Enhancing the energy storage performances of metal–organic frameworks by controlling microstructure

Metal–organic frameworks (MOFs) are among the most promising materials for next-generation energy storage systems. However, the impact of particle morphology on the energy storage performances of these frameworks is poorly understood. To address this, here we use coordination modulation to synthesise three samples of the conductive MOF Cu3(HHTP)2 (HHTP = 2,3,6,7,10,11-hexahydroxytriphenylene) with distinct microstructures. Supercapacitors assembled with these samples conclusively demonstrate that sample microstructure and particle morphology have a significant impact on the energy storage performances of MOFs. Samples with ‘flake-like’ particles, with a pore network comprised of many short pores, display superior capacitive performances than samples with either ‘rod-like’ or strongly agglomerated particles. The results of this study provide a target microstructure for conductive MOFs for energy storage applications.


Materials
All materials were purchased from commercial suppliers, used without additional modification and handled in air unless specified below.
Tetraethylammonium tetrafluoroborate (NEt 4 BF 4 ) was dried under vacuum at 100 °C for 48 h before being transferred to a N 2 -filled glovebox. Anhydrous acetonitrile (ACN) was purged with N 2 for 3 h before taking it into a N 2 -filled glovebox, where it was further dried by the addition of activated 3 Å molecular sieves. Molecular sieves were activated at 250 °C in a vacuum oven for 12 h prior to transferring into a N 2 -filled glovebox. 1-Ethyl-3-methylimidazolium tetrafluoroborate (EMIM-BF 4 ) was dried at room temperature under dynamic vacuum for 120 h before being transferred to a N 2 -filled glovebox.
Any unexpected observations and safety hazards are noted below.

A-CuHHTP, C-CuHHTP
Ammonia and pyridine-modulated Cu 3 (HHTP) 2 samples (A-CuHHTP and C-CuHHTP, respectively) were synthesised by modifying a recently published literature procedure. 1 For A-CuHHTP, 35% aqueous ammonia solution was used as the source of the modulator. For C-CuHHTP, pyridine was used as the modulator.
A solution of Cu(NO 3 ) 2 .3H 2 O (0.127 g, 0.526 mmol, 1.65 eq) and the modulator (50 eq) in distilled water (2 mL) was prepared. The resulting royal blue solution was added dropwise to a dispersion of H 6 HHTP (0.103 g, 0.318 mmol, 1.00 eq) in distilled water (8.2 mL) in a 40 mL screw-top vial. The vial was closed with a screw cap fitted with a septum and the resulting mixture was heated in a furnace oven at 80 °C for 24 h. The dark blue precipitate formed was separated by centrifugation and the supernatant layer discarded. The precipitate was then washed successively with water (3 × 30 mL), ethanol (3 × 30 mL), and acetone (3 × 30 mL). Washing was performed by centrifuging the precipitate with the desired washing solvent for 15 minutes before removing the supernatant layer, replacing with fresh washing solvent, shaking the centrifuge tube vigorously, and centrifuging once again. No soaking of the precipitate was performed. The precipitate was then filtered by vacuum filtration, and the resulting dark blue powder was dried at 80 °C under dynamic vacuum for 96 h and then stored in a N 2 -filled glovebox until used. No differences in the physical appearances of the reaction mixtures and products were observed between the syntheses of A-CuHHTP and C-CuHHTP.

B-CuHHTP
DMF-modulated Cu 3 (HHTP) 2 (B-CuHHTP) was synthesised by modifying a recently published literature procedure. 2 H 6 HHTP (0.117 g, 0.361 mmol, 1.00 eq) was added to a mixture of 1.5 mL (19.5 mmol, 54.0 eq) of DMF in 12 mL of H 2 O. The resulting solution was sonicated for 15 mins at room temperature, leading to partial dissolution of the solid and the formation of a dark brown solution. A solution of CuSO 4 .5H 2 O (0.205 g, 0.820 mmol, 2.27 eq) in distilled water (9 mL) was also prepared. Both solutions were heated in an oven at 80 °C for 10 mins prior to being mixed. The 40 mL vial was closed with a screw cap fitted with a septum, and the resulting mixture was then heated at 80 °C for 12 h. The dark blue precipitate formed was separated by centrifugation and the supernatant layer discarded. The precipitate was then washed successively with water (3 × 30 mL), ethanol (3 × 30 mL), and acetone (3 × 30 mL). Washing was performed as stated above. The precipitate was filtered by vacuum filtration and washed further on the filter with water (250 mL), ethanol (250 mL) and acetone (250 mL). The resulting dark blue powder was dried at 80 °C under dynamic vacuum for 96 h and then stored in a N 2 -filled glovebox until used.

X-ray Diffraction (XRD)
High resolution synchrotron XRD data were collected at the I11 beamline at Diamond Light Source.
Samples were loaded into borosilicate glass capillary tubes (0.5 mm outside diameter, 0.01 mm wall thickness; Capillary Tube Supplies Ltd.) in a N 2 -filled glovebox, and then sealed with Loctite EA 3430 epoxy adhesive. The adhesive was allowed to cure in the glovebox at ambient temperature for at least 72 h before removal of the capillaries. Diffraction patterns were collected under ambient conditions using a Mythen II position-sensitive detector (PSD) with two 5-second scans separated by an angular shift in detector position of 2.5 °. The wavelength and intrinsic peak-shape parameters were refined against a known Si 640c NIST standard. The refined wavelength for the PSD scans was 0.82683 Å (~ 15 keV). Simulated XRD patterns were produced using VESTA version 3. 3

Transmission Electron Microscopy (TEM)
TEM imaging was performed using Thermo Fisher Scientific F20 operating with at 200 keV with a beam flux of < 30 e − / Å 2 s. Images were taken using a Gatan OneView 4k Camera with an acquisition time of 0.5 s. The sample was aligned off the region of interest and images were taken immediately when on the region of interest to avoid the impacts of electron beam damage. Analysis of TEM images was performed in FIJI ImageJ. FFT image simulations were performed using CrystalMaker software.

Electrical Conductivity Measurements
The electrical conductivity of Cu 3 (HHTP) 2 samples was measured via a two-point probe method using a homemade set-up. Pellets composed of Cu 3 (HHTP) 2 were prepared by loading the material into a 13 mm Evacuable Pellet Die (Specac) and applying a force of 4 ton-force cm −2 for 1 min with a Specac hydraulic press. The areal mass loading of the pellets was approximately 50 mg cm −2 , and the thickness varied between 200 -300 μm. Samples were pressed between two stainless-steel electrodes using a hydraulic press (Specac), with PTFE disks used to prevent short-circuiting through the press. All

Elemental Analysis
Cu content was determined via inductively coupled plasma optical emission spectroscopy using a Thermo Scientific iCAP-7400 ICP spectrometer. C, H and N concentrations were determined via CHN combustion analysis using an Exeter Analytical CE-440, with combustion at 975 °C.

Gas Sorption
Low pressure N 2 isotherms (adsorption and desorption) were collected using an Anton Parr Autosorb iQ-XR at 77 K. An oven-dried sample cell (Type A long cell, 9 mm outer diameter, LG bulb) was tared before being loaded with the sample. Ex-situ degassing (80 °C, 24 h) was performed before the evacuated tube was weighed again to determine the sample mass. Isotherms were collected over 24 -30 h, and the samples reweighed following analysis to ensure accurate mass readings. Sorption isotherms were evaluated in AsiQwin version 5.21 software. All Cu 3 (HHTP) 2 samples displayed Type I N 2 isotherms, with high gas uptake below 0.1 P/Po, indicating microporosity. Material BET areas were calculated from isotherms using the BET equation and Rouquerol's consistency criteria implemented in AsiQwin. [5][6][7] All pore size distribution fittings were conducted in AsiQwin using N 2 at 77 K on carbon (cylindrical pores) quenched solid density functional theory (QSDFT) model with a bin pore width of 0.5 Å.

Electrochemical Characterisation Electrode Preparation
Freestanding composite Cu 3 (HHTP) 2 films were prepared using an existing literature method. 1 The masses of components were calculated so that the final films had a composition of 85 wt% Cu 3 (HHTP) 2 , 10 wt% acetylene black (nominal surface area stated by supplier = 75 m 2 g −1 , measured BET area = 62 m 2 g −1 ), and 5 wt% PTFE. The same procedure was used for all Cu 3 (HHTP) 2 samples, and all films had a thickness of ca. 250 μm. Films made with acetylene black as the only active material were prepared using the same method.
Neat Cu 3 (HHTP) 2 pellet electrodes were prepared using the same method as described above (see 'Electrical Conductivity Measurements' section).

Supercapacitor Assembly
Symmetric supercapacitors were prepared as coin cells in Cambridge Energy Solutions CR2032 SS316 coin cell cases. Film electrodes were cut from freestanding composite Cu 3 (HHTP) 2 and acetylene black films with areal mass loadings ranging between 7 -18 mg cm −2 . Pellet electrodes were used as prepared. The electrodes were dried in vacuo at 80 °C for at least 24 h prior to assembling the cell in a

Electrochemical Cell Characterisation
All electrochemical measurements were carried out using a Biologic SP-150 potentiostat and a Biologic BCS-800 Series ultra-precision battery cycler. EIS measurements were performed in the frequency range from 1 MHz to 10 mHz (decreasing frequency) at the open circuit voltage (OCV) using a singlesinusoidal signal with a sinus amplitude of 10 mV and drift correction applied. The specific capacitance, C g (F g −1 ), was calculated from GCD discharge profiles using the Supycap Python code (GitHub -AdaYuanChen/Supycap: Analysis tool for the CC and CV experiment of supercapacitors). C g values were determined using only the mass of active Cu 3 (HHTP) 2 material in the supercapacitors.
The internal resistance, R, was calculated from both Nyquist plots produced from EIS measurements, and from the voltage drop at the beginning of GCD discharge profiles. For the calculation from Nyquist plots, R was obtained from the low-frequency interception of the semi-circular response with the Re(Z) axis, as in the literature. 8 For the calculation from GCD discharge profiles, the Supycap Python code was used. Current densities were calculated by dividing the current applied during the GCD experiment, I, by the average mass of active MOF material per electrode, .
̅ m C' and C'' were calculated from EIS data using the following equations: Where C' is the real capacitance (F), C'' is the imaginary capacitance (F), ω is the angular frequency (rad), Z'(ω), also written as Re(Z), is the real impedance (Ω), Z''(ω), also written as Im(Z), is the imaginary impedance (Ω), and Z(ω) is the total impedance (Ω). Volume of Unit Cell / Å 3 1281 2660 Figure S1: Selection of SEM images from a range of A-CuHHTP samples. The measurements shown (summarised in Table S2 below) were used to determine the length-to-aspect ratio of the 'flake-like' particles.    Table S3 below) were used to determine the length-to-aspect ratio of the 'rod-like' particles.