Electrochemical deposition of zeolitic imidazolate framework electrode coatings for supercapacitor electrodes

Zn and Co electrodes have been successfully coated with ﬁ ve different zeolitic imidazolate frameworks ZIFs (ZIF-4, ZIF-7, ZIF-8, ZIF-14 and ZIF-67) via the anodic dissolution method. Careful control of the reaction conditions allows for electrode coating growth; in contrast to previous reports of electrochemical ZIF growth, which have not succeeded in obtaining ZIF electrode coatings. Coating crystallinity is also shown to be heavily dependent upon reaction conditions, with amorphous rather than crystalline material generated at shorter reaction times and lower linker concentrations. Electrochemical applications for ZIF-coated electrodes are highlighted with the observation of an areal capacitance of 10.45 mF cm (cid:1) 2 at 0.01 V s (cid:1) 1 for additive-free ZIF-67 coated Co electrodes. This is superior to many reported metal organic framework (MOF)/graphene composites and to capacitance values previously reported for additive-free MOFs.

Electrochemical techniques, in contrast, enable good quality coatings to be obtained at low temperature, with short reaction times and without specialised equipment. The only real limitation is that the growth substrates must be conductive, however this is also a potential advantage as the ability to obtain MOF coatings directly on electrodes could permit their applications in electrochemical energy storage to be studied more readily.
There are two main electrochemical methods for obtaining MOFs as coatings which can be broadly described as cathodic and anodic. In the cathodic method both the metal cation and linker species are present in the electrolyte solution. Electrochemical reduction of a pro-base generates a base, which goes on to reduce water to produce a high pH environment local to the cathode surface [58,59]. A high pH environment can promote MOF formation; basic solutions of trialkylamines are often added in traditional solvothermal syntheses. High pH favours MOF formation as it increases the equilibrium concentration of the deprotonated, anionic form of the linker, which interacts with the cationic metal species to produce the MOF. As the high pH environment generated in the cathodic electrochemical method is local to the cathode surface, formation of a MOF coating is favoured over formation of bulk MOF in solution [58,59]. Recent work by the same group has further expanded upon this mechanistic understanding [60] and the method has allowed for the production of coatings of Zn [58][59][60], Eu [21], Tb [23] and Zr [61] MOFs via this method.
Conversely in the anodic method only the linker species is present in the electrolyte solution; electrochemical oxidation of the metallic anode provides the source of metal cations for the MOF formation [62]. Having originally been developed as a method for bulk solution synthesis of MOFs [62] it was discovered that, dependent upon the reaction conditions, formation of a MOF coating on the anode could be favoured and used to coat defined electrode architectures [63,64]. Utilising this methodology, and in some cases incorporating minor modifications such as the use of galvanic displacement [65] or a high temperature, high pressure (HTHP) cell [19], coatings of Cu [63][64][65][66], Fe [19], Tb [22], Gd [22], Zn [67] and Zr [61] MOFs have been obtained. However attempts to synthesise Al MOFs and Zn Zeolitic Imidazolate Frameworks (ZIFs) as coatings via this methodology failed, with the material forming instead exclusively in solution [68]. This was attributed to the larger induction times for the synthesis of these materials, which favours homogenous solution growth over heterogeneous electrode surface growth [68].
ZIFs are a subset of MOFs that are so named due to the topologies they adopt being found in zeolites and the linkers used in their syntheses all being imidazole derivatives [69]. Obtaining ZIF coatings electrochemically is of interest as for many of the previously reported applications for MOF coatings stability to temperature, humidity and a variety of solvents is needed and ZIFs, unlike many of the MOFs made as electrochemical coatings to date, meet these requirements [69]. Combining these useful properties of ZIFs, with the previously mentioned advantages of the electrochemical methods for obtaining MOFs as coatings, is therefore an attractive proposition. Additionally there are over 150 reported ZIFs so far with a wide range of pore sizes and properties which further increase their attractiveness as coatings [70]. With a single exception, where a complicated indirect bipolar electrodeposition modification was used to obtain Zn ZIF-8 [71], no ZIFs have to date been synthesised as coatings electrochemically.
Recently there has been increasing interest in MOFs, particularly ZIFs, as materials for supercapacitor electrodes due to their intrinsically high surface areas, which should be ideal for the storage of charge in the electrical double layer. However due to the low electrical conductivity of most MOFs, in all cases reported hitherto conductive additives were used (such as graphene, graphene oxide, polyaniline or carbon black) or the MOF was pyrolysed to give porous carbons or metal oxides [26][27][28][29][30][31][32]. One report combined ZIF-67 with polyaniline to obtain exceptionally high areal capacitance values, however without the polymer additive the value was three orders of magnitude lower [27]. The much greater extent of integration between the MOF and the electrode surface provided by the anodic growth method, as compared to the dispersion of MOF crystals onto electrode surfaces, should mitigate the need for conductive additives and give rise to higher "pure" MOF based capacitance values. Here we present the successful electrochemical synthesis of five ZIF coatings via the anodic method encompassing two metals, Zn and Co, and four linkers, imidazole (IM), benzimidazole (bIM), 2methylimidazole (mIM) and 2-ethylimidazole (eIM). The identity of the coatings is confirmed by powder X-ray diffraction (PXRD), the morphology and quality of the ZIF coatings is analysed by SEM, and the apparent Brunauer-Emmett-Teller (BET) surface area is compared against literature values using N 2 (g) adsorption. We demonstrate how careful control of the synthesis conditions enables us to overcome the issue of the separation of the nucleation and growth processes and that it is possible to obtain topologies other than the most thermodynamically stable for a given metal linker combination, increasing the potential of the anodic method for generating a wide variety of thermally and chemically stable porous ZIF coatings. Finally we demonstrate the potential advantage of the anodic dissolution method for obtaining MOF coatings for electrochemical applications, utilising the Co and mIM example, by utilising them as supercapacitor electrodes.

ZIF coating synthesis
ZIF coatings were synthesised by immersing two metal foil electrodes (geometric area $16 cm 2 in the case of Zn and 2 cm 2 in the case of Co) held approximately 2 cm apart into a heated, deaerated electrolyte solution containing linker. A PGSTAT302 N potentiostat (Metrohm Autolab B.V., The Netherlands) was used to apply a fixed potential difference of 2.5 V between the two metal foil electrodes for a set time, generating a coating of the ZIF on the anode surface (see Scheme 1). Table 1 contains the range of conditions used to explore the synthesis of each of the five ZIF coatings. The coated anodes were rinsed three times with methanol post synthesis to remove unreacted linker and supporting electrolyte. Material is also formed in solution in many of these syntheses but, as we are only interested in the coating, this will not be considered further.

Characterisation of ZIF coatings
The morphology and quality of the ZIF coatings were characterised on the anode surface using a FEI Quanta 200 (Environmental) Scanning Electron Microscope (E)SEM. All images were obtained at 20 kV, under low vacuum with a water vapour pressure of 0.83 Torr and utilising both Secondary and Backscattered Electron Detectors.
The identities of the ZIF coatings were confirmed using a PANalytical X'Pert X-ray diffractometer. Powder X-ray diffraction (PXRD) patterns were obtained from ground samples of the removed coatings using Cu-Ka radiation at 40 kV and 30 mA, in the range 3-60 2u (with a step size of 0.017 2u and scan step time of 66 s) whilst spinning.
The N 2 adsorption isotherms of the ZIF coatings were obtained using a Micromeritics ASAP 2010 Volumetric Adsorption Analyser at 77 K. Ground samples of the removed coatings were first stirred in methanol overnight to remove excess linker and electrolyte from the pores and then degassed overnight at 100 C before analysis. The BET multipoint method [72] was used on the adsorption isotherm to calculate the apparent surface areas.
Electrochemical Impedance Spectroscopy (EIS) was performed over the frequency range 1 MHz À 200 mHz with a 10 mV RMS perturbation voltage. The experimental Nyquist plot was fitted with the standard Randles circuit.
Cyclic voltammetry was performed in the potential range 0 to 1.5 V, the step potential was 0.61 mV and the scan rate varied from 0.01 to 1 V s À1 .
Areal capacitance was calculated from the cyclic voltammograms using Eq. (1): Where E 1 is the cathodic potential, E 2 is the anodic potential, i(E) is the instantaneous current, the integral gives the total power obtained from the positive and negative sweeps in the cyclic voltammogram, n is the scan rate and A is the active electrode area of one electrode. Galvanostatic charge-discharge was performed with the charging and discharging current densities varied between 4.2 mA cm À2 to 40 mA cm À2 and À4.2 mA cm À2 to À40 mA cm À2 respectively, whilst the voltage the system was charged to varied from 1.5-0.25 V depending on the charging/discharging currents used. Areal Capacitance was calculated from the charge-discharge curves using Eqs. (2a) and (2b): Where I is the discharging current, dV/dt is the slope of the discharge curve, A is the active electrode area of both electrodes, t 1 is the time at the end of the discharge process, t 0 is the time at the beginning of the discharge process and V' is the voltage range of the discharge curve. Areal energy density (W h cm À2 ) was calculated using Eq. (3): Where C is the areal capacitance calculated from Eq. (2) (F cm À2 ), V' is the voltage range of the discharge curve (V) and the factor of 3600 converts W s cm À2 to W h cm À2 .
Areal power density (W cm À2 ) was calculated using Eq. (4): Where U is the areal energy density calculated using Eq. (3) (W h cm À2 ) and t is the discharge time (hours).

Results and Discussion
3.1. Characterisation of ZIF electrode coatings and their syntheses 3.1.1. ZIF À 4 PXRD shows that, of the many ZIF structures reported in the literature of Zn and IM, the coating material consists of ZIF-4 (Fig. 1A). ZIF-4 is unusual in that it is not truly a "zeolitic" framework as no known zeolite adopts the same structure. The cag topology that it adopts (named after the chemical formula, CaGa 2 O 4 , of a mineral of the same structure) is instead found in materials such as variscite [73]. That the coating is ZIF-4 is of particular interest as it is not the densest, most thermodynamically stable Zn and IM ZIF. The most stable Zn and IM ZIF is a nonporous material with the zni topology (named after zinc iodide which adopts the same structure) [74,75]. As proof of the relative instability of the cag topology compared to the zni topology, heating the dry powder samples at 125 C for 30 minutes converted the ZIF-4 coating into the non porous zni phase ( Fig. 2A). It also subsequently became apparent that over longer time scales at room temperature this conversion also occurred. Despite varying the linker concentration, the DMF:water ratio in the solvent, the supporting electrolyte concentration and reaction time, the only crystalline material obtained as a coating was ZIF-4. Under some conditions an amorphous phase was also formed. This demonstrates the importance of optimising the reaction conditions in obtaining MOF coatings. Whilst the main difference from previous literature syntheses of anodic MOF coatings was the $50% higher reaction temperature used in order to favour crystallisation [63] the importance of reaction time, effectively metal cation concentration, is also evident. When the Zn 2+ : IM ratio is too low (i.e. at short reaction times) an amorphous phase is seen to form, with ZIF-4 only being observed once the concentration of Zn 2+ in solution increases (Fig. 2B).
The coherency of the coating was assessed by SEM, utilising the Z contrast provided by backscattered electrons with brighter sections corresponding to areas with little or no coating. The abundance of brighter areas observed at low magnification  (Fig. 1B) demonstrates the poor quality of the coating in terms of coverage and extent of intergrowth. However some individual crystals were seen to adopt the orthorhombic dipyramidal crystal shape, at higher magnification, therefore matching the orthorhombic crystal system of variscite as seen in Fig. 1C.
The N 2 adsorption isotherm shows the expected type I behaviour and the low BET surface area of 75 m 2 g À1 is expected for a material with only 0.2 nm pore openings (Fig. 1D) [9]. The isotherm also clearly shows evidence of hysteresis, which to the authors knowledge has only previously been observed at higher pressures [76]. There appear to be no previous reports on the adsorption behaviour of ZIF-4 over smaller pressure ranges; but with such a small pore size it is not unreasonable that hysteresis may occur at lower pressures as well.

ZIF À 7
PXRD confirms the identity of the coating material as ZIF À 7 (Fig. 3A). ZIF-7 adopts the sodalite (SOD) structure and is denser and more thermodynamically stable than the other Zn and bIM ZIF, ZIF-11, which adopts the RHO structure topology [69]. It is not therefore surprising that ZIF-7, rather than ZIF-11, is obtained as the coating although this represents a contrast from the formation of the kinetically favoured ZIF-4, which was obtained when using IM. The breadth of the diffraction peaks, causing the overlap of the two most intense reflections at 2u = 7.2 and 7.7 , is indicative of poor crystallinity or small particle size, sub 1 mm.
The SEM images indicate that poor crystallinity is the cause as they show a highly irregular coating of crystals ! 1 mm with no clear morphology (Fig. 3B). A lack of a clear morphology is not uncommon in the MOF crystals obtained as anode coatings and, whilst not ideal, is not necessarily prohibitive to potential applications [19,61,66].
The probable reason for the lack of a clear morphology and poor crystallinity, versus other MOF crystals obtained as anode coatings [63], is the ca. 10 times greater concentration of linker utilised in this synthesis. A much greater concentration of linker was identified as one potential method for overcoming the problem that the induction time causes for obtaining ZIF anode coatings; a significantly higher linker concentration increasing the probability of nucleation. Whilst raising the concentration of benzimidazole was successful in obtaining ZIF-7 as a coating, the side effect of increasing the nucleation rate in any crystallisation is that smaller crystals, with no clear morphology, often form as is observed to be the case here. The ZIF-7 coating is mostly coherent with some brighter areas indicating that the coverage is not complete (Fig. 3C), an issue for some potential applications but not all.
The N 2 adsorption isotherm shows the expected type I behaviour and the calculated BET surface area of 358 m 2 g À1 matches well with literature values for material obtained by standard solvothermal synthesis [77] (Fig. 3D); demonstrating that despite the greater concentration of linker, the porosity of the coating is not compromised. As was found for ZIF-4 (Fig. 1D), the isotherm exhibits hysteresis but in this case this phenomenon has been documented previously at the same pressures as used here [78].

ZIF À 8
PXRD confirms that ZIF-8, which like ZIF-7 adopts the stable SOD structure, was synthesised as the coating when using Zn and mIM (Fig. 4A). This was expected as ZIF-8 is the only ZIF reported to date obtained using Zn and mIM [79]. The comparatively narrower peaks observed in the pattern, as compared to ZIF-7 (Fig. 3A), indicates a higher quality of crystal.
Initially SEM appears to indicate that the crystals constituting the coating are again small and lack clear morphology (Fig. 4B) but upon closer inspection the expected rhombic dodecahedral morphology of the ZIF-8 crystals can be observed (Fig. 4C). The coherency of the coating also appears to be better than in the case of ZIF-7, with no obvious brighter areas visible in the backscattered electron SEM image (Fig. 4B). Some potential applications for MOF coatings, such as acting as templates for electrodeposition [25], require coatings of a high coherency making this observation of interest for future work.
The N 2 adsorption isotherm shows the expected type I behaviour but the calculated BET surface area of 1730 m 2 g À1 is around 10% lower than some of the literature values quoted for material obtained by standard solvothermal synthesis [69] (Fig. 4D). This is attributed to the even higher concentration of linker, ca. 60 times greater, used in this synthesis as compared to the anodic MOF coating synthesis in the literature [63]. This high concentration was used in combination with an aqueous electrolyte which is unusual for ZIF synthesis. This was because previous work in the literature demonstrated that a vast excess of linker in aqueous solution gave a 80% yield of ZIF-8 after only 5 minutes reaction at room temperature [80]. This synthesis clearly overcame the induction time problem and it is believed that by combining this high linker concentration, with the additional application of the mild temperature utilised in all successful anodic MOF coating syntheses reported here, a ZIF-8 coating could be obtained and the problems outlined in the literature in forming coherent coatings would be overcome [68]. Whilst this combination was clearly successful in obtaining the coating, the use of such a vast excess of linker suggests that trapped linker may be responsible for the lower than expected BET surface area. It is worth noting that the aqueous, high linker concentration electrochemical ZIF-8 synthesis appears to produce a superior product to the solvothermal method it was based upon [80]. The calculated BET surface area is 60% higher, the expected crystal morphology for ZIF-8 is more clearly observed and the reflections in the PXRD are sharper [80].

ZIF À 14
PXRD shows that ZIF-14 was obtained as the coating when Zn and eIM were used (Fig. 5A). ZIF-14 adopts the analcime (ANA) zeolite structure, which is slightly denser and more thermodynamically stable than the only other reported Zn and eIM structure which adopts the RHO topology [81]. There is also evidence of a second phase which is identified as ZnO [82], the presence of which is logical considering the synthesis process is reliant upon the anodic oxidation of Zn. Sharp reflections indicate, as seen with ZIF-8, high crystallinity but this is not corroborated by SEM where it is not clear whether the expected icositetrahedral crystal morphology is observed or not [83] (Fig. 5B). Lower magnification SEM images show that the coating is of a generally poor quality (Fig. 5C). It is worth noting that, alongside the elevated reaction temperature, the concentration of linker is key to obtaining the desired material as the coating. When the Zn 2+ : eIM ratio was too high (i.e. the starting linker concentration was too low) ZnO became the predominant component of the coating, with ZIF-14 only potentially present as a very minor phase.
The porosity of ZIF-14 has been the source of recent controversy. The first paper to report on its synthesis found that it had only a very modest surface area ($ 29 m 2 g À1 ) [84] but very recently a surface area 20 times greater (649-701 m 2 g À1 ) has been reported [85]. It has been suggested that this large discrepancy may be due to difficulty in fully evacuating ZIF-14, requiring higher temperatures than expected, and the slow rate at which N 2 (g) penetrates the pore structure [85]. The adsorption isotherm we report here shows the type I behaviour indicative of true microporosity and the calculated BET surface area of 598 m 2 g À1 matches far better with the more recently reported higher values (649-701 m 2 g À1 ) and lends support to the theory of ineffective degassing and slow analysis gas penetration (Fig. 5D).

ZIF À 67
PXRD confirms that ZIF-67 was obtained as the coating when Co and mIM were used (Fig. 6A). To the best of our knowledge this is the first Co MOF to be obtained as a coating via the anodic method. As the Co analogue of Zn ZIF-8, ZIF-67 also adopts the thermodynamically stable SOD structure and displays a similarly high calculated BET surface area of 1521 m 2 /g (Fig. 6B) which is comparable to literature values [86].
The coherency of the film appeared to be quite high (Fig. 6C) and upon closer inspection the expected rhombic dodecahedral morphology was observed (Fig. 6D) Of the five coatings obtained ZIF-67 is potentially of most interest as a supercapacitor electrode material, as its bandgap of 1.98 eV is markedly lower than those of the other ZIFs obtained as coatings, such as ZIF-8 with a bandgap of 4.9 eV, which should give the former material a reasonable conductivity compared to the other ZIFs synthesised here [86]. EIS was performed, in order to assess the resistance of each of the five electrochemically obtained ZIF coatings.

Electrochemical Impedance Spectroscopy
The Nyquist plot (Fig. 7A) demonstrates clearly that whilst the solution resistance for all five ZIF coated electrodes is similar, as evidenced by the real axis values at the high frequency intercept for each material being within 1 V, that the charge transfer resistance of at least the four ZIF coated Zn electrodes varies significantly, as evidenced by the approximately 80 kV variation in the real axis values at the low frequency intercept. The Nyquist plot for ZIF-67 is only truly visible in the inset of Fig. 7A from which it is apparent that the shape of the plot is significantly different to that of the other ZIFs. The semi-circular shape of the plots for the four ZIF coated Zn electrodes shows that these materials acted as resistors over the frequency range used, with the width of the semicircle proportional to the charge transfer resistance. In contrast, the linear shape of the ZIF-67 plot makes clear that over the frequency range used ZIF-67 acted as a capacitor, with the 45 Fig. 6. PXRD (A) confirms the identity of the coating as ZIF-67 with the reflections in the experimental pattern (red) matching the predicted positions (black) [9]. N 2 adsoprtion (B) shows the expected type I isotherm behaviour. SEM images of (C) the coherency of the coating of ZIF-67 on the anode and (D) the lack of a clear morphology in the constituent crystals. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) slope indicative of diffusion being a limiting process. Diffusionlimited capacitance must be due to Faradaic processes which means that any capacitance observed from ZIF-67 coated Co electrodes would primarily be from pseudo-capacitance. Fitting the data for each of the five ZIF coated electrodes with the standard Randles circuit (with the addition of a Warburg impedance for ZIF-67 to account for the pseudo-capacitance) gave the solution and charge transfer resistances listed in Table 2. It should be noted that prior to this analysis the ZIF-4 coating had degraded to ZIF À zni. As stated the solution resistance values are very similar, within 1 V, whilst the charge transfer resistance values vary over approximately an 80 kV range. As predicted ZIF-67 exhibits a significantly lower charge transfer resistance than the other four ZIFs, between approximately 200 and 4300 times.
The Bode phase plots (Fig. 7C) show a similar trend to the Nyquist plots (Fig. 7A) with the four ZIF coated Zn electrodes displaying one plot shape and the ZIF-67 coated Co electrode showing another. The shape of the ZIF À 67 plot is characteristic of a pseudo capacitor, with the initial plateau in the phase at $ 45 followed by a steady decrease in phase towards 0 with increasing frequency. The frequency, which is proportional to discharge rate, at which the phase begins to decrease towards 0 is a measure of how quickly a supercapacitor electrode can operate and is dependent on the resistance of the material being used. The more resistive the material, the lower the frequency at which resistance begins to dominate over capacitance and the lower the maximum speed at which the supercapacitor can operate. The phase first begins to decrease at around 1 Hz and then decreases more rapidly after 1000 Hz, indicating the maximum speed at which the ZIF-67 based supercapacitor can operate. The shape of the plots for the four ZIF coated Zn electrodes show a different behaviour with the phase starting close to 0 at low frequency, before rising to plateau between 60 and 75 at medium frequencies and then finally decreasing again at higher frequencies. This behaviour can indicate corrosion of the underlying metal electrode [87] and could make the ZIF coated Zn electrodes unsuitable for supercapacitor electrodes, in addition to their high charge transfer resistances. Cyclic voltammetry (results not shown) is enough to confirm that Zn corrosion indeed occurs in this electrolyte system even at modest potentials, with the current from the oxidation making it impossible to make out any small potential capacitive effect from the ZIF coatings. For both these reasons only the ZIF-67 coated Co electrodes were taken forward for further capacitance analysis

Cyclic Voltammetry
The capacitive response of bare Co electrodes (with no ZIF-67 coating) was extensively analysed by cyclic voltammetry and as expected a low capacitance was observed. Cyclic voltammetry clearly demonstrates that the ZIF-67 coated Co electrodes possess a significantly higher capacitance than bare Co electrodes. This is attributed at least in part to the extensive double layer created by  the porous structure of ZIF-67 but primarily due to pseudocapacitance originating from redox processes occurring in the ZIF-67 coating (Fig. 8A). The area inside the curve also increases with scan rate as expected (Fig. 8B). The observed increase in the slope of the cyclic voltammagrams with scan rate reflects that, despite being markedly more conductive than the other ZIFs, ZIF-67 is still a poor conductor. At higher potentials, particularly above 1 V, there is an evident Faradaic process that contributes to the observed capacititve effect (Fig. 8C). The process appears to be essentially irreversible and its absence when bare Co electrodes are used indicates that it is related to the ZIF-67 coating and not due to species present in solution, for example dissolved oxygen. Using cyclic voltammetry data the calculated capacitance varies with scan rate as expected, with the maximum of 10.45 mF cm À2 at 0.01 V s À1 and 22% of that value retained at 1 V s À1 (Fig. 8D). This value is greater than that of many MOF/graphene composites [26] and seven times that of the highest reported value for "additive free" MOFs [27]. This is a crucial distinction as whilst there are many reports of much higher capacitances for various MOFs (including ZIF-67) they all, without exception, use some form of conductive additive to aid the performance of the material.

Galvanostatic charge À discharge
The capacitive response of ZIF-67 coated Co electrodes was also assessed using galvanostatic charge-discharge measurements in order to more accurately reflect the real working conditions of a supercapacitor device. Charge-discharge behaviour was assessed at a range of current densities over two orders of magnitude (Fig. 9A). The observed shape of the charge curves, with an initial sharp increase in potential with time followed by a slower increase up to E MAX , reflects the fact that the device is resistive with the effect becoming more pronounced at lower currents. This corroborates well with the cyclic voltammetry (Fig. 8B) that showed a marked increase in the slope of the cyclic voltammograms with scan rate. Only the three highest current densities used (4.17 mA cm À2 , 2.92 mA cm À2 and 1.67 mA cm À2 ) were sufficient to charge the device to 1.5 V, at the lower current densities of 0.42 mA cm À2 , 0.29 mA cm À2 , 0.17 mA cm À2 and 0.04 mA cm À2 the device was instead charged to 1.25 V, 0.75 V, 0.5 V and 0.25 V respectively. This is attributed to the still somewhat poor conductivity of ZIF-67 which for a given applied current limits the charging potential. Fitting the discharge curves (Fig. 9B) to a linear response gave fairly poor R 2 values between 0.92 and 0.98. This poor linearity is due to the pseudo-capacitive nature of the material. The capacitance was calculated from the slope of the discharge curves using Eq. (2a) to best compare with the literature. However, to more accurately assess the capacitance it was also calculated from the area under the discharge curve using Eq. (2b). The maximum capacitance calculated from Eq. (2a) was 2.61 mF cm À2 with a charging current density of 0.42 mA cm À2 , at E MAX equal to 1.0 V and a discharge time of 2.5 s. The corresponding capacitance calculated from cyclic voltammetry (E MAX = 1.0 V and Half cycle time = 2.5 s) was 4.09 mF cm À2 which compares reasonably well with a difference of $ 36%. The maximum areal energy and power densities obtained of 0.42 mW h cm À2 and 2.08 mW cm À2 (at 0.42 mA cm À2 and 2.9 mA cm À2 respectively) are comparable to those of recently reported MOF-graphene and MOF-polyaniline composites [26,27]. Using Eq. (2b) the maximum areal capacitance, energy density and power density decrease to 1.15 mF cm À2 , 0.18 mW h cm À2 and 1.60 mW cm À2 respectively. Fig. 8. Cyclic voltammetry shows the increased capacitive response due to the ZIF-67 coating at 1 V s À1 (A), the variation with scan rate (B), the change in charge storage mechanism from one assigned to double layer capacitance at lower potentials to a pseudo-capacitive process at higher potentials at 1 V s À1 (C) and the dependence of areal capacitance on scan rate (D).
Finally the durability of the ZIF-67 based capacitance was assessed by performing repeated charge-discharge measurements, with E MAX = 0.5 V and i = 0.2 mA cm À2 . Capacitance as a percentage of maximum capacitance (C/C MAX ) was still greater than 67% after over 600 cycles (Fig. 9C). Whilst this cycling performance is worse by an order of magnitude when compared to some MOF composite materials [26,27] it is also superior by an order of magnitude to others [33]. In order to assess the cause of the decrease in capacitance with repeated cycling the used ZIF-67 coated Co electrodes were characterised again by PXRD (Fig. 10A) and SEM (Fig. 10B).
The PXRD showed that the crystalline ZIF-67 coating had mostly degraded into an amorphous material but that there was still a small amount of crystalline material present. The SEM showed an increased prevalence of brighter areas (indicative of exposed metal) compared to that seen prior to electrochemical studies (Fig. 6C); demonstrating that not only was the coating degraded but the amount of material present on the electrode had   [9]. SEM images of (B) the coherency of the coating of ZIF-67 on the electrode after electrochemical testing. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) also decreased. Conversely it did also show that there was still a significant amount of material present, explaining how the capacitive behaviour of ZIF-67 coated Co electrodes was still retained after such extensive electrochemical testing.

Conclusions
Five ZIFs have been obtained as electrode coatings via the anodic growth method, four of which had not previously been obtained as coatings. These new coatings were obtained by tuning the parameters of the anodic growth method in order to favour coating growth over growth in solution. Surveying the reaction conditions in each case demonstrates that generally longer reaction times (a proxy for higher metal ion concentration), higher linker concentration and higher reaction temperature all favour coating growth over solution growth. Additionally these reaction conditions have a strong influence on the crystallinity of the coatings. These results also indicate that, for a given metal and linker combination, it is possible to obtain topologies other than the most thermodynamically stable, potentially expanding the number of structures that could be obtained as coatings via this method. Finally the potential of this additive free ZIF-67 coated Co as a supercapacitor electrode has been demonstrated and shown to outperform all reported supercapacitor materials based purely on MOFs and many MOF/graphene composite materials.