Continuous Graphene-like Layers Formed on Copper Substrates by Graphene Oxide Self-Assembly and Reduction

Continuous Graphene-like Layers Formed on Copper Substrates by Graphene Oxide Self-Assembly and Reduction Alba Marı  a Ferna ndez-Sotillo and Paloma Ferreira-Aparicio* Cite This: ACS Appl. Energy Mater. 2020, 3, 10023−10036 Read Online ACCESS Metrics & More Article Recommendations * sıSupporting Information ABSTRACT: Copper substrates are known to be able to assist the spontaneous assembly and reduction of graphene oxide in liquid medium, but very scarce information is known about the involved mechanism. This study analyzes a number of relevant factors affecting this process: copper substrate geometry, concentration of graphene oxide dispersion, reaction time, presence of ligands in the reaction medium, and reaction temperature. Characterization techniques such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and attenuated total reflectance−Fourier transformed infrared (ATR− FTIR) have been applied to characterize the reduced graphene oxide (rGO) coating on copper. In the light of the obtained results, a reaction mechanism is proposed. Open-area copper substrates favor not only the complete coating of the copper surface but also the formation of self-standing continuous rGO films on open areas. Removal of copper ions, which are originated in the redox process, takes place by means of Cu2+-rich subsurface streams flowing underneath the rGO layers toward the aqueous medium. This simple and cost-effective procedure paves the way to design graphenelike bidimensional layered structures from the copper-induced self-organization of GO platelets as building blocks. KEYWORDS: self-assembled reduced graphene oxide, graphene oxide reduction mechanism, suspended reduced graphene oxide layers, copper oxidation, Cu2O shape control, subsurface interfacial ionic flow ■INTRODUCTION Graphene has enormous potential as a corrosion barrier and conductive coating. Numerous attempts to produce protective layers on metals have been reported in the last decade. 1−5 However, inconsistent results have been found. 6−8 Ideally, graphene can be considered as a perfectly flat and infinite twodimensional sheet formed by trigonally bonded sp2carbon atoms, 9 without basal plane fluctuations and edge states, but even for experiments under the most controlled conditions, some microscopic ripples are found. 10 In contrast to such ideal graphene, real samples unavoidably contain edges, suffer from basal plane fluctuations, and contain defects due to vacancies and heteroatoms 11 or adsorbed impurities, which lead to an inevitable alteration of their electronic structure and increased chemical reactivity. 12 Some experimentally observed graphene properties can be reasonably explained by the model that considers free-standing two-dimensional sheets, but there is some deviation from this ideal concept in many other experimental phenomena. The origin of these differences is the presence of defects produced during the synthesis procedure. Up to now, several strategies have been developed to synthesize graphene, which could be categorized into six main groups. 13 i. Micromechanical cleavage of graphite flakes using the “Scotch”tape method, which allows reliable and easy preparation of high structural and electronic quality graphene but suffers from low yield; ii. Epitaxial growth of graphene on single-crystal substrates at high temperatures and in ultrahigh vacuum, which can grow large-size and high-quality graphene, but requires high-cost fabrication systems and suffers from the difficulty in transferring the graphene from the substrates as well as low yield; iii. Thermalor plasma-enhanced chemical vapor deposition (CVD) of graphene from the decomposition of hydrocarbons at high temperatures on metal substrates (such as Ni, Cu, Pt) or metal oxides (Al2O3, MgO), which allows fast, uniform, large-area, high-quality graphene production, but its disadvantages are the high cost of manufacture and the relatively low yield, although it has great potential for further improvement; Received: July 20, 2020 Accepted: September 28, 2020 Published: September 28, 2020 Articlewww.acsaem.org © 2020 American Chemical Society 10023 https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 Downloaded via CIEMAT on October 28, 2020 at 11:44:36 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. iv. Chemical exfoliation of graphitic materials involving oxidation, intercalation, exfoliation, and/or reduction of graphene derivatives, such as graphite, graphite oxide, expandable graphite, carbon nanotubes, graphite fluoride, and graphite intercalation compounds. These methods can potentially afford bulk quantities of graphene, especially from graphene oxide (GO); v. Bottom-up synthesis strategies from organic compounds suited to synthesize nano/micrographene and graphenebased materials from structurally defined precursors. These approaches are focused to precisely control the formation of molecular graphene (<5 nm), nanographene (5−500 nm), and integrated macrographene (4500 nm) with well-defined structures and high processability; vi. Other less common methods (i.e., electrochemical exfoliation of graphite or graphene growth from directarc discharge of graphite), which have more drawbacks in terms of scalability and quality, despite their advantages. Among all of these procedures, exfoliation of graphite by oxidation followed by the subsequent reduction of the obtained graphite oxide has been demonstrated to be a primary low-cost strategy that can yield large quantities of reduced graphene oxide (rGO) with high processability. Chemical exfoliation yields heavily functionalized GO sheets, which consist partly of tetrahedrally bonded sp3carbon atoms. The presence of covalently bonded functional groups originates the deformation of the graphene oxide platelets, which are atomically rough 14 since some of their carbon atoms are displaced slightly above or below the graphene plane. 10 The most attractive property of GO is that it can be reduced, to a certain extent, to graphene-like sheets. The removal of oxygen-containing groups and the recovery of a conjugated structure yields reduced GO sheets, which can be considered as a kind of chemically derived graphene. Usually, this type of structure generated from GO has been named as functionalized graphene, chemically modified graphene, chemically converted graphene, or reduced graphene oxide (rGO). Residual functional groups and defects substantially modify the structure of the carbon plane. Accordingly, it should not be appropriate to refer to rGO simply as graphene since the properties can be substantially modified by the presence of defects or oxygenated groups. One of the most interesting properties of graphene oxide is that it can be used for applications based on solutionprocessable technology. 15−19 Graphene oxide can be easily reduced by powdered metals. 20,21 In addition, GO flakes have the ability to become self-assembled from solution on several reducible substrates. 22,23 A large variety of elements (Fe, Co, Ni, Cu, Zn, Al) are able to grow three-dimensional rGO films on its surface, but the solubility and stability of the metal ions formed and their oxides may interfere in the process. As a matter of fact, the resulting metal ions are usually intercalated and distributed within the stacked rGO layers. 22,24 This type of nanocomposites is recently attracting a great deal of attention. In particular, Cu2O structures have many interesting physical, chemical, electronic, and magnetic properties, with applicability in many fields (i.e., hydrogen production, solar cells, electrodes, antimicrobial coatings or filters, catalysts for organic reactions, biosensors, magnetic storage devices, gas sensors, and supercapacitors). 25−27 Its combination with reduced graphene oxide layers in Cu2O/rGO nanocomposites is currently the focus of a recent research field for the development of a new generation of devices, which can be improved by controlling and tailoring the structures of their components. 28−32 Self-organization of reduced GO platelets has been demonstrated on conductive substrates such as copper, as well as in many other active and inert metals, semiconducting Si, nonmetallic carbon, and even indium-tin oxide glass. Apart from the preliminary studies reported by Cao et al. 22 and Hu et al., 23 this method of spontaneous reduction and assembly under mild conditions has received up to now very scarce attention. Improvement of its potential requires further research to determine the mechanism of the process and the characteristics of the synthesized films. The present study analyzes the graphene oxide assembly process and its reduction to produce continuous rGO films on planar copper substrates with increasing open areas. Parameters such as assembly time, concentration of GO dispersion, and pH or temperature of the reaction medium are evaluated. Electron microscopy in scanning and transmission modes, energy-dispersive X-ray analysis, Fourier transform infrared spectroscopy−attenuated total reflectance (FTIR−ATR), and X-ray photoelectron spectroscopy have been applied to elucidate the rGO film formation mechanism. ■EXPERIMENTAL SECTION A commercial dispersion of GO purchased from Graphenea (4.0 mg· mL−1) was used for the rGO coatings. The pH value of the asreceived suspension was ca. 2.0. For the experiments, the 4.0 mg·mL−1 dispersion was diluted to 1.0 mg·mL−1(pH 2.4) or 0.5 mg·mL−1(pH 2.8). By considering the influence of pH on the reaction between copper and graphene oxide, the dispersion pH was modified in some cases by adding H2SO4and NH4OH as ligands. Four different copper substrates were used for GO self-assembly, i.e., a continuous copper film from 3 M, and three flattened copper grids from Dexmet with different open areas: 2Cu5.5-020FA, Cu540FA, and 3Cu7-100F. The nomenclature of the grids indicates in the first number the original material thickness, then the material symbol, followed by the strand width, and finally the long way of the diamond aperture, all of those expressed in 10−3inch units. Optical and scanning electron microscopy (SEM) images of the surface of the applied substrates are presented in Figure 1. For the synthesis, the copper substrates were immersed in GO dispersions of determined composition for a controlled period of time. The rGO self-assembly produced by interaction with metallic copper was analyzed as a function of substrate morphology, immersion time, initial pH value in the suspension, and the type and concentration of the added ligands. Figure 1. Copper substrates used for GO self-assembly: (a) smooth and flat copper film; (b) copper grid with small open area (2Cu5.5020FA), (c) copper grid with intermediate open area (Cu5-40FA), and (d) copper grid with large open area (3Cu7-100F). ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10024 The pH was measured in the bulk GO dispersion before and after the synthesis with a Crison pH meter equipped with a special probe for all kinds of medium and pH values in the range between 0 and 12. The synthesized samples were analyzed by scanning electron microscopy (SEM) using a Hitachi FE-SEM SU-6600. The precipitation of Cu2O crystallites beneath the rGO layers was observed in some cases favored by the reaction conditions. The average size of the Cu2O crystals was estimated by image analysis. Numerous individual particles of several representative SEM images of a given sample have been measured for each estimated value. By considering the structure of the crystallites (cubic or octahedral geometry), the Feret diameter (i.e., the normal distance between two parallel tangent planes touching the particle outline) has been applied for bidimensional measurements. 33 By considering the formation of unimodal particle size distributions in the analyzed samples, the estimated average particle size corresponds to the mean value. Some additional images of the coated samples were taken using an Olympus laser confocal microscope LEXT OLS5000. To analyze in detail the structure of the resulting rGO films, some experiments were performed on bare tabbed transmission electron microscopy (TEM) copper microgrids (square 100 mesh). The effects of immersion time and GO concentration on the resulting rGO coating were investigated. A JEOL JEM 2100 transmission electron microscope from the Spanish National Center for Electronic Microscopy was used for the analysis. X-ray diffraction (XRD) measurements were carried out in a PANalytical X’Pert PRO diffractometer operating with Bragg− Brentano geometry and using a Cu Kαradiation source. Data were collected in a 2θrange of 5−120°. Attenuated total reflectance−Fourier transform infrared (ATR− FTIR) spectroscopic analyses were performed on coated copper grids for surface characterization using an Agilent Cary 630 ATR−FTIR spectrometer equipped with a germanium crystal. X-ray photoelectron spectroscopy was carried out on coated grid samples using a scanning X-ray photoelectron spectroscopy (XPS) microprobe PHI Quantera II, using a monochromatic Al Kαsource. An Ar-ion gun was used for depth profiling. The representation of the graphene structures for the schemes has been produced using the Avogadro software tools. 34 ■RESULTS AND DISCUSSION The study of the interaction of GO with copper has been analyzed using different planar copper surfaces with increasing open areas. They have been immersed separately in aliquots of aqueous suspensions of GO for a defined period length. Different parameters have been varied to evaluate the progress and characteristics of the rGO coating on them depending on the following variables: substrate geometry, reaction time, GO concentration, ligands in solution, and reaction temperature. Effect of Substrate Geometry. The samples in Figure 1 were immersed for a period of 150 min in an aqueous dispersion containing 1 mg·mL−1of GO. After drying the samples in air at room temperature, they were examined. Figure 2 shows the SEM images of different samples at increasing magnifications. It can be observed that, in all cases, copper becomes oxidized, simultaneously promoting the assembly of GO flakes on its surface to form large sheets of reduced graphene oxide (rGO). The inhomogeneous appearance of continuous planar surfaces without openings gives clear evidence of a redox reaction in solution. The micrographs of the samples reveal the preferential formation and precipitation of Cu2O crystallites beneath the assembled rGO film on low open-area substrates (Figure 2a4,b4), as confirmed by XRD on them. In the case of open surfaces (grids), the amount of oxide on the surface decreases with the open area, and precipitated crystallites are preferentially accumulated in areas far from the borders of the substrate surface. By observing the samples at a high magnification, a very thin skin of graphene covering the substrates surface and the cupric oxide crystallites can be appreciated on them as a silky tissue (Figure 2a4−d4). It is worthy to note that in open-area substrates, GO flakes can get assembled even when they are not in intimate contact with the copper surface; the grid substrate is able to promote the lateral growth of rGO sheets beyond its surface, even over the holes. Thus, a large part of the open areas becomes covered by a thin film of chemically converted graphene (Figure 2b1,c1 and b2,c2,d2). In general, the resulting graphene film covering the surface and the voids is hard to be distinguished in the absence of defects, but some of the micrographs reveal their presence on the voids with foldings (Figure 2c2) or tears (Figure 2b2 and d2), or on ripples or wrinkles at the copper surface at a high magnification (Figure 2a4,b4,c4,d4). By considering the low pH levels in the reaction medium (ca. 2.4), formation of soluble cupric ions (Cu2+) would be expected since it is the preferential form of copper at pH values below 6. The cupric ion precipitates most frequently as copper hydroxide [Ksp Cu(OH)2= 1019.32] at intermediate pH levels (typically pH 6.5−12) and that its precipitation depends on variables such as copper concentration, presence of other anions and cations, temperature, and time to thermodynamic equilibrium. 35 A possible explanation for the Cu2O crystallization on the substrate at the metal−solution interface is that, as the rGO film grows and spans its surface over the substrate, this coating performs as an impermeable barrier for cupric ions. The diffusion limitations of soluble Cu2+ ions beneath the graphene layers toward the borders of the assembled sheets lead to an increase of their concentration at the Cu−rGO interface to reach the bulk solution, and it causes precipitation of copper oxides. According to this, the presence of large aggregates of oxides is observed on the continuous flat sample (Figure 2a1−a4), Figure 2. SEM images obtained at different magnifications corresponding to rGO films grown by GO spontaneous self-assembly and reduction on different copper substrates after immersion in a solution containing 1 mg GO·mL−1for 2.5 h; (a1−a4) continuous and flat copper surface from 3 M; (b1−b4) copper grid perforated with elongated punctures and a small open area (2Cu5.5-020FA from Dexmet); (c1−c4) copper-expanded grid with intermediate open area (Cu5-40FA from Dexmet); and (d1−d4) copper-expanded grid with large open area (3Cu7-100F from Dexmet) complementary images in the Supporting Information (Figures S1−S4). ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10025 more dispersed crystallites on the low open-area grid (Figure 2b1−b4), and very scarce amount of crystallites on the expanded grids with larger open areas (Figure 2c1−c4 and d1−d4). The accumulation of oxides in certain zones probably indicates where the rGO sheets have begun to grow and where cupric ions have enhanced limitations for diffusion toward the coverage edges. These results suggest that the release of copper ions to the aqueous dispersion is favored in copper substrates with larger open areas. Furthermore, large voids surrounded by copper can be covered by rGO films as a drum skin (see Figures S5 and S6 in the Supporting Information). Effect of Reaction Time. The spontaneous self-assembly and reduction of GO on copper grids has been explored for longer immersion periods using a dispersion of 0.5 mgGO· mL−1. The progress of the redox reaction for rGO formation is associated with the oxidation of the substrate to produce copper ions, which may remain solvated in solution, adsorbed to GO, or precipitated as copper oxides. To better analyze the evolution with time of the grown rGO film and the precipitation of oxides at the interphase between the copper surface and rGO coating, a substrate of intermediate open area was selected: the Cu5-40FA expanded grid. The observation by SEM of the synthesized samples with synthesis periods up to 53 h revealed that, as the reaction progresses, Cu2O crystallites grow progressively in size beneath the rGO film, as shown in the images in Figure 3a. The pH of the reaction medium, which was measured in the bulk solution before and after the sample immersion, was observed to increase with time progressively from its initial value of 2.8 up to 3.8 after 53 h of reaction. Despite the considerable increase of the pH value in the medium, the final measured value would not be high enough to induce the precipitation of copper oxides if certain homogeneity was considered in the environment of the reaction interface and the bulk solution. Therefore, the precipitation of the copper oxides beneath the rGOassembled films is indicative of a large local increase of concentration of copper ions, where diffusion toward the bulk solution is hindered by the graphene film barrier. The average size of the Cu2O crystallites has been estimated from their measurement in several images for each sample and is presented in Figure 3b as a function of reaction time. After 2 days in contact with the GO dispersion, cuprous oxide crystallites reach average sizes above 1 μm. After 72 h, the rGO film obtained becomes so thick that its analysis becomes difficult by SEM imaging. X-ray diffraction analysis was performed on this sample, which confirmed the presence of Cu2O and rGO on the copper grid (Figure 3c). An optical image of the coated grid is shown in the inset of Figure 3c. The thickness of the coating can be better appreciated in the topographic images taken with a laser confocal scanning microscope in Figure 3d,e, which allows a comparison between the coatings in samples synthesized for 22 and 72 h. After 72 h, the amount of stacked rGO layers yields a smoother surface, which only allows sensing the original geometry of the copper grid. According to these results, the progress of the reaction is associated with the oxidation of the substrate and a gradual increase of the thickness and darkness of the rGO film. The amount of copper oxide in this thick sample, which is completely covered by rGO, continues increasing beneath the graphene-like film, as inferred from the Cu2Odiffraction peaks in the diffractogram in Figure 3c. Some optical images obtained for the largest open-area expanded grid (3Cu7-100F) at different reaction periods are available in the Supporting Information (Figure S7). To gain insight into the characteristics of the free-standing rGO film, which covers the grid holes, bare TEM copper microgrids were used for the synthesis periods in the range of 15−120 min. In this case, a lower GO concentration of 0.25 mg·mL−1was used. Just 15 min of GO immersion was enough to obtain unsupported films on the voids, which grew in thickness with the elapsed time, as shown in Figure 4. It can be appreciated in the TEM images that some GO flakes remain trapped in between the grown rGO layers. Its presence contributes to darken the films and introduce in them additional defects, which may contribute to break thick films due to internal tension forces after drying (Figure 4c). Effect of GO Concentration. The synthesis of rGO films on bare TEM copper microgrids has revealed two different types of growth in open-area supports. They are illustrated in the images in Figure 5 with a GO concentration of 0.25 mg· mL−1: (i) extended rGO layers covering a major part of the holes, in which trapped GO flakes can be observed (Figure 5a) Figure 3. (a) SEM images at the same magnification showing the formation of Cu2O crystallites beneath the rGO coating, which grow simultaneously with the GO self-assembly and reduction, for increasing immersion periods on the Cu5-40FA grid; (b) estimation of the average crystallite size as a function of time on the surface; (c) X-ray diffraction analysis obtained for an rGO-coated Cu5-40FA grid for 72 h (inset: image showing the dark coating of the grid; a detailed image is available in Figure S9); (d, e) confocal laser microscope images of the rGO-coated samples for 22 h (d) and 72 h (e) showing their topography in an area of 0.4 mm2. GO dispersion concentration for the synthesis: 0.5 mg·mL−1. Figure 4. Transmission electron microscopy images of the unsupported rGO films grown in a copper microgrid for increasing period of time: (a) 15 min; (b) 60 min; (c) 120 min. GO concentration in the dispersion: 0.25 mg·mL−1. ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10026 and (ii) monolayers spanning adjacent strands of the microgrid free from isolated GO platelets (Figure 5b). The lateral assembly and reduction of GO platelets are probably initiated at the copper surface, but, as previously observed, the formed rGO film can be extended beyond the strands and over the holes. The reduction process is assisted by the oxidation of the copper surface and uses the rGO ability as an electronic conductor. If the growth rate of the extended film is similar at the different strands surrounding the void, the suspended film can be easily completed. By considering that this process may proceed simultaneously at both sides of the grid, some GO flakes floating in the medium remain trapped between the layers grown from both faces, as shown in the images in Figure 5a and in Figure 4. On the other hand, if rGO does not grow laterally at the same rate all around the hole, the film gets folded onto itself at the internal angles and wraps the grid strands up toward the opposite face of the grid. In this case, an interdigital membrane between strands is obtained. As a matter of fact, for the used concentration of 0.25 mg GO·mL−1, the first growth mode of planar films, which retains GO flakes inside, prevails over the interdigitated membrane mode. However, this latter way, which becomes predominant when the GO concentration decreases, seems to correspond to the preliminary stages of the growth. By reducing the GO concentration in the initial dispersion by 10 and 100 times to 0.025 and 0.0025 mg·mL−1, respectively, the interdigitated membrane growth mode is favored, as illustrated in Figure 6. However, the extension of the interdigitated membranes becomes reduced as the GO concentration decreases (see the micrograph in Figure 6b). It is worthy to note that the interaction between stacked rGO layers is very weak since each individual layer can be clearly distinguished at the folding zones, where they are separated by relatively large distances. Each grown rGO layer contributes to smooth the substrate roughness on its surface and reduces the amount of protrusions and sinks on it. This is also an evidence of a weak interaction between copper and rGO. By considering the observed growth pattern, it is possible that the successive stacking of folded layers spanning strands is the antecedent for the formation of planar films covering completely the holes. Effect of Ligands and pH on the Synthesis of rGO Coatings. Addition of Sulfate Ligand. The support with the largest open area (3Cu7-100F) was further analyzed by immersion in a GO dispersion with increasing amounts of sulfate ion in the range of 1 ×10−9to 4 ×10−9mol·L−1. The pH decrease was very small in these solutions (all of them in the range of 2.55−2.50). However, the effect of the sulfate ligand could be appreciated on the images of the four samples in Figure 7. Images denoted as “I”show the self-assembled rGO films spanning the gap between strands, those denoted as “II”are taken from the copper surface near the open areas, micrographs marked with “III”show the copper surface in the strand’s center, and “IV”denotes detailed images of the corrosion holes on copper, the oxide crystallites, and their coating. The main difference among them is more evident in images II, which show the presence of numerous holes due to enhanced dissolution of the copper substrates and are mainly localized on the edges near the hollow areas. By increasing the amount of SO42−ligand, the rGO formation seems to proceed faster, since the corrosion and precipitation of oxides are more evident. Images denoted as III, which show the accumulation of Cu2O crystallites following the grooves of the grid surface, do not present significant differences. These images (III in Figure 7) can be directly compared to the micrograph in Figure 2d3, prepared under identical conditions without the addition of SO42−ions. It becomes clear that the addition of Figure 5. Transmission electron microscopy images of the unsupported rGO film grown in a copper microgrid for a 15 min period immersed in a GO concentration of 0.25 mg·mL−1. Unsupported film assembly covering a large part of the grid hole and retaining GO flakes inside (a). Unsupported film assembly free from GO flakes and spanning two strands of the grid (b). Figure 6. Transmission electron microscopy images of the unsupported rGO films grown for 60 min in copper microgrids immersed in a GO dispersion of 0.25 mg·mL−1(a) and diluted by 10 (b) and 100 times (c), respectively. Figure 7. SEM micrographs of rGO-coated samples at different magnifications showing the effect of increasing addition of sulfate ligand on the formation of cuprous oxide crystallites in the 3Cu7100F expanded grid. Sulfate concentration in 1 mg·mL−1GO dispersion: (a) 10−9mol·L−1, (b) 2 ×10−9mol·L−1, and (c) 4 × 10−9mol·L−1. Immersion conditions: 2.5 h at room temperature. Images denoted as (I) view of the unsupported rGO film formed from the grid; (II) details of the copper surface near the open area; (III) details of the copper surface on the strand; (IV) closer details of the rGO film covering the grid surface, with holes and Cu2O crystallites. ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10027 sulfate ions promotes the precipitation of oxides on the center of the strands probably due to the increased corrosion. As a matter of fact, it has been suggested that SO42−ions in GO structures favor proton conductivity along the films, and especially for reduced graphene oxide, in which epoxide groups favor mixed proton−electric conduction. 36 The details of the grid borders in the images denoted as IV in Figure 7 reveal the accumulation of Cu2O crystallites on those zones, which are covered in all cases by extremely thin rGO-assembled layers that can be distinguished by the wrinkles in the micrographs. From the images, it can be deduced that an internal stream of ions flows through the interface in the direction of the plane between the copper surface and the rGO overlayer toward open areas or discontinuities in the coating. This is probably the way to decrease their concentration beneath the coating driven by the gradient with the bulk solution. Accordingly, a barrier effect seems to be originated by rGO for diffusion of ions. Their high concentration in subsurface streams is probably the origin of the precipitation of the oxides and the accumulation of Cu2O crystallites in flat zones of the strands. To get information about the flux of ions in rGO-coated grids, a second series of samples was prepared similarly to that in Figure 7. It is well known that hydrochloric acid solutions can effectively dissolve copper oxides by the formation of chloride complexes. 37 As a matter of fact, previous works by Hu et al. 23 have applied this strategy to remove copper oxides present on chemically converted graphene. Aiming to investigate the ion pathways for the removal of oxides from the copper surface, the as-synthesized samples (before drying) were immersed in a diluted hydrochloric acid solution (10−2 M) for 5 min. As shown in Figure 8, cubic Cu2O crystallites have been mostly removed after this treatment, but some hints reveal the pathways for copper ions toward the bulk solution. Mass transport limitations due to the reduced graphene oxide barrier have hindered their compete removal. A detailed analysis of the sample’s structure indicates that solvated Cu2+ ions seem to be removed from the Cu−rGO interphase through subsurface streams flowing toward the bulk solution in rGO ripped zones. Thus, branched recrystallization patterns due to the redissolution of Cu2O and recrystallization processes can be observed in different images: in between rGO layers (images a.I and b.III in Figure 8, a detailed image is available in the Supporting Information in Figure S8) and over the grid strand at the interface between copper and the rGO assembly (image c.III in Figure 8). The formation of these structures probably occurs as a result of the competition between different physical phenomena, i.e., diffusion, capillary forces (due to surface energy effects), and crystalline anisotropy. 38 Torn rGO layers detached from the coated surface after the hydrochloric acid treatment become usually wrapped, curled, or rolled up over itself, as shown in images a.II, b.I, bII, and c.II in Figure 8. The oxidation of the copper surface leads to clearly defined polyhedral holes in some places (Figure 8bII,cII insets). Energy-dispersive X-ray spectroscopy analyses were performed at different regions of these samples to estimate their composition. The webbing formed over the holes between strands is mainly formed by reduced graphene oxide with a C/ O atomic ratio in the range of 4.1−4.5. Small amounts of copper have been detected in the regions of the webbing with branched structures (0.3% Cu/16.9% O/82.8% C for the sample in Figure 8aI), whereas in clear or transparent zones of the unsupported rGO film of the sample, no copper has been detected (0% Cu, 18.1% O, and 81.9% C in Figure 8aI). The copper surface is always covered by rGO, resulting in higher amounts of carbon for samples produced with a higher amount of sulfate ligand, probably corresponding to a higher number of assembled layers over the copper surface. The average compositions in atomic percentage (% Cu, % O, % C) obtained for the samples in Figure 8a−c are (68.2/3.1/28.8), (62.8/1.3/35.9), and (5.7/4.9/89.4), respectively. In these regions, the C/O ratios (above 9) are far superior to those found for the rGO films grown on voids. This can be attributed to the presence of numerous GO flakes trapped in between extended rGO layers in unsupported films. Finally, the atomic composition of dendrites appearing on the strands after HCl treatment (Figure 8cIII) reveals the presence of the three components (Cu, O, and C) that could correspond to recrystallized monoclinic CuO and rGO (8.7/24.9/66.4). Modification of pH with Ammonium Ligand. The addition of NH4OH to the GO dispersion has been also analyzed. The (NH4)+ligand is expected to substantially increase the pH of the GO dispersion, contribute to the formation of soluble copper complexes [Cu(NH3)4]+, modify the copper corrosion rate, and consequently influence the rate for assembly and reduction of the GO flakes. To analyze the effects of the ammonium ligand on the precipitation of oxides and the growth rate of rGO layers on copper, a long synthesis period of 53 h was applied. A series of samples were prepared by immersion of the 3Cu7-100F expanded grid in aqueous GO dispersions with increasing aliquots of NH4OH. A sample prepared without ammonium hydroxide addition was taken as Figure 8. SEM micrographs showing the effect of the removal of copper oxide crystallites from rGO-coated 3Cu7-100F expanded grid by immersion in HCl solution (10−2M). Sulfate concentration in 1 mg·mL−1GO dispersion: (a) 10−9mol·L−1, (b) 2 ×10−9mol·L−1, and (c) 4 ×10−9mol·L−1. Immersion conditions: 2.5 h at room temperature. Images denoted as (I) view of the unsupported rGO film formed from the grid or general view of the grid; (II) details of the copper surface near the open area; (III) details of the accumulation of copper ions retained in the interface between the grid and the rGO coating; (IV) closer details of the rGO film covering the grid surface or the holes produced by copper dissolution. ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10028 reference. The pH values in the reaction medium were measured before and after the synthesis period. Figure 9 compiles the SEM images obtained for the different samples and the modification of the pH values after the synthesis. It can be clearly appreciated that, although basic solutions favor copper oxide precipitation, the addition of ammonia ligand drastically reduces the presence of precipitated copper. In the absence of ligands, the formation of large Cu2O crystals is observed after 53 h of synthesis (Figure 9a). It is worthy to note that cupric oxide deposits begin to grow as cubic crystals (see the inset in Figure 9aII). This indicates that simple shapes, with well-developed facets, are formed under or near equilibrium conditions and can achieve a minimum surface energy, as previously seen for samples synthesized for shorter periods (Figure 3a). However, when a certain size is reached (above 0.5 μm), larger crystals seem to grow under conditions far from equilibrium, since they show more complicated shapes, which are not necessarily the most stable in terms of surface energy (Figure 9a, general view in a.II and detailed view in a.IV). Factors such as mass transport and surface kinetics seem to be playing a major role in determining their shapes under the conditions originated by the formation of a Figure 9. Micrographs of the 3Cu7-100F expanded grid after 53 h of immersion in 0.5 mg·mL−1GO aqueous dispersion with increasing amounts of NH4OH solution 0.1 M: (a) no addition of ligands, (b) 0.4·mmol·L−1, (c) 0.8·mmol·L−1, (d) 1.2·mmol·L−1, (e) 1.4·mmol·L−1. Values of pH in the GO dispersion are indicated before and after the synthesis process. (I) Images of the grid strands and the open area; (II) details of the strand surface; (III) details of the rGO near the strand border and the open area; (IV) high-magnification image of the crystals or holes in the strand surface. Figure 10. SEM micrographs obtained for the 3Cu7-100F expanded grid after 24 h of immersion in 0.5 mg·mL−1GO aqueous dispersion at increasing temperature: (a) 30 °C, (b) 40 °C, (c) 50 °C. (I) General image of the grid after the synthesis process; (II) images of the unsupported film spanning the grid strands; (III) details of the strand surface. ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10029 diffusion barrier between the copper surface and the bulk solution. Mass transport-limited growth is possibly promoted in the samples in Figure 9a,b (images II and IV), where cupric ions are probably depleted at the top of the crystal due to the rGO layers above. Therefore, large crystals grow preferentially in planes oriented toward directions different from that of the rGO coating (Figure 9aIV). Polyhedron-shaped Cu2O crystals, whoseapexesprotrudeintothezonesofhigherions concentration, can be clearly observed. The formation of oxides is inhibited at higher pH values in the presence of the NH4+ligand (Figure 9c−e). The final pH in the solutions of the samples after 53 h tends, in general, to values close to 5.2, both for lower and higher pH values. For the samples synthesized in solutions with pH values above that (Figure 9d,e), some wrinkles on the reduced graphene oxide layer become apparent on the copper surface as channels for subsurface streams of copper ions flowing toward open areas to reach the bulk solution and equilibrate its concentration gradient. This is clearly appreciated in Figure 9 d(I, II, and III), and particularly in Figure 9eI−II. The formation of channels for subsurface solvated ions is so intense for the latter sample that the unsupported rGO film spanning the strands is plenty of ripples and folds (Figure 9eIII) and looks like a filamentous film (Figure 9eI). Temperature Rise Effect. The effect of temperature rise on the rGO coating on copper grids was also analyzed. By increasing the temperature of the dispersion medium, the assembly and reduction process of rGO is accelerated simultaneously with the oxidation of the copper surface. Thicker films of stacked rGO layers are able to grow over the open areas by covering completely a significant part of the grid holes at higher temperatures, as shown in Figure 10. The SEM micrographs in it show the rGO coating obtained after 24 h immersion in 1 mg·mL−1GO dispersions maintained at 30 °C (Figure 10a), 40 °C(Figure 10b), and 50 °C(Figure 10c). A general image of the samples and details of the unsupported films and the strand surface are shown for each sample. The increased solubility of copper ions at higher temperatures seems to favor their diffusion toward the bulk solution through channels formed at the interface between the copper surface and the rGO films, since lower amounts of oxides are detected underneath the rGO coating. Some precipitated oxides with flower-like structures, different from those previously observed, appear at 40 °C, as shown in Figure 10bIII. Dubey et al. have found this type of structures by modulating the processing parameters of chemical oxidation of copper surfaces. 39 They have observed that the alkalinity level in the solution determines the morphology of the copper structures formed from the oxidation process. At the same time, the increase of temperature accelerates the GO reduction and assembly. It can be appreciated in images a.I, b.I, and c.I in Figure 10. The thickness of the rGO film increases notably with the increase of temperature by 10 °C. The expansion of the assembled rGO layers spanning the strands and covering the holes takes place in a relatively short time. However, as previously shown, tears, ripples, and breaks are frequently produced in grown films during the drying process. The accumulation of copper ions and GO platelets in between the rGO layers are probably the origin of tensions and fractures. As a matter of fact, numerous GO flakes can be appreciated between the continuous rGO-assembled films in the micrograph in Figure 10cII. As the number of stacked layers increases, the consistency of the unsupported films is improved, but, in any case, copper ions find channels to leave the layered structure over the grid by opening holes or tears in the structure or moving toward the adjacent grid cell. In some cases, it has been also observed that the coating has been grown preferentially over one of the faces of the grid, and copper ions are released through the opposite face. Furthermore, the high rate of copper corrosion originates the aperture of flow channels in some regions or the trapping of those ions in between the layers of the film, which finally precipitate as oxides. The higher consistency of thicker films also enhances their fragility. Figure 11 shows the images of fractures on the coating for release of ions. Although numerous Cu2O crystallites appear over the copper substrate at the rGO-teared zones, the metal surface concentrates hollow areas originated from the Figure 11. SEM images of an rGO copper grid synthesized at 50 °C for 2.5 h showing the rGO break areas with accumulation of Cu2O crystallites. ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10030 release of copper ions. Previous studies have shown that oxide islands on the Cu surface usually follow a three-dimensional growth mode. The initial oxide growth happens through the formation of cuprous oxide islands, which eventually coalesce and grow in size. The kinetics of the oxide formation depends on the temperature and the copper surface, and it is dominated by direct impingement and oxygen surface diffusion in the first instance and after coalescence by oxygen diffusion into the metal bulk. 40−42 The removal of solvated copper ions by subsurface streams creates indentations or geometric profiles on the copper surface. Surface Functional Groups in Reduced Graphene Oxide on Copper Substrates. X-ray photoelectron spectroscopy (XPS) analysis was used to probe the functional groups on some of the samples. The spectra in the C 1s and O 1s regions obtained for an rGO-coated sample synthesized for 15 min with a 0.5 mg·mL−1GO dispersion at 25 °C are presented in Figure 12 in comparison to those obtained for dry selfsupported films prepared from the concentrated commercial GO dispersion used for the synthesis before (GO) and after reduction in an ascorbic acid solution (rGO-AA). The C 1s spectrum for the rGO-coated copper grid shows peaks at 284.8 and 285.7 eV, which could be assigned to sp2C atoms and some C−C (sp3) atoms in a defective structure of graphene, respectively. 43 A third peak at 288.2 eV can be assigned to C atoms in carbonyl groups, indicating that ca. 15% of the carbon atoms are oxidized. 44 The C 1s spectrum obtained from GO shows peaks at 284.7, 286.8, 288.4, and 289.7 eV, which were assigned to the nonoxidized ring CC atoms, C atoms bonded to hydroxyl/ epoxy/ether functional groups, C atoms in carbonyl groups, and carbon in carboxylic groups, respectively. 44,45 In this case, the percentage of oxidized carbon on the basis of the assignment is ca. 51%. A similar GO sample reduced with ascorbic acid (rGO-AA in Figure 12) shows some different features from the rGO film reduced in solution assisted by the copper substrate. The peaks assigned hydroxyl/epoxy/ether functional groups, and the C atoms in carboxyl groups become reduced, whereas the signal corresponding to the carbonyl groups remains practically invariable, accounting for ca. 9% of the total carbon atoms. For this latter sample, the percentage of oxidized carbon on the basis of the assignment is ca. 33%. These data indicate that the copper-assisted reduction and assembly of GO is much more effective than reductants such as ascorbic acid. The spectra in the O 1s region reveal some more differences between the samples. Although hydroxyl groups predominate in the original GO sample (signal at ca. 533.1 eV) and are accompanied by chemisorbed intercalated water molecules trapped between adjacent graphene oxide sheets (contribution at ca. 534.7 eV), 43,45,46 the main contribution to the O 1s is shifted to lower binding energy (ca. 532.6 eV) after reduction reaction with ascorbic acid, which could be assigned to the formation of ether-type bonds C−O−C. The evolution of GO after reduction and assembly on the copper substrate is different, since there is a large increase of carbonyl groups (signal at ca. 531.9 eV), which account for 73% of the total oxygen content. A smaller contribution at lower binding energies (ca. 529.7 eV), corresponding to a 15% of the total oxygen, could be ascribed to quinone functional groups or surface cupric oxide. To reveal subsurface information, depth profiling was applied using an Ar+-ion beam to etch surface layers. Before removing material from the sample, a spectrum from the surface of the sample was recorded. Then, XPS spectra were registered from the etched surface after successive ion gun etch cycles over a selected area of the sample. Figure 13 shows the XPS depth profiles obtained for the rGO-Cu sample in Figure 12. The surface composition of the pristine-coated grid in atomic percentage is 1.5% Cu, 79.2% C, and 19.3% O, which confirms that the whole surface was completely covered by rGO and that the small amounts of copper found at the surface were probably due to copper ions released from the substrate to the solution during the synthesis. The application of ion gun etching cycles reveals that as the depth increases, the amount of carbon becomes reduced and the copper content increases. The depth profiling reveals that the inner rGO layers mainly correspond to sp2domains. The main difference in composition in subsurface layers is observed in the oxygen functional groups. A progressive decrease of the signal at 531.9 eV due to carbonyl groups is followed by a progressive increase of the signal at 529.7 eV. Considering that the copper signal at ca. 923.7 eV clearly indicates the presence of copper in metallic state, the oxygen contribution at the lowest binding energy should be assigned to quinone groups in subsurface layers. Additionally, FTIR−ATR spectra were registered for an rGO-coated grid (3Cu7-100F) and for the dried film obtained Figure 12. XPS spectra in the C 1s and O 1s regions for the rGO film produced on the copper substrate by self-assembly and reduction of GO (Cu−rGO), and for self-supported films obtained from drying the same GO dispersion before (GO) and after reduction in an ascorbic acid solution (rGO-AA). Figure 13. XPS spectra in the C 1s, O 1s, and Cu 2p3/2 regions for the rGO film produced by self-assembly and reduction of GO on a copper substrate synthesized for 15 min. Sets of spectra indicating the number of Ar+-ion bombardment cycles (n) applied to the pristine sample (n= 0, solid black line); n= 1 cycle (blue dashed line); n=10 cycles (green dash dotted line); n= 20 cycles (orange dotted line). Synthesis conditions: 0.5 mg·mL−1GO dispersion for 15 min at 25 °C. ACS Applied Energy Materials www.acsaem.org Article https://dx.doi.org/10.1021/acsaem.0c01722 ACS Appl. Energy Mater. 2020, 3, 10023−10036 10031