Rational design of MXene/activated carbon/polyoxometalate triple hybrid electrodes with enhanced capacitance for organic-electrolyte supercapacitors

Rational design of MXene/activated carbon/polyoxometalate triple hybrid electrodes with enhanced capacitance for organic-electrolyte supercapacitors Jun-Jie Zhu a, ⇑ , Avireddy Hemesh b , Jordi Jacas Biendicho b , Luis Martinez-Soria a , Daniel Rueda-Garcia a , Joan Ramon Morante b,c , Belen Ballesteros a , Pedro Gomez-Romero a,d, ⇑ a Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and BIST, Campus UAB, Bellaterra, 08193 Barcelona, Spain b Catalonia Institute for Energy Research (IREC), Jardins de les Dones de Negre 1, 08930, Sant Adrià de Besòs, Spain c Faculty of Physics, University of Barcelona, Barcelona, Spain d Consejo Superior de Investigaciones Científicas (CSIC), Spain highlights POMs were nano-dispersed within MXene with assistance of AC matrix. The hybrid combined the merits and compensate for the demerits of each component. Gravimetric capacitance of the hybrid was 2.2 times that of MXene. POMs remained anchored in the longterm leading to great cycle stability. Hybrid’s asymmetric cells outperformed AC//MXene cells by weight and volume. graphical abstract article info Article history: Received 13 January 2022 Revised 26 April 2022 Accepted 28 April 2022 Available online 5 May 2022 Keywords: Hybrid electroactive materials MXene Polyoxometalates Organic Supercapacitors abstract We report a triple hybrid electrode (MXene/activated carbon (AC)/polyoxometalates (POMs)) combining the merits of three materials: MXene (high volumetric capacitance), AC (high gravimetric capacitance) and Phosphotungstate (fast redox). Phosphotungstic acid (HPW12) and tetraethylammonium phosphotungstate (TEAPW12) were the two POMs used to prepare MXene/AC/POMs triple hybrids. MXene/AC/ TEAPW12 outperformed MXene/AC/HPW12 in 1 M tetraethylammonium tetrafluoroborate (TEABF4)/ acetonitrile. Nano-dispersion of POMs facilitates charge storage through surface capacitive processes (91% at 2 mV s 1 ). MXene/AC/TEAPW12 delivered significantly higher gravimetric capacitance (87F g 1 at 1 mV s 1 ) than MXene (40F g 1 at 1 mV s 1 ) in the same organic electrolyte, without sacrificing much volumetric capacitance (less than 10%). The gravimetric capacitance of the triple hybrid was similar to that of MXene/AC, whereas its volumetric capacitance was 1.5 times higher. Replacing TEA cations with 1-ethyl-3-methylimidazolium cations (EMIM + ), the capacitance improved by 21%. Coupled with AC positive electrodes in an asymmetric cell, MXene/AC/TEAPW12 delivered 4.6 times higher gravimetric energy density and 3.5 times higher volumetric energy density than a similar MXene asymmetric cell https://doi.org/10.1016/j.jcis.2022.04.170 0021-9797/Ó2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). ⇑ Corresponding authors at: ICN2, Campus UAB, 08193 Bellaterra (Barcelona), Spain. E-mail addresses: [email protected] (J.-J. Zhu), [email protected] (P. Gomez-Romero). Journal of Colloid and Interface Science 623 (2022) 947–961 Contents lists available at ScienceDirect Journal of Colloid and Interface Science journal homepage: www.elsevier.com/locate/jcis at relatively high-power densities. This study proves that MXene/AC/TEAPW12 combines the merits and compensates for the demerits of each component and is a promising electrode material for organicelectrolyte supercapacitors. Ó2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Energy storage devices with high energy density, power density and long lifespan are in urgent demand for a growing number of applications. Supercapacitors are a good starting point for this challenge since they feature intrinsically high power density and long cycle life. However, they run short of energy density. There are two main ways to increase the energy density of supercapacitors: i) increasing working voltage through the use of organic electrolytes, and ii) increasing storage capacity by a proper choice of electrode materials or through hybridization[1]. In this study, we followed both approaches based on the design of complex but unique hybrid materials leading to an optimal combination of charge storage capacity and high conductivity together with the high voltage range typical of organic electrolytes. We first needed a material with a very high volumetric capacitance, and MXenes excel in this respect as they constitute a family of two-dimensional materials, usually containing one layer of transition metal and one layer of carbide. MXenes have a general formula M n+1 X n T x , where M is a transition metal, X is carbon and/or nitrogen, n is an integer between 1 and 3, and T x represents surface functional groups[2]. A conductive inner transition metal carbide layer enables fast electron supply to electrochemically active sites, which makes of MXenes promising electrode materials for supercapacitors[3]. MXenes stand out from their competitors because of the high volumetric capacitance[4–6], the key parameter for the industry when compactness is required. In this scenario, titanium carbides (Ti 3 C 2 T x ) is the most widely studied and developed MXene, which has been found to deliver a volumetric capacitance as high as 900F cm 3 in an aqueous electrolyte[3]. Conventional aqueous electrolytes allow for a stable potential window of less than 1.5 V. Since energy density is proportional to the square of the voltage, other electrolytes like ‘‘water-insalt” electrolytes (with V = 2.2 V)[7], organic electrolytes (with V = 2.7 V)[8] or ionic liquids (with V > 3 V)[9,10] have been preferred for a rational optimized design. Indeed, several groups have already attempted to use MXene in organic electrolytes. Some examples include MXenes for lithium-ion capacitors [11,12], sodium-ion capacitors [13], or just supercapacitors with conventional organic electrolytes[8,13–15]. Some experimental drawbacks have already been described in these types of applications. For example, contrary to the previous hypothesis, Ti 3 C 2 T x MXene provides much lower specific capacitance in conventional organic electrolytes (less than 40 F g 1 or 70 F cm 3 in 1 M TEABF4 in acetonitrile[16]) than in aqueous electrolytes. Various approaches have been followed to enhance the energy storage capabilities of MXenes, including expanding interlayer distance by other nanomaterials or functional groups[5,17–22], increasing active sites by making porous MXenes[23,24], or trying to introduce active materials[25–31], including polyoxometalates (POMs)[19,32]. However, these attempts have failed to intercalate POMs into MXenes and have led to mixed phases instead[32]. POMs constitute a family of nanometric metal oxide clusters, able to provide fast reversible multi-electron redox reactions not limited by diffusion[33–35]. As such, POMs have been used as active materials, just by themselves[35,36] or as active components in hybrid materials[33,35,37–41] in many kinds of energy storage devices. In these devices, their proper dispersion at the nanoscale and long-term anchoring onto a conducting substrate are key issues for fully harnessing the fast reversible multielectron redox activity characteristic of POMs. Several porous materials, such as activated carbon[39,41,42], and 2D materials, for example, reduced graphene oxide (rGO)[38,40], have been shown to effectively anchor POMs[39,40]. The resulting hybrid materials have shown enhanced performance, including capacitance[38–42], and even increased potential window[39]. The synthesis of hybrid materials made of MXenes and POMs has been tackled by various authors. For example, Chen et al.[25] modified Ti 3 C 2 T x MXene with Keggin-type POMs by ionic liquid polymer linker but used it in low voltage aqueous electrolytes. Chao et al.[32] proposed the in-situ growth of Keplerate-type POMs on Ti 3 C 2 T x MXene, leading to a hybrid material with interesting performance in lithium-ion capacitors and sodium-ion capacitors. However, in their study, the first report needed the use of a linker to bond the acid form of a POM to MXene and its use was restricted to aqueous electrolytes. And while the second used organic electrolytes they failed to disperse POMs, leading to a material made of a mixture of phases, namely, MXene and crystallized POM. Thus, in both cases there was a failure to harness the full potential of POMs combined with MXene while working in a high-voltage organic electrolyte. Accordingly, this was our goal, and in order to achieve it, we made use of activated carbon (AC) as a ‘‘transporting” phase for POMs, since, in a previous study, we had shown that this dual hybrid material can work in organic solvents, with properly dispersed POMs with their electroactivity not limited by diffusion [33]. Our working hypothesis was supported by our earlier work on AC/POM hybrids, which showed proper dispersion of POM clusters through physical adsorption in micropores, leading to materials able to operate both in aqueous or organic electrolytes[39,42,43]. Likewise, ‘‘hybrid” MXene/AC flexible electrodes have also been reported, with activated carbon facilitating the charge transfer from the electrode[16]. With these considerations in mind, we decided that a triple hybrid material formed by MXene, AC and POM could provide a complex but versatile system, in a rational design to produce a multifunctional electrode for energy storage[44]. Herein, we report the results of our work, which demonstrate that the MXene/AC/ POM system can be made to optimize the functionality of each of the three components, leading to enhanced volumetric and gravimetric capacitances in an organic electrolyte symmetric supercapacitor. 2. Experimental 2.1. Synthesis of MXene (Ti 3 C 2 T x ) MXene (Ti 3 C 2 T x ) was synthesized through a LiF/HCl etching method[45]. Namely, 0.5 g MAX phase (Ti 3 AlC 2 , 99%, Y-Carbon ltd.) was added slowly into a mixture of 0.5 g LiF (>99.99%, Sigma-Aldrich) and 10 mL 9 M HCl solution (diluted from 37% HCl, Sigma-Aldrich). The etching process was carried out in a capped Teflon-vessel for 24 h at 35°C. For washing and delamination, the mixture was poured into a 50 mL centrifugation tube and Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 948 washed with deionized water by centrifugation at 3500 rpm for 5 min. The supernatant was discarded. This step was repeated for several cycles until the pH of the dark-green supernatant was > 6. The sediments at the bottom expanded upon washing, which indicated delamination. Finally, the sediments were dispersed in water by hand-shaking and deaerated with Ar for 20 min and stored in the fridge for future use. A certain amount of the MXene colloid was filtered-off on a Celgard 3501 membrane to determine the concentration and for further use. The asprepared electrode had a thickness of 11 ± 1 l m and an areal density of 3 ± 0.3 mgcm 2 . 2.2. Synthesis of MXene/HPW12 20 mL MXene colloid (around 1 mg mL 1 ) was mixed with 20 mL 10 mM phosphotungstic acid (H 3 PW 12 O 40 (HPW12), reagent grade, Sigma-Aldrich) solution, and probe-sonicated for 1 h under circulating cooling water at 10 °C. The mixture was filtered-off and washed with Milli-Q water until the filtrate was neutral. The product was peeled off from the membrane, dried in a vacuum oven at 60 °C and stored in an argon-filled glovebox. The as-prepared sample (MXene/HPW12) could serve as a free-standing electrode. The as-prepared electrode had a thickness of 13 ± 1 l m and an areal density of 3.3 ± 0.3 mg cm 2 . 2.3. Synthesis of MXene/AC/HPW12 0.05 g activated carbon (DLC Super 30 from Norit Chemical) and 1.44 g phosphotungstic acid (HPW12) were added into 50 mL MXene colloid (around 1 mg mL 1 ) and kept in probe-sonication for 1 h under circulating cooling water to keep the temperature below 10 °C. Subsequently, the mixture was filtered-off onto a Celgard 3501 membrane to get a free-standing electrode. The asprepared electrode had a thickness of 24 ± 2 l m and an areal density of 2.8 ± 0.2 mg cm 2 . 2.4. Synthesis of MXene/AC/TEAPW12 AC/TEAPW12 was prepared following our previously reported method[33]. Then, 0.05 g AC/TEAPW12 was added into 50 mL MXene colloid (around 1 mg mL 1 ) and kept in probe-sonication for 1 h under circulating cooling water to keep the temperature below 10 °C. The mixture was filtered-off onto Celgard 3501 membrane to obtain a free-standing electrode. The as-prepared electrode had a thickness of 21 ± 2 l m and an areal density of 2. 4 ± 0.2 mg cm 2 . For comparison, MXene/TEAPW12 and MXene/AC were also prepared and characterized (Figs. S1, S2 and S10). 2.5. Material characterization Powder X-ray diffraction (XRD) patterns were collected on a PANalytical X’pert Pro-MRD diffractometer with Cu K a radiationðk= 1.5406 Å) and PIXel detector. The interlayer distance (d) of the (002) planes was determined by Bragg’s law: d¼k 2sinhð1Þ where kis the X-ray wavelength,his the the diffraction angle. Scanning electron microscopy (SEM) images were taken on a Quante 650 FEG microscope. For energy-dispersion X-ray spectra (EDX) semi-quantitative analysis, we averaged the elemental analysis of five spectra to present the final results. High resolution transmission electron microscopy (HR-TEM) images, high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) images, and selected area electron diffraction (SAED) patterns were obtained from an FEI Tecnai G2 F20 microscope. Volumetric N 2 sorption isotherms were collected at 77 K (N 2 ) using an ASAP 2020 HD (Micromeritics). Pore size distribution was estimated using a density functional theory (DFT) model (N 2 - cylindrical pores-oxide surface) implemented in the Microactive 4.00 software with a regularization factor of 0.01. X-ray photoelectron spectra were collected on Phoibos 150 probe from SPECS. Peak fitting for the high-resolution spectra was performed using CasaXPS Version 2.3.1. Before the peak fitting, the background was subtracted using a Shirley function. 2.6. Electrochemical characterization T-type Swagelok cells were applied to fabricate three-electrode configuration. The working electrodes were cut into 10 mm round pieces and the overloaded activated carbon counter electrodes were cut into 12 mm round pieces. 0.01 M Ag/Ag + served as the reference electrode. 1 M TEABF4 in acetonitrile or 1 M EMIMTFSI in acetonitrile served as the electrolyte. CR2032 coin cells were used to fabricate asymmetric devices. The electrochemical tests were carried out in an Argon-filled glove box with the oxygen and water lever under 5 ppm. The gravimetric capacitance (C m ) and the volumetric capacitance (C v ) were calculated from cyclic voltammograms (CVs) according to the following equations: C¼RIVðÞdV 2 vD Vð2Þ C v ¼C V e ð3Þ C m ¼C mð4Þ where RIVðÞdV is the integral area of CVs, v is the scan rate, D Vis voltage window, V e is the volume (cm 3 ) of the electrode and mis the mass (g) of the electrode. The gravimetric energy density E m (Wh kg 1 ) and volumetric E v energy density E v (Wh cm 3 ), gravimetric power density P m (W kg 1 ) and volumetric power density P v (W cm 3 ) were calculated from the two-electrode cell discharge data using the following equations: E m ¼C D V 2 3:62m total ð5Þ E v ¼C D V 2 3600 2V total ð6Þ P m ¼1000 I D V m total ð7Þ P v ¼I D V V total ð8Þ where m total is the total mass (g) of the two electrodes, V total is the total volume (cm 3 ) of the two electrodes, and Iis the discharging current (A). 3. Results and discussion The crystal structure of the solid phases was characterized by XRD. Fig. 1 presents the XRD patterns of the pristine MXene and derived materials. Table 1 shows the positions of (002) diffraction peaks and the interlayer distance. All the XRD patterns confirm the Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 949 layered structure of the pristine MXene and the hybrid materials but with different interlayer distances and various crystallinity. As can be observed, the pristine MXene has its (002) peak centred at 7.44°, and an interlayer distance of 11.87 Å, which is in the typical range of LiF/HCl-etched Ti 3 C 2 T x MXene[45]. The (002) peak of MXene/HPW12 is broader, revealing a poorer crystallinity, probably associated with the powerful probe-sonication that broke down large nanosheets. In this case, the peak shifts towards smaller two theta angles. On the other hand, the interlayer distance of 12.42 Å, is only slightly larger than that of the pristine MXene. Arup Chakraborty et al. have recently pointed out that even small anions such as Cl – and Br – are difficult to insert into MXenes[46]. Since PW12 anions are 1 nm in diameter, significantly larger than pre-intercalated Li + , we conclude that the PW12 clusters are not intercalated into MXene but are mainly anchored on the surface. The XRD pattern of MXene/AC/HPW12 presents a broad peak at around 6.91°, corresponding to a slightly larger interlayer distance (12.80 Å) compared with MXene. The AC particles are micron scale. Thus, neither AC nor PW12 clusters can insert MXene layers. The slight variation of interlayer distance is supposed to be associated with the fact that high-power sonication slightly increases interlayer distance[47]. Furthermore, the sonication during the preparation and the amorphous AC reduced the overall crystallinity of the final material. For the same reasons, the XRD pattern of MXene/AC/ TEAPW12 presents a similarly broad peak, with minor shifting with respect to the pristine MXene. The XRD patterns of the hybrid materials (MXene/HPW12, MXene/AC/HPW12 and MXene/AC/ TEAPW12) do not exhibit any other peaks corresponding to HPW12 nor TEAPW12. This is very significant since it rules out the inclusion of the POMs as crystallized extended phases. Instead, XRD results are consistent either with the absence of POMs in the hybrid or with POM clusters being dispersed individually or at most in groups small enough to prevent bulk diffraction. The following EDX analyses in Table 2, confirm the existence of tungsten (PW12). In order to observe the morphology and evaluate the loading mass of PW12 clusters, the samples were characterized under the scanning electron microscope equipped with energydispersive X-ray spectroscopy. The pristine MXene (Fig. 2a) exhibits an accordion-like structure, which remains even after suffering high-power probe-sonication (MXene/HPW12, Fig. 2b). For the MXene/HPW12 material, we cannot detect or discern any other heterogeneous aggregates, but the EDX spectrum confirms the presence of PW12 clusters, not apparent most likely due to the small nanosize of the PW12 clusters (1 nm). MXene/AC/HPW12 (Fig. 2c) and MXene/AC/TEAPW12 (Fig. 2d) exhibit a similar morphology, different from pristine MXene with the particles spreading on or surrounded by nanosheets. Since in the XRD patterns of MXene/AC/HPW12 and MXene/AC/TEAPW12 we cannot discern any diffraction peaks except those from MXene, we confirmed that the particles were amorphous AC, with dispersed, not wellcrystallized PW12 clusters. We have previously shown that AC is an ideal matrix for absorbing PW12 and spreading them in nanopores[39]. We therefore propose that most PW12 clusters are absorbed on AC in these triple hybrid materials, as proven in the EDX spectra and HAADF-STEM images, shown in the following Table 2 and Fig. 4. Table 2 lists the weight percentage of tungsten and titanium derived from EDX spectra. MXene/HPW12 only contains 6 wt% tungsten even with the assistance of high-power sonication, revealing that MXene is not a good substrate for direct-anchoring of POMs. That is most likely the reason why most previous studies had to resort to linking groups[48,49] or in-situ synthesis[32]. Indeed, through the in-situ synthesis (synthesis of MXene/ TEAPW12, in supplementary materials), we could load a large amount of TEAPW12 on MXene (Table S1). As for MXene/AC/ HPW12, improvement in the loading mass of PW12 must be ascribed to the inclusion of AC in the material providing a microporous matrix for anchoring POMs. MXene/AC/TEAPW12 contains less W than Mxene/AC/HPW12 because we carried out an ex-situ synthesis, in which TEAPW12 clusters had been already immobilized in AC accounting for 37.5 wt% in AC/TEAPW12 precursor. Fig. 1. (a) XRD patterns of pristine MXene, MXene/HPW12, MXene/AC/TEAPW12 and MXene/AC/HPW12. (b) Zoom-ins of (a) in the 5–10°2hrange showing (002) peaks. Table 1 The positions of (002) peaks and their corresponding interlayer space derived from the XRD patterns. Sample 2hvalues for (002) diffraction peaks/° Interlayer distance/Å MXene 7.44 11.87 MXene/HPW12 7.09 12.42 MXene/AC/ TEAPW12 6.91 12.80 MXene/AC/ HPW12 6.95 12.71 Table 2 Semi-quantitative analysis result from EDX spectra. Sample Ti/ wt. % W/ wt. % Mxene/HPW12 51.2 ± 3.6 6.1 ± 1.1 Mxene/AC/HPW12 31.4 ± 2.1 17.5 ± 2.0 Mxene/AC/TEAPW12 39.7 ± 2.8 13.5 ± 1.8 Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 950 Since the SEM images reveal a remarkable expansion of the space among MXene nanosheets due to the AC particles, we carried out volumetric N 2 sorption experiments to investigate the influence of the inclusion of HPW12 or AC/TEAPW12 on the porous properties of MXene. Fig. 2e presents the nitrogen sorption isotherms of MXene, MXene/HPW12 and MXene/AC/TEAPW12. MXene/AC/TEAPW12 exhibits a significantly larger absorbed volume. The specific surface area of MXene/AC/TEAPW12 is 499 m 2 g 1 , 30 times larger than pristine MXene (15 m 2 g 1 ), which we ascribe to the microporous nature of AC/TEAPW12 (specific surface area = 1007 m 2 g 1 )[33]. The pore-size distribution (Fig. 2f) reveals that MXene/AC/TEAPW12 exhibits a hierarchical porous structure characterized by the dominance of AC micropores and the development of the mesopores inherited from MXene. While the inclusion of AC/TEAPW12 promotes mesopore formation, the addition of HPW12 leads to a decrease of specific surface area (from 15 m 2 g 1 to 7.8 m 2 g 1 ), the pore structure remains the same (Fig. 2f). This can be explained by the heavy molecular weight of HPW12 as well as the fact that HPW12 clusters occupy some mesopores. In order to compare the surface composition, bonding forming and element valence, we carried out XPS analysis on the pristine MXene, the double hybrid (MXene/HPW12) and the triple hybrid material (MXene/AC/TEAPW12). The low-resolution survey spectra (Fig. 3a) show that in addition to elements Ti, C, F, Cl and O that compose MXene, the hybrid materials (MXene/HPW12 and MXene/AC/TEAPW12) contain element W (Fig. 3a inset, W 4d peaks). The elemental compositions calculated by XPS are presented in Table S2, which shows that MXene/AC/TEAPW12 has a higher W atomic ratio than MXene/HPW12. Fig. 2. SEM images of (a) MXene, (b) MXene/HPW12, (c) MXene/AC/HPW12 and (d) MXene/AC/TEAPW12. (e) N 2 sorption isotherms and (f) pore size distribution curves of MXene, MXene/HPW12 and MXene/AC/TEAPW12. Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 951 Fig. 3. (a) XPS full spectrum of MXene, MXene/HPW12 and MXene/AC/TEAPW12 (inset is the zoom-in area to discern W 4d peaks). C 1 s spectra of (b) MXene, (c) MXene/ HPW12 and (d) MXene/AC/TEAPW12. Ti 2p spectra of (e) MXene, (f) MXene/HPW12 and (g) MXene/AC/TEAPW12. (g) W 4f spectra of MXene/HPW12 and MXene/AC/ TEAPW12 (Ti 3p overlap). Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 952 Fig. 3b-d illustrate the C 1 s spectra of MXene, MXene/HPW12 and MXene/AC/TEAPW12. The pristine MXene and MXene/ HPW12 show three peaks located at 281.85, 284.8 and 288.1 eV, which are matched to C–Ti, CAC and CAO, respectively. The MXene/AC/TEAPW12 shows an extra peak at 285.3 eV, which we attribute to the CAC bonds in AC. Fig. 3e-g depict the Ti 2p spectra of MXene, MXene/HPW12 and MXene/AC/TEAPW12, in which the Ti–C (2p 3/2 455.0 eV, 2p 1/2 461.1 eV), Ti(II) (2p 3/2 455.8 eV, 2p 1/2 461.5 eV), Ti(III) (2p 3/2 457.0 eV, 2p 1/2 462.9 eV) and Ti(IV)(2p 3/2 458.6 eV, 2p 1/2 464.4 eV) peaks agree well with results from previous research[50,51]. Ti(II) and Ti(III) are associated with surface terminating groups, while Ti(IV) is associated with TiO 2 . MXene/ HPW12 presents less intensive Ti(II) peaks and more intensive Ti (III) peaks than MXene, implying the inclusion of the small amount of HPW12 changed the valence of Ti. HPW12 is anchored on MXene through chemical bonds. By contrast, the Ti 2p spectrum of MXene/AC/TEAPW12 does not show such significant change except for stronger Ti(IV) peaks, which should be ascribed to the fact that AC is just physically constrained among MXene nanosheets and TEAPW12 are pre-immobilized on AC. The stronger Ti(IV) peaks in MXene/HPW12 and MXene/AC/TEAPW12 must be associated with the slight oxidation due to the high-power probe-sonication. We also collected high-resolution spectra from 25 to 45 eV to investigate the W 4f orbital in hybrid materials. However, since the binding energy of Ti 3p is very close to W 4f, the W 4f 7/2 and 4f 5/2 peaks are deformed. Nonetheless, it is still possible to discern the two main peaks at 36.1 and 38.2 eV, agreeing well with W(Ⅵ)4f 7/2 and 4f 5/ 2 peaks in PW12 (Fig. S2d). HAADF-STEM is an effective method to investigate the dispersion of heavy atoms, particularly the PW12 clusters. Fig. 4a-d present HAADF-STEM images of pure MXene and these hybrid samples. The corresponding HR-TEM images are presented in Fig. S3. Under HAADF-STEM, MXene (Fig. 4a) presents a homogeneous texture, while MXene/HPW12 (Fig. 4b) is slightly different. In some areas, the nanosheets are covered by a thin, bright layer, corresponding to PW12 confirmed by EDX (Fig. S4). We conclude that MXene is not an effective substrate for dispersing PW12 clusters because aggregation happens even in low loading mass. On the other hand, the dispersion states of PW12 clusters in triple hybrid materials MXene/AC/HPW12 (Fig. 4c) and MXene/AC/TEAPW12 (Fig. 4d) are quite different from the MXene/HPW12. The bright dots, around 1 nm (as confirmed by EDX element mapping and line scan in Fig. S5), spread homogeneously on the substrate, indicating that the PW12 clusters are well-dispersed at the nanoscale. Fig. 4ef present the STEM image and the EDX element mapping in an extended area, in which we can discern the MXene nanosheets surrounding the AC particles loaded with PW12 clusters. The SAED patterns (Fig. S6) prove the triple hybrid materials have mixed phases: crystallized MXene, amorphous AC and uncrystallized PW12 clusters. This is in good agreement with the fact that MXene/AC/HPW12 and MXene/AC/TEAPW12 do not show any diffraction peaks other than those of MXene in XRD. With the images of SEM and STEM, we illustrate how the three components combine in MXene/AC/TEAPW12 (Fig. 5). First, the PW12 nanoclusters are anchored on microporous AC matrix and dispersed at nanoscale. Then, the PW12-anchored AC particles mix with flexible MXene nanosheets and stack together under vacuum-assisted filtration. Finally, we obtain the PW12-anchored AC wrapped by MXene nanosheets. Cyclic voltammograms were carried out to characterize the electrochemical properties of the materials, as well as to determine the capacitance of the various electrodes studied and to analyze the energy storage mechanisms. The CV of MXene in 1 M TEABF4 in acetonitrile (Fig. 6a) is predominantly capacitive (ideally rectangular shape) but superposed a pair of broad redox waves. The reduction wave centred at –1.75 V vs. Ag/Ag + must be ascribed to the intercalation of TEA + . The corresponding oxidation peak, centred at –1.48 V vs. Ag/Ag + , is broader and weaker, indicating the Fig. 4. HAADF-STEM images of (a) MXene, (b) MXene/HPW12, (c) MXene/AC/HPW12 and (d) MXene/AC/TEAPW12. (e) STEM image at low magnification and (f) the corresponding element mapping of MXene/AC/TEAPW12. Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 953 intercalation–deintercalation process is not 100% reversible. This agrees well with the XRD patterns of cycled MXene, in which we observe a significant shifting of (002) peak to low angle (Fig. S7). In the CV of MXene/HPW12 (Fig. 6a), besides the broad waves from MXene, we can discern some new small peaks from PW12 clusters. Since the loading mass of HPW12 is very low, these peaks are weak and their contribution to the capacitance is not very substantial. The CV of MXene/AC/TEAPW12 (Fig. 6b) mainly shows four pairs of redox waves, similar to AC/TEAPW12[33]. Since TEAPW12 is immobilized on AC in advance, MXene/AC/TEAPW12 presents redox waves at the same positions. Fig. 5(b) illustrates how the triple hybrid MXene/AC/TEAPW12 works. The MXene nanosheets and porous AC particles construct the framework for electron flow. The hierarchical porous structure facilitates ion diffusion. As a result, all three components contribute to electrochemical energy storage: MXene and TEAPW12 store charges through reversible redox reactions; AC store charges through electric double-layer capacitance. On the other hand, the situation is more complicated for MXene/AC/HPW12. The CV of MXene/AC/HPW12 (Fig. 6b) shows more than six pairs of redox waves. In principle, PW12 clusters only present three or four pairs of redox waves in this potential range (more redox pairs are possible in a wider potential range, but they are not 100% reversible). The extra redox waves must be associated with different anchoring states: PW12 clusters anchoring on AC; and PW12 clusters anchoring on MXene (The classification of these redox waves is presented in Fig. S8a-c). To investigate the energy storage mechanism, we carried out CVs at various scan rates (Fig. S9). The dependence of the current response on the scan rate can provide insights into the chargestorage mechanism according to Eq9 as suggested by John Wang et al. [52]: iðVÞ v 1=2 ¼k 1 v 1=2 þk 2 ð9Þ where iðVÞis the current at the given potential V,vis the scan rate. k 1 is a variable characterizing the contribution from surface capacitive processes (e.g. pseudocapacitance, electric double layer capacitance), and k 2 is the other variable representing the contribution from diffusion-controlled processes (e.g. intercalation–deintercala tion, sluggish redox reactions) at a given potential. k 1 and k 2 can be determined by a linear fitting of Eq9. Fig. 6c shows the contribution of the diffusion-controlled process versus the surface capacitive process to the overall capacitance of MXene/AC/TEAPW12 from 0.2 to 2 mV s 1 . The surface capacitive process dominates the energy storage process in the whole range. At 0.2 mV s 1 , the surface capacitive process accounts for 78.9% of total capacitance; when the scan rate rises to 2 mV s 1 , this ratio increases to 91%. With the rise of the scan rate, the contribution of surface capacitance increases, revealing that the surface capacitive process prevails at faster scan rates. This trend is common in many other hybrid materials[53,54]. MXene/AC/ TEAPW12 maintains a higher surface capacitive contribution in the whole range, mainly benefiting from the excellent dispersion of PW12 clusters. Fig. 6d presents the gravimetric capacitance of MXene, MXene/ HPW12, MXene/AC/HPW12 and MXene/AC/TEAPW12 in 1 M TEABF4 in acetonitrile at various scan rates. MXene and MXene/ HPW12 almost deliver the exact equivalent gravimetric capacitance because the loading of PW12 on MXene/HPW12 is very low. MXene/AC/HPW12 and MXene/AC/TEAPW12 exhibit significantly higher gravimetric capacitance, thanks to AC with high gravimetric capacitance. Compared with the hybrid material Fig. 5. Schematic illustration of the combination of TEAPW12, AC and MXene. (b) Schematic illustration of the working mechanism of MXene/AC/TEAPW12. Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 954 MXene/AC of the same weight ratio (MXene/AC 1:1, 88F g 1 at 0.1 Ag 1 [16], or 77F g 1 at 1 mV s 1 (Fig. S10)), MXene/AC/TEAPW12 could deliver the equivalent gravimetric capacitance. The addition of TEAPW12 does not improve the gravimetric capacitance due to the large molecular weight of PW12 clusters (>2800 g mol 1 ). On the other hand, as other studies show, the anchoring of TEAPW12 can improve the volumetric capacitance, instead of the gravimetric capacitance in the same electrolyte[33]. Due to the porous nature of AC, MXene/AC delivers 33% lower volumetric capacitance than that of MXene[16]. In contrast, in our study, the volumetric capacitance of MXene/AC/TEAPW12 is only around 10% lower than pristine MXene (Fig. S10), but much higher than MXene/AC[16], AC/ TEAPW12[33] and AC[33]. To investigate the electrochemical response as well as diffusion properties of the triple hybrids, we carried out impedance tests (Fig. 6e). All the spectra are fitted by a modified Randles circuit (Fig. 6f). The fitted values are presented in Table S4. The intercepts at the very high-frequency region represent equivalent series resistance (R s ), which is a combination of the ionic resistance of the electrolyte, the intrinsic resistance of the active materials, and the contact resistance at the active material/current collector interface. Since all the electrodes were characterized in the same setup, their R s are similar. The arc at high frequency represents chargetransfer resistance R ct . The following transitory parts between the semicircles and the final linear parts represent the Warburg impedance, which is associated with mass transfer (diffusion). The R ct of MXene and MXene/HPW12 are smaller (around 50%) than MXene/AC/HPW12 and MXene/AC/TEAPW12, but the diffusion impedance values are significantly larger (5 times), revealing that different steps restrict the charge store process. In MXene and MXene/HPW12, due to the high electronic conductivity of MXene nanosheets, the active sites which have already been Fig. 6. (a) Cyclic voltammograms of MXene and MXeneHPW12 at 1 mV s 1 in 1 M TEABF4 acetonitrile. (b) Cyclic voltammograms of MXene/AC/HPW12 and MXene/AC/ TEAPW12 at 1 mV s 1 in 1 M TEABF4 acetonitrile. (c) Surface capacitive contribution and diffusion-controlled contribution at various scan rates. (d) Gravimetric capacitance of MXene, MXene/HPW12, MXene/AC/HPW12 and MXene/AC/TEAPW12 at various scan rates. (e) Electrochemical impedance spectra of MXene, MXene/HPW12, MXene/AC/ HPW12 and MXene/AC/TEAPW12. (f) Equivalent circuit for fitting impedance spectra. Jun-Jie Zhu, A. Hemesh, Jordi Jacas Biendicho et al. Journal of Colloid and Interface Science 623 (2022) 947–961 955