Experimental and Theoretical Studies on Corrosion Inhibition of Niobium and Tantalum Surfaces by Carboxylated Graphene Oxide

The corrosion of two different metals, niobium and tantalum, in aqueous sulfuric acid solution has been studied in the presence and absence of carboxylated graphene oxide. Potentiodynamic measurements indicate that this nanomaterial inhibits corrosion due to its adsorption on the metal surfaces. The adsorbed layer of carboxylated graphene hinders two electrochemical reactions: the oxidation of the metal and the transport of metal ions from the metal to the solution but also hydrogen evolution reaction by acting as a protective barrier. The adsorption behavior at the molecular level of the carboxylated graphene oxide with respect to Nb, NbO, Ta, and TaO (111) surfaces is also investigated using Molecular Dynamic and Monte Carlo calculations.


Introduction
In general, metals are prone to oxidation when exposed to different aggressive solutions in the course of industrial cleaning or other diverse treatments (acid pickling, descaling and cleaning). Although the use of paints and polymers offers a simple approach to decrease the effect of corroding species on the surface, the major drawbacks of this method are the weak interactions among the metal surface and the coating layer so the physical organic barrier is deteriorated after a short time of exposure [1]. The widespread method of the protection of metals against corrosion remains the use of inhibitors molecules [2][3][4]. These moieties when added at relatively small concentrations in the corrosive media where the metal is exposed can interact with its surface, forming an organic coating and therefore decreasing the corrosion rate. Other approaches for corrosion inhibition rely on the use of surface modification strategies through the formation SAM's (Self Assembled Monolayers) using phosphonic acids [5,6], silanes, aryl diazonium salts [7], etc. Graphene is considered as one of the most important classes of modern materials-nanomaterials [8], which are extensively explored for use in different fields of science from drug delivery [9] to photovoltaics [10]; however, graphene remains almost unexplored towards applications aimed at the corrosion inhibition of materials. Only limited scientific research that uses graphene nanomaterials is devoted to corrosion studies [11][12][13], and to our knowledge there is no reported study on the use of carboxylated graphene oxide (GO-COOH) as a corrosion inhibitor. The production of nanomaterials is steadily increasing [14,15] using low-cost, abundant natural materials [16]; therefore, the future use of nanomaterials in corrosion protection is promising.
Quantum chemical methods are widely employed for the study of the physico-chemical properties of the inhibitors in order to understand at the molecular level their interaction with the metal surface [1,4,17,18]. Density Functional Theory (DFT), as a powerful technique for performing calculations on many bodied systems, permits the correlation between the adsorption ability of the inhibitor molecule on the metal surface and its electron donating tendency that is expressed by the value of E HOMO . Higher values for this orbital suggest an increasing donor-acceptor interaction between the inhibitor molecule and the vacant d-orbitals of the metal surface atoms [19,20]. DFT calculations also make possible the estimation of other parameters like the dipole moment, electronegativity and ionization potential energy [21]. For example, the inhibitor efficiency increases when the metal surface is covered with higher dipole moment molecules due to the electrostatic interaction between the charged metal surface and charged centers of molecules [22]. Molecular dynamics simulation provides more insight into the corrosion inhibition mechanism and the different interactions at the interface inhibitor-metal surface [23]. However, for larger systems, these methods are computationally costly to employ.
Sulphuric acid 97%, pro analysis, MERCK, UN-No.1830 (Darmstadt, Germany) is used as the electrolyte for the potentiodynamic measurements. Its concentration in whole measurements was 0.1 mol/dm 3 .
The electrodes were made of commercial niobium (d = 1 mm) and tantalum (d = 0.5 mm) wires. Prior to electrochemical measurements, metal surfaces were mechanically polished with emery paper, cleaned with distilled water and degreased in ethanol, washed with distilled water and finally dried in air. The electrochemical studies and linear sweep voltammetry (LSV) were performed using an Autolab potentiostat (Metrohm Autolab, Utrecht, Netherland) and employing a three-electrode cell assembly. A saturated calomel electrode (SCE) and platinum electrode were used as reference and auxiliary electrodes, respectively. All solutions were prepared from analytic grade chemicals and bidistilled water. The metal electrodes were allowed to stabilize their open circuit potential (OCP) until the potential stabilization criteria of dE/dt limit 10 −6 V/s is reached. Potentiodynamic polarization was carried out by scanning the potential ±250 mV from the evaluated OCP using a scan rate of 1 mV s −1 . Polarization experiments were carried out using a potentiostat from Eco Chemie BV, the Netherlands, Autolab PGSTAT128N (Metrohm, Utrecht, The Netherland), and the obtained data were analyzed with Autolab software (NOVA 2.0.1, Metrohm, Utrecht, Netherland).

Molecular Dynamic (MD) Simulations
The MD simulations with the Forcite Module (Materials Studio) were performed using Universal Forcefield. Prior to the use of MD simulations, the surfaces were optimized using the Smart optimization algorithm with the energy convergence criteria of 10 −4 kcal/mol and force criteria of 5 × 10 −3 kcal/mol/Å. The atom charges were assigned using the QEq method with the atom-based electrostatic and Van der Waals summation method (truncation: cubic spline, cutoff distance 15.5 Å, and spline width 1 Å). The optimized structure of GO-COOH was added to the optimized slab surface along the perpendicular directions. A "vacuum layer" with a height of 50 Å was placed upon the surface with PBC (periodic boundary conditions). The MD was conducted using an NVT (constant-temperature, constant-volume) canonical ensemble at 298 K. The time step for MD was 1 fs with a total simulation time of 300 ps (3 × 10 5 steps). The system temperature was maintained using a Berendsen Thermostat (0.1 ps decay constant). For the data analysis, 300 ps of trajectory frames were used.

Adsorption Energy and Radial Distribution Function
The adsorption energy was calculated using the following equation: The radial distribution function (RDF) [22,23,27,28] defines the probability of finding adsorbate molecules at distance r from the surface atoms of the slab models. It was calculated as follows: g(r) = (n(r))/2πr∆rρ (2) where r (A) is the given distance from surface atoms from the slab, n(r) is a time-averaged number of atoms in the area of r ± ∆r, ρ is the number density of the system.

Potentiodynamic Polarization
Polarization plots obtained for the Nb and Ta electrodes in a 0.1 M H 2 SO 4 aqueous solution in the presence and the absence of carboxylated graphene oxide are shown in Figure 1a,b.
Potentiodynamic measurements, in the case of the Ta and Nb bare electrodes, showed a fast increase of the anodic current, after the corrosion potential. In both cases, there is an active dissolution of tantalum and niobium due to the formation of Ta 5+ and Nb 5+ ions, respectively, which slow down around −0.05 V.
According to the Pourbaix diagrams for the Ta-H 2 O and Nb-H 2 O [29,30], in aqueous sulphuric acid solution at 25 • C at pH value > 0.5, Ta and Nb are oxidized to Ta 2 O 5 and Nb 2 O 5 ; therefore, the formation of passive oxide layers prevents further oxidation of metals when they are under positive potential [31,32].
The inhibition behavior of the GO-COOH layer is evidenced by the decreased anodic and cathodic current densities.  Table 1.
Even though the corrosion current density decreases in the presence of GO-COOH, the corrosion potential does not change significantly in the presence of the inhibitor. The anodic Tafel slope is somewhat larger than that of the cathodic one, and the corrosion potential is slightly displaced anodically, showing that this protective barrier acts mainly as an anodic inhibitor. Graphene layers are well known for the creation of adhered coatings on metal surfaces and affect the slope of anodic and cathodic reactions and hinder the corrosion reaction due to the large specific surface area and excellent mechanical properties [11,33]. These layers may hinder the diffusion of water molecules to the electrode surface [33]. The inhibition efficiency (IE) was calculated from the corrosion current density of the metal electrodes in the absence and presence of the inhibitor using the following equation: it reached 70% and 64% for Nb and Ta, respectively. IE (%) = (I unhib ) − (I inh )/(I unhib ) × 100 We compared these experimental results with those of the differential simulations. The Universal force field correctly describes the C-C, C-O distances in the GO-COOH structure, Figure 2    The adsorption energies presented from the MC simulations are summarized in Figure 5. This figure shows the distribution of potential energy as a function of adsorption energy, and the minimum adsorption is located on the left side of the axis. The adsorption energy for Nb (111) and Ta (111) surfaces divided per surface area of the used slab models is comparable for both the gas phase and in the presence of water molecules. The presence of water, as expected, favors the adsorption of GO-COOH, as it decreases the adsorption energy (the energy gain is around −1.8 kcal/mol·Å2). The presence of water molecules drastically changes the adsorption geometry of the GO-COOH, Figure 6, top-view. In the absence of water molecules, the distance of GO-COOH (H atoms) from the surface plane of the material is around 1.8 Å; this distance increases drastically in the presence of water molecules: for the Nb (111) it is 5.7 Å, whereas for Ta (111), it is is 6.4 Å. The water molecules are intercalated between the metal (oxide) surface and GO-COOH. These molecules increase the interaction between the two partners and therefore lowers the interaction energy. These water molecules provide common solvation of both partners. Similarly, the O atoms of the GO-COOH move further away from the material to host the intercalating water molecules. The O atoms of the side GO-COOH are closer to the surface of the metal. Energies are lower than for the metal counterpart surfaces, both in the gas phase and in the presence of water molecules. This is related to the increased GO-COOH atom distances from the oxide surfaces with regard to metal ones ( Table 2).  The GO-COOH on the Ta (111)/50H 2 O is adsorbed on the side, whereas in the case of Nb (111)/50H 2 O, we observe a different adsorption geometry-the planar one. This is in good agreement with the experimental data obtained concerning the corrosion inhibition efficiency difference between Nb and Ta (the planar adsorption of GO-COOH covers a greater surface compared with the side adsorption geometry). The evaluation of the adsorption of oxides is important in the case of initial surface corrosion but also for non-uniform corrosion although we have not tested the inhibition efficiency for the Ta and Nb oxides.

Molecular Dynamics
The MD snapshots of GO-COOH adsorbed onto the Ta and TaO (111) surfaces with the upright direction at different simulation times are displayed in Figures 7 and 8.  After several ps of simulation time, on both of the surfaces (Ta and TaO), the GO-COOH has a tendency for surface interaction. The interaction with the TaO surface appears to progress slightly faster than with the Ta surface (GO-COOH that sits on the surface after 70 ps); in this case after 45 ps of the simulation time, GO-COOH sits flat on the surface, reaching the lowest energy conformation on surfaces takes some hundreds ps after which there is a tendency for the GO-COOH to adsorb closest to the surface.
From Figure 9, the nearest distance distribution for most of the C (GO-COOH)-Ta (surface) atoms during the 300 ps MD simulation time has a very minor contribution at 1.03 Å, and a major one at the side -COOH group at 2.89 Å. Most of the atoms from the top COOH group during MS pass time at 2.88 Å; for the two COOH groups, there is a small RDF at a distance 2.86 Å that arises during the GO-COOH interaction with the surface in the interface vicinity.  The MD adsorption snapshots of GO-COOH on Nb and NbO (111) (Figures 11 and 12) surfaces are similar to those of Ta and TaO.  The GO-COOH rapidly reaches the surfaces; after 45 ps, it lies flat on the surfaces and tends to find the most favorable adsorption geometry that leads to the lowest energy of the systems.
To evaluate the interaction, RDF calculations involving the surface atoms of slab models and the GO-COOH molecule were analyzed. In Figure 13, the nearest distance distribution for most of the C (GO-COOH)-Nb (surface) atoms during the 300 ps MD simulation time is at 1.41 Å, the side -COOH group has major distance distributions at 2.84 and 3.45 Å, whereas most of the atoms from the top COOH group during MS spend time at 2.83 Å, for the two COOH groups, there is a small RDF at a very short distance of 1.1 Å that arises during the GO-COOH contact with the surface and its folding at the contact interface. The small distributions at high r values above 5 Å arise from the fact that the MD is started at a distance 20 Å from the surface. Regarding the interface distances in the case of NbO (111), as seen from Figure 14, the C (from GO-COOH) is positioned at 1.85-1.95 Å (two RDF peaks), the side-COOH group at 1.1 Å (important contribution) and the top COOH group fluctuates at 1.94 and 2.97 (meaning that the search for the most stable position has more complex dynamics than in the case of the Nb surface; this is also evidenced by the RDF spikes with decreased intensity from 20 to 4 Å).
In general, these short distances lead to relatively strong interactions of GO-COOH with Ta, Nb, and TaO, NbO surfaces. The calculated adsorption energies for the MD systems are, as in the case of MC calculations, very similar for both of the systems. In the case of Nb (111), the adsorption energy is −0.380 kcal/mol·Å2, whereas for the Ta (111) surface, this value is −0.382 kcal/mol·Å2.

Conclusions
This work describes the corrosion inhibition of niobium and tantalum by modified graphene oxide. Experimental results from potenciodynamic measurements will be subjected to theoretical calculations.
The results of the inhibition efficiency showed that the use of GO-COOH for corrosion inhibition in the aqueous acidic media, in the case of Nb, can decrease the corrosion by up to 70%, whereas for the Ta, the inhibition efficiency reaches 60%. These data are important as a starting point to see if the MC and MD simulations will be in agreement and offer some insights regarding the molecular behavior of GO-COOH at the studied material interface. Thus, the interaction of GO-COOH on the Nb (111), NbO (111), Ta (111) and TaO (111) surfaces was investigated by MC and MD simulations with the aim to better understand the interface interaction of GO-COOH. The MC simulations are in agreement with the experimentally observed behavior of GO-COOH; the calculated interaction energy was a bit stronger between the Nb (111)/50H2O//GO-COOH than the Ta (111)/50H2O//GO-COOH. The MD and MC simulation can be helpful to design and explore GO structures prior to experimental measurements that tend to exhibit more pronounced inhibition efficiencies, leading to decreased corrosion cost.