Nanoporous Microtubes via Oxidation and Reduction of Cu – Ni Commercial Wires

Metallic porous microtubes were obtained from commercial wires (200–250 μm diameter) of Cu-65Ni-2Fe, Cu-44Ni-1Mn and Cu-23Ni, alloys (wt. %) by surface oxidation at 1173 K in air, removal of the unoxidized core by chemical etching, and reduction in annealing in the hydrogen atmosphere. Transversal sections of the partially oxidized wires show a porous layered structure, with an external shell of CuO (about 10 μm thick) and an inner layer of NiO (70–80 μm thick). In partially oxidized Cu-44Ni-1Mn and Cu-23Ni, Cu2O is dispersed in NiO because the maximum solubility of Cu in NiO is exceeded, whereas in Cu-65Ni-2Fe, a Cu2O shell is present between CuO and NiO layers. Chemical etching removed the unoxidized metallic core and Cu2O with formation of porous oxide microtubes. Porosity increases with Cu content because of the larger amount of Cu2O in the partially oxidized wire. After reduction, the transversal sections of the metallic porous microtubes show a series of f.c.c.-(Cu,Ni) solid solutions with different compositions, due to the segregation of CuO and NiO during oxidation caused by the different diffusion coefficients of Ni and Cu in the respective oxides. Pore formation occurs at each step of the process because of the Kirkendall effect, selective phase removal and volume contraction.


Introduction
Microtubes and nanotubes attract interest both from the scientific and technological point of view because of the properties that allow their use as catalysts, sensors, drug-delivery carriers, acoustic insulators, biomedical diagnosis agents, and chemical reactors in Micro Electro Mechanical Systems (MEMS), lasers, magnetic, optical, semiconductor and biomedical devices [1][2][3][4].In particular, metallic nanotubes attract attention because of magnetic properties [4], surface enhanced Raman scattering (SERS) [5] and catalytic performance [6].The synthesis of microtubes and nanotubes is typically run by self-organization of constituent atoms or molecules, template-assisted method (assisted by electroless deposition, electrodeposition or atomic layer deposition), Kirkendall-based route or hydrothermal method [1,[7][8][9].
Metallic porous materials are of great interest because of their functional and structural features.Due to their mechanical properties, metallic porous materials can solve the problem of intrinsic brittleness of ceramic materials [10,11].Current applications of metallic porous materials are in the field of energy and environmental protection (e.g., vehicle exhaust gas, coal combustion) [12], in the chemical industry (e.g., microreactors) [13] and biomedical science (e.g., carrier of biological agents, bone and tissue substitute materials) [14].Recently, Cu-Ni foam films [15] and nanoporous gold obtained from a metallic glass precursor [16] attracted attention because of the improved electrocatalitic activity and surface-enhanced Raman scattering, respectively.
In the case of metallic microtubes and nanotubes, the synthesis is usually performed by template method via electrochemical deposition [1,7,8,17].This approach allows a fine tuning of the wall thickness and leads to the formation of compact and smooth surfaces.On the other hand, the fabrication of nanoporous microtubes is desirable because of the enhanced capillarity properties and the specific surface properties, which lead to potential applications in MEMS for fluid transportation or sensing [18][19][20].
The development of nanoporous materials, starting from the reduction of oxides, was investigated by several authors, and it is generally achieved by the selective removal of one phase by chemical routes [21,22].In a previous work, a simple method for the synthesis of a nanoporous metallic microtube starting from a commercial constantan (Cu-Ni alloy) wire was presented [23].The processing route consists of the following three steps: (1) partial oxidation of the alloy; (2) chemical etching for the removal of the un-oxidized inner core; and (3) reduction of the oxides in hydrogen atmosphere.
In this work, the same method was applied using commercial wires of Cu-Ni alloys with different Cu:Ni ratio and wires of the pure elements (i.e., Cu and Ni).The Cu/Ni ratio affects the phase selection and microstructure of the oxidized wires and leads to different levels of porosity in the final reduced microtubes.
In the first step, each wire was cut in 3-5 cm long pieces and placed in air in a pre-heated muffle oven at 1173 K inside alumina crucibles.In order to obtain similar oxide layers, the wires were exposed to the oxidation atmosphere for different times ranging from a minimum of 5 min to a maximum of 1440 min, on the basis of the composition.
In the second step, chemical etching with 65% nitric acid aqueous solution at room temperature allowed removal of the inner metallic core, together with some oxide phases, leaving intact the external oxide layer, with formation of an oxide microtube.
Finally, reduction of the microtubes was performed by heating the samples from room temperature up to 723 K at 10 K/min under 10 bar of hydrogen inside a High Pressure Differential Scanning Calorimeter (DSC) DSC HP-204 (Netzsch, Selb, Germany), using aluminium pans as sample holders.
Thermodynamic properties of the reactions involved were calculated using the Thermocalc software and the database "Substance" [24].
The thickness of the oxide layer after oxidation for different times was measured with a DMLM optical microscope (Leica, Wetzlar, Germany) across the transversal section of the samples.Scanning Electron Microscopy (SEM) observations were performed using a Stereoscan 410 (Leica, Wetzlar, Germany) equipped with an INCAx-sight probe (Oxford Instruments, Abingdon, UK) for Energy Dispersive X-ray analysis (EDX).
X-ray diffraction (XRD) measurements were performed with a X'Pert Pro (Cu Kα) (PANalytical B.V., Almelo, The Netherlands) using a capillary spinner technique with a 0.3 mm diameter glass capillary.Rietveld refinement was performed using the software MAUD (version 2.55 64 bit, Università degli Studi di Trento, Trento, Italy) [25].The March-Dollase model was used to take preferential crystallographic reflections into account in the oxidized and chemically etched samples.

Oxidation
Figure 1 shows the thickness of the oxide layer as a function of the square root of the time.Data points are calculated by averaging thickness measurements of the oxide in different positions in the same cross section and in different cross sections of the same sample.The error bars are the standard deviation of the averages.In the case of pure elements, a single linear behaviour is observed.In the case of Cu-Ni alloys, two oxidation steps are observed (continuous and dotted lines in Figure 1).Both steps can be described by a parabolic law, x = K p t (where x, t and K p are the oxide thickness, the oxidation time and the parabolic rate constant, respectively).As shown in Figure 1, the second oxidation step is faster than the first one and both oxidation steps become faster when the Ni content in the starting alloy decreases, in accordance with the literature [26][27][28].K p values for both steps are summarized in Table 1.

Oxidation
Figure 1 shows the thickness of the oxide layer as a function of the square root of the time.Data points are calculated by averaging thickness measurements of the oxide in different positions in the same cross section and in different cross sections of the same sample.The error bars are the standard deviation of the averages.In the case of pure elements, a single linear behaviour is observed.In the case of Cu-Ni alloys, two oxidation steps are observed (continuous and dotted lines in Figure 1).Both steps can be described by a parabolic law,  = √    (where x, t and Kp are the oxide thickness, the oxidation time and the parabolic rate constant, respectively).As shown in Figure 1, the second oxidation step is faster than the first one and both oxidation steps become faster when the Ni content in the starting alloy decreases, in accordance with the literature [26][27][28].Kp values for both steps are summarized in Table 1.
Figure 1.Oxide thickness as a function of the square root of oxidation time at 1173 K for Cu (black squares), Cu-23Ni (red circles), Cu-44Ni (green up triangles), Cu-65Ni (blue down triangles) and Ni (cyan diamonds) samples.In the case of pure elements, experimental points were fitted with a single parabolic law (continuous lines),  = √, where x, t and Kp are the oxide thickness, the oxidation time and the parabolic rate constant, respectively.In the case of Cu-Ni alloys, experimental points show the presence of two steps (continuous and dotted lines, respectively).For Cu-23Ni and Cu-65Ni samples, each step was fitted with a parabolic law, leading to different values of the constant rate, Kp′ and Kp′′, as reported in Table 1.

Table 1.
Values of the parabolic rate constants at 1173 K for oxidation of Ni, Cu-65Ni, Cu-44Ni, Cu-23Ni and Cu samples.Kp′ and Kp′′ values are related to the first oxidation step (present in all the samples) and the second oxidation step (present only in the Cu-Ni alloys), respectively.According to Refs.[27][28][29][30], oxidation of Cu-Ni alloys occurs through several reactions.On one hand, Cu rapidly oxidizes to Cu(I), and, subsequently, it transforms to Cu(II) with a slower kinetics, forming an outer layer.On the other hand, an inner layer, consisting of a NiO and Cu2O mixture, forms between the outer Cu(II) oxide layer and the metallic core.Oxide thickness as a function of the square root of oxidation time at 1173 K for Cu (black squares), Cu-23Ni (red circles), Cu-44Ni (green up triangles), Cu-65Ni (blue down triangles) and Ni (cyan diamonds) samples.In the case of pure elements, experimental points were fitted with a single parabolic law (continuous lines), x = K p t, where x, t and K p are the oxide thickness, the oxidation time and the parabolic rate constant, respectively.In the case of Cu-Ni alloys, experimental points show the presence of two steps (continuous and dotted lines, respectively).For Cu-23Ni and Cu-65Ni samples, each step was fitted with a parabolic law, leading to different values of the constant rate, K p and K p , as reported in Table 1.

Sample Kp′ (g
Table 1.Values of the parabolic rate constants at 1173 K for oxidation of Ni, Cu-65Ni, Cu-44Ni, Cu-23Ni and Cu samples.K p and K p values are related to the first oxidation step (present in all the samples) and the second oxidation step (present only in the Cu-Ni alloys), respectively.According to Refs.[27][28][29][30], oxidation of Cu-Ni alloys occurs through several reactions.On one hand, Cu rapidly oxidizes to Cu(I), and, subsequently, it transforms to Cu(II) with a slower kinetics, forming an outer layer.On the other hand, an inner layer, consisting of a NiO and Cu 2 O mixture, forms between the outer Cu(II) oxide layer and the metallic core.The occurrence of two steps with different oxidation rates (see Figure 1 and Table 1) is due to the recrystallization of the outer CuO layer and to the stress relaxation of the inner layer [27].Thus, diffusion of oxygen through anion vacancies in the inner layer becomes faster, leading to a larger macroscopic oxidation rate [27].
On the basis of the kinetic data, we tuned the oxidation time in order to obtain partially oxidized wires with similar oxide thickness (i.e., 80-100 µm) and obtain mechanically self-standing oxide layers.Thus, wires of Cu-23Ni, Cu-44Ni, Cu-65Ni and Cu were oxidized for 30 min, 160 min, 300 min and 1440 min, respectively.In the case of pure Ni, the oxidation kinetics is extremely slow as a consequence of passivation (see, for instance, Ref. [31]).Therefore, we decided to exclude this sample from the prosecution of the work.
Figure 2a shows the XRD patterns of the oxidized samples.The results of the Rietveld refinement related to the XRD patterns of Figure 2a are reported in Table 2.In the case of pure copper, oxidation for 1440 min leads to the formation of Cu 2 O and CuO.In the case of Cu-23Ni and Cu-44Ni (oxidized for 30 min and 160 min, respectively), the following phases can be identified: Cu 2 O, CuO and NiO.As expected, the amount of copper oxides (i.e., Cu 2 O, CuO) increases as the Cu content in the starting alloy increases.Finally, in the case of Cu-65Ni, oxidized for 300 min, the main phases are CuO and NiO.A residual amount of the unoxidized f.c.c.-(Cu-Ni) starting alloy is present due to the slower kinetics of oxidation for this alloy (Figure 1 and Table 1), while the amount of Cu 2 O is below the detection limit of the technique.The occurrence of two steps with different oxidation rates (see Figure 1 and Table 1) is due to the recrystallization of the outer CuO layer and to the stress relaxation of the inner layer [27].Thus, diffusion of oxygen through anion vacancies in the inner layer becomes faster, leading to a larger macroscopic oxidation rate [27].
On the basis of the kinetic data, we tuned the oxidation time in order to obtain partially oxidized wires with similar oxide thickness (i.e., 80-100 m) and obtain mechanically self-standing oxide layers.Thus, wires of Cu-23Ni, Cu-44Ni, Cu-65Ni and Cu were oxidized for 30 min, 160 min, 300 min and 1440 min, respectively.In the case of pure Ni, the oxidation kinetics is extremely slow as a consequence of passivation (see, for instance, Ref. [31]).Therefore, we decided to exclude this sample from the prosecution of the work.
Figure 2a shows the XRD patterns of the oxidized samples.The results of the Rietveld refinement related to the XRD patterns of Figure 2a are reported in Table 2.In the case of pure copper, oxidation for 1440 min leads to the formation of Cu2O and CuO.In the case of Cu-23Ni and Cu-44Ni (oxidized for 30 min and 160 min, respectively), the following phases can be identified: Cu2O, CuO and NiO.As expected, the amount of copper oxides (i.e., Cu2O, CuO) increases as the Cu content in the starting alloy increases.Finally, in the case of Cu-65Ni, oxidized for 300 min, the main phases are CuO and NiO.A residual amount of the unoxidized f.c.c.-(Cu-Ni) starting alloy is present due to the slower kinetics of oxidation for this alloy (Figure 1 and Table 1), while the amount of Cu2O is below the detection limit of the technique.In the case of Cu-23Ni, Cu-44Ni and Cu-65 Ni, the cross section of the partially oxidized wires reveals the presence of an external darker oxide layer and an internal brighter unoxidized metallic core.In the case of fully oxidized Cu, Figure 3d, concentric layers of two oxides are visible.As mentioned before, Cu rapidly oxidizes to Cu(I), and, subsequently, it transforms to Cu(II) with a slower kinetics.Thus, in agreement with the XRD patterns, Figure 2a, and the compositional contrast, the external layer can be related to CuO, whereas the internal one can be related to Cu 2 O.
Figure 3a-d shows the SEM micrographs (backscattered electron detector) of the transversal cross sections of the oxidized wires.In the case of Cu-23Ni, Cu-44Ni and Cu-65 Ni, the cross section of the partially oxidized wires reveals the presence of an external darker oxide layer and an internal brighter unoxidized metallic core.In the case of fully oxidized Cu, Figure 3d, concentric layers of two oxides are visible.As mentioned before, Cu rapidly oxidizes to Cu(I), and, subsequently, it transforms to Cu(II) with a slower kinetics.Thus, in agreement with the XRD patterns, Figure 2a, and the compositional contrast, the external layer can be related to CuO, whereas the internal one can be related to Cu2O. Figure 4a-c shows magnified views of the cross sections reported in Figure 3a-c, respectively, together with the corresponding EDX map.All the three partially oxidized alloys show a Cu-rich external oxide layer (red pixels) that corresponds to CuO which is detected in the corresponding XRD patterns.The intermediate oxide layers of the three samples show a different contrast of red and green pixels corresponding to Cu and Ni, respectively.In Cu-65Ni, an almost uniform green intermediate oxide layer is observed, Figure 4a, suggesting the presence of almost only NiO, in accordance with the XRD results, Figure 2a.In Cu-44Ni and Cu-23Ni, the intermediate oxide layer shows a mixture of green and red pixels, indicating the coexistence of NiO and Cu2O in agreement with the corresponding XRD patterns.In the case of Cu-65Ni, Figure 4a, and Cu-44Ni, Figure 4b, between the intermediate oxide layer and the unoxidized metallic core, there is a thin red ring (Cu-rich), which has a metallic appearance when observed at the optical microscope [23].The formation of this metallic copper layer occurs according to the reaction [29]:   4a, and Cu-44Ni, Figure 4b, between the intermediate oxide layer and the unoxidized metallic core, there is a thin red ring (Cu-rich), which has a metallic appearance when observed at the optical microscope [23].The formation of this metallic copper layer occurs according to the reaction [29]: which is spontaneous at 1173 K (∆G = −51.3kJ/mol).
which is spontaneous at 1173 K (ΔG = −51.3kJ/mol).Figure 5a shows the composition profiles of the cross sections of the partially oxidized wires (Figure 4).In the case of Cu-65Ni (red circles), the abrupt decrease of Cu content corresponds to the sharp transition between the thin external CuO layer and the intermediate Ni-rich oxide layer, Figure 4a.The amount of Cu in the Ni-rich oxide layer ranges between 3 at.% and 28 at.%.On the basis of the lattice parameter (a = 0.41923 nm) and the Vegard law for Ni1−xCuxO [32], an average content around 25 at.% of Cu in NiO can be estimated.According to Ref. [33], the maximum solubility of Cu in NiO is about 25 at.%; therefore, it can be concluded that the intermediate Ni-rich oxide layer consists of almost only (Cu,Ni)O.Finally, the rapid increase of Cu content at an oxide/metal interface corresponds to the thin metallic copper ring observed in Figure 4a.In the case of Cu-44Ni (black triangles), a similar profile is observed.However, the amount of Cu in the Ni-rich oxide is above the solubility limit, leading to the precipitation Cu2O particles and their coexistence with NiO in the intermediate Ni-rich oxide layer, as confirmed by the green and red regions in Figure 4b and by the XRD patterns, Figure 2a, respectively.The occurrence of a saturated (Cu,Ni)O solid solution is supported by the value of the lattice parameter (a = 0.41920 nm), which corresponds to about 25 at.% of Cu.Finally, in the case of Cu-23Ni (green squares), the profile shows a thicker Cu-rich layer, corresponding to the external CuO ring, followed by an intermediate layer with a copper content between 45 at.% and 60 at.%.In this case, the value of the lattice parameter is Figure 5a shows the composition profiles of the cross sections of the partially oxidized wires (Figure 4).In the case of Cu-65Ni (red circles), the abrupt decrease of Cu content corresponds to the sharp transition between the thin external CuO layer and the intermediate Ni-rich oxide layer, Figure 4a.The amount of Cu in the Ni-rich oxide layer ranges between 3 at.% and 28 at.%.On the basis of the lattice parameter (a = 0.41923 nm) and the Vegard law for Ni 1−x Cu x O [32], an average content around 25 at.% of Cu in NiO can be estimated.According to Ref. [33], the maximum solubility of Cu in NiO is about 25 at.%; therefore, it can be concluded that the intermediate Ni-rich oxide layer consists of almost only (Cu,Ni)O.Finally, the rapid increase of Cu content at an oxide/metal interface corresponds to the thin metallic copper ring observed in Figure 4a.In the case of Cu-44Ni (black triangles), a similar profile is observed.However, the amount of Cu in the Ni-rich oxide is above the solubility limit, leading to the precipitation Cu 2 O particles and their coexistence with NiO in the intermediate Ni-rich oxide layer, as confirmed by the green and red regions in Figure 4b and by the XRD patterns, Figure 2a, respectively.The occurrence of a saturated (Cu,Ni)O solid solution is supported by the value of the lattice parameter (a = 0.41920 nm), which corresponds to about 25 at.% of Cu.Finally, in the case of Cu-23Ni (green squares), the profile shows a thicker Cu-rich layer, corresponding to the external CuO ring, followed by an intermediate layer with a copper content between 45 at.% and 60 at.%.In this case, the value of the lattice parameter is 0.41841 nm, which corresponds to about 13 at.% of Cu (i.e., below the solubility limit of (Cu,Ni)O).Thus, the intermediate oxide layer contains a larger amount of Cu 2 O with respect to the other two samples, as shown by the red/green contrast in Figure 4 and in accordance with the quantitative analysis of the XRD patterns (Table 1).
Metals 2017, 7, 46 8 of 14 samples, as shown by the red/green contrast in Figure 4 and in accordance with the quantitative analysis of the XRD patterns (Table 1).

Etching
Figure 2b shows the XRD patterns of the samples after etching with HNO3 solution.The corresponding quantitative analysis of the crystalline phases is reported in Table 1.In the case of pure Cu sample, only the reflections of CuO are observed whereas the peaks of Cu2O disappear.The dissolution of Cu2O leads to formation of an oxide microtube formed by the external layer of CuO, which is unaffected by the HNO3 attack, Figure 3h.In the case of Cu-Ni samples, XRD patterns show the presence of CuO and NiO, while the peaks of Cu2O and residual f.c.c.solid solutions disappear.Thus, microtubes with two concentric layers of CuO (external) and (Cu,Ni)O (internal) are obtained (Figure 3e-g).

Reduction
Figure 6 shows the HP-DSC traces related to the reduction of Cu, Cu-23Ni, Cu-44Ni and Cu-65Ni.
In the case of Cu, a single broad exothermic signal, starting around 440 K, is visible.In the case of the Cu-Ni samples, the onset temperature of the first sharp peak (indicated by the arrows on the

Etching
Figure 2b shows the XRD patterns of the samples after etching with HNO 3 solution.The corresponding quantitative analysis of the crystalline phases is reported in Table 1.In the case of pure Cu sample, only the reflections of CuO are observed whereas the peaks of Cu

Reduction
Figure 6 shows the HP-DSC traces related to the reduction of Cu, Cu-23Ni, Cu-44Ni and Cu-65Ni.
Metals 2017, 7, 46 9 of 14 In the case of Cu, a single broad exothermic signal, starting around 440 K, is visible.In the case of the Cu-Ni samples, the onset temperature of the first sharp peak (indicated by the arrows on the left in Figure 6) increases with the Ni content (from 460 K for Cu-23Ni and Cu-44Ni to 510 K for Cu-65Ni).In some cases (i.e., Cu-23Ni and Cu-44Ni), the first signal is split into two overlapped The splitting of the first peak can be likely due to the inhomogeneous morphology of the oxide layer (Figure 4), which can affect the kinetics of the reduction reaction.After the main sharp peak, a second broad exothermic signal (indicated by the arrows on the right in Figure 6) appears around 550 K, which is absent in the reduction of the pure Cu sample.XRD analysis (not shown) performed before and after the beginning of the second broad DSC signal reveal that the first sharp double signal is related to the reduction of CuO, while the second broad signal around 550 K corresponds to the reduction of NiO.
Metals 2017, 7, 46 9 of 14 left in Figure 6) increases with the Ni content (from 460 K for Cu-23Ni and Cu-44Ni to 510 K for Cu-65Ni).In some cases (i.e., Cu-23Ni and Cu-44Ni), the first signal is split into two overlapped peaks.The splitting of the first peak can be likely due to the inhomogeneous morphology of the oxide layer (Figure 4), which can affect the kinetics of the reduction reaction.After the main sharp peak, a second broad exothermic signal (indicated by the arrows on the right in Figure 6) appears around 550 K, which is absent in the reduction of the pure Cu sample.XRD analysis (not shown) performed before and after the beginning of the second broad DSC signal reveal that the first sharp double signal is related to the reduction of CuO, while the second broad signal around 550 K corresponds to the reduction of NiO. Figure 2c shows the XRD patterns of the samples after reduction in H2 atmosphere.Obviously, in the case of Cu sample, only the diffraction peaks of Cu (f.c.c.) are present (not shown).In the case of Cu-Ni samples, reflections of two f.c.c.solid solutions (Cu-and Ni-rich, respectively) are observed.The crystallographic reflections of the Ni-rich f.c.c.solid solution appear broader than those corresponding to the Cu-rich f.c.c.solid solution.The morphology of metallic microtubes obtained after reduction is shown in Figure 3k-l.In the case of Cu-65Ni and Cu-44Ni samples, Figure 2k,i, respectively, uniform microtubes are obtained, while, in the case of Cu-23Ni wire, Figure 3j, two concentric microtubes are obtained.Figure 5b shows the composition profile of the microtubes shown in Figure 3k-j.In all of the samples, the observed progressive decrease of the Cu content from the external surface, which consists of almost pure copper, towards the internal surface, which consists of almost pure nickel, derives from the multi-layered oxide structure.In the case of Cu-23Ni, the external and internal microtubes are Cu-and Ni-rich, respectively.In this sample, the external microtubes can be detached from the internal one, so that an Ni-rich microtube is obtained, as shown in Figure 7.The red pixels indicate the presence of residual copper on the surface of the Ni-rich microtube.Figure 2c shows the XRD patterns of the samples after reduction in H 2 atmosphere.Obviously, in the case of Cu sample, only the diffraction peaks of Cu (f.c.c.) are present (not shown).In the case of Cu-Ni samples, reflections of two f.c.c.solid solutions (Cu-and Ni-rich, respectively) are observed.The crystallographic reflections of the Ni-rich f.c.c.solid solution appear broader than those corresponding to the Cu-rich f.c.c.solid solution.The morphology of metallic microtubes obtained after reduction is shown in Figure 3k-l.In the case of Cu-65Ni and Cu-44Ni samples, Figure 2k,i, respectively, uniform microtubes are obtained, while, in the case of Cu-23Ni wire, Figure 3j, two concentric microtubes are obtained.Figure 5b shows the composition profile of the microtubes shown in Figure 3k-j.In all of the samples, the observed progressive decrease of the Cu content from the external surface, which consists of almost pure copper, towards the internal surface, which consists of almost pure nickel, derives from the multi-layered oxide structure.In the case of Cu-23Ni, the external and internal microtubes are Cu-and Ni-rich, respectively.In this sample, the external microtubes can be detached from the internal one, so that an Ni-rich microtube is obtained, as shown in Figure 7.The red pixels indicate the presence of residual copper on the surface of the Ni-rich microtube.The compositional gradient of the reduced microtubes, Figure 5b, shows the presence of a series of f.c.c.-(Cu-Ni) solid solutions with different compositions.As shown in Ref. [23] for the Cu-44Ni sample, the broad crystallographic reflections of the Ni-rich f.c.c.solid solution can be successfully fitted by Rietveld refinement assuming the coexistence of f.c.c.-(Cu-Ni) solid solutions with different lattice parameters, as a consequence of the continuous variation in composition along the cross section of the microtubes.On the basis of the experimental results in Figures 2c and 5b, the same explanation can be given for the broadening of the crystallographic reflections of the Ni-rich f.c.c.solid solution in Cu-23Ni and Cu65-Ni samples.Thus, the values of the lattice parameters reported for the Ni-rich f.c.c.solid solution in Table 1 are just representing an average value of the lattice parameters of the various solid solutions with different composition.
The morphology of internal walls of the obtained microtubes show different features, depending on the composition of the starting alloy.In Cu-65Ni, Figure 8a, the metallic particles have irregular shapes, whereas in Cu-44Ni, Figure 8b, the shape of the metallic particles becomes smoother.In both samples, the size of the metallic particles is similar (average size of 280 ± 90 nm and 270 ± 70 nm for Cu-65Ni and Cu-44Ni, respectively).They appear isolated and not interconnected pores are visible between the metallic particles on the surface.In the case of Cu-23Ni, the internal surface is characterized by interconnected pores separated by metallic ligaments, as shown in Figure 8c.In this case, the ligament width is 430 ± 100 nm and the pore size is 280 ± 80 nm.The compositional gradient of the reduced microtubes, Figure 5b, shows the presence of a series of f.c.c.-(Cu-Ni) solid solutions with different compositions.As shown in Ref. [23] for the Cu-44Ni sample, the broad crystallographic reflections of the Ni-rich f.c.c.solid solution can be successfully fitted by Rietveld refinement assuming the coexistence of various f.c.c.-(Cu-Ni) solid solutions with different lattice parameters, as a consequence of the continuous variation in composition along the cross section of the microtubes.On the basis of the experimental results in Figures 2c and 5b, the same explanation can be given for the broadening of the crystallographic reflections of the Ni-rich f.c.c.solid solution in Cu-23Ni and Cu65-Ni samples.Thus, the values of the lattice parameters reported for the Ni-rich f.c.c.solid solution in Table 1 are just representing an average value of the lattice parameters of the various solid solutions with different composition.
The morphology of internal walls of the obtained microtubes show different features, depending on the composition of the starting alloy.In Cu-65Ni, Figure 8a, the metallic particles have irregular shapes, whereas in Cu-44Ni, Figure 8b, the shape of the metallic particles becomes smoother.In both samples, the size of the metallic particles is similar (average size of 280 ± 90 nm and 270 ± 70 nm for Cu-65Ni and Cu-44Ni, respectively).They appear isolated and not interconnected pores are visible between the metallic particles on the surface.In the case of Cu-23Ni, the internal surface is characterized by interconnected pores separated by metallic ligaments, as shown in Figure 8c.In this case, the ligament width is 430 ± 100 nm and the pore size is 280 ± 80 nm.The compositional gradient of the reduced microtubes, Figure 5b, shows the presence of a series of f.c.c.-(Cu-Ni) solid solutions with different compositions.As shown in Ref. [23] for the Cu-44Ni sample, the broad crystallographic reflections of the Ni-rich f.c.c.solid solution can be successfully fitted by Rietveld refinement assuming the coexistence of various f.c.c.-(Cu-Ni) solid solutions with different lattice parameters, as a consequence of the continuous variation in composition along the cross section of the microtubes.On the basis of the experimental results in Figures 2c and 5b, the same explanation can be given for the broadening of the crystallographic reflections of the Ni-rich f.c.c.solid solution in Cu-23Ni and Cu65-Ni samples.Thus, the values of the lattice parameters reported for the Ni-rich f.c.c.solid solution in Table 1 are just representing an average value of the lattice parameters of the various solid solutions with different composition.
The morphology of internal walls of the obtained microtubes show different features, depending on the composition of the starting alloy.In Cu-65Ni, Figure 8a, the metallic particles have irregular shapes, whereas in Cu-44Ni, Figure 8b, the shape of the metallic particles becomes smoother.In both samples, the size of the metallic particles is similar (average size of 280 ± 90 nm and 270 ± 70 nm for Cu-65Ni and Cu-44Ni, respectively).They appear isolated and not interconnected pores are visible between the metallic particles on the surface.In the case of Cu-23Ni, the internal surface is characterized by interconnected pores separated by metallic ligaments, as shown in Figure 8c.In this case, the ligament width is 430 ± 100 nm and the pore size is 280 ± 80 nm.

Discussion
Figure 9 shows a scheme of the cross section of the wires of the different alloys after oxidation, etching and reduction, based on the experimental results presented in the previous sections.In all cases, after oxidation, the external layer consists of CuO.Only the case of Cu-23Ni, a pure Cu2O layer is present beneath the external CuO layer.Subsequently, in all samples, a mixed layer consisting of (Ni,Cu)O and Cu2O is observed, where the amount of Cu2O increases as the Cu content of the alloy increases.Finally, between the (Ni,Cu)O + Cu2O oxidized layer and the unoxidized metallic core, a thin metallic ring containing pure Cu is observed in Cu-65Ni and Cu-44Ni.According to the literature [28], the Cu ring is expected to also be present in Cu-23Ni; however, it cannot be detected in the EDX map, Figure 4c, and the composition profile, Figure 5b.A possible explanation is that the results of the EDX microprobe are affected by the surrounding matrix: the small composition difference between the unoxidized metallic core and the expected Cu ring likely leads to a limited contrast between these two regions in the EDX map.

Discussion
Figure 9 shows a scheme of the cross section of the wires of the different alloys after oxidation, etching and reduction, based on the experimental results presented in the previous sections.In all cases, after oxidation, the external layer consists of CuO.Only in the case of Cu-23Ni, a pure Cu 2 O layer is present beneath the external CuO layer.Subsequently, in all samples, a mixed layer consisting of (Ni,Cu)O and Cu 2 O is observed, where the amount of Cu 2 O increases as the Cu content of the alloy increases.Finally, between the (Ni,Cu)O + Cu 2 O oxidized layer and the unoxidized metallic core, a thin metallic ring containing pure Cu is observed in Cu-65Ni and Cu-44Ni.According to the literature [28], the Cu ring is expected to also be present in Cu-23Ni; however, it cannot be detected in the EDX map, Figure 4c, and the composition profile, Figure 5b.A possible explanation is that the results of the EDX microprobe are affected by the surrounding matrix: the small composition difference between the unoxidized metallic core and the expected Cu ring likely leads to a limited contrast between these two regions in the EDX map.
After oxidation, all of the Cu-Ni samples show some degree of porosity.Cu and Ni are segregated during oxidation of Cu-Ni alloys as a consequence of the larger diffusion coefficient of Cu in Cu 2 O with respect to the one of Ni in NiO [34].The different mobility of the metallic ions in the corresponding oxides creates an excess of vacancies that coalesce, with consequent formation of pores by the Kirkendall effect [35].
After chemical etching of the partially oxidized wires, the unoxidized metallic core and Cu 2 O are leached and a self-standing oxide microtube is obtained.The removal of Cu 2 O from the (Ni,Cu)O matrix induces the formation of further porosity.The formation of pores is proportional to the Cu 2 O content in the alloy (Cu-23Ni > Cu-44Ni > Cu-65Ni).In the case of the Cu-23Ni sample, the amount of Cu dissolved in (Ni,Cu)O (13 at.%) is lower than the one foreseen by the equilibrium.This is likely due to kinetic reasons, since the oxidation time (30 min) is probably not enough to complete the solubilisation of Cu ions in (Ni,Cu)O, leaving an amount of Cu 2 O larger than expected in equilibrium conditions.Additional porosity forms during reduction because of the contraction due to the different densities of f.c.c.Cu-Ni solid solutions (8.9 g/cm 3 ) and Cu-Ni-oxides (about 6.5 g/cm 3 ).After oxidation, all of the Cu-Ni samples show some degree of porosity.Cu and Ni are segregated during oxidation of Cu-Ni alloys as a consequence of the larger diffusion coefficient of Cu in Cu2O with respect to the one of Ni in NiO [34].The different mobility of the metallic ions in the corresponding oxides creates an excess of vacancies that coalesce, with consequent formation of pores by the Kirkendall effect [35].
After chemical etching of the partially oxidized wires, the unoxidized metallic core and Cu2O are leached and a self-standing oxide microtube is obtained.The removal of Cu2O from the (Ni,Cu)O matrix induces the formation of further porosity.The formation of pores is proportional to the Cu2O content in the alloy (Cu-23Ni > Cu-44Ni > Cu-65Ni).In the case of the Cu-23Ni sample, the amount of Cu dissolved in (Ni,Cu)O (13 at.%) is lower than the one foreseen by the equilibrium.This is likely due to kinetic reasons, since the oxidation time (30 min) is probably not enough to complete the solubilisation of Cu ions in (Ni,Cu)O, leaving an amount of Cu2O larger than expected in equilibrium conditions.Additional porosity forms during reduction because of the contraction due to the different densities of f.c.c.Cu-Ni solid solutions (8.9 g/cm 3 ) and Cu-Ni-oxides (about 6.5 g/cm 3 ).
The morphology of the inner wall of the reduced microtubes depends on the Cu-content in the starting alloy and the relative amount of Cu2O in the partially oxidized wire.In the case of de-alloying of metallic alloys, interconnected porosity forms when the most noble element is below the "parting limit", so that, after removal of the less noble elements, the most noble element atoms reorganize themselves by surface diffusion forming hillocks separated by pores [36,37].In our case, the selective etching of a compound (i.e., Cu2O) in a corrosion resistant matrix, (Ni,Cu)O, is carried out.Only, in Cu-23Ni, the amount of Cu2O is larger than the one of (Ni,Cu)O (see Table 1) to allow the percolation of the etching agent through the (Ni,Cu)O matrix, leaving a continuous network of interconnected ligaments and pores, as shown in Figure 8c.In the case of Cu-44Ni, the presence of rounded metallic particles on the internal surface of the reduced microtube, Figure 8b suggests the occurrence of surface diffusion of the atoms in the (Ni,Cu)O matrix.However, the amount of Cu2O is lower than the one of (Cu,Ni)O (see Table 1), thus no interconnected ligaments and pores form.The morphology of the inner wall of the reduced microtubes depends on the Cu-content in the starting alloy and the relative amount of Cu 2 O in the partially oxidized wire.In the case of de-alloying of metallic alloys, interconnected porosity forms when the most noble element is below the "parting limit", so that, after removal of the less noble elements, the most noble element atoms reorganize themselves by surface diffusion forming hillocks separated by pores [36,37].In our case, the selective etching of a compound (i.e., Cu 2 O) in a corrosion resistant matrix, (Ni,Cu)O, is carried out.Only, in Cu-23Ni, the amount of Cu 2 O is larger than the one of (Ni,Cu)O (see Table 1) to allow the percolation of the etching agent through the (Ni,Cu)O matrix, leaving a continuous network of interconnected ligaments and pores, as shown in Figure 8c.In the case of Cu-44Ni, the presence of rounded metallic particles on the internal surface of the reduced microtube, Figure 8b suggests the occurrence of surface diffusion of the atoms in the (Ni,Cu)O matrix.However, the amount of Cu 2 O is lower than the one of (Cu,Ni)O (see Table 1), thus no interconnected ligaments and pores form.Finally, in the case of Cu-65Ni, the very low amount of Cu 2 O with respect to (Cu,Ni)O (below the detection limit of X-ray diffraction) is not enough to promote the reorganization of the atoms in the (Cu,Ni)O matrix by surface diffusion, leaving particles with irregular shapes, as shown in Figure 8a.

Conclusions
In this work, porous metallic microtubes were obtained by partial oxidation of Cu-Ni wires with different composition (Cu-23Ni, Cu-44Ni, Cu-65Ni), chemical etching of the partially oxidized wires and reduction of the oxide microtubes.
In all samples, after oxidation, the external layer consists of CuO and the internal oxide layer is a mixture of Cu 2 O and NiO.Only in the Cu-23Ni sample is a layer of pure Cu 2 O present between the CuO and the Cu 2 O/NiO layers.Furthermore, the amount of Cu 2 O increases as the amount of Cu in the starting alloy increases.
In the etching step, the unoxidized metallic core and Cu 2 O are selectively removed, leaving a porous oxide microtube, where the external and internal layers are CuO and NiO, respectively.
In the final step, the oxide microtube is reduced to a metallic microtube.In the case of Cu-44Ni and Cu-65Ni wires, the microtube section presents a continuous concentration gradient, which shows an increase of the Ni content from the outer wall towards the inner wall.In the case of the Cu-23Ni sample, two concentric and separated microtubes, respectively, rich in Cu and Ni, form because of the dissolution of the intermediate Cu 2 O layer.In this way, a microtube of almost pure Ni can be obtained starting from a Cu-rich alloy.
Finally, the porosity of the metallic microtubes depends on the amount of Cu 2 O formed during the oxidation step.Only in the case of the Cu-23Ni sample is the amount of the corrosion resistant NiO below the "parting limit", leading to the development of a continuous network of interconnected ligaments and pores by leaching of Cu 2 O.

Figure 1 .
Figure1.Oxide thickness as a function of the square root of oxidation time at 1173 K for Cu (black squares), Cu-23Ni (red circles), Cu-44Ni (green up triangles), Cu-65Ni (blue down triangles) and Ni (cyan diamonds) samples.In the case of pure elements, experimental points were fitted with a single parabolic law (continuous lines), x = K p t, where x, t and K p are the oxide thickness, the oxidation time and the parabolic rate constant, respectively.In the case of Cu-Ni alloys, experimental points show the presence of two steps (continuous and dotted lines, respectively).For Cu-23Ni and Cu-65Ni samples, each step was fitted with a parabolic law, leading to different values of the constant rate, K p and K p , as reported in Table1.

Figure 2 .Table 2 .aaa
Figure 2. X-ray diffraction (XRD) patterns of partially oxidized wires (a); oxide microtubes obtained after chemical etching of the partially oxidized wires (b); metallic microtubes obtained after reduction of the oxide microtubes (c) for Cu-65Ni, Cu-44Ni, Cu-23Ni and Cu samples.Black dots and red lines represent experimental points and refined patterns, respectively.Vertical lines at the bottom of each figure indicate the positions of the crystallographic reflections for the different phases.Table2.Quantitative phase analysis (wt.%) and corresponding lattice parameters of Cu-65Ni, Cu-44Ni, Cu-23Ni and Cu samples after oxidation, chemical etching and reduction, obtained by Rietveld refinement of the X-ray diffraction (XRD) patterns shown in Figure2.Sample ProcessStep

Figure 2 .Table 2 .
Figure 2. X-ray diffraction (XRD) patterns of partially oxidized wires (a); oxide microtubes obtained after chemical etching of the partially oxidized wires (b); metallic microtubes obtained after reduction of the oxide microtubes (c) for Cu-65Ni, Cu-44Ni, Cu-23Ni and Cu samples.Black dots and red lines represent experimental points and refined patterns, respectively.Vertical lines at the bottom of each figure indicate the positions of the crystallographic reflections for the different phases.

Figure
Figure4a-cshows magnified views of the cross sections reported in Figure3a-c, respectively, together with the corresponding EDX map.All the three partially oxidized alloys show a Cu-rich external oxide layer (red pixels) that corresponds to CuO which is detected in the corresponding XRD patterns.The intermediate oxide layers of the three samples show a different contrast of red and green pixels corresponding to Cu and Ni, respectively.In Cu-65Ni, an almost uniform green intermediate oxide layer is observed, Figure4a, suggesting the presence of almost only NiO, in accordance with the XRD results, Figure2a.In Cu-44Ni and Cu-23Ni, the intermediate oxide layer shows a mixture of green and red pixels, indicating the coexistence of NiO and Cu 2 O in agreement with the corresponding XRD patterns.In the case of Cu-65Ni, Figure4a, and Cu-44Ni, Figure4b, between the intermediate oxide layer and the unoxidized metallic core, there is a thin red ring (Cu-rich), which has a metallic appearance when observed at the optical microscope[23].The formation of this metallic copper layer occurs according to the reaction[29]:

Figure 4 .
Figure 4. Backscattered electron micrographs (left) and elemental maps (right) of the cross section of the oxidized Cu-65Ni (a); Cu-44Ni (b) and Cu-23Ni (c) wires.In the elemental maps, red and green dots indicate Cu and Ni, respectively.

Figure 4 .
Figure 4. Backscattered electron micrographs (left) and elemental maps (right) of the cross section of the oxidized Cu-65Ni (a); Cu-44Ni (b) and Cu-23Ni (c) wires.In the elemental maps, red and green dots indicate Cu and Ni, respectively.

Figure 5 .
Figure 5. Energy Dispersive X-ray analysis (EDX) composition profiles of Cu as a function of the distance from the outer surface for the partially oxidized wires (a) and the reduced metallic microtubes (b).Red circles, black triangles and green squares refer to Cu-65Ni, Cu-44Ni and Cu-23Ni samples, respectively.Dashed lines are a guide to the eye.

Figure 5 .
Figure 5. Energy Dispersive X-ray analysis (EDX) composition profiles of Cu as a function of the distance from the outer surface for the partially oxidized wires (a) and the reduced metallic microtubes (b).Red circles, black triangles and green squares refer to Cu-65Ni, Cu-44Ni and Cu-23Ni samples, respectively.Dashed lines are a guide to the eye.
Figure 2b shows the XRD patterns of the samples after etching with HNO 3 solution.The corresponding quantitative analysis of the crystalline phases is reported in Table 1.In the case of pure Cu sample, only the reflections of CuO are observed whereas the peaks of Cu 2 O disappear.The dissolution of Cu 2 O leads to the formation of an oxide microtube formed by the external layer of CuO, which is unaffected by the HNO 3 attack, Figure 3h.In the case of Cu-Ni samples, XRD patterns show the presence of CuO and NiO, while the peaks of Cu 2 O and residual f.c.c.solid solutions disappear.Thus, microtubes with two concentric layers of CuO (external) and (Cu,Ni)O (internal) are obtained (Figure 3e-g).

Figure 6 .
Figure 6.High Pressure Differential Scanning Calorimetry (HP-DSC) traces of the oxide microtubes under hydrogen atmosphere.

Figure 6 .
Figure 6.High Pressure Differential Scanning Calorimetry (HP-DSC) traces of the oxide microtubes under hydrogen atmosphere.

Figure 7 .
Figure 7. Secondary electrons image and EDX map of the Ni-rich microtube obtained after detachment of the external Cu-rich microtube.

Figure 7 .
Figure 7. Secondary electrons image and EDX map of the Ni-rich microtube obtained after detachment of the external Cu-rich microtube.

Metals 2017, 7 , 46 10 of 14 Figure 7 .
Figure 7. Secondary electrons image and EDX map of the Ni-rich microtube obtained after detachment of the external Cu-rich microtube.