High catalytic activity of oriented 2.0.0 copper(I) oxide grown on graphene film

Metal oxide nanoparticles supported on graphene exhibit high catalytic activity for oxidation, reduction and coupling reactions. Here we show that pyrolysis at 900 °C under inert atmosphere of copper(II) nitrate embedded in chitosan films affords 1.1.1 facet-oriented copper nanoplatelets supported on few-layered graphene. Oriented (1.1.1) copper nanoplatelets on graphene undergo spontaneous oxidation to render oriented (2.0.0) copper(I) oxide nanoplatelets on few-layered graphene. These films containing oriented copper(I) oxide exhibit as catalyst turnover numbers that can be three orders of magnitude higher for the Ullmann-type coupling, dehydrogenative coupling of dimethylphenylsilane with n-butanol and C–N cross-coupling than those of analogous unoriented graphene-supported copper(I) oxide nanoplatelets.

M etal nanoparticles (MNPs) supported on large surface area solids are widely used as heterogeneous catalysts for a large variety of organic reactions including reductions [1][2][3][4] , oxidations [5][6][7] and couplings 8-10 among other processes. One of the key issues in this area is how to control the preferential facets of the MNPs exposed to the reaction, since theoretical calculations as well as experimental data suggest that different crystallographic planes may exhibit specific activity in catalysis 11 . Among the numerous types of large area materials that have been used as supports, there are abundant data in the literature showing that graphene (G) as support of MNPs exhibits unique features not found in other solid supports 12,13 that make these graphene-based composites containing MNPs as suitable catalysts 14 and photocatalysts 15,16 . G being a one atom thick layer of sp 2 carbons in hexagonal arrangement constitutes the physical limit of two-dimensional materials in which all constitutive atoms are accessible to interact with substrates and reagents. Theoretical calculations predict that the overlap between the extended p orbital of G and the d orbitals of transition metals may result in a strong metal-support interaction that can modulate the electronic density at the MNPs-G interface [17][18][19][20] . In addition, the large surface area of G and its high adsorption capacity should cooperate to the reaction mechanism by adsorbing substrates near the active MNP sites. Both properties of G, its ability to interact with MNPs and its high adsorption capacity, are highly desirable to act as support for MNPs. In a related precedent to the present work, the formation of powders of 1.1.1 oriented CuPd nanoplatelets (NPs) on G has been reported starting from (Cu 2 þ -Pd 2 þ )-containing graphene oxide that was reduced at room temperature with NaBH 4 (ref. 21). The resulting oriented CuPd/G was used as catalyst for Pd-catalysed transformation of glycerol into lactic acid 21 . The same group has also reported the prior preparation of oriented 1.1.0 Cu 2 O NPs using sodium dodecylsulfate as template and subsequent adsorption of these preformed oriented Cu 2 O NPs on graphene oxide 22 .
Herein, we go forward in this direction showing a one-step synthesis of oriented (1.1.1) Cu NPs on few-layered graphene (fl À G) films as well as their selective oxidation to Cu(I) that becomes also facet oriented in the (2.0.0) plane. The key issue is the oriented growth of Cu NPs as G starts to develop in the one-step synthesis of the G-supported Cu NPs. Activity data show that as consequence of this preferential orientation these Cu 2 O-containing few-layers (fl) G films exhibit a remarkably high catalytic activity compared with related Cu catalysts with random crystalline orientation of Cu 2 O NPs for three coupling reactions.

Results
Synthesis of Cu=fl À G. Our preparation procedure for oriented Cu NPs on top of fl À G films (ðCuÞ=fl À G, Cu meaning 1.1.1 oriented Cu NPs) is inspired in the procedure of preparation of high electronic quality G films by chemical vapour deposition [23][24][25] . In this procedure of G preparation, a clean Cu or Ni metallic surface is used to grow G films by pyrolysis at temperatures about 1,500°C employing methane as carbon source in hydrogen atmosphere. It is accepted that under these conditions, methane is decomposed into hydrogen and carbon atoms that start to deposit on the metal surface 23,26 . The low solubility of C in Cu makes this metal especially suited for the preparation of G, since metal carbides are not formed [27][28][29] . Up to six carbon atoms can fit around a single metal atoms. As result, the atomically flat metal surface acts as template in the formation of G and each hexagon is formed crowning a metal atom of the surface. In chemical vapour deposition (CVD) studies, it has been proved that the 1.1.1 facet of Cu films is more suited to form high-quality G compared with the 1.0.0 facet that matches worse with the symmetry and dimensions of G (refs [30][31][32]. Our leading hypothesis is that the same principles should also apply to the reverse process and G sheets formed spontaneously in the pyrolysis of natural polysaccharides could drive the growth of Cu NPs to a preferential crystallographic plane. In the present case, preparation of films was performed by pyrolysis of Cu 2 þ -chitosan films supported on quartz as substrate (see experimental procedure for the preparation of Cu 2 þ -chitosan). Previously, we have found that films of fl À G can be obtained by pyrolysis under inert atmosphere and temperature about 1,000°C of films of few nanometres thickness of chitosan and other filmogenic natural biopolymers [33][34][35] . Chitosan is able to form continuous, crack-free nanometric films with subnanometric roughness on arbitrary substrates. Owing to the tendency of polysaccharides to render graphitic carbon residues, subsequent pyrolysis of these chitosan films renders G or fl À G films conformal to the substrate. In the present case, it was anticipated that, on pyrolysis, Cu 2 þcontaining chitosan films will form Cu NPs on G due to the spontaneous segregation of the two components during the pyrolysis. Chitosan and other polysaccharides are well known agents to trap metal ions in aqueous solution, including Cu 2 þ , and for this reason they are widely employed in water purification [36][37][38] . On the other hand, the low solubility of Cu and C and the lack of formation of the corresponding metal carbide will determine that Cu atoms present in the Cu 2 þchitosan film will segregate in a different phase as they become reduced to Cu NPs. In a prior study, pyrolysis of hybrid Ni 2 þ -Mn 4 þ hydrotalcite/alginate solids led to a spontaneous carbochemical reduction of Ni 2 þ metal ions resulting in the segregation of Ni NPs on the carbon residue 39 . In view of these precedents, we anticipated that nanometric films of chitosan embedding Cu 2 þ could lead on pyrolysis to G-supported Cu NPs. Also it was expected that in the process, fl À G film could act as template determining the preferential growth of a certain facet of Cu NPs as they are formed on G. Experimental evidence supports that our hypotheses were in general terms correct. Figure 1 shows the atomic force microscopy (AFM) images of the resulting Cu=fl À G obtained after pyrolysis.
One of the key points in the properties of the resulting Cu=fl À G is the high quality of the Cu 2 þ -chitosan film of nanometric thickness (filmogenecity). In these nanometric Cu 2 þ -chitosan films the presence of Cu 2 þ cannot be distinguished by electron microscopy or AFM, since Cu 2 þ ions should be highly dispersed, probably as individual ions, on the polysaccharide matrix by interaction with the amino groups of the polymeric fibrils. The presence of Cu 2 þ on chitosan can be, however, ascertained by analytical tools such as inductively coupled plasma optical emission spectrometry (ICP-OES) or X-ray photoelectron spectroscopy (XPS). Pyrolysis of chitosan films embedding Cu 2 þ ions results in a shrinking to less than one half of the initial film thickness as consequence of the formation and stacking of G sheets in the pyrolysis. According to these measurements, the Cu=fl À G is constituted by about 10 G sheets (4 nm height). On top of these G sheets, the presence of Cu particles can be observed (see Fig. 1). The subnanometric vertical resolution of the AFM equipment shows that Cu particles are flat and have a morphology as nanoplatelets, with an average high of about 3 nm and the frontal view shows that these thin Cu plates are homogenously distributed on top of the fl À G film.
Field emission scanning electron microscopy (FESEM) images of Cu-films are presented in Fig. 2. In agreement with AFM images, FESEM also confirms the high regularity of the morphology of Cu NPs, their uniform distribution and relatively small lateral particle size (from 5 to 20, average 8 nm, insets of Fig. 2), particularly considering the high preparation temperature (1,000°C). FESEM images for Cu=fl À G suggest a high crystallinity and ordering of the metal atoms in the Cu nanoplatelets. Unfortunately, the nanometric films corresponding to Figs 1 and 2 (panels a and b) do not exhibit X-ray diffraction (XRD), due to the low thickness and Cu content and size in these films. The regularity of Cu nanoplatelets and the type of crystallographic facet preferentially grown in the Cu NPs obtained by this procedure were determined indirectly by analysing the XRD pattern of analogous samples obtained by pyrolysis of chitosan embedding Cu 2 þ as thicker films. Pyrolysis of thicker Cu 2 þ -chitosan films (50 nm thickness in comparison with 10 nm for Cu=fl À G) render Cu=ml À G samples (ml: multi layer) with sufficiently high Cu content to allow recording XRD. Figure 3 shows a XRD pattern obtained for pyrolysed Cu 2 þ -chitosan films (50 nm thickness) as compared with the pattern of unoriented Cu NPs obtained by the polyol method 40 .
As it can be seen in Fig. 3, XRD of the pyrolysed Cu 2 þ -chitosan films exhibits almost exclusively a single peak corresponding to the 1.1.1 plane in contrast to the XRD of conventional Cu NPs prepared by the polyol method that can be indexed undisputedly to cubic Cu (JCPDS no. 04-0836). The XRD recorded for thick Cu/ml-G films constitutes firm evidence that Cu NPs exhibit oriented facets when grown following the chitosan pyrolysis method. We propose that the preferential growth of this crystal plane is due to the epitaxial growth of Cu NPs templated by evolving G sheets, in a similar way as the proposed mechanism to explain the formation of G by CVD on Cu films already commented in the introduction. Figure 4 summarizes our proposal. In support of this proposal, the evolution of Cu particles and G as a function of the pyrolysis temperature was followed by FESEM (see Supplementary Fig. 1). It was observed that fl À G starts to develop at temperatures about 800°C. At lower temperatures, no significant conductivity characteristic of G was observed and FESEM images were not possible or were of low quality. At 800°C the presence of some Cu particles is already observed, although their number and density on the surface are very low compared with the number of nanoplatelets formed at 1,000°C. The low density of Cu particles observed at 800°C could be probably due to the fact that most Cu 2 þ ions are still not reduced and are mostly embedded on the amorphous carbon evolving from chitosan fibrils. From 800 to 900°C, the degree of graphitization increases significantly as already reported and also the number of Cu particles that segregates increases considerably as the FESEM images show. The process seems to be complete at 1,000°C. Thus, according to these images evolution of Cu nanoplatelets and G formation are relatively synchronous, allowing the preferential growth of Cu facets due to templation by G. An additional point to consider with respect to Cu-G interaction is the relatively small particle size of Cu NPs taking into account the high pyrolysis temperature (1,000°C) and time (over 6 h) to which the samples are submitted. This relatively small Cu particle size suggests a certain control of particle growth by metal-support interaction as claimed in precedents in where an unexpectedly small particle size of metal oxides in the pyrolysis of precursors has been observed 41 .
XRD of freshly prepared thick Cu=ml À G samples correspond cubic Cu metal (JCPDS no. 04-0836). It is, however, frequently observed that Cu NPs undergo spontaneous oxidation on storage at the ambient. In the present case, the partial oxidation of Cu=fl À G films in contact with the atmosphere was determined by Raman spectroscopy. Figure 5 shows the Raman spectra of originally Cu=fl À G film that has been stored at the ambient and before its use as catalyst. Raman spectroscopy of the ambientexposed Cu=fl À G films allows detecting together with the G additional peaks at 850, 435 and 180 cm À 1 attributable to the presence of Cu 2 O based on abundant literature data 42,43 . It should be commented that Cu(0) cannot be detected by Raman spectroscopy.
This spontaneous oxidation of Cu(0) also occurs for Cu=ml À G for which the oxidation process can be followed by XRD (Fig. 6). It was observed that on oxidation of Cu=ml À G, the XRD of the film shows the formation of oriented Cu 2 O=ml À G as determined by the relative intensity of the 2.0.0 peak that is much higher than that of the 1.1.1 facet. The presence of small peaks corresponding to CuO was also detected, but their intensity was much smaller than those of oriented Cu 2 O (see Fig. 6). Even an oxidation treatment consisting in heating at 300°C under air flow for 1 h is not able to promote the complete oxidation of Cu(I) to Cu(II). One important observation was not only that Cu 2 O NPs were also oriented as the Cu NPs, but also that the process is reversible and reduction treatment by hydrogen restores from 2.0.0 oriented Cu 2 O the original Cu(0) NPs with preferential orientation as the initially formed NPs (Fig. 6 plot c). This cycling oxidation/reduction starting from oriented Cu=ml À G was performed twice observing that preferential orientation for the 1.  Thus, the information from vibrational spectroscopy for Cu=fl À G together with the XRD for Cu=ml À G confirm the changes in the oxidation state of Cu and the presence of oriented To provide direct evidence of any preferential morphology and facet orientation of Cu 2 O particles, Cu 2 O/fl À G films were studied directly by transmission electron microscopy (TEM). TEM imaging requires the prior detachment of Cu 2 O/fl À G from the quartz substrate without disturbing orientation of the particles, a process that is not obvious. To avoid any possible influence on the orientation of Cu 2 O particles supported on fl À G in the detachment from quartz, the substrate was submitted to consecutive mechanical polishing up to 100 mm thickness, dimpling grinding and argon ion milling until complete removal of quartz 44 . The resulting self-standing Cu 2 O=fl À G film was introduced directly in the TEM chamber without deposition on a holley copper grid. The images a-c presented in Fig. 7 and Supplementary Fig. 2 revealed that the Cu 2 O=fl À G sample is constituted also by nanoplatelets as in the case of Cu=fl À G. In addition, these Cu 2 O nanoplatelets present a preferential 2.0.0 facet orientation as in the case of thick Cu 2 O=ml À G films determined by XRD. Figure 7 shows three representative images to illustrate the preferential 2.0.0 facet orientation of Cu 2 O=fl À G as deduced by TEM. In panel d of XPS also shows the presence of Cu(I) and Cu(II) on the outermost layers of oriented Cu=fl À G and Cu 2 O=fl À G films (Fig. 8). The shallow penetration depth of XPS allows probing exclusively the outermost layers of the nanoplatelets and this technique in combination with the Auger peak reveals that the external layers of the nanoplatelets are constituted by Cu(I) and Cu(II), based on the observation of a major component at a binding energy of 933 eV contributing in a 0.3 atom%, together with the component corresponding to Cu(II) at 935 eV contributing in a 0.8 atom%. It should be noted that this atomic proportion corresponds to the shallow region probed by XPS. Importantly, the influence of the presence of Cu on G is revealed by the fact that, even though chitosan contains N and it has been reported that its pyrolysis renders N-doped G (refs 33,34), in the present case no peak corresponding to residual N is observed in the XPS. Moreover, the C1s peak in notably narrow centred at 284.5 eV, with components due to C atoms with dangling bonds at holes and a residual, minor population of C-O (o5 atom%), indicating that the quality of G is high. Thus, apparently, the presence of Cu nanoplatelets also influences G as support healing defects by removing completely N doping atoms from G and leaving a minor proportion of oxygenated functional groups.
Catalytic activity. As commented in the introduction, one of the main applications of MNPs supported on G is their use as catalysts. The ability to prepare oriented Cu 2 O nanoplatelets allows us to test the influence of facet orientation on the activity for certain reactions catalysed by Cu(I). To illustrate how preferential orientation can influence the catalytic activity, three different reaction types characteristic of Cu(I) sites, namely: (i) the Ullmann self coupling of iodobenzene (Eq. 1), (ii) the dehydrogenative coupling of dimethylphenylsilane with nbutanol to form the corresponding n-butoxysilane (Eq.  Figure 5 | Raman spectroscopy of Cu=fl À G films. Raman spectra of Cu=fl À G films supported in quartz substrates (2 Â 2 cm 2 ) immediately before (a) using them as catalyst for the Ullmann-type coupling of iodobenzene and after (b) being used in the reaction. The inset shows and expansion of the low wavenumber region that allows distinguishing specific vibrational peaks for Cu(I) and Cu(II). Comparison of the two spectra show that while the bands corresponding to G undergo minor alterations under the reaction conditions, the peaks due to Cu species disappear from the quartz as consequence of Cu 2 O=fl À G detachment from the quartz plate (see text for explanation). optical emission spectrometry analysis was 0.23±0.05 mg of Cu per 1 Â 1 cm 2 plate. The activity of these oriented Cu 2 O=fl À G films was compared with that of a synthesized Cu 2 O NPs (average particle size 5-7 nm) obtained by the ambient oxidation of Cu NPs prepared by the polyol method that have been supported on fl À G (Cu 2 O/fl À G). There are numerous reports in the literature showing that reduction of transition metals by thermal treatment in ethylene glycol renders MNPs of homogeneous particle size around 5 nm (refs [48][49][50][51][52]. This polyol method to obtain unoriented Cu NPs has been already employed for the preparation of MNPs supported on G and other carbon nanoforms and used as heterogeneous catalyst 53 . In the present case, Cu NPs supported on fl À G were prepared at 1 and 0.1 wt% loading. Characterization data of the fl À G sample used as support that is coincident with that previously reported in the literature is provided in Supplementary Fig. 3 (ref. 34). Importantly XRD of the as-prepared Cu/fl À G sample shows the presence of Cu without any preferential facet exposed (see Fig. 3a). Spontaneous oxidation of this sample by exposure to the ambient renders again Cu 2 O/fl À G lacking any preferential orientation in the NPs. The formation of Cu 2 O in Cu 2 O/fl À G is, therefore, analogous to Cu 2 O=fl À G. Note that some core of metallic Cu can remain in both oriented Cu 2 O=fl À G and unoriented Cu 2 O/fl À G samples. Besides this Cu 2 O/fl À G at two different loadings, a third Cu 2 O/fl À G sample was prepared by adsorbing commercial Cu 2 O NPs on fl À G at 0.1% loading (commercial Cu 2 O/fl À G). Cu 2 O/fl À G and commercial Cu 2 O/ fl À G are the analogous catalysts to determine the influence of particle orientation.
The Ullmann-type self coupling of iodobenzene was carried out at a Cu-to-substrate mol ratio of 2.24 Â 10 À 6 mol% for facetoriented Cu 2 O=fl À G catalyst and for the three Cu 2 O/fl À G analogues. Product formation at this low Cu/substrate mol ratio was only observed for Cu 2 O=fl À G. The results obtained are presented in Table 1.
Control experiments in the absence of any Cu or G under optimized conditions using KOCH 3 as base did not afford any product and the use as catalyst of fl À G in 1,4-dioxane and KOCH 3 as base barely afforded a minimal conversion (Table 1, entry 1). In contrast, the use of oriented Cu 2 O=fl À G films as catalyst and KOCH 3 as base in 1,4-dioxane afforded a substantial conversion of 26.4%, leading to a mixture of biphenyl (2) and o-and p-iodobiphenyl (3a,b). The use of dimethylsulphoxide as solvent and K 2 CO 3 , KOH or KOCH 3 as base was unsatisfactory resulting in no conversion (K 2 CO 3 and KOH) or 1 was converted in a small percentage (KOCH 3 ) without forming the expected 2. Importantly, when Cu 2 O/fl À G at 1 or 0.1 wt% loadings or commercial Cu 2 O/fl À G containing unoriented Cu 2 O NPs were used as catalyst at a Cu-to-substrate mol ratio of 1.8 Â 10 À 4 mol%, conversion of 1 was minimal or zero. XRD of commercial Cu 2 O/fl À G that was prepared by supporting commercial Cu 2 O in ethanol on fl À G is provided in Supplementary Fig. 4. When the Ullmann coupling of 1 was carried out with a Cu-to-1 mol ratio of 7.85 Â 10 À 2 mol%, then, a distribution of products similar to that obtained for fl À G in the absence of Cu (Table 1, entry 4) was observed. When the catalytic activity of Cu 2 O/fl À G ARTICLE with 0.1 wt% Cu 2 O loading was tested at 7.85 Â 10 À 3 Cu-to-1 mol%, then, conversion of 1 was also significantly decreased (Table 1, entry 5). All these catalytic data show that oriented Cu 2 O=fl À G is active at much lower Cu content than other analogous unoriented Cu 2 O=fl À G catalysts. Importantly, a turnover number (TON) value for iodobenzene conversion using oriented Cu 2 O=fl À G film of 1.45 Â 10 5 was estimated for the reaction with a minimal turnover frequency (TOF) value calculated at final reaction time of 6,100 h À 1 . These numbers are much higher than those determined for conventional, randomly oriented Cu 2 O/fl À G or commercial Cu 2 O/fl À G catalysts under the same conditions. For Cu 2 O/fl À G (1 wt% Cu 2 O loading) the estimated TON and TOF values were 75 and 3.6 h À 1 , respectively.
During the course of the reaction, it was observed that Cu 2 O=fl À G film detaches from the quartz substrate and appears as a self-standing film in the liquid phase. Accordingly, Raman spectroscopy after the reaction reveals that although G is still present on the quartz plate, the characteristic Raman peaks associated to the presence of Cu are absent in the quartz. This detachment probably reflects exfoliation of fl À G during the course of the reaction. However, if the detached Cu 2 O=fl À G film is recovered (now released from the quartz plates), it has been possible to perform a second use as catalyst of the Cu 2 O=fl À G film, reaching essentially similar values for 1 conversion and product distribution (Table 1, entry 3).
A plausible reaction mechanism for this iodobenzene coupling based on the knowledge of the presence of Cu(I) is proposed in Fig. 9. According to this proposal oxidative insertion of Cu(I) ions on the C-I bond will give rise to a phenyl copper and copper iodide species on the surface of Cu 2 O nanoplatelets. Phenyl copper could undergo homocoupling with a similar intermediate present of the surface of Cu 2 O NP to form 2 or, alternatively, may attack iodobenzene in the liquid phase forming positional isomers 3. The role of the base should be the removal of I À ions from the surface of Cu 2 O NPs. An important issue that remains open is the potential role of the electrical conductivity in the catalytic activity. In this regard, it is worth noting that the electrical conductivity of fl À G on quartz is 82 O& (ref. 34) and this conductivity is not much affected by the presence of oriented Cu or Cu 2 O.
To further demonstrate the superior catalytic activity as consequence of the preferential crystal orientation, Cu 2 O=fl À G films were also tested as catalyst for another typical Cu(I) catalytic reaction, namely the dehydrogenative silylation of n-butanol (see Eq. 2 in Table 1) 54,55 . It has been recently reported that G supported Cu nanoplatelets catalyze this dehydrogenative coupling reaction 14 . As in the previous case of the Ullmanntype reaction, it is of interest to determine whether or not oriented Cu 2 O NPs with preferential exposed 2.0.0 facets exhibit enhanced catalytic activity with respect to conventional randomly oriented Cu 2 O NPs supported on the same type of G. With this purpose in mind, the catalytic activity of two related samples with oriented or random NPs, namely Cu 2 O=fl À G and Cu 2 O/fl À G (1 or 0.1 wt% Cu 2 O loading), was compared. The dehydrogenative coupling was carried out at a Cu-to-substrate mol ratio of 2.5 Â 10 À 4 %.
As expected in view of the activity of Cu(I) for the dehydrogenative coupling 54,55 , both catalysts Cu 2 O=fl À G and Cu 2 O/fl À G promote the reaction with almost complete selectivity towards the expected coupling product 6 at low conversion (5%). A summary of the catalytic activity data are shown in Table 1. However, as conversion increased the selectivity towards the coupling product using the Cu 2 O=fl À G as catalyst decreased gradually up to 60 at 34% of conversion, a fact that is not observed in the case of Cu 2 O/fl À G. This selectivity decrease is due to the appearance of the corresponding disiloxane, whose formation from silane 4 requires the reaction with oxygen gaining access from the ambient and this is generally observed for slow reactions (note differences in times in Table 1). As previously indicated for the Ullmann-type coupling, also in this case, detachment of the Cu 2 O=fl À G film from the quartz plate during the course of the reaction was observed.
Control experiments using as catalysts fl À G or Cu 2 O NPs obtained by the polyol method adsorbed on fl À G at the low concentrations corresponding to the total amount of Cu present in the two 1 Â 1 cm 2 quartz plates (0.46 mg of Cu) showed no conversion of compound 4, indicating again the high catalytic activity of oriented Cu 2 O=fl À G. The TOF of Cu 2 O=fl À G measured at 1 h reaction time gives a value of 22,700 h À 1 while the TOF value of Cu 2 O/fl À G under similar conditions was 1.4 h À 1 . No significant influence of the Cu 2 O loading on fl À G, either 1 or 0.1 wt%, on the TON was observed (compare entries 6 and 7 in Table 1). This remarkable difference in the TOF value is a consequence of the minute amount of Cu present on the quartz film (0.46 mg of Cu) as compared with the amount of Cu present of Cu 2 O/fl À G 160.000 mg, while still being able to convert 4. We propose that this very high TOF value for Cu 2 O=fl À G films reflects the high catalytic activity of the material due to the preferentially exposed catalytically more active (2.0.0) facet present in Cu 2 O=fl À G.
The third coupling reaction that was tested was C-N coupling of aniline (7) and bromobenzene (see Eq. 3 in Table 1). This coupling has also been reported to be promoted by Cu(I) sites in the presence of strong bases 56,57 . The results presented in Table 1 show that Cu 2 O=fl À G is more efficient than Cu 2 O/fl À G (either at 1 or 0.1 wt% Cu 2 O loading) or commercial Cu 2 O/fl À G promoting C-N coupling. The estimated TON value for the disappearance of 7 using Cu 2 O=fl À G as catalyst was 2.76 Â 10 5 . The estimated TOF value at 5 h reaction time is 55,200 h À 1 . Moreover, in the last case, the high activity of Cu 2 O=fl À G is also reflected in the selective formation of the product of the exhaustive double C-N arylation triphenylamine (10). In addition, Cu 2 O=fl À G was reusable under the reaction conditions. Although unoriented Cu 2 O/fl À G also exhibits activity for C-N coupling, the TON value considering 7 disappearance (TON of Cu 2 O/fl À G at 1 wt% Cu 2 O loading 369) is about 750 times lower than that of oriented Cu 2 O=fl À G or more than three orders of magnitude lower (TON of 74 for formation of compound 10) if formation of product 10 is considered. Furthermore, as in the case of the Ullmann coupling, the catalytic activity of Cu 2 O/fl À G at 0.1 wt% Cu 2 O loading decreases almost proportionally with the diminution of the Cu 2 O content ( Table 1, entry 4 in Eq. 3), resulting in no significant influence of the TON as a function of the Cu 2 O content. Overall, the results shown in Table 1 indicate again that the preferential 2.0.0 facet orientation of Cu 2 O NPs increases the catalytic activity of Cu 2 O=fl À G for this reaction.

Discussion
In the present work it has been shown that pyrolysis of Cu(NO 3 ) 2 embedded in chitosan films forms oriented Cu=fl À G constituted by 3 nm height Cu nanoplatelets with preferential (1.1.1) facets. We propose that this preferential growth is a result of the epitaxial templation of G on the nascent Cu nanoplatelets during phase segregation occurring at temperatures higher than 800°C. Cu=fl À G undergoes spontaneous oxidation on exposure to the ambient to render oriented Cu 2 O=fl À G constituted by Cu 2 O nanoplatelets with preferential 2.0.0 facets (about 82%), as determined by comparison with the raw TEM image with the image showing exclusively 2.0.0 oriented nanoplatelets. Activity data of Cu 2 O=fl À G for three typical Cu(I) catalysed couplings show that the intrinsic activity of oriented Cu 2 O nanoplatelets is about four orders of magnitude more active than analogous catalyst containing small Cu 2 O NPs on fl À G (1,933, 68,000 and 3,730 times for the Ullmann-like reaction, dehydrogenative silane coupling with alcohols and C-N coupling, respectively). It is proposed that this higher activity is a reflection of the intrinsic catalytic activity oriented Cu 2 O nanoplatelets. Further work in progress is aimed at showing the general scope of this procedure for the templation of other MNPs on fl À G. Activity data of other MNP supported on fl À G should allow disclosure of the specific catalytic activity of some preferred facets in oriented MNPs for other reactions.

Methods
Synthesis of fl À G. Alginic acid sodium salt from brown algae (Sigma) was pyrolysed under argon atmosphere using the following oven program: annealing at 200°C for 2 h and, then, heating at 10°C min À 1 up to 900°C for 6 h. The resulting graphitic powder was sonicated at 700 W for 1 h in water and the residue removed by centrifugation to obtain fl À G dispersed in water.
Cu NPs deposition on fl À G. fl À G from alginate pyrolysis (100 mg) was added to ethylene glycol (40 ml) and the mixture was sonicated at 700 W for 1 h to obtain dispersed fl À G. CuCl 2 (10.6 or 1.06 mg for the preparation of the sample at 1 or 0.1 wt% Cu 2 O) was added to the reaction mixture and Cu metal reduction was then performed at 120°C for 24 h under continuous stirring. The Cu/fl À G were finally separated by filtration and washed exhaustively with water and with acetone. The resulting material was dried in a vacuum desiccator at 110°C to remove the remaining water.
Preparation of commercial Cu 2 O/fl À G (0.1 wt%) was carried out by dispersing 60 mg of G from pyrolysis of alginate into 60 ml of ethanol using a ultrasound source (tip of 700 W) during 1 h and then addition of 4.5 ml of commercial ethanolic suspension of Cu 2 O (Aldrich, Ref: 678945, 1.5% (w/v)) and the mixture stirred for 12 h to achieve deposition of the nanoparticles on fl À G.
Synthesis of oriented Cu NPs over fl À G films (Cu=fl À G). Chitosan (0.5 g) from Aldrich (low molecular weight) was dissolved in a copper(II) nitrate aqueous solution (18 mg of Cu(NO 3 ) 2 Á 2½ H 2 O in 25 ml of water). A small quantity of acetic acid (0.23 g) is necessary for complete dissolution of chitosan. The solution was filtered through syringe of 0.45 mm diameter pore to remove impurities present in commercial chitosan. The films were supported on a quartz plate (2 Â 2 cm 2 ) by casting 300 ml of filtered solution at 6,000 r.p.m. during 1 min. The pyrolysis was performed under argon atmosphere using the following oven program: heating rate at 5°C min À 1 up to 900°C for 2 h. The amount of copper present on the films was determined by inductively coupled plasma optical emission spectrometry (ICP-OES) by immersing the plates into aqua regia at room temperature for 3 h and analysing the Cu content of the resulting solution.
Ullmann-like reaction procedure. All reagents were purchased from Sigma-Aldrich and used as received without any further purification. To a solution of iodobenzene, 1 (2.0 mmol) in 4 ml of solvent (1,4-dioxane or dimethylsulphoxide) 2 mmol of base (KOCH 3 , NaOH or Na 2 CO 3 ) and catalyst were added. The resulting mixture was stirred in an autoclave for 24 h at 160°C. After the reaction, the catalyst was collected by filtration and the reaction products were analysed and identified by gas chromatography mass spectrometry (GC-MS; THERMO Electron Corporation instrument), Trace GC Ultra and DSQ, TraceGOLD: TG-5SilMS column with the following specifications: 30 m Â 0.25 mm Â 0.25 mm, working with a temperature program that starts at 50°C maintained for 2 min and afterwards increasing the temperature at a rate of 10°C min À 1 up to 250°C that was maintained for 10.00 min, resulting in a total run time of 32 min. The pressure of He used as the carrier gas was 0.38 Torr. Mass spectra of the products were acquired at 70,000 resolution. Biphenyl (2) Dehydrogenative silylation reaction procedure. The catalyst Cu 2 O=fl À G (2 plates of 1 Â 1 cm 2 , load 2.5 Â 10 À 4 mol% Cu versus substrate) prepared no longer than 2 days before the reaction and stored at the ambient was introduced into a 5 ml reinforced glass reactor equipped with a magnetic bar. Then, n-butanol was added (10.9 mmol) under dry N 2 atmosphere and the system purged for 15 min with N 2 . Finally, 4 (3.27 mmol) was introduced into the reactor with a syringe. The reaction mixture was stirred at 110°C in an oil bath. At the final time, the system was allowed to cool to room temperature. The reaction was carried out in triplicate using three Cu 2 O=fl À G samples prepared independently, obtaining consistent results with a deviation o8%.
In the case of powdered, unoriented Cu 2 O/fl À G, the catalyst was added in a 5 ml reinforced glass reactor equipped with a magnetic bar. The reactor was purged by N 2 and the alcohol was added (10 mmol). Under inert atmosphere of N 2 the catalyst was sonicated for a 1 h and, then, introduced in a preheated oil bath. Finally, dimethylphenylsilane 4 was introduced in the reactor with a syringe. The reaction was stirred at 110°C. At the final reaction time the reaction mixture was allowed to cool to room temperature and the catalyst removed by filtration. In both cases (plates and powders), a known amount of n-dodecane was added as internal standard and the reaction was followed by monitoring periodically the reaction mixture by GC. 1 H-NMR and MS spectra and analytical (GC retention time) data of dimethylphenylbutoxysilane were coincident with those reported in the literature 14 .
General procedure for the C-N coupling. To a solution of bromobenzene (1.2 mmol) and aniline (1 mmol) in 4 ml of 1,4-dioxane, potassium tert-butoxide (2.1 mmol) was added as base and two pieces of 1 Â 1 cm 2 of Cu 2 O=fl À G on quartz (0.24 mg of Cu total) or unoriented Cu 2 O/fl À G as powder (10 mg for solid catalyst containing 1 or 0.1 wt% Cu). The resulting mixture was submitted to mechanical stirring in an autoclave for 24 h at 200°C. The products were analysed and identified by using GC-MS (THERMO Electron Corporation instrument).
Physicochemical characterization. Powder XRD patterns were recorded on a Schimadzu XRD-7000 diffractometer using Cu Ka radiation (l ¼ 1.5418 Å, 40 kV, 40 mA) at a scanning speed of 0.20°per min in the 10-80°2Y range for the ex situ experiments. The in situ experiments were carried out in the 6-60°2Y range, with a 10°C min À 1 heating rate, and a flow of hydrogen or air of 10 ml min À 1 . On each plateau, the temperature was kept for 30 min before the acquisition of the diffractogram.
Raman spectra were collected with a Horiba Jobin Yvon-Labram HR UV-Visible-NIR (200-1,600 nm) Raman Microscope Spectrometer, using a laser with the wavelength of 632 nm. The spectra were collected from 10 scans at a resolution of 2 cm À 1 .
TEM images of an oriented Cu 2 O=fl À G sample were recorded at the Electron Microscopy Center of the Universitat de Valencia after abrasion of the quartz support by consecutive treatments consisting in mechanical polishing from the backside of the substrate until B100 mm thickness, followed by backside dimpling with a dimple grinder GATAN Model 656 and final low-angle, ion milling using an argon gun and plishing system Fishione Model 1010. Ref. 44 provides the fundamentals and detailed description of the methodology.