POTENTIODYNAMIC BEHAVIOUR OF GRAPHITE AND PLATINUM ELECTRODES IN MOLTEN SODIUM NITRATE-POTASSIUM NITRATE EUTECTIC

-The electrochemical behaviour of graphite and platinum electrodes has been studied in molten alkaline nitrates using cyclic voltammetry. The voltammograms obtained with graphite exhibit two current peaks not observed on platinum under similar conditions. An electrochemical reaction involving 0; ion and NO has been shown to proceed via an oxide on the graphite electrode surface which can be electrochemically reduced. It was assumed the formation of the NO3 intermediate is mainly responsible of the corrosion of the graphite in this melt. The thermal and electrochemical properties of graphite in the molten alkaline nitrate have been correlated.


INTRODUCTION
In a previois study of the sodium nitrate-potassium nitrate melt at temperatures from 230 to 470°C marked changes in the electrode surface and electrode reaction of graphite anodes were observed[l, 21.They became more marked at higher temperatures.suggesting the formation of an intermediate oxidised graphite surface.A similar explanation has been given of the anodic processes of graphite in molten potassium bisulphate [3].
Considering the technical importance of graphite as an electrode material it is interesting to study its chemical and electrochemical behaviour under both stationary and non-stationary conditions.The present research attempts to explain the surface reactions occurring on graphite in molten alkaline nitrates by using the linear potential sweep technique and to compare the behaviour of graphite with that of platinum electrodes in the same melt.

EXPERIMENTAL
The electrolysis cell consisted of a gas-tight Pyrex glass cylindrical container with a flanged cover supporting the different electrodes.Spectroscopic graphite rods (AGI(SP) of either 3 or 6 mm dia were used as working electrodes.The apparent electrode areas were from 0.15 to 4 cm2, the area boundary being sometimes delimited by painting the rest of the electode surface with boron nitride.The counterelectrode was a platinum spiral placed into a separate fritted disc compartment.The reference electrode was a Ag/Ag+ (AgNO, 0.07 M) electrode connected through the usual Luggin-Haber capillary arrangement.The pseudo-ohmic drop was always lower than 1 mV.The electrolyte (1: 1 NaNOs-KNO,) was prepared from A.R. chemicals vacuum dehydrated at 220°C and kept under pure nitrogen.A.R. quality gases were employed in the experiments (N,, 02, COz, NO, and NO).Both NOz or NO were collected at the cell outlet.
Experiments were carried out in the temperature range 24&350 k 1°C.The linear potential sweep technique is described elsewhere [4].

Current-potential ctqves
Semilogarithmic plots of stationary potentiostatic anodic E-Z curves with graphite anodes fit, at potentials <On6 V, reasonable straight lines with slopes slightly higher than 2+3(RT/F).At @6 V a limiting current is observed, related to the oxidation of the nitrite ion which is present in the melt, and proportional to its concentration.
At potentials 1=-o-6 V, another straight line is exhibited, with a slope %2+3(2RT/F).This is the potential region where the NO; ion discharge takes place.

Voltammograms with graphite electrodes
Prdintirtrrry wsdts.When a graphite electrode previously baked in air at 360 'C is dipped into a dehydrated nitrate melt, a rest potential of 015-0.20 V is attained.A voltammogram obtained by sweeping the potential from -@3 to 0.6 V, at 50 mV/s, shows during the first sweep an anodic current peak at 0575 V and during the returning half-cycle a cathodic current peak at 0.475 V.During repetitive potential sweeps an appreciable increase of the voltammogram area is noticed and a new anodic wave located at ca 0*32-0.35V occurs.

Potential, V
Fig. 1.Voltammogram obtained with the graphite/&olten nitrate system at 10 mV/s and 244°C.The numbers correspond to the successive potential sweeps.
After 2 days in the melt the graphite electrode attained a rest potential of O-07 V.Under this condition, the voltammograms obtained either at 5, 10 or 50 mV/s, from -0.3 to 0.8 V, exhibit an anodic current peak between 0.30 and O-35 V and a cathodic current peak at 0,225 V.These current peaks are better defined at lower potential sweep rates, as shown in Fig. 1.
There is an anodic current plateau between -0.3 and 02 V which, as further referred, is related to the electrochemical oxidation of OH-ions.The voltammograms run with graphite efectrodes previously aged in the melt present an area much larger than the area of the votammograms obtained with fresh graphite electrodes.When graphite electrodes were not previously baked in air, the effect was not found.The increase of area starting only when the electrolysis proceeds at anodic potentials larger than that corresponding to the NO; ion discharge.

Anodic oxidation of&rite
To identify the anodic current peak related to the NO; ion oxidation, KN02 (up to 15 x lo-* M) was added to the melt.The latter always contained some nitrite, probably in part formed by decomposition of nitrate during the dehydration treatment and in part present as an initial impurity.Single sweep voltammograms show an increase of the anodic current peak at ca O-575 V and a simultaneous shift of the peak potential towards more positive values with the increase of NO; ion concentration.This potential shift can be due either to a non-compensated ohmic drop, which is rather unlikely because of the currents involved and the large electrical conductance of the system, or to the fact that the electrochemical oxidation of NO; ion becomes more irreversible as its concentration -is increased, as was discussed elsewhere [5].The peak current related to the anodic oxidation of NO; ion occurs in a similar way on platinum electrodes.

Comparison of the system graphite/molten nitrate to the system platinum/molten nitrate
For reference purposes voltammograms were recorded with the system platinum/nitrate melt from 0 to 0.8 V at 244"C, where only the NOT/NO, couple is observed (Fig. 2).Further, similar to what was reported by ZamboninC6I.the following observations weremadeonvoltammogramsrunfrom -2.1 to -03 V, (Fig. 3): (i) One cathodic current peak which is related to the electrochemical reduction of NO;-NO; and the simultaneous formation and precipitation of an alkali metal oxide on the electrode.The maximum current shifts towards more cathodic potentials when the potential sweep rate increases from 50 to 300 mV/s; (ii) An anodic current peak at -1.0 V, assigned to the electrochemical dissolution of the oxide, which also shifts with the potential sweep rate; (iii) An anodic current peak which corresponds to the discharge of the 0; ion.As seen in Fig. 3, the voltammogram shape obtained for the first potential sweep differs from the following ones, but a stationary shape is reached after the third cycle.Furthermore, no appreciable change of the voltammogram area is noticed in subsequent Potential.V Fig. 3. Voltammograms obtained with the platinum/molt~n nitrate system starting from cathodic potentials at 300 mV/s and 245°C.
The anodic current peak related to 0, ion reduction is evidenced only when the melt is electrolysed at cathodic potentials (-1.5; -1.2 V).This current peak appears equally well during the first potential sweep either with platinum or with graphite, and its location (ca -0.6 V) seems to be independent of the potential sweep rate.However, on graphite electrodes in the same potential region a cathodic current peak is observed, probably related to the reduction of 0, to 0;.The anodic current peak for OH ion oxidation is recorded after a previous cathodisation of the melt (Fig. 4).The pre-cathodisation is made at -1.0 V, its duration affecting the height of the OH-ion oxidation peak, as shown in Fig. 4 for the graphitemolten nitrate system at 350°C.According to Zambonin[6].the occurrence of this peak is due to the Oi-ion and to the presence of traces of water which react yielding 0; and OHions.Both reaction products are picked up in the voltammogram.The detection of the OH-ion current peak was made by adding known quantities of NaOH to the melt.The peak height increases as the NnOH concentration dots; the peak on graphite is less defined than that on platinum.FuCthermore, the peak potentia1 shifts both with increasini the OH-ion concentration and with increasing the potential sweep rate towards more anodic potentials.Apparently linear potential-log (potential sweep rate) and linear peak height-square root ofpotential sweep rate relationships are obeyed for the anodic OH-ion current peak, as described further on.Consequently, current peaks formerly identified on platinum are also found on graphite together with two new peaks, one comprising an anodic current at a potential (0*>@4V) preceding the anodic NO; ion discharge and another involving a cathodic current within the potential region 0.2-03 V.

Ident$cation of'new current peaks on graphite
(i) Eflct of'anodisution.The two new current peaks are not observed during the first potential sweep when new electodes and electrolyte are used if the potential amplitude is restricted to 0.7 V and the region where the electrochemical oxidation of NO; ion takes place is avoided.The peaks, however, appear definitely during the subsequent potential sweeps, andincrease markedly from one sweep to the following one (Fig. 5).When potential sweeps from 0 to 0.4 V are made, this effect is not observed.Both current peaks are reduced Potential, V Fig. 5. Effect of the repetitive potential sweeping on the voltammograms of the graphite/molten nitrate system; 50 mV/s; 345°C.when Nz is bubbled but they increase markedly after anodisation at 1.0 V for 1 min (Fig. 6).This effect indicates these peaks relate to substances yielded by the anodic reaction of either NO; or NO; ions or both.
(ii) E@xt of nitrogen dioxide.When NOz is bubbled through the electrolyte to the point of saturation, the graphite rest potential is 0,610 V at 243°C.(Saturation of the melt with NO2 requires #24 h.)If the NO, is removed, the rest potential decreases to -0.19 V. Sweeping the potential from 0 to 0.4 V, to avoid the interference of NO2 produced in the discharge of the NO; ion, the two new current peaks increase mark-* edly in the presence d NO2 (Figs. 7a and 7b).Furthermore, when NO, is removed by bubbling N2, an increase of the voltammogram area is observed (Fig. 8).On platinum electrodes the saturation of the melt with NO, produces an increase of the current peaks related to the NO;/N02 couple.
(iii) Efict of cathodisution on graphite and platinum.
When a graphite electrode is cathodised during 1 min at -1.0 V and afterwards a voltammogram is run from 0 to 0*4V, both new current peaks increase as compared to the blank (Fig. 9).The increase, however, is OF-----7  (1) is taken into account, the question is whether NO, or NO react at the electrode.Saturation with nitric oxide has no appreciable effect on the voltammogram on the platinum/molten nitrate system.There is only an increase of the NO; ion current peak, due to equilibrium (I ).But after the system has been cathodised at -I*2 V, the anodic current peak appears as already mentioned in (iii) and it is observed only during the first potential sweep.
(v) Effect of sodium peroxide.When NazOz is added to the melt it becomes turbid and gas (0,) is evolved and the graphite rest potential becomes more negative and stabilised at -0.7 V at 242°C.Voltammograms swept between 0 and 0.7 V, reproduce the blanks but those run between -1.0 and 0.7 V exhibit two well-defined oxidation current peaks, one at -0.6 V, corresponding to the oxidation of 0; ion to 02. and another at 0 V corresponding to the oxidation of OH-ion.There is also one cathodic current peak at -0.7 V which involves a reaction complementary to that of the former anodic current peak.Both anodic current peaks are clearly established during the first potential sweep and dccruw after successi& cycling.The OH-ion oxidation peak is the one which decreases more markedly.
(vi) Effect of sodium peroxide and nitric oxide.The effects described in (v) indicate that the 0; ion and NO (or NO,) may have a simultaneous effect on the electrochemical process (Fig. 11).The 0; ion may, in principle, be generated through 02-ion formation during the chemical oxidation of graphite by NOz, according to : The anodic current peak on platinum is exhibited by any voltammogram at any cycling.The peak height decreases by N2 bubbling.Parallel results can be obtained with the graphite/nitrate melt.as shown in Fig. 12.Apparently there is no interference from the chemical reaction between graphite and NO.
Returning now to the graphite/nitrate melt, one must conclude that the existence of the anodic current peak is probably originated by 0; ions which are chemically as well as electrochemically produced.On platinum, this peak occurs during the first sweep after ;I previous cathodisation, since on this metal there is no possibility of regenerating the 0, ions if the potential sweep covers only the anodic potential region.This is confirmed by the fact that additions of 0; ions and NO induce the occurrence of the anodic current peak at any cycle.
(vii) Increase of voltammogram area.The increase of voltammogram area with potential cycling of the graphite implies an increase of the electrode area produced by the surface attack during the electrochemical reaction when the anodic potential sweep exceeds the potential of the NO;/NO, couple or when NO? is bubbled through the melt.One must conclude, therefore, that NO1 is the entity mainly responsible for the corrosion of graphite.
(viii) Efict of sodium peroxide and nitrogen dioxide.To determine if either NO or NO, produces the new anodic oxidation current peak, experiments described under section (vi) were repeated but in the presence of NO, instead of NO.In this case the NO; ion concentration was minimized by NO, bubbling and, subsequently NO and NO2 were eliminated by bubbling N2 through the melt.Then Na,02 was added and the melt saturated with NO,.Under these conditions the anodic current peak of the 0; ion was no longer observed and the anodic current peak related to the OH-ion decreased as NO2 was bubbled.A similar effect was provoked by the presence of NO, probably <because NO, was produced as an intermediate.It is reasonable, therefore, to conclude that the new anodic current peak is due to an electrochemical interaction of NO and 0; ion.
(ix) Other additions to the system It has also been proved that additions of COZ.CO, 0, or Na2C0, cause no difference in the voltammogram display for the graphite/nitrate melt.

Temperature effects
(i) The platinum/molten nitrate system.Voltammograms obtained with this system from -05 to 07 V exhibit only the current peaks of the NOJNO, couple because of the nitrite present as a trace.At any constant potential, larger currents are observed as the temperature is raised and, as seen in Fig. 13, a new current peak at 0~35-0~40 V is observed.The potential difference between this current peak and that of NO; ion oxidation is the same as that noted between the oxi-dation current peak just described and the NO; ion current peak.Hence, the temperature increase is equivalent to the addition of basic ions to the melt.
(ii) The graphite/molten nitrate system.With increasing temperature marked changes occur (Fig. 14).First a net gas evolution is observed without any external current flow.The rest potential, which becomes more negative when the temperature is increased, does not return to its original value if the temperature decreased.The anodic current peak corresponding to OH-ion oxidation is split, one peak located at -0.2 V and the other at 0 V (Fig. 14).The former increases more markedly as temperature is increased This effect, which has also been observed on platinum (Fig. 13) indicates that its occurrence is independent of the electrode material.With graphite electrodes, no change in the anodic peaks is observed but the cathodic current during the returning potential sweep is greatly lowered.Probably the current peaks exist but they are masked by the proper electrochemical reactions of nitrate ions, whose overvoltages decrease appreciably as temperautre is increased.If the graphite electrode is left immersed in the melt at 350°C for a long time, a net increase of the OH-ion oxidation peak is observed (Fig. 15).In general.for both systems as the equilibrium superoxide z+ peroxide = oxide, shifts towards the right as temperature increases [6], the current peak related to the 0; + NO clcctrodc reaction is no longer clearly observed at 345°C.

Drpendences oJ' the various current peaks on the potential sweep rutr
(i) The graphite/molten nitrate system.The heights of the new oxidation and reduction current peaks increase linearly with v 'I2 (Fig. 16) and u respectively (Fig. 17).The oxidation current peak was corrected for the contribution of the OH-ion oxidation current.The plot of Fig. 17 involves the total peak current without any further correction.
The uncorrected peak height related to the NO; ion oxidation also depends linearly on u1j2 (Fig. 18).Simi- The potential of the NO; ion oxidation current peak is nearly independent of v at low NO; ion concentration, but shifts towards more anodic values at the highest concentrations employed.A similar situation obtains for the NO, reduction current peak.Here.at L' > 50 mV/s, the current peak is no longer distinguishable.The current peak related to the OH-ion discharge increases linearly with Y'/' and the corresponding potential depends linearly on log v (Fig. 21).(ii) The platinum/nwlten nitrate system.In this case the height of the new oxidation current peak increases linearly with v 'I2 (Fig. 22) and its corresponding potential depends linearly on log u (Fig. 23).Both the anodic and cathodic peak heights of the NO;/NO?couple change linearly with v"~, Finally, the height of the OH-ion oxidation peak on platinum (Fig. 24) and its corresponding potential fits the type of relationship just reported for the graphite electrodes.

Preliminary considerations
The thermodynamics of the nitrate/graphite system indicates there are some processes related to the spontaneous corrosion of graphite at melt temperature (Table 1).The more likely final product, as far as graphite is concerned, is carbon dioxide.The electrochemical behaviour of graphite and platinum electrodes show several coincident features.Thus, the known cur-   ( The existence of the three oxygen ions would explain the voltammetric and chronopotentiometric behaviour of molten nitrates, either in solutions containing acids (SO:-, HSO;, NO*) or in solutions containing bases (NO;, NO, CO;-, 02-, OH-, O,), without the presence of the NO: ion [7].The addition of a strong acid to the molten nitrate produces an effect similar to NOz, which can be reduced in a single electron transfer reaction, but cannot be electrochemically oxidised [7].The presence of nitrite initially in the nitrate melt has been evidenced [23] and it should be a consequence of equilibria (15~(18) involving the different oxygen ions.

The anodic oxidation of nitrate ion
It is clear that the electrochemical oxidation of NO; ion, on graphite as well as on platinum, occurs at an appreciable rate at the highest potentials and is at least preceded by four other reactions.In both cases no anodie current peak has been observed for the NO; ion oxidation, but the current increases exponentially with potential.
Kinetic data available from previous work allow the calculation of the stationary current-potential curves for the anodic discharge of NO; ion.On platinum[26, 271 as well as on graphite electrodes, their Tafel slope is equal to 2RT/F at 250°C characterising thus an activated electrochemical process.The final products for the former are NO, and O2 while for the latter they are carbon oxides and nitrogen oxides [1,2].
The kinetics of this reaction on platinum changes with temperature as seen by values of the Tafel slope which goes from 2RTJF at 250°C to RTJF at 35O"C [1,27].At lower temperatures the main anodic reaction on graphite is: (23b) Reaction schemes ( 22) and ( 23) involve charge transfer and thermal steps.Both mechanisms present alternative rate determining steps.It is unlikely that the energetics of the reaction rate would be mainly related to the thermal steps, as deduced from thermodynamic data given in Table 2. Furthermore, if they were significant rate limiting reactions no simple explanation could be advanced for the high anodic Tafel slope above mentioned.Therefore, only reactions ( 22a) or (23a) should be considered as rate determining and the anodic current density becomes: where k' is a complex rate constant.The previous mechanisms should be modified now on the basis that species such as 0; and O:-ions predominate over 0' -ion in the nitrate melt and it may probably comprise the NO; ion discharge through the equilibria involving the various oxygen anions, and obeying a Temkin-type adsorption isotherm.

The oxidation of the nitrite ion
Nitrite ions exist in the nitrate melt and participate in various ionic equilibria [5,29-311.Either with graphite or platinum electrodes its discharge occurs at a potential lower than that of the nitrate ion oxidation.As suggested by Lyalikov[32], the nitrite electrode diluted in alkaline nitrate melt, under certain conditions behaves as a reversible system.
On platinum, when the nitrite concentration increases[5], the activated process becomes more sluggish and under steady state conditions the E-Z curve yields a Tafel line with a slope of R T/F within the temperature range 20@4OO"C [33,34].The following mechanism was put forward to explain these results: where K, = (k&.).
The nitrite ion oxidation on graphite in the temperature range 24%340°C yields N02.The process is char-acter&d by a Tafel slope equal to 2R T/3F which was interpreted with the following reaction scheme: reaction (33b) being the rate determining step.The rate equation is: As the concentration of nitrite in the nitrate melt is small its oxidation might be expected to be diffusion controlled and indeed the observed current peak is expressed by the usual relationship: -84.899 -88561 700 -92.231NO,(g) + Qgraphite) = NO(g) + CO(gas) 500 -32442 600 -36.444 700 -40.395 rected for the residual current on both sides of the peak potential.

The cathodic reduction of nitrate ion
The electrolysis at 250°C of the scrupulously dried melt at high cathodic potentials lead to 0; and NO; ions but if the melt contains water, then OH-and NO; ions are the reaction products.All these species were properly identified in the voltammograms.
Various authors advanced mechanistic interpretations for the electrochemical reduction of nitrate in melt.Hills and Johnston[36] recorded cathodic polarograms using NaN03-KNOJ melt on platinum mio roelectrodes at 280 and 400°C.A sharp reduction peak was observed as ca -1.5 V, with respect to a platinum electrode of large area, followed by an abrupt current increase at ca -3OV.This fact is in agreement with the discharge of alkaline metals in molten alkaline nitrates earlier reportedC37.381,and with the reactions NO; +2e=NO;+02- or NO; + e = NO, + 02-.
The process is irreversible and this agrees with the interpretation of Bartlett and Johnston[39].Swofford and Laitinen[40] investigated the matter further and concluded that NO; ion reduction on platinum and other microelectrodes is limited by the Na,O precipitation on the electrode surface.
Although it should be noted that the 02-ion is normally not present in the melt [ 19,20,41], the reduction of NO; ion will, therefore, produce species such as O:-, 0; and NO;.The O:-ion concentration decreasing with time, while the 0; and NO; ion concentrations increase.The reaction O;-+ 2NOZ = 2NO; (38) produces a "hidden limiting current" which make difficult the straightforward determination of NO; ion reduction limiting current.This can be understood in terms of the electrochemical reaction involving 0; and NO, and also the equilibrium: O:-+ 2NO; = 20; + 2NO; (39) for which the equilibrium constant at 229"C, is 6.7 x lo-".
The mechanism postulated by Zambonin for the NO; ion reduction can be summarised as follows (41)

Oxide formation and reduction at the electrode SWface
The anodic current peak found between 0.25 and 0.35 V, with both platinum and graphite electrodes, increases with the square root of v, and the dependence of its potential on u fits the equation for an irreversible process under diffusion control The slope of the potential at current peak us log v, is RT/F for both graphite and platinum, yielding an cz, value equal to 0.5, where za is the number of electrons used up to the rate determining step [35].The relevant fact is the occurrence of the peak only when NO and 0; ion are simultaneously present and for graphite only, its occurrence implies the existence of a new reduction current peak.
The surface oxide may be formed during the anodic discharge of NO; ion where the main oxidising species must be the NO3 radical.which acts as intermediate both on graphite as well as platinum The final products.however.are different, ie, while graphite corrodes, platinum is unchanged.Thus, only for graphite the reduction current peak is observed.The corresponding reaction can be written as follows: The reduction of platinum surface oxide, if any, in the absence of oxidising species, would occur probably at more cathodic potentials.
The reduction current peak potential exhibits no measurable shift with the potential sweep rate.Its height depends approximately linear on the potential sweep rate.This suggested the reduction process is irreversible, a fact which resembles the characteristics of platinum oxide reduction in aqueous solutions [41,42] as well as in acid bisulphate melts [43].As in these cases it is likely that in the present circumstances the electrochemical process yielding 0; ion is also a second order irreversible process.Unfortunately, the voltammogram in this potential region is too complex to allow further conclusions on the reaction pathway.

The electrochemical oxidation of hydroxyl ion
The reduction of NO; ion in the presence of water yields OH-ions, both on graphite and platinum electrodes.The height of the corresponding current peak depends on vii2 and its potential changes linearly with log v, with a slope of RT/F.Its rate is probably determined by an initial irreversible single electron transfer step which is followed by a chemical reaction, such as, C+OH-=C(OH)+e (47a) The first step comprises the OH-ion discharge and the OH radical adsorption.A comparable adsorption process is known to occur for the thermal adsorption of water on graphite [44].If this step is rate-determining, the rate equation, on the assumption that its symmetry factor is 0.5, is i = zFk, a, u,,"exp(qF/2RT).
(48) This equation fits the results reasonable well as will be seen further on.Furthermore, this reaction scheme resembles the one recently discussed for the reduction of OH-ion in nitrate melts on platinum [45].

The increase qf the graphite electrode area
The increase of the voltammogram area during repetitive potential cycling within the potential range where the oxidation occurs, is a consequence of the corrosion of graphite in the nitrate melt.The area increase depends on the amount of charge passed, but as seen in Fig. 25, the area increases up to a limiting value which is about 1.5 times the initial electrode area.This factor may represent the roughness of the electrode acquired during stationary experiments and should be taken into account in defining the corrosion current densities.

The graphite rest potential
The rest potential for thermally treated graphite electrodes is O-l-O.2V, but for electrodes used without any special treatment, the initial rest potential is between -0-l to -0.2 V.With increasing temperature the electrode rest potential becomes more negative.This effect which is also observed with platinum, is irreversible.
The addition of basic species also makes the rest potential more negative, while the addition of acid species has the reverse effect.Therefore, the increase of the temperature is equivalent to an increase of the concentration of basic species in the melt.The corrosion of graphite itself may contribute for the formation of basic ions, as seen in Fig. 15.Certainly the rest potentials of either graphite or platinum dipped into the melt at high temperature decrease when a graphite rod is placed in contact with the melt.

Correlation between chemical and electrochemical behaviour of graphite
The electrochemical'behaviour of graphite is to a large extent related to its thermal oxidation in the temperature range studied here.At 200°C or even lower temperatures oxygen reacts with activated carbons through steps of irreversible adsorption and desorption, yielding as final product either CO or CO, [46].At least two functional type of sites exist on carbons, each being related respectively to the desorption as CO and CO,[47-511.Carbons treated at T < 500°C exhibit acidic properties, while those treated at higher temperatures are considered as basic.The surface oxides accumulate at lower temperatures, but their thermal stability depends strongly on the nature of carbons The electrochemical oxidation of graphite occurs in aqueous electrolytes [7685] as well as in other molten electrolytes containing oxb-anions [3,86-891.Most of the work done referred to either carbon or graphite electrodes used in aluminium production cells and all this information leads to a better understanding of the electrochemical behaviour of graphite.
The corrosion of graphite occurring during the electrochemical oxidation of NO; ion was explained through the participation of the NO, intermediate, the final reaction products depending on the temperature.A comparable situation probably arises when the discharge of the 0; ion in the presence of NO takes place.One can think that NO acts as a radical scavenger assisting the graphite oxidation through NO, formation, by keeping a high surface concentration of C(0) groups.Hence, any hinderance in NO3 formation, either by the absence of 0; ion or NO molecules must supress the graphite corrosion process.The electronic structure of 0, ion, NO and graphite are favourable to the formation of an intermediate such as NO, at the graphite surface.Then, the fIna reaction products will be determined principally by the concentration of strongly oxidising species in the melt.Thus, the temperature effect, as far as the final reaction products is concerned, can be understood.
The formation of NOa requires the initial electrochemical oxidation of one of the following species: 0; where again the initial electron transfer step is rate-determining.

Reconstruction of the anodic voltammogram
On the bases of the individual electrochemical reactions just discussed one can attempt to build up a whole anodic voltammogram.
For the purpose it is convenient to proceed from one of the extremes, either anodic or cathodic.Calculations start with the aid of equations ( 21), ( 24), ( 27) and ( 30), the E-1 voltammogram related to the NO; ion oxidation can be easily drawn, both for platinum and graphite.Another possibility consists in obtaining each voltammogram independently starting from the low anodic potential region.Then, at potentials beyond 0.7 V the calculated volhmmograms are sub&acted from the experimental one to obtain the E-I curve corresponding to the NO; ion discharge, which is finally compared to those already known (Figs. 26 and 27).The above discussion attempts to explain aiso the different electrochemical behaviour of graphite and platinum in strongly oxidising melts, as well as the observation made by Laitinen and Bathia that the reversible oxide electrode obtained on platinum using oxygen saturated chloride melts, was no longer observed when carbon (graphite) was employed [89].

Fig. 8 .
Fig. 8. Change of the electrode area of the graphite due to NO*.(a) blank; (b) after NO, bubbling and elimination by nitrogen ; 50 mV/s ; 245 "C.

Fig. 10 .
Fig. 10.Influence of the pre-electrolysis time at ~ I V; platinum/nitrate melt: 100 mV/s: 244°C.not as large as that observed after a previous anodisation.If the same experiments are repeated with a platinum electrode, only during the first potential sweep is the anodic current peak observed (Fig. IO).However, by running the voltammogram from -1.0 to 0.5 V, the same current peak appears at any potential sweep.(iv) Effect qfnirric oxide.The NO does not react with graphite which under NO saturation exhibits a rest potential equal to 0,342 V at 244°C.When formation of NO, occurs the potential becomes more positive up to O-6 10 V. A voltammogram obtained under NO saturation shows the height increase of the two peaks already mentioned and the peak potentials become more anodic (U 0*1 V).According to this.if the equilibrium[7], NO + NO; = NO, + NO; 2C + 2NO; + NOz.= 3N0 + 2C02 + O'-(2) or 4C + 2NO; + NOI = 3N0 + 4C0 + O2 -. (3) These reactions are energetically favoured as deduced from the free energy change at 5OO"K, of the following reactions: 2C(gr) + 2NaNO,(l) + NO,(g) = 3NO(g) + Na,O(s) + 2CO,(g), (4) AGO = -89.657kcal/mole and 4C(gr) f 2NaNO,(l) + NO,(g) = 3NO(g) + Na?O(s) + 4CO(g), (5) AC" = -44-628 kcal/mole.Therefore, should O:-ion be supplied to the melt by adding Na202, the corresponding oxidation current peak would be observed on platinum during any cycle by sweeping within the anodic potential region.The addition of Na20, to the platinum/nitrate system produces a shift of the rest potential towards more negative values.A similar effect should occur with the potential of the NOJNOI couple.If the melt is saturated with NO the rest potential increases up to -0.04 V at 244*C anel that of the NO;?JO, couple returns to its former value.

Fig. 11 .Fig
Fig. 11.Effect of Na,O + NO addition on the voltammograms run with the platinum nitrate melt at 245°C; 50 mV/s.(a) NO,/NO; couple;(b) after the addition of Na,O, (containing traces of OH ion); (c) after the addition of Na,O, and NO bubbling.The voltammogram is unchanged after the fourth potential sweep.

Fig. 20 .Fig. 21 .Fig. 23 .
Fig. 20.Plot of the potential corresponding to anodic peak located at ca 0.3 V US log v. Graphite/molten nitrate; 245°C.The slope of the plotted line is 2.3 (R T/F).

Fig. 24 .
Fig. 24.Plot of the aaodic current peak corresponding to the OH-ion vs the square root of the potential sweep rate.Platinum/nitrate melt; 245°C.
and the increase of the voltammogram area and the rest potentials, are characteristics of the graphite eleectrodes[ 11.2.The anion composition qf the meltThe acid-base theory byLux and Flood[991 I], can be expressed as: O* -+ acid = conjugate base, (6)and was applied with success to molten silicates, carbonates and phosphates.In nitrates, the following equilibria may be expected: the strong oxidising power of molten nitrates should be related to NO; or NO,[7,12].A reaction such as (7) suggests the exis tence of a reversible NO,/NOl electrode reaction[12,13]    the reversible nitrate electrode in the molten nitrate.Now, with presence of either water of OH ion the following reaction is postulated[17] :40H-= O2 + 2H20 + 4e.62)Reactions (10) and (12) are related to the equilibrium:O*-+ H,O = 20H-.(13)The values reported for the equilibrium constant of reaction (13) cover the range from nearly infinity[17]   down to approximately zero[ IS].More recently the oxidation of 02-ion has been interpreted with the following reaction:

C
C + NO,= NO2 + CO, + e (20~) where (20~) is the rate-determining step under conditions of a high degree of surface oxidation.The rate equation in terms of current density, i, is: i = zFk, aNo; exp(u qFJRT), (21)where k, is the specific rate constant of step (20c); aNor is the nitrate ion activity at the interface, a is the transfer coefficient of the anodic reaction (c( = 0.5) and the rest of symbols have their usual meaning.At higher temperatures the situation is somewhat different.The primary step is still the NO; ion dis charge, but the reaction mechanism changes according to i = zFk, aNO; e exp(a q+FJRT),(24)    where ac is the activity of the electrode surface.Taking again CI = 0.5, the anodic Tafel slope is 2RT/F as found in the stationary runs.From the kinetic viewpoint, at present both reactions (22a) and (23a) appear as indistinguishable steps.At low temperature, on platinum, the total anodic reaction is:2NO; = 2N02 + O2 + 2e(25) and the mechanism of the reaction is[lS, 271: NO; + Pt = (NO,)Pt + e (26a) (NO,)Pt = (0)Pt + NO, (26b) l=V(O) + Pt(0) = 01 + 2Pt (26~) Pt(O) + NO; = NO, + 0, '+ Pt + e (26d) where reaction (26a) is the rate determining step.The corresponding rate equation becomes: i = zFk,, uNOz tr, exp(qF/ZRT).; = (N,O,) = 2N02 + +02 (28d) where the O2 -ion would be solvated as the orthonitrate ion : NO; i-02-= NO:-.(29) At high temperatures the Tafel slope approaches the ratio RT/F.The reaction was tentatively interpreted assuming the initial discharge of O*-ion.The corresponding rate equation is: i = rFk' exp(vFJRTX (30) 31b) is rate determining, under a low degree of surface coverage by the (NO,) intermediate.The rate equation derived from this reaction scheme is: i = zFk, K, aNo; exp(qF/R T) (32) i = zF c,,; W&O; aP2 x(4 (35) II = rFo/RT: DNOi is the diffusion coefficient of nitrate ion and x(at) is a function of the potential which is tabulated[35].The observed peak height was cor-Table 2. Standard free energy change of different reactions involving NO2 Na,O at the interface inhibits a further reduction of the NO; ion.The equilibrium of the different oxygen ions has been omitted in the reaG tion mechanism Nitrate melts containing water react according to : NO; + Hz0 + 2e = NO; + 20H-.
C(0)+2e=C+OZ-(42)    or better suited to the melt composition recently sug-precedent facts, the probable reaction pathways for graphite are stated as follows: + NO + 0; = PtO(N0,) + e (46a) 2PtO(NO,) = 2Pt(O) + 2N02 + 0, (4W Fig.25.Plotofthevoltammogramareaagainsl thenumberof repetitive potential sweeps. [X&55].Reviews about these works up to recent years are respectively given in references[5659], including the various mechanisms proposed for the gasification of carbons[60-66].An appreciable difference of kinetic .behaviour between monolayer and multilayer graphite surface oxides has been observed[67,70].During the oxidation process of graphite an intermediate mobile surface oxide, before CO, evolution, was postulated[71-731 and in the thermal oxidation by means of strong oxidising agents, such as ozone, a transient surface oxide has been proposed.From the point of view of this report it should also be mentioned that NO adsorbs irreversibly on fresh graphite surfaces yielding a (NO)C adsorbed species[74].The amount adsorbed decreases on oxidised graphite surface.Water also adsorbs on graphite[44,72].
ion, NO or graphite at the surface.Therefore, three possibilities arise for the initiation of graphite corrosion in the nitrate melts:)C = NO, + (0)C (5ld) 2(O)C = c -I-co, (5le) (NO,)C = NO + CO, (51f.l(CO? + c = 2CO) (51g)The contribution of step (51f) becomes more significant when the temperature is increased.Equilibria comprising nitrogen oxides.such as (33~) and (33e) also participate in the reaction schemes.Any of these reaction schemes lead to the Tafel slope equal to 2RT/F as found through the different relationships between the experimental parameters derived from the voltammograms, if the initial electron transfer step is rate determining.:' The anodic peak due to the 0; ion + NO reaction on platinum can be explained in terms of a mechanism resembling initially mechanism I, but involving a different fate for the reaction intermediate: 0; + Pt = (OI)PPt + r (52a) (0,)Pt + NO = (NO,)Pt (52b) Pt(N0,) -Pt + fOz + NO2 (52~)

Fig
Fig. 26.Experimental and calculated voltammograms at 10 mV/s for graphite/molten nitrate at 244°C.(+) independent reactions; (x) theoretical voltammogram; full line corresponds to the experimental one.NO; ion discharge was not considered in the calculations.