Role of Metal Centers in Tuning the Electronic Properties of Graphene-Based Conductive Interfaces

A major bottleneck in the fabrication of efficient bio-organic nanoelectronic devices resides in the strong charge recombination that is present at the different interfaces forming the complex system. An efficient way to overcome this bottleneck is to add a self-assembled monolayer (SAM) of molecules between the biological material and the electrode that promotes an efficient direct electron transfer whilst minimising wasteful processes of charge recombination. In this work, the presence of a pyrene-nitrilotriacetic acid layer carrying different metal centers as SAM physisorbed on graphene is fully described by mean of electrochemical analysis, field emission scanning electron microscopy, photoelectrochemical characterisation and theoretical calculations. Our multidisciplinary study reveals that the metal center holds the key role for the efficient electron transfer at the interface. While Ni2+ is responsible for an electron transfer from SAM to graphene, Co2+ and Cu2+ force an opposite transfer, from graphene to SAM. Moreover, since Cu2+ inhibits the electron transfer due to a strong charge recombination, Co2+ seems the transition metal of choice for the efficient electron transfer.


Introduction
The creation of optically responsive materials is nowadays of high importance, since it allows to obtain a clean source of energy taking advantage of the conversion of light into electrons. Among the vast variety of methods developed over the years, the interfacing of light harvesting proteins (LHP) on a metal substrate is becoming more and more effective, thanks to technological advances in nanostructing the photoactive modules on the electrode surface. 1,2 In this devices, light is absorbed by the LHP, converted into electrons and transported to the electrode through the metal, in a variety of biotechnological applications, such as biosensors, biofuel cells, solar-to-fuel devices and biomolecular nanoelectronics. 3,4,5 Yet, despite the recent success in manufacturing such devices, a poor direct electron transfer (DET) is a major bottleneck, making the efficiency of such devices not competitive for practical applications. 6,7,8 The main factor responsible for a low DET is the charge recombination at the biological-metal interface, which strongly affect the overall performance of the devices. A second factor which impedes an efficient DET is the conformational flexibility of the LHP, which is essential in order to retain the protein function and ability. Thus, a deeper understanding of the transfer of charges at these critical interfaces in essential to overcome the DET efficiency limitations as well as the use of more complex interfaces in which at least three or more components are nanoengineered. To this end, a self-assembled monolayer (SAM) of well-ordered molecules is added to the LHP/metal interface to enhance the DET while maintaining the much-needed conformational flexibility to the protein. 9 Finally, the choice of the redox active metal center, that contributes to the final charge flow direction and efficiency, cannot be underestimated. Traditional metals used as electrode materials are gold or hematite, 10,11,12,13 but recently a new class of organic materials has been introduced as the next generation semimetal for bio-organic applications.
This semimetal of choice is graphene, a monolayer of sp 2 hybridized carbon atoms linked together to form a fully conjugated honeycomb lattice. 14,15 Its unique electronic properties, such as the linear dispersion of the valence and conduction bands at the high K symmetry point and the point 4 degeneration of the two bands, make graphene an ambipolar material, able to efficiently transfer either electrons or vacancies with measured mobilities for a suspended monolayer exceeding 50000 cm 2 V -1 s -1 in ambient conditions. 16 The tunability of charge carriers and an impressive mobility make graphene the material of choice to enhance the DET in bio-organic interfaces. Yet, being a metal, graphene lacks a bandgap, which is needed for operating the electronic device. Many strategies have been developed over the years in order to open a bandgap in graphene, and among them, the addition of functional groups by either covalent or non-covalent interactions is one of the most robust and reliable approaches. 17,18,19,20,21 One successful strategy to overcome the zero-bandgap problem is to build an interface in which the SAM is physisorbed on graphene, allowing for the opening of a gap in graphene and the increase of DET either from or to the graphene. Such a strategy provides fine tuning of the energy level of the frontier orbitals since the SAM-graphene interaction governs the charge flow direction. Moreover, if the SAM carries multiple functionalization groups, this 'orbital' effect can be strongly enhanced and finely tuned. The physisorption of SAM on the graphene surface allows the preservation of the high mobility of graphene by a non-covalent functionalization with molecules containing common πsystems, such as pyrene and its derivatives, to finally create a molecular interface between the functional molecules and the graphene surface. 17,18 As a consequence, on one hand the π-conjugated structure of graphene is preserved and a bandgap is opened but, on the other hand, the reversibility of the interaction can result in the desorption of the molecules from the graphene surface. Pyrenederivative molecules have been extensively used as good candidates for organic conductive interfaces containing redox active catalysts and photoactive enzymes, after their immobilization on graphene. 22 In this paper, we designed and studied, both theoretically and electrochemically, several types of single layer graphene (SLG) devices containing pyrene derivatives functionalized with a nitrilotriacetic acid group (NTA), coordinated with different divalent metal cations (M 2+ ): Ni 2+ , Co 2+ and Cu 2+ (SLG/pyrNTA-M 2+ , see Figure 1). The aim of our study was to determine the role of 5 different metal centers present within the pyrene SAM for an improved DET. Nowadays the most common metal center used for oriented binding of His6-tag engineered proteins is Ni 2+ . Nevertheless, considering an improved efficiency of DET within nanostructured bioelectronic devices, other metal centers, such as the neighbouring metals in the periodic table, should be explored. Yet, both Co 2+ and Cu 2+ cations present a radical character which might, in turn, strongly change the final DET output at the interface. Therefore, we first applied density functional theory (DFT) calculations to assess the change in the work function and the direction of the charge transfer (CT) flow depending on the different metal cation used. We then verified our theoretical CT models with electrochemical investigations of the constructed SLG/pyrNTA-M 2+ electrodes. We would like to emphasize that the final device obtained is a full-solid state device in which no solvent is present. Our DFT calculations showed that despite a similar positive shift of the graphene work function in the case of either Co or Cu redox centers, the DET from graphene to the SAM is enhanced only when the Co 2+ is considered, while a strong charge recombination is found for the Cu 2+ -containing interface. An opposite CT direction, i.e. from SAM to graphene, is observed when the Ni 2+ cation is coordinating the pyrNTA SAM. 23 The electrochemical data confirm the theoretical CT models and stress the importance of the imidazole molecule, used to complete the coordination sphere of the M 2+ center, as the main attenuator of charge flow.
A similar set up as reported in our previous study has been adopted. 23 The geometry optimization of the full, periodic interfaces has been performed at the DFT level of theory, with the PWscf package of the Quantum Espresso suite of programs. 24 PBE functional coupled with the vdw-DF2 term 25 for the exchange and correlation was used to account for van der Waals interactions together with ultrasoft pseudopotentials 26 with a cut-off of 50 and 200 Ry for the expansion of the wave function and density, respectively. Since PBC conditions are used, the final structure should be neutral; thus, to assure a null net charge, one carboxyl group has been protonated. As a consequence, the geometry of the system is now distorted and does not reflect the expected octahedral coordination, but a square planar geometry. Although this might not be the lowest energy conformation of the SAM, all three interfaces have been built in the same way, thus assuring consistency for the calculations. To consider the radical nature of the systems when Co and Cu are present, spin magnetization was included during the calculations. After optimization at the DFT level of theory with the PBE functional, the distance between pyrene and graphene was measured to be between 3.7 and 3.8 Å. The work function analysis has been carried considering the methodology reported in our previous study. 23 Briefly, the averaged electrostatic potential along the axis normal to the interface (i.e. z-axis) is used to directly estimate the work function shift, comparing the potential on the bare side and on the SAM-covered side of the surface. The total work function is casted down into two contributions: Where the molecular contribution (ΔΦSAM) arises from the dipole moment of the SAM backbone, and the charge transfer contribution (ΔΦCT) from the interfacial electronic reorganization upon 7 physisorption of the SAM on graphene. ΔΦSAM is obtained by computing the electrostatic potential profile across the molecules without graphene, while keeping the coordinates of the system frozen.
The ΔΦCT contribution is then the difference of the two terms. Only when a dipole moment is present in the SAM the shift in the work function is observed.
where Q is the electric charge obtained by integration of area under the reduction peak of CVs curves, n is the number of electrons transferred in the redox reaction (n=1) and F is the Faraday constant (9.648  10 4 C·mol -1 ), and A is the geometric area of the electrode (0.4185 cm 2 ).

Metal coordination alters charge transfer properties of the SLG/pyrNTA-M-IM assemblies
After the geometry optimization, the whole SAM structure assumes an elongated conformation on the graphene surface, with an all 'anti' conformation of the alkyl backbone, and with pyrene oriented on the graphene surface in a AA-stacking pattern at an optimized distance of 3.7 Å -3.8 Å (see Supporting Information for more details). The adsorption energy has been obtained as: where the ESLG and ESAM are the contributions of the two components of the system calculated at the Second, while for Co 2+ and Cu 2+ the ΔΦSAM and the ΔΦCT contribution act to enhance the total work function shift, the presence of Ni 2+ alters the trend with an opposite interaction, which is the main factor responsible of the smaller total calculated work function shift. cation, which is the key to quantify the contribution of ΔΦSAM to the total work function. In fact, although all three metal centers considered in the study are the first row transition metals and neighbours in the periodic table, Co 2+ and Cu 2+ are formally radicals, which may contribute to the strong difference in dipole moment calculated for each SAM structure used in this study. Table 1 summarizes the results of the work function analysis. The second important parameter to consider is the CT contribution ΔΦCT, which is substantial, and which has different sign for the different metal centers. As observed in our previous work, 23 were / is the charge density of the full system and and the charge densities on the two non-interacting fragments. We found an excess of electrons of -0.08 |e| on the SAM for both SLG/pyrNTA-Co-IM and SLG/pyrNTA-Cu-IM interfaces, while for the system with the Ni 2+ cation this value is reversed, with an excess of electrons (-0.05 |e|) on the graphene surface. Thus, when the coordination of Ni 2+ is considered, the charge flow is opposite, going from the SAM to graphene.
Thus, we suggest that the strong positive shift of the work function for both systems with Co 2+ and Cu 2+ coordination arises from the synergic interaction of the CT and the molecular backbone  Figure S4) also confirms the localization of the Fermi level and the VBM level over the NTA moiety of the SAM, with the CBM of the system is delocalized over the graphene layer. Interestingly, the LUMO of the SAM is localized over the pyrene moiety of the molecule. The strong different localization of this frontier levels is the main phenomenon responsible for the high CT observed for these interfaces.
A different scenario occurs when the SLG/pyrNTA-Ni-IM interface is considered (Figure 3). In this case the contribution at the Fermi energy level is due to the SAM only, while the CBM the contribution is mixed, arising from both SAM and graphene. In striking contrast with respect to the previous interfaces, the PDOS analysis reveals that the Ni 2+ cation is the main element responsible for the VBM and CBM peaks, together with a strong contribution of the carbon atoms. Again, the shape of the orbitals involved in these levels clarify the different behaviour of this interface. In fact, now the Fermi level (which correspond to the VBM), lying at -3.38 eV, is characterized by one orbital localized over the NTA moiety as well as over the Ni 2+ metal center, while for the conduction band the situation is completely different. A double contribution is now present, with the LUMO of the SAM lying at -1.81 eV and the CBM very close in energy, at -1.77 eV, but while the first is fully localized over the NTA and Ni 2+ parts of the SAM, the latter is delocalized over the graphene layer and the pyrene group of the SAM (Figure S4).  and green (c) spots represent the Co 2+ , Ni 2+ and Cu 2+ ions, respectively, present on the SLG electrode surface. The red (a,c) and green (b) spots represent the N atoms that are present within the moieties.
Right: Quantitative elemental analysis of each selected EDX maps and chemical structures of all the molecules forming SAMs on SLG.
The broad-scan XPS analysis of pyrNTA-M (Co-Ni-Cu)-functionalized SLG on the FTO surface is shown in Figure 5. The XPS spectra reveal the presence of carbon, nitrogen and cobalt in the SLG/pyrNTA-Co and the signals corresponding to C1s, N1s, Co2p3/2 were determined at 283.1 eV, 403.3 eV and 780 eV, respectively (Figure S5 a1, a2, a3). Carbon, nitrogen and nickel in the SLG/pyrNTA-Ni and the signals corresponding to C1s, N1s, main Ni2p3/2 were determined at 283.4 eV, 402.9 eV and 853 eV, respectively (Figure S5 b1, b2, b3). Finally, the presence of carbon, nitrogen and copper in the SLG/pyrNTA-Cu and the signals corresponding to C1s, N1s, main Cu2p3/2 were determined at 283.4 eV, 402.9 eV and 933.2 eV, respectively (Figure S5 c1, c2, c3). In details, C1s (sp2) contribution ( Figure S5 a1, b1, c1) is the main component for SLG and pyrene moieties, N is a main element in the NTA molecule and the presence of N1s additive is clearly shown in Figure S5 a2, b2, c2. The signals corresponding to Sn3d5/2 and O1s which originate mostly from FTO surface, were determined at around 484 eV and 531 eV, respectively.
Similar  values were reported for various (bio)organic and inorganic monolayers on various electrode materials including graphene. 28,29,30,31 19 The CV characterization clearly shows a well-defined pair of redox peaks for the SLG/pyrNTA-Cu electrode. These two redox peaks, appearing at around -0.05 V in the anodic and -0.3 V at the cathodic scan, correspond to reversible Cu II /Cu I redox couple, as shown previously. 32 The voltammetric profiles for the SLG/pyrNTA-Co and the SLG/pyrNTA-Ni systems also illustrate peaks characteristic for oxidation and reduction of particular active metal centres, albeit their definition is somewhat less defined compared to the Cu-based system. Nevertheless, the electrochemical signals for redox Co II /Co III couple appear at around -0.25V and -0.45V, respectively, as well as for Ni II /Ni III couple, at around -0.2 and -0.5V, respectively. In order to obtain a deeper insight into electrochemical properties of the redox pairs investigated in this study, a DPV analysis was carried out for each system ( Figure   6B). This type voltammetry is much more sensitive than CV and it is used in the present study to confirm the presence of a very low surface concentration of metal centers integrated with pyrene-NTA moiety. Indeed, differential pulse voltammograms showed well-defined redox active peaks obtained for the three analyzed electrodes that are derived from the respective metal centres coordinated to pyrene-NTA groups. Figure S6 presents the photocurrents recorded from all the three electrode configurations containing Cu 2+ , Co 2+ or Ni 2+ redox centers during prolonged illumination.
The data confirms that the photocurrents are rather stable up to at least 1 hour of continuous illumination. The SLG/pyrNTA-Cu electrode is generated the highest currents whilst similar smaller current values were obtained from the SLG/pyrNTA-Co and SLG/pyrNTA-Ni nanosystems. This data also confirms the CV and DPV data recorded for the three electrode nanoarchitectures (see Figure 6).

20
In order to experimentally verify our DFT modelling of the directionality of electron flow within the SLG/pyrNTA-M assemblies, photochronoamperometric measurements were conducted on all types of the functionalized SLG electrodes. Figure 7 presents the comparison of the current densities (j) at various potentials, obtained from the graphene monolayer modified with pyrNTA SAM coordinated with three distinct M 2+ cations, in the absence of (SLG/pyrNTA-M) ( Figure 7A) and presence of imidazole (SLG/pyrNTA-M-IM) ( Figure 7B). circuit potential (OCP), the highest current of 12.8 nA·cm -2 was obtained from the SLG/pyrNTA-Co electrode compared to SLG/pyrNTA-Ni (10 nA·cm -2 ) and SLG/pyrNTA-Cu (5.3 nA·cm -2 ) samples in the absence of imidazole ( Figure 7A). On the other hand, when the coordination sphere of the metal centers is completed with imidazole the trend is reversed for Ni and Cu ions, and now reads pyrNTA-Co-IM > SLG/pyrNTA-Cu-IM > SLG/pyrNTA-Ni-IM ( Figure 7B). The photocurrents were clearly enhanced when a more negative bias was applied. At -300 mV the current density value was 86.6 nA·cm -2 for the SLG/pyrNTA-Co electrode, which represents over 2.5-fold increase of the cathodic photocurrents compared to the SLG/pyrNTA systems coordinated with Ni 2+ and Cu 2+ cations, with values of 43.6 and 32.1 nA·cm -2 respectively.
The same mode of electron transfer, from graphene to the pyrNTA SAM, should be observed for the SLG/pyrNTA-Cu interface. Indeed, at an OCP as well as negative bias of -100 mV the currents generated from the Co-and Cu-containing samples were similar (see Figure 7A). However, the currents obtained for the Cu-containing assemblies were generally lower compared to the Cosamples, indicating the presence of charge recombination processes occurring within SAM-Cu, especially at a high negative bias (see Figure 7A).
Similar characteristics of photocurrent generation can be observed when the coordination sphere of the metal center is completed with imidazole. Overall, imidazole modification of SLG/pyrNTA films resulted in considerable lowering of the photocurrents for all the electrode assemblies ( Figure 7B), in agreement with our previous study. 23 At an OCP, the highest currents were observed for the samples with Co 2+ (4.8 nA·cm -2 ), the lowest ones for the assemblies with Ni 2+ (2.1 nA·cm -2 ) and an intermediate situation when Cu 2+ is present (3.1 nA·cm -2 ). Similarly, the highest cathodic photocurrents were recorded for the samples with Co 2+ (43 nA·cm -2 at -300 mV), and the lowest ones for Cu-containing samples (17.2 nA·cm -2 , see Figure 7B). On the other hand, the highest anodic photocurrent densities were recorded for the samples with Ni 2+ and imidazole (18.1 nA·cm -2 ), confirming that Ni 2+ cation promotes the charge transfer from SAM to graphene, as already observed in [23]. Nevertheless, the redox behaviour of the assemblies is more complex in the presence of Ni 2+ cation compared to the other metal centers investigated in this study, as this cation seems to promote generation of also cathodic currents at a negative bias of at least -200 mV (see Figure 7B). Moreover, a similar behaviour is observed for the Cu-and Co-containing interfaces, in which an enhancement of the anodic current is measured, up to 8.8 and 7.2 nA·cm -2 , which is in apparent contrast with the charge flow direction. Both behaviors can be explained by the unbalance between the DET and the overpotential applied, in which the last one overturns the (small) ground state charge transfer for the cathodic current and decreases the electron injection barrier by applying negative external field. For generation of the anodic currents, the situation is more complex, since the unexpected trend is observed only for Co-and Cu-containing interfaces; yet, the explanation holds true also in this case and smaller external electric field might be applied.

Conclusions
In this study we report the quantum mechanical and electrochemical characterizations of three various SLG assemblies functionalized with pyrNTA moiety conjugated with three different metal centers (M 2+ : Co 2+ , Ni 2+ , Cu 2+ ) exerting different effects on the Fermi energy levels of the conductive pyrNTA SAM. Both theoretical and electrochemical data confirm the strict relationship between the directionality of the charge flow between graphene and pyrNTA interface and the electronic properties of the coordinated metal center. In particular, we observed that the SLG/pyrNTA-Ni-IM interface favours the electron transfer from SAM to graphene, while the presence of the other two metal cations favours an opposite flux of electron, from graphene to SAM. In addition, the presence of Cu 2+ hampers the DET due to strong charge recombination at the interface, while Co 2+ seems an ideal metal of choice.
Our combined quantum and electrochemical data points towards the rational design for the optimized functionalized graphene photoelectrode, whereby careful selection of the metal center has the 23 profound effect on the preferred directionality of the electron flow. This study clearly shows that the most promising system for the pyrene-NTA-SLG photocathode assembly should incorporate Co 2+ cation as the ligand to NTA for the most efficient graphene-to-SAM charge transfer. On the other hand, for the opposite CT (from pyrNTA SAM to graphene) Ni 2+ cation should be utilized, since it promotes the highest anodic photocurrents in accordance with the electrochemical data and DFT calculations of the energy levels.
In summary, the present work paves the way for the optimal design of the highly oriented biophotoelectrode assemblies (incorporating His6-tag as the protein binding site), in which the directionality of the electron flow can be fine-tuned within the conductive interface composed of the pyrNTA moiety by introducing the specific metal redox center depending on the desired configuration of the electrode.

Supporting Information
Geometrical analysis of the three interfaces, Cartesian coordinates of the optimized systems at the DFT level of theory, shape of the frontier molecular orbitals and XPS spectra. The Supporting Information is available free of charge on the ACS Publications website.