**2. Results and Discussion**

The electrodeposition of Ni-Cu was realized under potentiostatic conditions at E = −0.8 V for 130 minutes on niobium discs; linear sweep voltammetry was firstly performed in the electrodeposition solution at 5 mV s<sup>−</sup>1, starting from open circuit potential (OCP) up to <sup>−</sup>0.8 V, to favor a slow deposition of a thin film onto niobium surface.

Figure 1 displays the scanning electron microscopy (SEM) images and the Auger mappings of the Ni-Cu deposit onto the niobium surface; as can be seen, a homogeneous distribution of both the elements, with some copper agglomerations, characterized the surface. The chemical composition of the film calculated by energy dispersive X-ray (EDX) analysis indicated an average molar fraction of 30% of copper and 70% of nickel.

**Figure 1.** SEM image (**a**) and Auger mappings (**b**) of Ni-Cu deposit onto niobium substrates: Cu in green and Ni in red. The separated Auger mappings of Cu and Ni are reported in (**c**) and (**d**), respectively.

The SEM images of the samples obtained using different corrosion conditions (Figure 2), show the presence of porous structures in each sample. Table 1 reports the conditions adopted for the corrosion of each sample, denoted as Stcorr-trelax where *tcorr* and *trelax* are the corrosion and relaxation times used in the pulsed corrosion steps, respectively. Table 1 also reports the ratio φ = *tcorr*/*trelax* together with the range of pore diameters and molar fractions of Cu and Ni (oxygen was found as third element up to unit).

As can be seen from SEM images, different morphologic features can be recognized depending on the experimental conditions; samples prepared with the highest φ as S1–5 (Figure 2a) and S0.1–0.5 (Figure 2b) were featured by a tubular structure, while the samples prepared using the lowest φ (S1–50 in Figure 2d, S0.1–5 in Figure 2e and S0.01–0.5 in Figure 2f) presented a cauliflower-like structure with

multi-dimensional pores. Only the sample S0.01–0.05 (Figure 2c) is characterized by a poor pore density with very small diameters.

**Figure 2.** SEM images of samples submitted to different conditions of anodic dissolution: (**A**) S1–5; (**B**) S0.1–0.5; (**C**) S0.01–0.05; (**D**) S1–50; (**E**) S0.1–5; (**F**) S0.01–0.5. Insets report the magnifications of the SEM images of the samples.

Data in Table 1 and Figure 2, indicated a combined effect of *tcorr* and *trelax* on the pore size and morphology of the resulting samples: in fact, *trelax* being the same, a decrease in *tcorr* leads to lower pores diameter (See Figure 2 for sample S1–5 compared with S0.1–5, and S0.1–0.5 compared with S0.01–0.5) and higher residual copper amount.

**Table 1.** Range of diameters d (nm) and molar fractions of Cu and Ni (XCu and XNi ) of porous nickel electrodes prepared using different *tcorr* and *trelax* during the anodic pulsed voltage dissolution experiments.


Conversely, an increase in *trelax* leads to a decrease of the pores size, *tcorr* being the same, as for samples S1–5 and S1–50 (*tcorr* of 1 s), or for samples S0.1–0.5 and S0.1–5 (*tcorr* of 0.1 s) (See Figure 2). Moreover, in the case of *tcorr* of 0.01 s, both samples present the highest copper residual amount.

As it was firstly proposed by Erlebacher et al. [39], these results can be explained considering that the mechanism involved in the formation of porous structures, by dealloying of binary systems under potentiostatic conditions, resulted from two concurrent processes occurring at the alloy/electrolyte interface: the chemical dissolution of the most reactive metal atoms and the atom rearrangement of the most inert atoms, which expose the underlying more reactive metal atoms to further dissolution.

Using pulsed potential, it is possible to influence the interplay between atoms rearrangement and chemical dissolution [38]. Thus, the increase in the ratio φ favors the rearrangement of nickel with respect the copper dissolution, which in turn leads to more dendritic samples with high content of residual copper, as observed comparing samples S1–5 with S0.1–5 or S0.1–0.5 with S0.01–0.5. Moreover, very short relaxation times, as in the case of sample S0.01–0.05, limit the nickel atoms rearrangement thus hindering the exposure of underlying copper and its progressive dissolution: few and very small pores with very high residual copper content was obtained under this condition.

To gain insight into the structural features of the deposited films, X-ray diffraction patterns were acquired, as shown in Figure 3. The pattern of the initial deposit (Figure 3a) shows the occurrence of sharp peaks due to the niobium substrate and additional broader reflections which can be ascribed to the occurrence of nanocrystalline nickel and copper phases, based on comparison with reference PDF cards (Figure 3c). After the electrochemical corrosion, changes in the X-ray pattern can be observed in Figure 3b (sample S1–5): in particular, the relative intensity of the peaks due to Ni and Cu is decreased, suggesting the occurrence of a lower contribution from nanocrystalline phases as compared to the original film. Additional peaks are also observed which can be tentatively ascribed to a copper-rich oxide (Cu4O3).

In order to characterise the electrodes, electrochemical impedance spectroscopy (EIS) measurements were performed in KOH 1 M solutions at open circuit potential (OCP). The related Nyquist and Bode plots are reported in Figures 4 and 5: as can be seen, samples S1–5, S0.01–0.05 and S1–50 are characterized by a wide flattened semicircle in the low frequency (LF) region of Nyquist plot, and by one wave in the phase angle Bode plot. This behavior is like that of smooth electrodes covered with flat pores, indicating that the surface behaves like a flat one, i.e. the pores are well accessible at all the frequency values [40,41]. For the samples S0.1–0.5, S0.1–5 and S0.01–0.5 a small arc is visible at high frequency (HF), followed by a second branch in the LF region of the Nyquist plot, as well as the beginning of a second wave appears in the phase angle Bode plot, indicating a behavior typical of porous and rough electrodes [42], which respond differently, depending on the frequency region.

**Figure 3.** XRD pattern of the films as-deposited (**a**) and of the sample S1–5 after the corrosion (**b**) and relevant PDF cards for reference structures.

**Figure 4.** Nyquist diagram (**a**), Bode phase (**b**) and Bode modulus (**c**) recorded in KOH 1 M at open circuit potential. Inset: equivalent circuit used to model the EIS spectra.

To model the experimental data, an electrochemical equivalent circuit, which involves two time constants, was used (inset of Figure 4): the model is a slightly modified version of that originally proposed by Armstrong and Henderson [43], in which the capacitances were replaced by the constant phase elements (CPE) [44–46] which represent a deviation from the purely capacitive behaviour, related to surface in-homogeneity, or to variations of properties in the direction that is normal to the electrode surface. Such variability may be attributed, for example, to changes in the conductivity of oxide layers, or to porosity and surface roughness [47].

**Figure 5.** Nyquist diagram (**a**), Bode phase (**b**) and Bode modulus (**c**) recorded in KOH 1 M at open circuit potential.

The impedance ZCPE is described by the following equation:

$$\mathcal{Z}\_{\rm CPE} = \frac{1}{\mathcal{Q}(j\omega)^{\rm n}} \tag{1}$$

where Q is a capacitance parameter and n is a parameter characterizing the rotation of the complex plane impedance plot [47].

This two time constants model was widely used to describe the response of Ni-based porous electrodes during HER; when the semicircle at HF is potential-independent, it can be related to the electrode surface porosity response, while the potential-dependent LF semicircle can be related to the charge transfer resistance process [44,48,49]. On the other hand, when both semicircles change with overpotential, the LF time constant is associated to the hydrogen adsorption on the electrode surface, while the HF time constant is related to the charge transfer resistance [48,50].

In the present work, the same electrical circuit was used to model the response of the porous electrodes at OCP and the different time constants were correlated to the different pore size distribution. Analogous approach was adopted with hierarchical structures by Abouelamaiem et al. [51]: the smaller the pores, the higher the constant time values.

As can be seen from Figures 4 and 5, in the present case, the proposed model was able to interpret the Nyquist and Bode plots of the selected samples. The fitting parameters are presented in Table 2, along with the values of chi-squared (χ2), which were always in the order of 10−4.


**Table 2.** Electrical circuital parameters obtained from the fit of EIS spectra at open circuit potential for the synthesized samples, recorded in KOH 1 M.

Table 2 reports also the values of capacitance (C) calculated as proposed by Brug et al. [52] for non-Faradaic system:

$$\mathcal{C}\_{i} = \frac{(Q\_{i}\mathcal{R}\_{i})^{1/n\_{i}}}{\mathcal{R}\_{i}} \tag{2}$$

and the time constant (τ)

$$
\pi\_{\bar{i}} = \mathbb{C}\_{\bar{i}} \mathbb{R}\_{\bar{i}} \tag{3}
$$

Depending on the samples, different time constant values are calculated from the relevant circuital parameters, that can give information on the processes occurring at the electrode surface.

So, for example, the values of τ<sup>1</sup> between 0.10 and 0.38 s, are in the order of magnitude typically related to the charge transfer between electrode/electrolyte interface. Time constants in the order of half second were reported by Cardona et al [48,53], for porous Ni and Ni-Cu electrodes, and were attributed to the charge transfer kinetics. On the other hand, as suggested by other references [46,50,53], the τ<sup>2</sup> values 10-fold higher, calculated for samples S0.1–0.5, S0.1–5, S0.01–0.5, could be instead associated to the diffusion inside the smaller, less accessible pores. Finally, a very fast charge transfer inside the bulk material should be the reason of the very low values of τ<sup>2</sup> (in the order of ms) calculated for samples S1–5, S0.01–0.05, S1–50.

This different behavior can be explained considering the different morphology of the samples: when a tubular structure with little variations of the pore diameter is obtained, as in the case of sample S1–5, the response of the impedance is dominated by accessible surface inside the pores. Conversely, when a multidimensional pores structure, such as a cauliflower-like surface is obtained (see for example sample S0.01–0.5), the slow diffusive process in the smaller pores may become important, given that it is revealed when low frequency region is explored.

The values of *Cdl* were used, in turn, to compare the real surface area of the samples accessible to the electrolyte: the surface roughness factor Rf, was determined relating the *Cdl* of the samples with that of a smooth Ni electrode equal to 20 μF/cm2 [46] an in turn, the values of real active surface area (Ar) were calculated (see Table 2).

The results reported in Table 2 highlight that the group of samples prepared using a lower φ presents higher values of C*dl* and of roughness factor (Rf). Moreover, at the same φ values, an increase in the porosity is recorded with the decrease of the corrosion and relaxation time. This is not verified in the case of the sample S0.01–0.05, where very low pore density was obtained. Moreover, Rf values of 8 and 13 were obtained by EIS at commercial nickel plate and at Ni-Cu co-deposit, respectively.

In the last part of the work attention was paid on the evaluation of the catalytic activity of the samples toward HER. As is well known, especially when heterogeneous reactions are involved, the achievement of high specific area is of a great concern. However, also the morphology, and then the exploitability of the surface, may affect the material performances, particularly when the reactions involve the production of gas, as in the case of HER. Moreover, a synergistic effect on the catalytic activity was often attributed to the presence of copper, either in alloy, or co-deposited with nickel.

In order to study these effects, two samples were selected, which were characterized by different roughness factor, morphology and residual copper content: in particular, S1–5 was selected among those with φ = 0.2, and S0.01–0.5 among those with φ = 0.02 (see Table 1). Their electrocatalytic performance and their photoactivity were investigated, and their behavior compared with those of commercial nickel plate and Ni-Cu co-deposit.

Figure 6 reports the results of linear sweep polarization curves, performed in 1 M KOH solution: the cathodic current densities are reported as a function of the overpotential (η) together with the corresponding Tafel linearization, as inset.

**Figure 6.** Linear sweep voltammetries in 1 M KOH solution; inset: magnification of the linear Tafel polarization curves.

If compared with commercial nickel plate, the samples S1–5 and S0.01–0.5 present higher catalytic activity, the highest being obtained for sample S0.01–0.5. This enhancement in the catalytic activity can be connected to the increased specific surface area. Moreover, the trend of the linear sweep polarization of the Cu-Ni co-deposit indicates the positive effect of the presence of copper.

The kinetic parameters of the related processes were determined by considering both the geometric and the real surface area of the electrodes. The polarization curves are represented by Tafel equation [54]:

$$
\eta = a + b \log j \tag{4}
$$

where η is the overpotential, *b* is the Tafel slope, *j* is the current density and *a* is the intercept of the curve related to the exchange current density *j*<sup>0</sup> through equation:

$$a = \frac{2.3 \text{ RT}}{\beta nF} \log j\_0 \tag{5}$$

where *n* represents the number of electrons exchanged, *F* is the Faraday constant, β is the symmetry factor and *R* is the gas constant.

Values of exchange current density *j*0, and Tafel slope *b*, estimated from the linear polarization curves using Equations (4) and (5), are listed in Table 3.

As can be observed, the values of Tafel slope (*b*) obtained for the synthesized samples range from 126 to 140 mV/dec, indicating that HER proceeds via the Volmer-Heyrosky mechanism [44,55]; similar values were reported in the literature for porous Ni-based samples also in presence of copper [46,48]. Both samples S1–5 and S0.01–0.5 present exchange current density values considerably higher with respect to the commercial nickel, indicating a considerable improvement of the apparent electrocatalytic properties of the fabricated electrodes.


**Table 3.** Kinetic parameters for HER obtained from the polarization curves recorded in 1 M KOH.

The value measured for the Ni-Cu co-deposit is instead indicative of the effect of the presence of copper in the deposit. Normalizing the exchange current densities for the active surface area of the electrodes, the real exchange current density *j*0*<sup>r</sup>* was calculated.

The higher *j*0*<sup>r</sup>* value for the sample S1–5 with respect to that for S0.01–0.5, suggests that the inner porous surface of this sample is not totally exploitable during HER, due to the gas bubbles shielding. This behavior can be explained by considering the more open structure of sample S1–5 with respect to S0.01–0.5, thus confirming the strong effect of the pores (size, shape and distribution) on the resulting electrocatalytic performance. Comparing the *j*0*<sup>r</sup>* values for sample S1–5 with that of commercial nickel, the effect of the composition of the sample can be observed: in fact, the enhancement in the performance is not only related to the increased surface area, but also to the copper amount which affects the overall electroactivity of the electrodes. This effect can be confirmed at Ni-Cu co-deposit, where the enhancement of *j*0*r*, compared to a commercial nickel electrode, can be connected essentially to the presence of a rough dual Ni-Cu system.

In order to investigate possible applications of the developed samples as photocathodes, they were submitted to thermal annealing at 500 ◦C in order to allow the formation of NiO.

Figure 7 shows, as an example, the cathodic photocurrent response measured at sample S0.01–0.5: as can be seen, under irradiation a remarkable increase in photocurrent is observed.

In order to compare the photocurrent of samples S1–5, S0.01–0.5 and commercial nickel, in Figure 8 the values of photocurrent are reported for three values of applied potential; as can be seen, applied potential being the same, samples S1–5 and S0.01–0.5 present higher values of photocurrent with respect to the commercial nickel. These values are comparable with those reported in the literature and measured in similar conditions: for NiO photocathodes fabricated by alkaline etching, anodizing nickel foil in an organic-based electrolyte, about 400 μA/cm<sup>2</sup> was measured irradiating the samples with 300 W arc xenon lamp [15]. The values in the order of magnitude of the tens were indicated in a recent review, where sensitized NiO photocathodes for water splitting cells were used: the authors state that photocurrent values varied consistently with the specific surface area of the electrode, suggesting that NiO electrodes made under different conditions should possess comparable photoelectrochemical performance [56].

**Figure 7.** LSV of sample S0.01–0.5 in 0.1 M KNO3. The potential was ramped (5 mV s<sup>−</sup>1) from the OCP to −0.8 V. Data were recorded under dark and irradiation condition (AM0 filter).

**Figure 8.** Cathodic photocurrent density recorded in KNO3 0.1 M solution for different samples, at different applied potential: −0.4 V (blue bars), −0.6 V (orange bars) and −0.8 V (grey bars).

If data at different applied potential are considered, a different behavior of S1–5 and S0.01–0.5 samples can be observed: as the cathodic potential is increased, while a regular increase in performance is measured at sample S1–5, sample S0.01–0.5 shows lower value of photocurrent when the most cathodic potentials is applied. This trend may be explained considering that at the highest cathodic potential, gas can be generated which may limit the exploitability of the whole pore structure, especially when small pores are involved in the structure, as at sample S0.01–0.5. Of note is that if the photocurrent values are normalized by the real surface area (Ar) of the samples, the highest performance is measured at sample S1–5, at all the potentials; moreover, at the highest cathodic potential, the performance of S0.01–0.5 sample decreases, and it becomes even worst than the commercial Ni sample.
