Sadaf
Saeedi Garakani
a,
Dongjiu
Xie
b,
Atefeh Khorsand
Kheirabad
a,
Yan
Lu
*bc and
Jiayin
Yuan
*a
aDepartment of Materials and Environmental Chemistry, Stockholm University, Stockholm 10691, Sweden. E-mail: jiayin.yuan@mmk.su.se
bDepartment for Electrochemical Energy Storage, Helmholtz-Zentrum Berlin für Materialien und Energie, Hahn-Meitner Platz 1, Berlin, 14109, Germany. E-mail: yan.lu@helmholtz-berlin.de
cInstitute of Chemistry, University of Potsdam, 14476 Potsdam, Germany
First published on 25th June 2021
This study deals with the facile synthesis of Fe1−xS nanoparticle-containing nitrogen-doped porous carbon membranes (denoted as Fe1−xS/N-PCMs) via vacuum carbonization of hybrid porous poly(ionic liquid) (PIL) membranes, and their successful use as a sulfur host material to mitigate the shuttle effect in lithium–sulfur (Li–S) batteries. The hybrid porous PIL membranes as the sacrificial template were prepared via ionic crosslinking of a cationic PIL with base-neutralized 1,1′-ferrocenedicarboxylic acid, so that the iron source was molecularly incorporated into the template. The carbonization process was investigated in detail at different temperatures, and the chemical and porous structures of the carbon products were comprehensively analyzed. The Fe1−xS/N-PCMs prepared at 900 °C have a multimodal pore size distribution with a satisfactorily high surface area and well-dispersed iron sulfide nanoparticles to physically and chemically confine the LiPSs. The sulfur/Fe1−xS/N-PCM composites were then tested as electrodes in Li–S batteries, showing much improved capacity, rate performance and cycle stability, in comparison to iron sulfide-free, nitrogen-doped porous carbon membranes.
To be highlighted is the doping of carbons with nitrogen (N), which is to date the most popularly used heteroatom to dope carbon due to its wide presence in various organic compounds to be used as carbon precursors, and its similar atomic size to a carbon atom which facilitates the easy formation of C–N covalent bonds inside the carbon matrix. The doped N atoms have the capability of altering the physicochemical characteristics of carbon materials and supplying carbon materials with target-specific functions, such as high conductivity, oxidation resistance, and catalytic activities.4 The application scope of nitrogen-doped porous carbons (NPCs) has been expanded drastically in the past two decades. They have been utilized in batteries,5,6 fuel cells,7 catalysis,8 membrane separation,9 and supercapacitors.10 Particularly, NPCs have received considerable interest in lithium-ion batteries and lithium–sulfur (Li–S) batteries.11,12 Tao et al. showed that N-doped carbons enhanced the Li2S conversion in Li–S batteries. Moreover, the existence of pyridinic N atoms in porous carbon was reported to be highly effective in the capture of lithium polysulfides and the further reduction towards Li2S.13 Wei et al. demonstrated that the hierarchically structured NPCs reduced the Li+ diffusion pathway and increased the active sites for Li+ storage in lithium-ion batteries. Furthermore, such NPCs could improve the electrical conductivity and modify volume alterations that occur in the cycling tests.14 To be highlighted here is the nitrogen-doped porous carbon membrane (N-PCM), which is a unique form of NPCs containing a hierarchical interconnected porous structure in a membrane state. There has been recently increasing interest in synthesizing and applying N-PCMs as an electrode in various electrochemical devices.15
The chemical structure of the precursor for the synthesis of NPCs plays a crucial role in determining the final shape, carbonization yield, chemical composition, and physicochemical properties. Dai and Antonietti et al. are the pioneers who synthesized NPCs based on nitrile-functionalized ionic liquids (ILs).16,17 Thereafter, ILs have been receiving considerable interest for their use in the synthesis of NPCs with a high nitrogen content, elevated pyrolytic yield, and high specific surface area.18 Our group applied poly(ionic liquid) (PIL), the polymerization product of ILs, as a precursor to produce porous carbons in a variety of defined shapes.19 In PILs, the polymer nature allows them to be processed into well-defined structures or shapes, and the cation–anion pair is one of the critical factors in creating the micro-/mesopores.19 Compared to other classes of polymers, PILs have some unique advantages. First and foremost, PILs can be of high thermal stability up to 450 °C and thus their carbonization yield is usually higher than their neutral counterparts.20 Secondly, using rich heteroatoms in the PIL's chemical structure helps control the heteroatom type and content, mainly nitrogen but also S, B and P, in the final carbon product. Thirdly, PIL structures in various shapes, such as membranes, nanotubes, and spheres, due to a high carbonization yield, can be better preserved during the carbonization at high temperature, e.g. at 1000 °C.15 Thus, PILs are a good platform to produce shaped porous carbon structures.
One of the important usages of NPCs is their energy application. Li–S batteries are a type of rechargeable batteries with noticeable features of high specific energy density, low materials cost, environmentally friendliness, and abundant availability of sulfur, making them promising as a future energy storage set.21 However, some obstacles, including limited electrical conductivity of sulfur, the rapid capacity declining induced by the dissolution of lithium polysulfides (LiPSs), and significant volume alterations through discharging/charging cycles hinder their practical usage.22–24 To mitigate these problems, NPCs are one of the promising solutions due to their high electronic conductivity and the hierarchically porous structure to buffer volume alterations.25 To this end, some metal compounds, in particular transition metal sulfides, nitrides, and carbides, reveal a strong chemical bonding potential with LiPSs, which can improve the diffusion and catalytic reduction of soluble LiPSs.26,27 Recently, metal sulfides MxS, such as FeS,5 FeS2,28,29 NiS,30 and WS231 were investigated as cathodes for Li–S batteries. The results disclosed that high LiPSs adsorption and following redox reaction on these catalysts enhanced the sulfur usage. Their structural defects and poor electrical conductivity can be solved to a large extent by incorporation into NPCs that are conductive when produced at high temperatures,21,32 making the MxS/NPC system attractive cathode materials for Li–S batteries.
In this contribution, we succeeded in synthesizing N-PCMs containing iron sulfide nanoparticles (termed Fe1−xS/N-PCMs), via vacuum carbonization of an iron-containing porous hybrid PIL membrane as a sacrificial template at 900 °C. The Fe1−xS/N-PCM sample carrying a specific surface area of 274 m2 g−1, 5.0 wt% of N and 15 wt% of well-dispersed iron sulfide nanoparticles of 25 ± 7 nm in size was successfully applied as the cathode in Li–S batteries. In comparison with iron sulfide-free N-PCMs, the Fe1−xS/N-PCMs were found to be effective in decreasing the shuttling behaviour of LiPSs, showing a higher capacity and extended cycle life.
Fig. 1 depicts the synthetic pathway towards Fe1−xS/N-PCMs from the PIL/FDA-based hybrid porous polymer membrane. Briefly, the cationic PIL and the diacid compound, FDA were dissolved in DMSO and cast onto a glass plate to form the porous PIL membrane. A similar method was reported by us to prepare porous PIL membranes from trimesic acid.36 The fabricated porous polymer membrane was then converted to N-doped PCMs containing Fe1−xS (0 < x < 0.125)37 nanoparticles via vacuum carbonization at 900 °C. It is worth mentioning that the carbonization temperature plays a vital role in determining the physicochemical properties of the final porous carbon product. In this context, the carbonization of the hybrid porous polymeric membrane was repeated at temperatures ranging from 300 to 900 °C, so we could systematically study the effect of the carbonization temperature on the properties of the carbon products. The corresponding carbon samples were termed Fe1−xS/N-PCM-y, where y denotes the final carbonization temperature.
Fig. 1 Schematic illustration of the synthetic procedure of Fe1−xS/N-PCM-y from a hybrid porous polymer membrane derived from a cationic poly (ionic liquid) and 1,1′-ferrocene dicarboxylic acid. |
The obtained Fe1−xS/N-PCM-y products were collected and characterized comprehensively by thermogravimetric analysis (TGA), nitrogen sorption, Raman spectroscopy, and elemental analysis. Fig. 2a shows the oven yield (the mass ratio of the residue carbon to the polymer template) as a function of the carbonization temperature. As anticipated, the oven yield decreased stepwise with rising temperatures. The most significant mass drop due to structural fragmentation occurred from 300 to 400 °C; over 400 °C, only a gradual weight loss due to structural rearrangement was seen. Through TGA, we investigated the thermal degradation of the polymeric membrane under a N2 atmosphere (Fig. 2b). The mass falls rapidly at a temperature between 250 °C and 400 °C, followed by a gradual weight loss above 400 °C. The massive weight loss at around 300 °C is associated with the cyclization reaction of the cyano groups of PIL, that is, any fragment that is not covalently connected to the newly formed, thermally stable s-triazine network will be volatilized in this stage. After that, the established s-triazine network will stabilize the carbon product and only lose its weight slowly with increasing temperature. This phenomenon has also been observed similarly in nitrile-containing ILs reported by Dai et al.16 In general, the weight loss according to the TGA result illustrates the same trend as the oven yield vs. the carbonization temperature curve in Fig. 2a. The high oven yield at 900 °C of 23% could be explained by the high thermostability of the ionic liquid-based polymeric precursor, the Fe-based nonvolatile inorganic component, and the crosslinkable cyano groups of polymers that could develop a thermally stable network during the carbonization. In parallel, an iron-free N-PCM, which was obtained from FDA-free porous PIL membranes, was produced as reported previously.38 Generally speaking, the PIL chemical structure is the key factor to control the oven yield of the hybrid porous polymer membrane.
Assisted by the TGA data (Fig. S2, ESI†), the oxidation resistance of the Fe1−xS/N-PCMs-y products towards air at different temperatures can be investigated. From 100 to 300 °C, none of the Fe1−xS/N-PCM-y samples change their weight explicitly. The rapid weight loss of samples occurred at 300 to 550 °C due to oxidative burning. For a clear comparison, the temperature at a 10% loss of their weight was plotted in Fig. S3 (ESI†) against the carbonization temperature, at which the sample was produced. The result recommended that the oxidation resistance of the Fe1−xS/N-PCM-y samples was improved at a higher carbonization temperature. That is, carbons produced at a higher temperature are more oxidation resistant.
The N2 sorption isotherms of the carbon samples prepared at different temperatures from 300 to 900 °C are presented in Fig. 2c. It is clear that the samples prepared at 300 °C and 450 °C are poorly or non-porous thus have little N2 uptake up to p/p0 = 1, while the samples prepared above 450 °C show a type-I isotherm, and all have significant N2 sorption even below p/p0 = 0.05, indicative of the microporous nature of these samples. The specific surface areas of the carbon products calculated by BET equation (SBET) are found to change with carbonization temperature, as plotted in Fig. 2d (data in Table S1, ESI†). The low SBET value (<50 m2 g−1) of the products carbonized below 450 °C, illustrates the non-porous nature; however, the SBET increases rapidly once the carbonization temperature goes beyond 450 °C. Considering the large mass loss of the precursor in the range of 250–400 °C and the formation of a s-triazine network at 200–300 °C, it is reasonable to judge that the polymer first builds up the thermally stable s-triazine network and then the fragments volatize to leave pores behind. The highest SBET (∼401 m2 g−1) is obtained by pyrolysis at 700 °C and decreases above 700 °C, which can interpret that some of the smaller pores formed at 700 °C will start to collapse and merge into mesopores at a carbonization temperature above 700 °C.
The SBET of iron-free N-PCM samples prepared at different temperatures have been reported previously.38
The chemical composition of the carbon products was measured by combustion elemental analysis. The nitrogen content vs. the oven temperature was plotted in Fig. 2e, depicting a temperature-dependent nitrogen content change in carbons. The plot disclosed that the nitrogen content first increases from 11.7 wt% at 300 °C to a maximum of 14.5 wt% at 450 °C due to the thermal decomposition of N-poor species, e.g. the carboxylate groups, and the continuous formation of the s-triazine network that is rich in N. Above 450 °C it is followed by a reduction of nitrogen content with a rapid drop after 600 °C because any nitrogen atoms, if not connected to carbon atoms in an sp2 hybridized state, will be mostly kicked out of the graphitic plane.18,33,39 The iron content of the carbon products was calculated by the mass residue as Fe2O3 at 900 °C in the TGA tests conducted in synthetic air (Fig. S2, ESI†). According to Fig. 2f and Table 1, the iron content is augmented by increments of temperature up to 700 °C, after which the iron content remains practically the same until a final content of 9.4 wt% at 900 °C. The degree of graphitization and the phase structure information of the carbons in Fe1−xS/N-PCMs-y could be probed by Raman spectroscopy (Fig. 2g). The two separate carbon peaks detected for all samples at around 1350 and 1570 cm−1 are assigned to D- and G-band, respectively. The disorder in carbon atoms and structural defects is related to the D-band, while the G-band can be ascribed to the ordered carbon structures.40 The D-band intensity was slightly higher than the G-band for all samples (Fig. S4, ESI†), indicating that the ordered and disordered structure had almost the same contribution. Similar consequences were achieved by Paraknowitsch and He et al. on using IL and PIL, respectively, as a precursor for producing nitrogen doped carbon materials.33,41 Our result could be explained by the nitrogen doping-induced local disorder, which supports the D-band intensity, as nitrogen atoms are considered as “structural defects” in the carbon matrix.
Carbonization temperature/°C | 300 | 450 | 600 | 700 | 800 | 900 |
---|---|---|---|---|---|---|
Carbonization yield (%) | 63 | 35 | 29 | 27 | 24 | 23 |
S BET (m2 g−1) | 31 | 18 | 214 | 401 | 390 | 274 |
N content (wt%) | 11.7 | 14.5 | 13.6 | 10.02 | 7.97 | 5.02 |
Iron content (wt%) | 3.7 | 6.5 | 6.6 | 9.5 | 9.8 | 9.4 |
XRD structure | am. | am. | am. | Fe1−xS | Fe1−xS | Fe1−xS |
The phase structure of the iron compound was monitored by X-ray diffraction (XRD) tests of the Fe1−xS/N-PCMs-y samples (Fig. S5, ESI†). A clean Fe1−xS phase emerged once the carbonization temperature went above 700 °C. As a representative example, the XRD pattern of Fe1−xS/N-PCMs-900 is shown in Fig. 2h, where two broad peaks located at 23.5° and 43.6° can be identified and attributed to the (002) and (100) planes of a graphitic phase, respectively. The diffraction peaks at 29.9°, 33.8°, 43.7°, and 53.1° can be assigned to the (200), (204), (208), and (220) planes of pyrrhotite Fe1−xS, respectively (JCPDS No. 22-1120).
The elemental composition and the valence states of Fe1−xS/N-PCMs-900 were analysed by XPS. As presented in Fig. 2i, the typical iron 2p, sulfur 2p, nitrogen 1s, and carbon 1s peaks can be evidently detected in the survey spectrum at 715, 160, 401, and 285 eV, respectively. A high-resolution N 1s spectrum of the sample was fitted into four specific subpeaks determined at 398.38, 400.92, 399.88, and 404.48 eV (Fig. S6, ESI†), in accordance with the pyridinic (26.3 atom%), graphitic (53.5 atom%), pyrrolic (10.5 atom%), and oxidized-N (9.7 atom%), respectively.42,43 The C 1s spectrum certified the presence of three peaks (Fig. S7, ESI†), corresponding to the graphite-like carbon (284.65 eV), the nitrogen binding carbon (C–N, 285.2 eV), and a small peak at 287.7 eV which is the oxidized carbon (C–X).33 The S 2p XPS spectrum indicated three peaks at 163.8, 168.2 eV and 164.45 eV, which are in acceptable agreement with Fe1−xS44 (Fig. S8, ESI†). The high-resolution Fe 2p spectra exhibited the existence of three peaks (Fig. S9, ESI†); the two peaks located at 711.6 (39.1 atom%) and 724.9 eV (46.8 atom%) were attributed to Fe 2p2/3 and Fe 2p1/2, which can prove the presence of the Fe1−xS phase and the other peak at 716.8 (14.1 atom%) was associated with Fe3+.45
A photograph of the final intact carbon membrane sample Fe1−xS/N-PCMs-900 of 8.5 × 11.0 mm in size is presented in Fig. 3a. Since the manufacturing procedure is straightforward and simple, it can be easily scaled up to fabricate even larger ones. Fig. 3b and c represent an overview and a close view, respectively, of the cross-sectional scanning electron microscopy (SEM) images of the hybrid porous polymer membrane before pyrolysis. They revealed a three-dimensional interconnected macroporous structure with an average pore size of 262 ± 73 nm. Nitrogen sorption isotherms confirm that neither micropores nor mesopores exist in it (Fig. 2c). After carbonization, the macroscopic membrane shape is overall maintained but with a significant change in the microstructure. A high magnification cross-sectional SEM image of Fe1−xS/N-PCMs-900 (Fig. 3d) confirms the membrane shape with a thickness of around 50 μm. A further high-magnification SEM image (Fig. 3e) shows the dense packing of the macropores, where the pore size distribution histogram is displayed in Fig. S10 (that of the polymer membrane is shown in Fig. S11, ESI†), showing an average pore size of 101 ± 41 nm. Keep in mind that the SEM images visualize only the macropores and large mesopores, in previous discussions around Fig. 2c, the micropores have been identified by gas sorption measurements. Combining the SEM and N2 sorption isotherms, Fe1−xS/N-PCMs-900 combines both micro- and macropores. The bimodal pore size distribution of macropores and micropores in Fe1−xS/N-PCMs-900 makes a good condition for Li–S batteries. It is a perfect environment for efficient circulation of the electrolyte's active species in the macropores, where they could readily diffuse to and from the catalytic sites packed on the micropore surface.
To extract further structural information, transmission electron microscopy (TEM) is used. In Fig. 3f, the energy-dispersive X-ray (EDX) mapping proves the microscopic uniform distribution of Fe, S, N, and C elements across Fe1−xS/N-PCMs-900. The TEM image exhibits a random distribution of the Fe1−xS, nanoparticles in the carbon membrane (Fig. 3g). A high-resolution TEM (HRTEM) image of Fe1−xS/N-PCMs-900 and its enlarged view (Fig. 3h and i) show the layered texture with a spacing of 0.38 nm, corresponding to the graphitic phase. These crystalline areas are immersed in a large amorphous region, indicative of a turbostratic form of carbons with only a short-range order. At an even higher resolution, the dark dots corresponding to inorganic nanoparticles reveal the periodic lattice fringes with an interplanar distance of 0.23 nm and 0.19 nm, matching well the (208) and (220) faces of the hexagonal pyrrhotite Fe1−xS, respectively.21 The selected area electron diffraction (SAED) pattern measured for the Fe1−xS particles is presented in Fig. S12 (ESI†). The diffraction rings from the center toward the outside could be allocated to the (220), (208), (204), and (200) planes for the pyrrhotite Fe1−xS crystals.
The multimodal porous Fe1−xS/N-PCMs-900 will be a good candidate as a sulfur host material to suppress the shuttle effects of LiPSs. To test its electrochemical performance, the sulfur/Fe1−xS/N-PCMs-900 composite was prepared by a melting diffusion method. The specific sulfur contents inside the composites are around 70.2 wt% and 71.5 wt% for the sulfur/Fe1−xS/N-PCMs-900 and sulfur/N-PCMs-900 composites, respectively, as seen in Fig. S13 (ESI†). Fig. 4a shows the CV curves of the Li–S batteries in the CR2025 coin cell with Fe1−xS/N-PCMs-900 as the sulfur host material, which were measured in the voltage range of 1.7–2.8 V vs. Li/Li+ with a scan rate of 0.1 mV s−1. Two cathodic peaks can be observed at 2.23 and 1.95 V, which correspond to the reduction of sulfur to the soluble LiPSs (Li2Sn, 4 ≤ n ≤ 8), and to the further reduction towards short-chain Li2S2/Li2S, respectively. The anodic peak at 2.50 V could be attributed to the oxidation of Li2S2/Li2S to sulfur.27,46,47 The galvanostatic charging–discharging curves of Li–S batteries with Fe1−xS/N-PCMs-900 as host materials at 0.1C (1C = 1675 mA g−1) are illustrated in Fig. 4b. Two typical reduction plateaus are observed at 2.3 and 2.1 V in the discharge curve, which can be ascribed to the conversion reaction of S8 to LiPSs (Li2Sn, 4 ≤ n ≤ 8) and then to short-chain Li2S2/Li2S, respectively. The charge curves contain only one plateau at 2.35 V, which is assigned to the oxidation of Li2S2/Li2S to Li2S8/S.48,49 The cycling stability of the sulfur/Fe1−xS/N-PCMs-900 and sulfur/N-PCMs-900 cathodes at 0.1C is shown in Fig. 4c. The sulfur/Fe1−xS/N-PCMs-900 cathode delivers an initial specific discharge capacity of 1180.9 mA h g−1. After 100 cycles, the remaining specific capacity is 768.3 mA h g−1, indicating a stable cycling performance with a capacity declining rate of 0.2% per cycle. As a comparison, the sulfur/N-PCMs-900 cathode only shows a discharge capacity of 627.9 mA h g−1 after 100 cycles.
Furthermore, the rate capability tests were also measured by raising the discharge/charge current density from 0.05 to 1C per every 10 cycles and returning to 0.1C. The initial specific discharge capacities of the sulfur/Fe1−xS/N-PCMs-900 are determined as 1181.7, 802.4, 761.2, 701.4, and 490.5 mA h g−1 at 0.05, 0.1, 0.2, 0.5, and 1C, respectively (Fig. 4d), which is much higher than that of the sulfur/N-PCMs-900 electrode at low current density. Increasing the current density leads to a decrease in the specific capacity of Li–S batteries, which could be related to the polarization caused by the poor lithium diffusion at the high current density.50 When the current density is changed to 0.1C, the specific capacity of sulfur/Fe1−xS/N-PCMs-900 shows a durable discharge capacity of 798 mA h g−1, demonstrating the excellent reversibility of the electrode. The improved electrochemical performance of the Li–S batteries with Fe1−xS/N-PCMs-900 as the sulfur host material is ascribed to the following features: firstly, the multimodal pore structure and well-dispersed iron sulfide (Fe1−xS) nanoparticles in the N-PCMs can more efficiently confine LiPSs. Secondly, the Fe1−xS nanoparticles can facilitate the conversion reaction of LiPSs into Li2S.21 Additionally, the existence of the macropores in the carbon material contributes to a high sulfur loading (70%) and modulates the volume changes during cycling. As a result, the porous sulfur/Fe1−xS/N-PCMs-900 composite is a promising candidate as a sulfur host material for Li–S batteries.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ma00441g |
This journal is © The Royal Society of Chemistry 2021 |