Preparation, characterization and application in cobalt ion adsorption using nanoparticle films of hybrid copper–nickel hexacyanoferrate

Different mole ratios (nCu : nNi = x : y) of hybrid copper–nickel metal hexacyanoferrates (CuxNiyHCFs) were prepared to explore their morphologies, structure, electrochemical properties and the feasibility of electrochemical adsorption of cobalt ions. Cyclic voltammetry (CV), field emission scanning electron microscopy (FE-SEM), Fourier transform infrared spectroscopy (FTIR) and X-ray diffraction (XRD) indicated that the x : y ratio of CuxNiyHCF nanoparticles can be easily controlled as designed using a wet chemical coprecipitation method. The crystallite size and formal potential of CuxNiyHCF films showed an insignificant change when 0 ≤ x : y < 0.3. Given the shape of the CV curves, this might be due to Cu2+ ions being inserted into the NiHCF framework as countercations to maintain the electrical neutrality of the structure. On the other hand, crystallite size depended linearly on the x : y ratio when x : y > 0.3. This is because Cu tended to replace Ni sites in the lattice structure at higher molar ratios of x : y. CuxNiyHCF films inherited good electrochemical reversibility from the CuHCFs, in view of the cyclic voltammograms; in particular, Cu1Ni2HCF exhibited long-term cycling stability and high surface coverage. The adsorption of Co2+ fitted the Langmuir isotherm model well, and the kinetic data can be well described by a pseudo-second order model, which may imply that Co2+ adsorption is controlled by chemical adsorption. The diffusion process was dominated by both intraparticle diffusion and surface diffusion.

Recently, MHCFs have received extensive attention due to their unique electrochemical properties as modied electrode materials, such as PB (FeHCF), 15 copper hexacyanoferrate (CuHCF), 16 cobalt hexacyanoferrate (CoHCF), 17 nickel hexacyanoferrate (NiHCF), 18 and zinc hexacyanoferrate (ZnHCF). 19 Each MHCF has its own unique electrochemical and spectral properties; however, materials with diversied traits are preferred. Research into hybrid-metal hexacyanoferrates (h-MHCFs) is becoming a trend due to their multiple characteristics and properties. Previous researchers reported that the prepared h-MHCFs existed as a new phase rather than a simple mixture of corresponding single metal hexacyanoferrates. 20,21 Ghasemi et al. 22 found that NiCoHCF has a higher capacitance than NiHCF or CoHCF; Li et al. 23 revealed that CuNiHCF has a higher capacity retention than CuHCF as a cathode for sodium-ion batteries; it was also reported that the mixed MHCF showed an enhanced catalytic efficiency to H 2 O 2 as compared to that of FeHCF. 24 The surface coverage of CoNiHCF was proved to be higher than those of CoHCF and NiHCF. 25 In an aqueous electrolyte containing alkali salts, MHCFs exhibit reversible redox reactions (such as Fe III /Fe II ) accompanied by reversible intercalation and deintercalation of hydrated alkali cations, low cost, and low toxicity. 26 Within the family of MHCFs, an NiHCF modied electrode showed an excellent cation exchange capacity in redox reactions due to its open structural framework, [27][28][29] with uniform, compact and stable lms. The network structures with macropores 26 of CuHCF lms were characterized by nearly ideal current-potential curves. 30 We also found that a CuNiHCF lm modied electrode showed higher electrochemical stability in comparison to CuHCF and NiHCF in phosphate buffer solutions (PBS). 25 To the best of our knowledge, few papers on different proportions of Cu x Ni y HCF nanoparticle lms as modied electrode materials have been published, which inspired us to synthesize Cu x Ni y -HCF lms that may inherit the virtues of CuHCF and NiHCF.
Generally, MHCF lm modied electrodes are prepared by electrodeposition, and are difficult to produce on a large scale, 25 and fall off easily. Moreover, it is hard to control the proportion of metal phases (M 1 : M 2 ) in the products (h-MHCFs), which is why few researchers have focused on the characteristic differences among varied proportions of h-MHCFs, such as morphology, structure, or electrochemical properties. Chen et al. 31 prepared exact ratios of CuHCF nanoparticle ink by a wet chemical coprecipitation method, and printed them onto the electrode surface to remove cesium electrochemically; the cesium removal was comparable in magnitude to that by an electrodeposited lm. Wang et al. 25 prepared h-MHCF lms by the same method; the surface coverage of redox-active sites of the CuCoHCF lms was 35 nmol cm À2 , which was much higher than the 9.02 nmol cm À2 of electrodeposited lms; 32 and the Co/Ni atomic ratio in CoNiHCF compounds had been proved to be controlled by the reactant ratio in the reaction mixture.
Herein, we prepared a series of Cu x Ni y HCF nanoparticle lms with different molar ratios (n Cu : n Ni ¼ x : y) by using a wet chemical coprecipitation method, and explored the characteristics of each ratio and the variations among them by multiple characterization methods. Recently, the wide application of lithium-ion batteries (LIBs) has resulted in a rapid increase in the amount of spent LIBs, containing valuable metals such as cobalt, lithium, nickel, and copper. Therefore, we also attempt to recover cobalt ions (Co 2+ ) from solutions by using Cu x Ni y HCF nanoparticle lms. The feasibility of electrochemical adsorption of cobalt ions onto hybrid copper-nickel hexacyanoferrate lms is discussed, as well as the adsorption isotherms and kinetics properties.

Materials and apparatus
All reagents, such as Cu(NO 3 6 ]$3H 2 O and KNO 3 , used in the operations were of analytical grade or better without further purication. Ultrapure water (18 MU cm) was used throughout all the operations. All the experiments were carried out at room temperature.
Ion concentrations were determined by inductively coupled plasma mass spectrometry (ICP-MS, Thermo Fisher Scientic iCAP Q, America). The surface morphologies of Cu x Ni y HCF lms were observed by a eld-emission scanning electron microscope (FE-SEM, Hitachi SU-8010, Japan). Their structures and compositions were characterized by X-ray powder diffractometry (XRD, Rigaku SmartLab, Japan) and Fourier transform infrared spectroscopy (FTIR, Bruker Vertex 70, Germany). Cyclic voltammetry of Cu x Ni y HCF electrode lms was performed by using an electrochemical analyzer (Chen Hua CHI660e, China).

Preparation of modied electrodes
As described in our previous studies, 25 h-MHCF nanoparticle inks with different metal molar ratios were prepared by wet chemical coprecipitation. Firstly, mixed solutions with different molar ratios (x : y ¼ 1 : 4, 1 : 2, 1 : 1, 2 : 1, and 4 : 1) were obtained from Cu(NO 3 ) 2 $xH 2 O and Ni(NO 3 ) 2 $6H 2 O, to which a solution of K 3 [Fe(CN) 6 ] was added; the reactant mole ratio of total metal nitrates to K 3 [Fe(CN) 6 ] was kept at 3 : 2. Secondly, one group of parallel samples were dried in an oven at 60 C to obtain powder samples for subsequent chemical analysis and characterizations. The remaining samples were surface-treated by K 4 [Fe(CN) 6 ]$3H 2 O to dissolve well in water. Thirdly, the 5% (ratio of mass to volume) ink-like solutions of nanoparticles were prepared and shaken for one week (25 C, 150 rpm). Indium tin oxide (ITO) was washed with acetone as a support substance, and Cu x Ni y HCF nanoparticle inks were coated on it using a KW-4A spin-coater instrument (2000 rpm for 10 s followed by 2500 rpm for another 10 s). Finally, the modied electrodes were dried in an oven at 120 C for 2 h.

Characterization
In order to investigate the composition ratios, the prepared Cu x Ni y HCF powders were decomposed in aqua regia by using microwave digestion. Aer microwave digestion, dilution and ltration, the elements were determined by ICP-MS. The obtained x : y ratio is linearly tted to the theoretical values in the synthesis process. The tness (R 2 ) is given by eqn (1): y*, y and y 0 stand for the theoretical ratio values, the actual ratio values and the mean value of y, respectively. For convenience, the ratio of x : y is converted to the percentage of Cu (W Cu ) in the nanoparticle lms, with the following expression: where n Cu and n Ni stand for the number of moles of Cu and Ni in Cu x Ni y HCF nanoparticle lms, respectively. A three-electrode system was used with a saturated calomel electrode (SCE, Hg/Hg 2 Cl 2 /KCl) as the reference electrode, a Pt wire as the counter electrode and the ITO plate loaded with nanoparticle lms as the working electrode. The cyclic voltammograms (CVs) of Cu x Ni y HCF lms in the different cation solutions were obtained, using a CHI660e electrochemical analyzer. The electrochemical properties of Cu x Ni y HCF nanoparticle lms were characterized by CV, formal potential (E f ), accumulated quantity of electric charge (Q) and surface coverage of redox-active sites (G). E f is calculated as half of the sum of potentials of the anodic (E pa ) and cathodic (E pc ) current peak. Q is the charge obtained by integrating the cathodic peak aer subtraction of the background. The calculation equation of G is as follows: Q is obtained by integrating the cathodic peak from the 2nd cycle under the background correction at a scan rate of 0.05 V s À1 , and other symbols have their usual meanings. The electrodes used here are modied by 3% ink. The stability of the Cu x Ni y HCF lms was investigated by long-term potential cycling (500 cycles), expressed by the rate of decrease (R d ) of the integrated Faraday charge Q with the increase in the number of cycles.
Q 1 and Q n stand for the integrated charge recorded from the 1st and nth scan cycles, respectively.

Electrochemical adsorption of Co 2+
2.4.1 Removal efficiency. Co 2+ removal was performed in a three-electrode cell containing 40 mL solutions stirred at 500 rpm by a magnetic stirrer (DL Instruments Ltd., MS-H280-Pro, China). The lms were rstly pretreated by an applied potential of +1.3 V (vs. SCE) for 30 min in 1 mg L À1 KNO 3 solution to discharge the residual K + during the surface treatment. The relevant reaction is as follows: Aerwards, the lms were washed with ultrapure water and transferred to 1 mg L À1 Co 2+ solution to adsorb Co 2+ by applying a reduction potential of À0.4 V (vs. SCE).
Contrast tests were conducted at the same time without applying potentials.
The results of electrochemical adsorption were given as removal efficiency (R), which was calculated according to eqn (7): where C 0 (mg L À1 ) and C t (mg L À1 ) stand for the Co 2+ concentration at the initial stage and at time t, respectively.

Isotherm and kinetic studies.
Batch adsorption experiments were conducted in Co 2+ solutions at various initial concentrations (0.5 mg L À1 , 1 mg L À1 , 2 mg L À1 , 5 mg L À1 , and 10 mg L À1 ) to investigate the adsorption mechanism. The batch adsorption data were tted with the Langmuir model (8), and the Freundlich isotherm model (9).
Here, C e (mg L À1 ) is the equilibrium concentration; Q e (mg cm À2 ) and Q m (mg cm À2 ) are the saturation adsorption capacity and the maximum adsorption capacity at the pseudoequilibrium, respectively. K L , K F and n (dimensionless) are the adsorption isotherm constants for the Langmuir isotherm and the Freundlich isotherm.
The intraparticle diffusion model, pseudo-rst-order and pseudo-second-order adsorption kinetic models were used to analyze the dynamic adsorption process, as represented by eqn (10)-(12), respectively.
where, Q e (mg cm À2 ) and Q t (mg cm À2 ) are the adsorption amounts of Co 2+ at the adsorption equilibrium and time t, respectively. k p,i (i represents the different stages), k 1 and k 2 are the rate constants. C i is the intercept at different stages.

Results and discussion
3.1 Characterizations of Cu x Ni y HCF nanoparticle lms 3.1.1 Component analysis. The actual contents of Cu and Ni of the prepared Cu x Ni y HCFs were determined by ICP-MS, as listed in Table 1.
The tness of the theoretical and actual values is 0.9840, showing that the two sets of values are very close. In other words, unlike electrodeposition, 33,34 the x : y ratios in the Cu x -Ni y HCF nanoparticle lms can be controlled by the preparation method. This is because hexacyanoferrate (II) ions are generated rstly in the vicinity of the electrode surface and then react primarily with one kind of metal ion in the electrolyte 27 when prepared by electrodeposition. For coprecipitation, metal ions and hexacyanoferrate (III) ions have been mixed and reacted evenly during the preparation stage of the ink. The results mean that Cu x Ni y HCF lms with appropriate metal ratios can be synthesized as required.
3.1.2 Morphology. The surface structures of the nanoparticle lms are shown in Fig. 1. It is found that NiHCF lm exhibits a compact structure consisting of homogeneous nanoparticles, with the formation of evident cracks and the agglomeration of NiHCF nanoparticles. While abundant pores obviously grow on CuHCF lms, which might facilitate the mass transfer and diffusion rate of ions within lms. 25 The nanoparticle size of CuHCF lms is obviously larger than that of NiHCF lms, which is consistent with previous research results. 26,35 The nanoparticle size and porosity of Cu x Ni y HCF lms varied from pure NiHCF lms to pure CuHCF lms. The In order to further explore the effect of x : y ratio on the properties of Cu x Ni y HCFs, Cu x Ni y HCF powders were characterized by XRD and FTIR. As shown in Fig. 2 The crystallite size (D) can be deduced from the diffraction peak according to Sherrer's formula: where K is the shape factor, l is the X-ray wavelength (0.154056 nm), b is the full width at half maximum of the diffraction peak, and q is the Bragg diffraction angle. As presented in Fig. 3, the crystallite sizes of NiHCF and CuHCF are estimated to be about 15 nm and 30 nm, respectively; the sizes of Cu x Ni y HCFs at different ratios of Cu/Ni are in between these values. The value of D rarely changes with the increase of copper content when W Cu < 30%; then it increases linearly with the W Cu values. We suppose that the lattice of NiHCF (as shown in Fig. 4a) is practically unchanged when W Cu < 33%. Cu 2+ inset in NiHCF lattice instead of K + as countercations to maintain the electrical neutrality of the structure in a situation such as that exhibited by Fig. 4b. The crystal structure begins to change as the Cu content increases to around 33%, while the Cu replace part of the Ni to occupy the lattice positions illustrated by Fig. 4c. Fig. 5 illustrates the FTIR spectra of Cu x Ni y HCF powders. In particular, there is a small shoulder at 2165.21 cm À1 of NiHCF, while the other samples show only one major peak in the vicinity of 2096 cm À1 . The other bands of Cu x Ni y HCFs are very    close to each other, owing to the similar absorption peaks of NiHCF and CuHCF. The broad band at around 3330 cm À1 refers to the O-H stretching vibrations. 38 The vibrational band near 2096 cm À1 corresponds to the stretching vibration absorption band of the -C^N group. 39,40 The bands in the region of 1400-1680 cm À1 are related to the asymmetric and symmetric stretching vibrations of COO À groups. 41,42 The stretching vibration absorption band of the free -C^N is at about 2060 cm À1 in aqueous solution; 43 the blue shi of the absorption band observed together with the two smaller peaks at about 595 cm À1 and 491 cm À1 are attributed to the formation of Fe-CN-M. 39, 44 3.1.4 Electrochemical properties. CV plots with distinct redox couple peaks of Cu x Ni y HCF lms in 0.5 M K + solution are shown in Fig. 6a. One set of symmetrical redox couples of CuHCF lms is around 0.723 V, while the voltammetric patterns of the NiHCF lms have two pairs of redox peaks at a lower potential range of 0.45-0.65 V, assigned to the presence of two phases within the NiHCF lms. 45,46 The intensity and location of the response curve peaks of Cu x Ni y HCFs depend on the value of x : y, which is consistent with the change in structure from XRD and FTIR results. As indicated in a previous study, 47 Cu x Ni y HCF is a new phase rather than a simple mixture of CuHCF and NiHCF. The typical CV response of each Cu x Ni y HCF is attributed to its respective unique structure. In particular, two sets of redox peaks of Cu 1 Ni 4 HCF and Cu 1 Ni 2 HCF are similar to NiHCF, and, the redox peaks at 0.45 V are severely weakened, which might be because of their special structures, as illustrated in Fig. 4b. The E f values of Cu x Ni y HCF lms were plotted as a function of W Cu , as approximated in Fig. 3. At a low molar fractions of Cu, the E f value hardly changes with the addition of  Cu, then depends linearly on the content of copper when W Cu rises to about 33%. These observations conrm our previous hypothesis that Cu exists in two forms: as countercations substituted for K + , under which condition the E f is close to that of NiHCF, 48 or occupying lattice positions instead of Ni. This is similar to the electrodeposited lms studied by Kulesza et al. 27 It can be found from Fig. 6a that the redox peaks at more negative potential disappear when W Cu > 30%, which also demonstrates that Cu started to occupy positions of the lattice.
The ratio value of cathodic current to anodic peak (i pc /i pa ) approaches one, implying the reversibility of the electrochemical redox reaction. The existence of Cu greatly improved the property of NiHCF lms in this respect. The long-term CV of Cu x Ni y HCF lms was carried out in 0.5 M K + at a scan rate of 0.05 V s À1 to study the stability of Cu x Ni y HCFs. R d values were plotted against the cycle numbers, as depicted in Fig. 6b. NiHCF lms show comparatively high stability compared with CuHCF lms, which is in accord with the materials synthesized previously from chlorides. 25 The stability of Cu x Ni y HCF lms increases with a decrease in x : y values in the order: Cu 4 Ni 1 HCF < Cu 2 Ni 1 HCF < Cu 1 Ni 1 HCF < Cu 1 Ni 2 HCF < Cu 1 Ni 4 HCF. It is noteworthy that the cycling stabilities of Cu 1 Ni 4 HCF, Cu 1 Ni 2 -HCF and Cu 1 Ni 1 HCF lms are notably better than that of NiHCF lms, indicating that the stability of Cu x Ni y HCF lms will be enhanced on the basis of NiHCF lms when the copper content of copper is lower than or even equal to the nickel content. In contrast, the cycling stability will get worse when the proportion of copper is higher.
The results of the surface coverage G calculated from the integrated charge in 0.5 M K + solution are shown in Fig. 6c. The G value of NiHCF lms is apparently higher than that of CuHCF lms. The G values of Cu 1 Ni 4 HCF lms and Cu 1 Ni 2 HCF lms are 1.2-1.4 times as much as the G values of other Cu x Ni y HCFs.

Electrochemical behaviors in cationic solutions.
In addition to K + solution, the CV responses of Cu x Ni y HCF lms in Li + , Co 2+ , Ni 2+ and Ga 2+ solutions are also discussed. The electric quantity Q of the reduction current of Cu x Ni y HCF lms in the Li + , Co 2+ , Ni 2+ and Ga 2+ solutions were integrated (as shown in Fig. 6c). The Cu 1 Ni 2 HCF lms show the maximum value of the integrated charge in hybrid lms, which may be bound up with the G.
The ne CV responses imply the possibility of electrochemical adsorption of metal cations. In the electrochemical adsorption experiment, Co 2+ were effectively removed by Cu x -Ni y HCFs. That is why Co 2+ was chosen to carry out the experiment on the electrochemical adsorption behavior of Cu x Ni y HCF lms. The Cu 1 Ni 2 HCF lms were used for comparison with CuHCF lms and NiHCF lms since they have excellent electrochemical performances. In order to study information about the mechanism, CV and electrochemical impedance spectroscopy (EIS) experiments were conducted. The CVs of Cu x Ni y HCF modied electrodes at various sweep rates from 0.005 to 0.15 V s À1 in 0.5 M Co 2+ solutions are shown in Fig. 7a-c. The logarithmic cathodic peak current increases linearly with an increase in the logarithmic scan rate (shown as insets). The slopes of the Cu x Ni y HCF lms are 0.75-0.88 (listed in Table S1 †), indicating that the reaction at the thin lms is controlled by an ion diffusion process and a surface electron transfer process, with a tendency towards the latter. 49 The DE gradually increases with the variation in the scan rate, indicating a limitation arising from the charge-transfer kinetics 50 which weakened the reversibility. 51 The Nyquist plots (Fig. 7d) are composed of a distorted semicircle in the high-frequency region assigned to the chargetransfer reaction of the electrode and a linear part at the lowfrequency end due to a diffusion-controlled process. 29,52 The CuHCF lm is characterized by the lowest diameter of the semicircle, standing for the quickest electron transfer rate. 53 The charge transfer resistance decreases in the order: Cu 1 Ni 2 -HCF > NiHCF > CuHCF.
3.2.2 Removal efficiency of Co 2+ . Fig. 8a compares the results of Co 2+ adsorption in electrochemical and conventional systems. The removal efficiency of electrochemical adsorption is 3-4 times higher than that of conventional adsorption, verifying that an electrochemical removal process had played an important role. And, the process of Co 2+ adsorption involved both surface adsorption and cation intercalation/ deintercalation.
The electrochemical adsorption capacity of CuHCF lms is much higher than that of NiHCF lms, verifying the results of integrated charge and EIS. The high adsorption capacity can be interpreted by the large lattice size of CuHCF. This is consistent with the fact that the sorption capacity of CuHCF is larger than that of NiHCF, as mentioned by Thierry et al. 54 It is remarkable that the equilibrium removal efficiency of Cu 1 Ni 2 HCF lms is higher than that of NiHCF lms, and slightly lower than that of CuHCF lms, which may due to the well-ordered porous structures of CuHCF lms. The redox peaks of CuHCF lm in Co 2+ solutions are sharper than those of Cu 1 Ni 2 HCF lm, indicating a faster dynamic process. The cycling stability of Cu 1 Ni 2 HCF lm is obviously better than that of CuHCF lm, when the equilibrium removal efficiency of Cu 1 Ni 2 HCF lm is close to that of CuHCF lm. That is why we chose Cu 1 Ni 2 HCF for the next step.
The crystal structure and surface morphology of Cu 1 Ni 2 HCF lm before and aer removing Co 2+ were investigated by XRD and SEM; the results are depicted in Fig. 8b and c. Compared with the peaks of (In 1.88 Sn 0.12 )O 3 (PDF # 89-4598) coated on substrates, the new peaks generated at around 17.95 and 25.35 and the sharper peaks at 35.48 and 37.65 are caused by Cu 1 Ni 2 HCF lm. There is no signicant change in the results of SEM and XRD aer Co 2+ -loading, indicating that the structure of the nanocrystals had not been changed by the adsorption process. The concentrations of Cu 2+ and Ni 2+ before and aer pretreating the lms were determined by ICP-MS (Fig. S2 †). The results show that there is no signicant release of Cu 2+ or Ni 2+ from the lms, indicating that the prepared nanoparticle lms were stable as well. These phenomena are consistent with the expected cation insertion process, during which the essential feature of crystal structure is preserved. This may also due to the dilute solution and trace amount of Co 2+ adsorbed on Cu 1 Ni 2 HCF. The same conclusions can be drawn from NiHCF and CuHCF lms (as shown in Fig. S1 †).
3.2.3 Isotherm and kinetic studies. The effect of initial concentrations on the adsorption capacity of Cu 1 Ni 2 HCF lms for Co 2+ was studied and the results are shown in Fig. 9a. It has been found that the plot reected a two-stage process, a steep ascent portion followed by a plateau extending to the equilibrium. The main reason is that the concentration gradient between the solution and the interface of the adsorbent increased with a rise in initial concentration, which enhanced the driving force for the diffusion of Co 2+ from solution to the lms. 55,56 The Langmuir isotherm model and the Freundlich isotherm model were used to t the experimental data. The Langmuir isotherm model assumes that only one monolayer of adsorbate molecule can be adsorbed on homogeneous sites of the adsorbent and the adsorbates do not react with each other. The Freundlich isotherm model is based on multilayer adsorption on a heterogeneous surface. The calculation results of the Q e , constants and correlation coefficient (R 2 ) are listed in Table 2. The linear correlation coefficient (R 2 ) from the Langmuir isotherm model is 0.9989, which is higher than that calculated from the Freundlich isotherm model, and the value of Q m from the Langmuir isotherm model is close to the practical Q e obtained from experiment. It can be concluded that the Langmuir isotherm model can describe the adsorption process better than the Freundlich isotherm model, which is in agreement with the redox reaction. The results from the data in 1 mg L À1 Co 2+ solutions tted to the intraparticle diffusion, pseudo-rst-order and pseudo-second-order kinetic models are given in Table 2. The sorption process involves bulk diffusion, lm diffusion and intraparticle diffusion (or pore diffusion). 35 Fig. 8b shows that two linear regions exist in the intraparticle diffusion plot. The adsorption system was stirred at a speed of 500 rpm. Thus, the two curves may indicate that the adsorption of Co 2+ onto Cu 1 -Ni 2 HCF lm occurred via two phases: 57 lm diffusion in the rst linear part followed by intraparticle diffusion during the second linear portion. The linear portions of the two stages do not pass through the origin, and, k p,1 is much greater than k p,2 , which suggests that both lm and intraparticle diffusion processes are the probable rate-controlling steps. 58 The constant C 1 increased with an increase in ion concentration, indicating the growth of the thickness of the boundary layer. The probability of external mass transfer was reduced, and the probability of internal mass transfer increased conversely. 59 Both the pseudo-rst-order kinetic model and the pseudosecond-order kinetic model show a good t to the experimental data with well-matching correlation coefficients at a low initial concentration. At a higher initial concentration, the correlation coefficient (R 2 ) obtained from the pseudo-second-order kinetic model is higher than that of the pseudo-rst-order kinetic model, Fig. 8 (a) Removal efficiency of Co 2+ onto Cu x Ni y HCF films vs. contact time (Co 2+ concentration: 1 mg L À1 , electrode area: 15 mm Â 25 mm); (b) XRD patterns and (c) SEM images of Cu 1 Ni 2 HCF film before and after adsorption (Cu 1 Ni 2 HCF-Co 2+ ). Fig. 9 (a) The effect of initial concentration on the adsorption capacity. The inset is the fitted result with the Langmuir isotherm model; (b) adsorption kinetic data in 1 mg L À1 and 10 1 mg L À1 Co 2+ solutions fitted with the intraparticle diffusion model.
suggesting that the pseudo-second-order model is more suitable for describing the adsorption process. This may predict that electrochemical adsorption involved a complicated process, and chemical adsorption might be the rate-limiting step. 60

Conclusions
In summary, the proportion of metals in Cu x Ni y HCF nanoparticle lms can be controlled by the reactant ratio during the preparation process. With an increase in the Cu ratio, the crystallite size and E f of Cu x Ni y HCF lms varied from NiHCF lms to CuHCF lms. The results of SEM and XRD, together with the shape of the CV curves supported Cu 2+ being inserted into the NiHCF lattice as countercations to maintain the electrical neutrality of the structure when the proportion of copper is less than 33%; at higher molar fractions of Cu, Ni was partially replaced by Cu. Cu 1 Ni 2 HCF showed superior electrochemical properties of stability and current responses. The electrochemical adsorption behavior of Co 2+ followed the Langmuir isotherm model and the pseudo-second order model. Our study reveals the potential of Cu x Ni y HCF nanoparticle lms as a promising technology for the recovery of Co 2+ from spent LIBs.

Conflicts of interest
There are no conicts to declare.