KINETICS AND MECHANISM OF THE ELECTROCHEMICAL OXIDATION OF GRAPHITE IN BISULPHATE MELTS*

The kinetics of the electrochemical oxidation of graphite to volatile compounds by electrolysing molten bisulphates in the temperature range from 180 to 320°C has been studied by conventional steady-state and non-steady-state measurements. CO*, CO and traces of SO, are formed during the electrochemical reaction. Assuming 4 Faraday/m01 of CO, the anodic efficiency is about 90%. The CO&O ratio is about 2 and is independent of the cd in the range considered. The electrochemical reaction is compared to the thermal oxidation of graphite. The reaction is interpreted through a scheme involving consecutive reactions, where the rate-determining step, under Ten&in conditions, is a thermal process involving a desorption of intermediates following either a first-order or second-order process depending on the type of adsorption obeyed. The activation energy derived from the rate constant at the rest potential is about 42-5 Kcal/mol. R&nmMn a dtudi6, en utilisant m&hodes stationnaires et non-stationnaires conventionelles. la cin&ique de l’oxydation Clectrochimique du graphite pour former des produits volatiles pendant 1’8lectrolyse des bisulphates fondues & des tbmperatures de 180 B 320°C. Dans la reaction blectrochimique COI, CO et des traces de SOI sont form&. En supposant 4 Faraday pour chaque mole de COI le rendement anodique de ce demier est d’environ 90%. Le rapport CO&O est environ 2 indkpendement de la densit de courant pour l’intervalle consider& On a 6tablie une comparison entre le r&action Blectrochimique et l’oxidation thermique du graphite. La r&action a 6t6 interpret6 par un schema qui comprend des r&actions con&cutives ou l’&ape rbgulatrice, d’aprks Temkine, est un pro&s thermique bas& sur la dhorption des intermtiiaires selon un pro&s de premier ou second ordre qui dbpend du type d’adsorption observ&e. L’dnergie d’activation d&iv&e de la constante de vitesse au potential de r&pos est d’environ 42,s Kcal/mol. Zusammenf&sung-Es wurde die Kinetik der elektrochemischen Oxydation des Graphits bei der Elektrolyse von geschmolzenem Bisulfatem untersucht. Der Temperaturbereich war von 180 bis 320°C. Es wurden station&e und nicht-stationlre Messmethoden verwendet. Bei der Elektrolyse wurden COB, CO und SO,-spuren gebildet. Unter der Annahme von 4 Faraday pro mol CO%, war die Anodischeausbeute ungef&hr 90%. Der Mittelwert des Verhiiltnisses CO&O war cu 2 und unabhtigig von der Stromdichte im untersuchten Bereich. Die elektrochemische Reaktion wird mit der thermische Oxydation von Graphit verglichen. Die Bruttoreaktion wird mittels eines Mechanismus tit einer thermischen Reaktion als Geschwindigkeitbestimmende Stufe gedeutet. Bei Ten&in-Bedingungen, umfasst dieser therm&he Prozess eine Desorption von Zwischenprodukten, die, abhiingig von der Adsorptionsart, entweder einem Vorgang erster zweiter Ordnung gehorcht. Die von Geschwindigkeitskonstant bei Ruhmpannung bestimmte Aktivierungsenergie betrlrgt 42,5 Kcal/mol.

The electrolysis cell used was of Pyrex glass, similar to that described in an earlier publications Graphite electrodes of 3-mm diameter and of different lengths corresponding to geometrical areas between 2 and 5 cm2 were used as anodes. The cathode consisted of a bright platinum sheet placed into a separate compartment. Two reference electrodes were employed. In the first set of experiments the reference electrode was a platinized platinum electrode dipped into the melt and continuously saturated with purified hydrogen gas at 1 atm. In the last set of experiments an AgJAg+ electrode was used. In both cases the reference-electrode compartment was connected to the working electrode by a Luggin-Haber capillary. Pure potassium bisulphate and sodium-potassium-bisulphate eutectic previously vacuum dried at room temperature in a desiccator for several weeks were used. Temperature was varied from 180 to 320°C.
Conventional steady-state and non-steady-state measurements were made, including CurrentJvoltage curves recorded either galvanostatically or potentiostatically. In each curve the current or the potential was raised and lowered two or three times to estimate any hysteresis effect. The non-steady measurements involved the determination of build-up and decay of the electrode potential, recorded in the latter case with an oscilloscope for the shorter times and with a potentiometric recorder for the times larger than 5 s. Details of the instrumentation used have been described in previous works.1*9

Identijcation of reaction products
The gaseous products of the anodic reaction were analysed by ir spectrometry. A suitable electrolysis cell, previously described, was employed for collecting gaseous products of the anodic and cathodic reactions by means of a conventional vacuum line.
Before sampling, the eIectrolyte in the anodic compartment was properly saturated with reaction products.
Samples were obtained either by evacuation or by passing argon or nitrogen through the anodic compartment.
COs and CO were identified as the main reaction products of the anodic reaction. SOa was also observed in their spectrum. Quantitative evaluation of the products was made in the way usual for ir spectrometry,

Anodic ejficiency
To determine the number of electrons related to each mole of product, coulometric determinations were made. COs was collected by passing the gases through a series of weighed traps containing Ascarite and calcium chloride.
Nitrogen was employed as carrier gas for the purpose. The gases were previously cooled to -78°C to condense any acid gases formed during the reaction. Blank runs were also made to correct for any impurity contained in the carrier gas.
As the reaction proceeded, a consumption of the graphite electrode was observed, best noticed at the higher electrolysis currents.
Blank runs without electrolysis performed in the presence of argon and in the absence of any reacting gases as oxygen, carbon dioxide etc, showed no appreciable consumption of the graphite electrode under these conditions.
The quantitative analysis results of the anodic gases are shown in Table 1. It is interesting to notice that the ratio between COz and CO is about 2 and no conclusive dependence of this figure on current density was observed at least between the range of 0.8 x 1O-3 to 45 x 1O-3 A/cma. Thf: ratio between CO, and SO, is also presented in Table 1. These results indicate that a reduction of SO, or any anion containing SO3 groups by CO, yielding CO, and SOa is negligible. Consequently the main gaseous reaction products are CO, and CO, be they either electrochemically or thermally produced. To calculate the efficiency of the anodic reaction two possibilities are considered (i) the products of the electrochemical reaction are CO, and CO (ii) the product of the electrochemical reaction is CO,. For case (i), taking into account that the ratio CO,/CO is about 2, 10 F are required for the formation of 3 moles of products, while for (ii) 4 F are required for each mole of CO, formed.
If the first possibility is assumed, the anodic efficiency with respect to CO, would always be higher than loo%, which is quite improbable.
If the anodic efficiency is evaluated according to possibility (ii), the result is about 90x, Table 2. Thus, the second possibility seems more reasonable since for a CO.&0 ratio of 2, the anodic efficiency for CO, must be 80 %. Although it is known that the spontaneous oxidation of graphite in the presence of gases such as CO, also occurs, this reaction is important only at higher temperatures. lo Therefore the formation of CO, in the temperature range of 180-320°C is due to a different process, which is discussed further on. Fig. 1. The overpotential q is defined as the difference between the electrode potential related to the cd i, and the rest potential of the graphite electrode when no net current fiows. The pseudo-ohmic overvoltage correction has been included. The cd has been calculated taking the apparent electrode area. In the whole range of current investigated a satisfactory Tafel line is observed, particularly for experiments performed at low temperatures. Current/voltage curves were obtained changing either the current or the potential upwards or downwards, when the experiments were made either galvanostatically or potentiostatically. In the whole range of temperature the first current/voltage curve recorded by changing the current upwards gives, for any tied value of i, a value of 7 lower than that observed when the current was lowered. The same effect may be referred to a fixed potential by comparing the corresponding cds. This hysteresis effect is not appreciable at the lower temperatures but becomes larger at tne higher.

Current/voltage curves show a linear relationship between overvoltage and logarithm of cd,
When current/voltage curves were recorded either by waiting a long time between readings or after repeated cycles, the hysteresis disappeared. This is a clear indication that a steady composition of the electrode surface had been reached after a prolonged electrolysis. From the Tafel lines the experimental Tafel slope, bT, and the apparent cd i,,, at q = 0, were evaluated. These figures are assembled in Table 3 and for the sake of comparison the values of the ratio 2*3(RT/F) at different temperatures were also included. This effect is probably due to the magnitude of the overvoltage concerned, although we should not discard a probable interference of the product formed during the thermal decomposition of the melt, which is rather appreciable at this temperature.
The temperature effect on current/voltage curves is a very sensitive decrease of overvoltage with increasing temperature.
The effect is reflected in the increase of the i. values when increasing the temperature. By plotting log iO against l/T a good Arrhenius plot is obtained.
From the slope of this plot the average experimental activation energy is 42.5 & 5 Kcaljmole.

Decay of emf at current intermption
After interrupting the electrolysis current the emf decay follows the already known law predicted for electrochemical processes under activation control, the emf decaying linearly with the logarithm of time. Typical emf decay curves are shown in Figs.

2, 3 and 4.
At the lowest temperatures the emf-decay curves present a linear region covering about three logarithmic decades. As the temperature increases, immediately after the current interruption, the emf tends to stay at nearly a constant value for a rather long time, and at slightly longer times the plot approaches a linear relationship in the q/log t diagram, involving a slope close to 2*3(RT,F).
Between 190 and 290°C there is always a good range in the q/log t plot where a 2,3(RT/F) slope is obtained.    In the temperature range from 290 to 320°C the semilogarithmic plot shows a more extended initial range where the emf does not change appreciably, and the linear region involving the above-mentioned slope covers only one logarithmic decade. Furthermore the slope 2.3 (RT/F) in these experiments is approached only with experiments performed at the highest cds. At low cds the slope of the linear portion is usually slightly lower. These results, however, must be considered taking into account the magnitude of the emf decay observed at higher temperatures, which is certainly much lower than that observed at lower temperatures.
These results from the transient experiments agreed satisfactorily well with the results obtained under steady-state conditions.
At the lower temperatures it is observed that the emf-decay slope, bd, is independent of the current used in the previous electrolysis, as is concluded from data assembled in TabIe 4. Contrarily to the experiments at higher temperatures the slope depends on the initial value of i, approaching the value 2.3(RT/F) at high initial currents.
From the t' value obtained from the interception of the emf-decay slope with the initial potential, the electrode capacitance, C, at overvoltage q was evaluated.ll Data obtained from decay curves are assembled in Table 4. The electrode capacitances are large, their magnitude being related quite likely to an appreciable contribution of the electrode roughness, because of the consumption of the electrode. The electrode capacitance values go from about 3 x IO3 ,uF/cm2 up to a maximum value of about lo* pF/cm2.
A slight dependence of the electrode capacitance on potential is observed, particularly at the low overvohage region and at the highest one. The lower values are obtained at the rest potential and beyond 0.6 V. There is an intermediate overvoltage region covering about O-3 V where the capacitance does not change appreciably with the overvoltage. At the lower temperatures the capacitance decreases with the electrode potential, while the reverse effect is observed at the highest temperatures. Figure 5 shows the electrode capacitance dependence on the overvoltage for one emf-decay experiment, the electrode capacitance being calculated by the differential method-l1 As the data presented in Fig. 5 correspond to an electrode surface probably involving a constant roughness factor, their dispersion is appreciably lower than that reported in Table 4.

Build-up curves
From the initia1 sIope of the build-up curves obtained under different experimental conditions, the electrode capacitance at the rest potential is obtained in the usual way.ll The average electrode capacitance is over 3 x IO3 pF/cm2, and is temperatureindependent. Taking into account that at the rest potential, and in the absence of any reacting gas, the corrosion of the electrode is negligible, this figure can be used to estimate the electrode roughness factor, as ca f02.

DISCUSSION
The electrochemical oxidation of graphite in bisulphate melts is a process closely related to the thermal oxidation of graphite in the presence of oxygen and other oxidizing gases such as ozone or carbon dioxide. The thermal oxidation of graphite has been studied in detail, both from the thermodynamic and kinetic standpoints.lO In the thermal reaction the role played by different surface groups and surface oxides has been shown, and the subject has been reviewed recently.la The electrochemical reaction is characterized as an activated electrode process, since over a wide range of cd the current/potential curves fit a Tafel line with a slope equal to 2_3RT/F and the apparent cd extrapolated to q = 0 is of order of magnitude characteristic of a slow electrochemical reaction. Furthermore, the average experimental activation energy is much larger than that usually found in electrode processes. In the non-steady measurements a satisfactory linear q/log t plot is obtained from emfdecay curves, with slopes coinciding with the Tafel slopes. The agreement is in principle an indication that the same reactions take place on graphite electrodes either steady or non-steady conditions.
For discussing this reaction we shall begin by giving an explanation of the Tafel slope in terms of a reaction mechanism.
The main products formed during the electrochemical reaction are also produced in the thermal oxidation of graphite. The COZ and CO formation takes place through one or more surface oxides and eventually through the participation of intermediate oxygen complexes on the graphite surface related to them.13 To interpret the electrode reaction we must first decide if the reaction intermediates of the oxide type obey either a Langmuir or a Temkin isotherm. As a diagnostic criterion of the type of adsorption, we must try to correlate results obtained by electrochemical measurements with those obtained in the chemisorption of gases on carbon. The chemisorption of different gases on carbon involves the coverage of only a small fraction of the total surface. This conclusion was clearly established, for instance, for oxygen adsorption on graphite. 14-16 Thus, at 2OO"C, the oxygen surface coverage on amorphous carbon is about 6%. At 300°C the amount of chemisorbed oxygen on Graphon, previously subjected to burn-off, constitutes only about 5 % of the BET area. Consequently, the number of sites on which a higher state of oxidation can be achieved yielding CO2 and CO, is rather small. This means that the degree of surface coverage must be lower than O-1 in the conditions prevailing during the thermal oxidation, as well as on graphite electrodes when no applied potential exists.
The decay slopes and the electrode capacitances calculated from t', over a large range of potential, are nearly independent of the potential at the current interruption.
However, for interpreting the total electrode capacitance two different contributions must at least be distinguished.
(i) there is a non-faradaic term the value of which is of the order of 3 x lo3 ,uF/cm2 independent of temperature.
It gives an idea of the actual active electrode area. This capacitance was measured at the rest potential, where the situation must be directly compared to the conditions prevailing during thermal oxidation, so a low degree of coverage is admitted; (ii) there is a faradaic contribution to the capacitance due to the accumulation of intermediates on the electrode surface because of the electrode reaction. This is generally potentialdependent, reaching a maximum value when the degree of surface coverage is O-5. The electrode capacitance determined at the rest potential agrees with the value obtained for the limiting case of a degree of coverage equal to one, as previously assumed in the electrochemical oxidation of nitrates on graphite electrodes at the same temperature. 2 Therefore, the behaviour of the electrode capacitance with potential resembles the case of an electrode process where intermediate compounds are formed during the reaction obeying a Temkin isotherm rather than a Langmuir, involving an appreciably large rate of change of the free energy of adsorption with degree of coverage.

On the basis of this explanation the coincidence of the steady Tafel slopes with the emf-decay slopes is immediately deduced-l'
On the assumption that the Ten&in isotherm is obeyed by the reaction intermediates, and considering that the process on steady conditions is the same as that taking place in the transients, we can postulate the probable reaction path yielding CO,. As a unique slope exists in the whole range of potential investigated, it is reasonable to interpret the kinetics of the reaction with a scheme of consecutive reactions. As Langmuir adsorption is discarded, we cannot consider any reaction mechanism involving electron-transfer reactions as rate-determining step. Neither is there an explanation for the slope 2*3(RT/F) under the limiting conditions derived from a Langmuir isotherm-l* Another possibility which is in principIe also discarded is the participation as ratedetermining step of any reaction involving the simultaneous transfer of two electrons. (3A) This reaction scheme is a rather simple one since only two intermediates, one containing the radical HS04 and another containing an oxygen atom attached to the graphite are supposed. The product formed in the first reaction may be considered as a kind of "graphite bisulphate". Actually graphite bisulphate has been formed under different circumstances.21 Both intermediates are produced in steps comprising a transfer of one electron and the COz formation is due to a thermal reaction of two radicaIs, according to step 3A. The kinetic analysis of this mechanism in terms of Temkin conditions, assuming that the activity of bisulphate ions is unity, gives the following rate equations for steps 1A to 3A,

v-1 = k-la, exp {-[AG__r* -yr,X + (1 -/?)FE]/RT), (2) us = k,a, exp (--[AGs* -yr& + (1 -y)r,X -DW/RT), (3) V-2 = k-zazaso,aH+ exp (-[AG_2* + (1y)rlX -p-,X + (1 -~)FE]/RT),
Here u, is the velocity of the step i and ki its specific rate constant, a, is the surface activity of the reacting intermediate and X is the total degree of surface coverage. AGi* is the molar free energy of activation of adsorption for step i. rl and r, refers to the rate of change of the free energy of adsorption with degree of coverage and the sub-indices 1 and 2 applied for adsorbed reactives and products in each stop in the anodic direction respectively. y is a symmetry factor for the adsorption co-ordinate and p is the symmetry factor for the reaction co-ordinate related to each step of the electrochemical reaction. E is the potential applied at the interface. From the kinetic standpoint none of the steps 1A and 2A may be considered as rate-determining because they are related to Tafel slopes equal to 2*3(2RT/fl, if the symmetry factors are taken equal to O-5. The only possibility is then restricted to step 3A. If it is assumed that the preceding steps are in quasi-equilibrium, that is vI = ZJ_~ and vs = v-s, then : r,X = Ki + FE. (6) That is, the degree of coverage by reactants in any rate-determining step wiI1 give the same dependence of the product r,X on potential, except for the constant K,, which comprises a different ratio of rate constants according to the step which is considered. Another reasonable assumption is to admit that in any step the rate of change of the free energy of adsorption with degree of coverage for the reaction in the forward direction is approximately the same as for the reverse direction (rl M rs)_ Under these conditions, if step 3A is rate-determining, this being the only one that under Ten&in conditions yields a theoretical Tafel slope lower than 2*3(2RTIF), the rate equation assuming the quasi-equilibrium of the preceding steps, becomes vQ = k3az2 exp (-AG,*/RT) exp (2yK,/RT) exp (2yFEjRT) = K' exp (2yFElRT).

(7)
According to (7), if the adsorption is a non-activated process (r = 1) the Tafel slope turns out to be 2.3(RT/2F), while for the case of activated adsorption (y = O-5), the Tafel slope results 2*3(RT/Q Another simple reaction scheme involving the initial participation of oxidized graphite can also be postulated, The existence of a certain degree of surface oxidation on graphite when the steady state has been reached in the electrode reaction is evidenced by the hysteresis observed at low cd through the current/voltage curves. This effect can be explained by assuming that an appreciable time is required to reach a rather high degree of surface oxidation. This effect has been earlier observed by employing graphite electrodes in cryolite melts. 22B23 Then, let us write the following reaction scheme : (1%  Step IC comprises the discharge of a bisulphate ion originating an intermediate which further dissociates according to step 2C. The product formed in this step is oxidized according to step 3C, yielding as a product an oxidized graphite involving two neighboring oxygen atoms. A further discharge of bisulphate ions on these oxidized sites yields an intermediate which dissociates according to step 5C. The product formed in this step decomposes electrochemically according to step 6C, yielding another intermediate of peroxidic type which finally decomposes according to step 7C, yielding CO, and regenerating a partially oxidized graphite surface that again enters the electrode reaction.

Again two limiting cases can be derived from (8). If the adsorption of the intermediate is a non-activated process (y = 1) the theoretical Tafel slope is equal to 2_3(RT/F), and for an activated adsorption process (y = O-S), the theoretical Tafel slope is 2*3(2RT[F). Let us turn now to an interpretation which is based upon possible structural aspects of intermediates involved in the reaction path. Thus, taking into account previous assumptions involved in schemes
It is evident that scheme C comprises various electron-transfer steps which depend on the distribution of surface oxides on graphite. This means that for highly oxidized graphites the reaction may start with step 4C, while for a nonoxidized graphite surface, an oxidation step preceding reaction 1C has to be added.
The interesting feature of scheme C is the participation of intermediates of peroxidie type which are similar to those already postulated for the mechanism of oxidation of carbon blacks by dry mixtures of oxygen and ozone.a4 In the latter reaction carbon dioxide is the reaction product.
After the kinetic analysis of this mechanism in terms of a Temkin isotherm it is also concluded that any step such as those involving either electron-transfer or ionization reactions can be discarded as rate-determining steps. Considering that step 7C is rate determining, the rate equation results In this mechanism the same intermediates already postulated in scheme C are assumed. The final step comprising the decomposition of the peroxidic intermediate is considered as rate-determining. The conclusion drawn from the kinetic analysis of this mechanism are the same as above mentioned for scheme C with reaction 7C rate-determining. If we assume that the rate-determining step is reaction 3A, the value of r can be estimated from the capacitance/potential dependence. This value should be about 20 Kcal/mole or even higher. l1 With this figure let us try to evaluate the range of potential where Ten&in conditions should be obeyed. Let E1 be the potential related to the degree of coverage X1 and E, related to X,. If the limits of Ten&in isotherm are supposed in the range O-1 to O-9, the potential range is (10) Hence, for r = 20, Temkin conditions would prevail at about 1.1 V. This potential range apparently exceeds the potential range found from the electrode capacitance dependence on potential, but supports the fact that emf-decay slopes coincide with the Tafel slope in the whole range of potential investigated.
If a more complex kinetic model is assumed for interpreting these results, such as surface-induced heterogeneity, 25*26 the same theoretical parameters are obtained.
From the previous discussion it follows that the reaction mechanisms proposed to explain the electrochemical oxidation of graphite in molten bisulphates are rather similar to the mechanism postulated for the discharge of oxide ions on graphite at much higher temperatures. 27 In this case, at low and high cds the reaction mechanism also involves thermal reactions as rate-determining steps. We therefore conclude that under Ten&in conditions there are various possibilities equally probable for interpreting the kinetics of the electrode reaction, which offers the same ambiguity already stressed for other electrode reactions interpreted in terms of a Temkin isotherm, such as the evolution and dissolution of oxygen on platinum in aqueous solutions2* However, the important conclusion is that whatsoever the first-or second-order process, the kinetics of the electrode reaction is determined by a thermal reaction.

The likely mechanism for the formation of CO
There are at least two possibilities to interpret the formation of CO. The first one is the interconversion of CO, to CO on the graphite surface promoting graphite oxidation by means of a reaction like that expressed by the Boudouard equilibrium.29 However, from the analysis of thermodynamic and kinetic data related to the thermal oxidation of carbons,30*31 it seems unlikely that in the temperature region of the present experiments any appreciable interconversion according to the Boudouard equilibrium takes place. Furthermore, the variation of CO, diagram with different types of carbons, the approximately fixed value of the CO&O ratio obtained either in the slow or in the fast decomposition and the appearance of only one reaction product when the starting material is a previously oxidized carbon, sustain the idea that even in the thermal reaction no appreciable interconversion of reaction products occurs.31 The ratio CO,/CO in the electrochemical reaction indicates that COz is the main product, as is also the case, for instance, for the thermal oxidation of Spheron-6 in the temperature range from 25 to 400"C. 31 Another possibility, which is supported by the existence of two reaction products at a nearly constant ratio, is that CO comes from the thermal decomposition of any oxidized intermediate formed during the electrochemical reaction, giving a further support to the presence of at least two oxidized intermediates on graphite. This reaction being restricted, as in the case of CO2 formation, only to edge atoms of graphite,32 would explain why the anodic efficiency for CO, formation is less than 100%.
The participation of different oxidized intermediates is related to the existence of different types of oxides on the graphite surface. It was showP that at temperatures between 25 and 400°C the peripherical fragments of carbon involving one or two atoms can react with oxygen forming lactonic rings, while structures of carbonyl type exists on adjacent carbon atoms. From the study of chemisorption of oxygen on very clean carbon surfaces, 2o it was concluded that the adsorption of oxygen atoms on more reactive sites may lead to the formation of lactonic groups, while those adsorbed on adjacent carbon atoms forms carbonyl groups. According to this structure, car-bony1 groups under certain conditions, for example during electrochemical reaction, may lead to anhydride group formation, which by further thermal decomposition at intermediate temperature, would yield CO and CO,, according to the reaction path recently proposed by Coltharp and Hackerman for the thermal reaction. 31 The existence of various types of adsorption sites of different energy on carbons has been the subject of much discussion. They have been postulated to explain the thermal reaction of graphite with oxygen as well as other oxidation reactions on carbons.34-41 Graphite electrode oxidation in cryolite melts has also been explained by the existence of stable and metastable complexes formed between graphite and oxygen.42-44

The activafion energy of the reaction
The experimental activation energy for the electrochemical oxidation of graphite in bisulphate melts is high as compared to the activation energy values found frequentry in electrode processes, but the figure is not surprising when it is compared with the experimental activation energies obtained in the thermal oxidation of graphite to volatile compounds, between 250 and 45O"C, which are of the same magnitude. These figures are dependent on the number of existing defects and on the degree of graphitization of the sample. For highly graphitized carbons it lies between 40 and 50 Kcal~mo1.45-60 The more probable value recently given is 44 kcal/mo1. 47 The coincidence of the experimental activation energies for the thermal and electrochemical oxidation of graphite within the same temperature range is another indication that the kinetics of these processes are closely related.