev 1st Qirr 1st Qrev 10th (mA h/g) (mA h/g) (mA h/g)
Qirr 10th (mA h/g)
Qrev 100th (mA h/g)
Qirr 100th (mA h/g)
OMC-77
209 206 228 222 241 237
3 4 5 5 2 5
172* 174* 178 177 171 177
0.04* 0.7* 5 1 0.7 4
OMC-100 OMC-110
*
After the 70th cycle.
435 449 497 496 486 472
177 178 182 176 190 192
is thus an inverted replica of the starting ordered mesoporous silica. In our case, the OMC were prepared via a direct soft-templating route, implying the formation of the polymer carbon network around surfactant micelles, leading to materials with quite thick mesopore walls. As proposed by Li et al., the elimination of the silica in the hard-templating route could cause the formation of microcrystalline defects within the carbon walls, which are absent in the direct preparation pathway [13]. These defects could account for the very high capacity observed in nanocasting-derived OMC, although the mechanisms involved are not understood yet. It should also be mentioned that, in our case, no conducting carbon was added to the preparation of the slurry before coating, which could also account for smaller observed capacities. Morevover, there was no pressing step involved in the electrode preparation, which results in a lower tap density. This has to be kept in mind regarding the interpretation and comparison of data since, for car manufacturers, the volumetric capacity is more important than the gravimetric capacity. If we calculate the irreversible/reversible capacity ratio (Qirr/Qrev) for the first insertion–deinsertion cycle, a value of 2.0–2.2 is obtained. It is worth mentioning that this value is lower than that observed for other hard carbons, like carbon xerogels for instance. Indeed, our studies reveal that this value is generally comprised between 3.0 and 3.5 for carbon xerogels (unpublished results). Since the microporosity of OMC is quite close to that of carbon xerogels (0.22 cm3/g vs. 0.28 cm3/g), the lower irreversible losses should be attributed to other parameters, like for instance the structural regularity of the OMC. Regarding the long-term cycling performance, Fig. 8 and Table 4 clearly show that the OMC prepared in the present work feature excellent charge/discharge stability after the initial formation of the SEI layer. From cycles 10–100, the capacity retention exceeds 90%, and, after 100 cycles, capacities of over 75% that of the first cycle were measured. Such a good cycling behavior could be related to the uniform mesoporous structure that is favorable for impregnation by the electrolyte, where Li+ ions can move quickly to be intercalated in the carbon walls [13]. Also, the presence of this porosity can account for the absorption of volume changes caused by the intercalation/deintercalation of lithium ions within the carbon framework. In this case, one can expect an improved long-life behavior of the electrode by avoiding the loss of contact between the active material and the current collector. Note that the cycling of the half-cell with OMC-77 was stopped after 70 cycles due to a technical problem, but its behavior is obviously the same as for the other materials.
4. Conclusions Ordered mesoporous carbons were successfully prepared via a direct templating pathway that features important benefits such as short synthesis time, moderate temperature and an aqueous medium. To solve reproducibility issues, the effect of preparation temperature on the structural and textural parameters has been studied in detail. Highly ordered mesoporous carbons could be obtained at temperatures lower than 100 °C whereas the increase of synthesis temperature tends to decrease the overall mesopore volume, thereby reducing benefits such as good impregnation of electrolyte in the case of anode materials or, more generally, diffusion in the case of heterogeneous catalysis, where high mesoporosity is required. In Li-based half-cells, these OMC show reversible and irreversible capacities in line with those generally reported for high-specific surface area hard carbons, but most importantly, a very good long-term cycling behavior has been shown for over 100 charge–discharge cycles, which could be attributed to the very regular stacking of homogeneous mesoporous channels. The
A.F. Léonard et al. / Microporous and Mesoporous Materials 195 (2014) 92–101
optimization of the synthesis route reported in this work will certainly open-up a wide range of perspectives in the preparation of OMC-based composites for Li-based batteries and also in other domains such as heterogeneous catalysis for example. Acknowledgements The authors are grateful to Prof. Bart Goderis and Mrs. Maja Vanhalle (KULeuven) for measuring the SAXS patterns. This work was realized within the framework of the SOMABAT Project that is financially supported by the 7th Framework Programme of the European Commission (Grant Agreement No. NMP3-SL-2010266090). CJG and MLP thank the FNRS for their ‘‘Chercheur Qualifié’’ and ‘‘FRIA’’ fellowship grants respectively. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.micromeso. 2014.04.028. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
B. Scrosati, J. Garche, J. Power Sources 195 (2010) 2419–2430. P.G. Bruce, Solid State Ionics 179 (2008) 752–760. M. Armand, J.-M. Tarascon, Nature 451 (2008) 652–657. P.G. Bruce, B. Scrosati, J.M. Tarascon, Angew. Chem. Int. Ed. 47 (2008) 2930– 2946. J.M. Tarascon, M. Armand, Nature 414 (2001) 359–367. F. Cheng, Z. Tao, J. Liang, J. Chen, Chem. Mater. 20 (2008) 667–681. J. Li, C. Daniel, D. Wood, J. Power Sources 196 (2011) 2452–2460. T. Yamamoto, T. Hara, K. Segawa, K. Honda, H. Akashi, J. Power Sources 174 (2007) 1036–1040. C. Cai, Y. Wang, Materials 2 (2009) 1205–1238. S. Flandrois, B. Simon, Carbon 37 (1999) 165–180. M. Yoshio, T. Tsumura, N. Dimov, J. Power Sources 146 (2005) 10–14. H.S. Zhou, S.M. Zhu, M. Hibino, I. Honma, M. Ichihara, Adv. Mater. 15 (2003) 2107–2111. H.Q. Li, R.L. Liu, D.Y. Zhao, Y.Y. Xia, Carbon 45 (2007) 2628–2635. R. Ryoo, S.H. Joo, S. Jun, J. Phys. Chem. B 103 (1999) 7743–7746. T.Y. Ma, L. Liu, Z.Y. Yuan, Chem. Soc. Rev. 42 (2013) 3977–4003. I. Muylaert, A. Verberckmoes, J. De Decker, P. Van Der Voort, Adv. Coll. Interf. Sci. 175 (2012) 39–51. C.D. Liang, Z.J. Li, S. Dai, Angew. Chem. Int. Ed. 47 (2008) 3696–3717. Y. Meng, D. Gu, F.Q. Zhang, Y.F. Shi, L. Cheng, D. Feng, Z.X. Wu, Z.X. Chen, Y. Wang, A. Stein, D.Y. Zhao, Chem. Mater. 18 (2006) 4447–4464.
101
[19] C.D. Liang, K.L. Hong, G.A. Guiochon, J.W. Mays, S. Dai, Angew. Chem. Int. Ed. 43 (2004) 5785–5789. [20] S. Tanaka, N. Nishiyama, Y. Egashira, K. Ueyama, Chem. Commun. (2005) 2125–2127. [21] T. Wang, J. He, J. Zhou, X. Ding, J. Zhao, S. Wu, Y. Guo, Micropor. Mesopor. Mater. 134 (2010) 58–64. [22] R. Liu, Y. Ren, Y. Shi, F. Zhang, L. Zhang, B. Tu, D.Y. Zhao, Chem. Mater. 20 (2008) 1140–1146. [23] J. Tang, T. Wang, X. Sun, Y. Guo, H. Xue, H. Guo, M. Liu, X. Zhang, J. He, Micropor. Mesopor. Mater. 177 (2013) 105–112. [24] T. Wang, C.X. Zhang, X. Sun, Y. Guo, H. Guo, J. Tang, H. Xue, M. Liu, X. Zhang, L. Zhu, Q. Xie, J. He, J. Power Sources 212 (2012) 1–12. [25] J. Tang, T. Wang, X. Pan, X. Sun, X. Fan, Y. Guo, H. Xue, J. He, J. Phys. Chem. C 117 (2013) 16896–16906. [26] C.D. Liang, S. Dai, J. Am. Chem. Soc. 128 (2006) 5316–5317. [27] X.C. Zhao, A. Wang, J.W. Yan, G.Q. Sun, L.X. Sun, T. Zhang, Chem. Mater. 22 (2010) 5463–5473. [28] Y. Huang, H. Cai, D. Feng, D. Gu, Y. Deng, B. Tu, H. Wang, P.A. Webley, D.Y. Zhao, Chem. Commun. (2008) 2641–2643. [29] F.Q. Zhang, Y. Meng, D. Gu, Y. Yan, C.Z. Yu, B. Tu, D.Y. Zhao, J. Am. Chem. Soc. 127 (2005) 13508–13509. [30] F.Q. Zhang, Y. Meng, D. Gu, Y. Yan, Z.X. Chen, B. Tu, D.Y. Zhao, Chem. Mater. 18 (2006) 5279–5288. [31] A.H. Lu, B. Spliethoff, F. Schüth, Chem. Mater. 20 (2008) 5314–5319. [32] D. Gu, H. Bongard, Y. Meng, K. Miyasaka, O. Terasaki, F.Q. Zhang, Y. Deng, Z.X. Wu, D. Feng, Y. Fang, B. Tu, F. Schüth, D.Y. Zhao, Chem. Mater. 22 (2010) 4828– 4833. [33] D. Liu, J.H. Lei, L.P. Guo, K.J. Deng, Carbon 49 (2011) 2113–2119. [34] D. Liu, J.H. Lei, L.P. Guo, D. Qu, Y. Li, B.L. Su, Carbon 50 (2012) 476–487. [35] J.M. Kim, Y. Sakamoto, Y.K. Hwang, Y.U. Kwon, O. Terasaki, S.E. Park, G.D. Stucky, J. Phys. Chem. B 106 (2002) 2552–2558. [36] P. Li, Y. Song, Z. Tang, G. Yang, Q. Guo, L. Liu, J. Yang, J. Coll. Interf. Sci. 401 (2013) 161–163. [37] Q. Huo, D.I. Margolese, G.D. Stucky, Chem. Mater. 8 (1996) 1147–1160. [38] J.L. Blin, A. Léonard, B.L. Su, J. Phys. Chem. B 105 (2001) 6070–6079. [39] A. Galarneau, H. Cambon, F. Di Renzo, F. Fajula, Langmuir 17 (2001) 8328– 8335. [40] C.J. Gommes, B. Goderis, J. Appl. Crystallogr. 43 (2010) 352–355. [41] S. Yin, T. Sato, J. Mater. Chem. 15 (2005) 4584–4587. [42] J.R. Dahn, W. Xing, Y. Gao, Carbon 35 (1997) 825–830. [43] S. Inagaki, Y. Fukushima, K. Kuroda, J. Chem. Soc. Chem. Commun. (1993) 680– 682. [44] A.J. Lecloux, in: J.R. Anderson, M. Boudart (Eds.), Catalysis, Science and Technology, vol. II, Springer-Verlag, Berlin, 1981, pp. 171–230. [45] J. Robertson, Mat. Sci. Eng. R 37 (2002) 129–281. [46] C.J. Gommes, H. Friedrichs, M. Wolters, P.E. de Jongh, K.P. de Jong, Chem. Mater. 21 (2009) 1311–1317. [47] J. Prass, D. Mueter, P. Fratzl, O. Paris, Appl. Phys. Lett. 95 (2009) art n° 083121. [48] C.Y. Chen, S.Q. Xiao, M.E. Davis, Micropor. Mater. 4 (1995) 1–20. [49] Z. Hou, F. Zeng, B. He, W. Tao, C. Ge, Y. Kuang, J. Zeng, Mater. Lett. 65 (2011) 897–900. [50] K. Guerin, A. Fevrier-Bouvier, S. Flandrois, B. Simon, P. Biensan, Electrochim. Acta 45 (2000) 1607–1615.