Chronopotentiometric Approach of CO₂ Reduction in Molten Carbonates

Chronopotentiograms corresponding to CO₂ reduction in four alkali carbonate eutectics, Li-Na (52:48 mol%), Li-K (62:38 mol%), Li-Na-K (43.5:31.5:25.0 mol%) and Na-K (58:42 mol%) are depicted in Figs. 2a–2d. Even though these curves can roughly be seen as similar, they differ by the potentials of the plateaus and by their shape. In all cases, as expected, the transition time, τ (time at which the concentration of the active reactant at the interface is consumed²⁰), tends to decrease with the absolute value of the imposed current (higher currents consuming more rapidly the electroactive species). Another general feature is that quarter time potentials, E_τ/4 (related to the time elapsed since the application of current is equal to one-fourth of the transition time^21–23), in most of the cases (Figs. 2a, 2c, 2d) become slightly more negative as the absolute value of the current increases, which indicates that the corresponding electrochemical reaction is not fully reversible.²² This phenomenon is quite clear in the case of Li-Na-K at 600°C (Fig. 2c), but less pronounced at higher temperatures, 650°C in Li-Na (Fig. 2a) and 750°C in Na-K (Fig. 2d). In Li-Na (Fig. 2a), a first plateau is observed at around −1.1 to −1.2 V vs Ag/Ag⁺, corresponding to the main reduction peak of CO₂ observed at the same potential by cyclic voltammetry (CV) and that has been attributed to the reduction of soluble CO₂ into soluble CO.¹⁸ Another plateau at around −1.4 V vs Ag/Ag⁺ is due to the reduction of carbonates from the electrolyte. A similar CO₂ reduction plateau (II) is observed in Li-Na-K (Fig. 2c) but at slightly lower potentials, which is in accordance with a slower electrochemical reaction (E_τ/4 potentials are more affected at higher currents). The second plateau due to the reduction of carbonates, occurring at potentials values more negative than −1.6 V vs Ag/Ag⁺, shows again that systems are less reversible in Li-Na-K (600°C) (Fig. 2c) than in Li-Na (650°C) (Fig. 2a), which is mainly an effect of temperature. However, when zooming into lower currents of Li-Na-K chronopotentiograms, we find small plateaus at around −0.75 V vs Ag/Ag⁺, which are related to the CV peak observed at nearly the same potential in in previous work.¹⁸ Li-K (Fig. 2b), under a lower partial pressure of CO₂ (of 0.05 atm. instead of 0.5 atm.) shows clearly the same plateau, with a sharp shape which is probably due to adsorbed CO₂ reduction (E_τ/4 with a constant value); another plateau follows at less than −1.2 V vs Ag/Ag⁺ corresponding to the main CO₂ reduction peak observed by CV in previous work.¹⁸ The same tendency is observed for Na-K (Fig. 2d) at higher temperature but with a slight slope for the first plateau. The plateau beginning at −0.8 V vs Ag/Ag⁺ could be due to the reduction of pre-adsorbed CO₂, but without strong evidence. In the case of Li-Na carbonate eutectic at 700°C, at very low imposed currents, as shown in Fig. 3, chropotentiograms display a new reduction plateau at around −0.35 V vs Ag/Ag⁺, which was hypothetically attributed in a previous paper¹⁸ to the reduction of CO_2ads into CO_ads; in effect, by CV in Na-K at 750°C, a reduction peak clearly appeared at a similar potential, associated to the reversible CO_ads oxidation peak. The influence of temperature is evident when comparing Fig. 2a and Fig. 3, the plateau appearing at −1.2 V vs Ag/Ag⁺ in Fig. 2a appears only from −6.0 to −7.0 mA in the form of a poorly defined transition, which means that CO₂ at the electrode is depleted faster at 700°C than at 650°C. Moreover, an equally poorly defined transition appears between −4.5 and −5.5 mA and also disappears rapidly at more negative currents. Finally, in Li-K at 650°C at a graphite electrode (Fig. 4), two reduction plateaus can be observed: one at around −0.95 V vs Ag/Ag⁺ and the other at around −1.2 V vs Ag/Ag⁺ stable in the practical conditions used since the reduction of carbonates under these conditions is not observed until a potential between −1.6 and −2.0 V vs Ag/Ag^{+ 2}. These two plateaus probably correspond to Reactions 1 (around −0.95 V vs Ag/Ag⁺) and 4 (around −1.2 V vs Ag/Ag⁺), with two steps of one electron followed by the neutralization of  CO^{2 −}₂.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Fig. 5 shows the transition time dependence on applied current as a function of temperature, oxoacidity of the carbonate melt (), nature of the electrolyte and working electrode; the legends PI and PII correspond to the so called Plateau I and Plateau II that are not always present under the different conditions tested herein. As already mentioned, the quarter potential E_τ/4 tends to decrease with the increase in the transition time (related to the decrease of the absolute value of current), which indicates that the electrochemical systems involved are quasi reversible, as analyzed before by CV.¹⁸ Fig. 5a (Li-Na with = 0.5 atm. at a gold electrode) shows the evolution of E_τ/4 with temperature, and is related to different reduction processes, as mentioned before; around −0.1 V vs. Ag/Ag⁺ (plateaus I) and around −1.2 to −1.4 V vs. Ag/Ag⁺ (plateaus II). For plateau II (Fig. 5a), when increasing the temperature from 600 to 650°C, the transition time becomes larger, which is consistent with the fact that both the solubility and the diffusion coefficient of CO₂ increase with the temperature. Another observation is that the potentials (E_τ/4) are more negative at lower temperature (Figs. 2c and 5a, 5c), which seems to be related to a slightly slower kinetics. For plateaus I (Fig. 5a), the behavior is quite different at the two studied temperatures, with a constant transition time at 600°C, showing probably an adsorption phenomenon that is not present at 700°C. Thus, it is more difficult to conclude about plateau I (Fig. 5a), but as we mentioned it could be related to adsorbed species (CO_2ads reduced into CO_ads).¹⁸ Fig. 5b shows the influence of on the transition times in Li-Na at 650°C. The four series displayed can be divided in two groups; in the first one, the quarter potentials belonging to plateau I (PI) are between −0.75 and −0.85 V vs Ag/Ag⁺ and, in the second, the potentials of plateau II (PII) varying between −1.0 and −1.2 V vs Ag/Ag⁺. Transition times tends to increase with the partial pressure of carbon dioxide in the case of plateau II which represents the reduction of soluble CO₂ into soluble CO; this is in agreement with the fact that increasing the amount of CO₂ requires higher currents for being entirely consumed. Situation is more complex in the case of plateau I and we can just note that the transition times are larger at CO₂ 0.1 atm. than at 1 atm., which is difficult to explain. Evolution of the quarter time potentials is in favor of a quasi reversible system. In the case of a graphite electrode in Li-K eutectic at = 0.05 atm. (Fig. 5c), the influence of temperature for plateau II (E_τ/4 at around −1 V vs Ag/Ag⁺) is similar than in Fig. 5a: with potentials lower at 600°C and systems that are quasi reversible.

Zoom In Zoom Out Reset image size

Current reversal chronopotentiometry (CRC) tests were carried out for exploring the reversibility and the characteristics of the reduction product. CRC graphs at a gold electrode in Li-K eutectic at 650°C, = 0.1 atm., displayed in Fig. 6a, show that two plateaus are successively observed in reduction (P I and P II), as well as in re-oxidation (P III and P IV). However, when a reversal current is applied immediately after P I (Figure not shown), no reverse plateau is observed and P I appears to correspond to an irreversible reaction. Since the potential continues to decrease after P II (Fig. 6a), it is possible that P III (Fig. 6a inset) corresponds to the re-oxidation of the last product generated between 0.5 and 1 s and P IV to the re-oxidation of product P II; thus, it seems that P III and P IV are both linked to P II. The observation of two reverse plateaus when P II is re-oxidized may correspond to a quasi reversible two-step oxidation reaction, or to the oxidation of soluble CO and adsorbed CO as proposed by Chery et al.¹⁸ Another CRC at a gold electrode in Li-Na at 650°C, = 0.1 atm., shown in Fig. 6b, displays a well-defined plateau in reduction, corresponding to the potential of P II, and another one in oxidation P IV. The transition times ratio is : the reduction product is soluble.

Zoom In Zoom Out Reset image size

Transition time τ, for a reversible electron transfer, is linked to the current by Sand's law:²⁰

If the reaction presents adsorption steps involving more than one electrochemical species, τ expression will extend beyond Sand's law relation. To take into account adsorption phenomena, the relations listed in Table II could be considered for elucidating the predominant reaction mechanism and the number of electrons n transferred in each step. It should be outlined that to determine n, accurate values of C_OD^1/2_O are required as well as careful measurements. For the analysis of τ we have considered the chronopotentiograms taken at a graphite electrode immersed in Li-K eutectic at 650°C and of 1, 0.5 and 0.2; (e.g. Fig. 2b) as well as those taken at a gold electrode immersed in Li-Na-K electrolyte at of 0.5 atm. at 550 and 600°C (e.g. Fig. 2c). These data were the most reliable chronopotentiometric measurements among the experiments performed, in the sense that τ shows a clear linearity with one or more of the proposed models. Each series of plateaus were analyzed for each of the four proposed mechanisms by plotting different combinations of the data as shown in Table II. If any of the plots results in a straight line, the mechanism represented by that plot is dominant (e.g. Fig. 7). From the linear regression with the best linearity, we can deduce the possible predominant mechanism and from the slope the value of n. In the case of Li-K at 650°C, we took the value of from Meléndez-Ceballos et al.²⁴ (1.1 × 10⁻⁶ cm²·s⁻¹) and the concentration C⁰_O was calculated from the solubility obtained in previous research³: 13.5 mol L⁻¹ atm⁻¹. We also used other values of diffusivity and solubility given by Nishina et al.,²⁶ but the results were impossible to interpret, proving the necessity of counting on new sets of experimental data. In the case of Li-Na-K eutectic, as we did not have values from our own, the values of and C⁰_O for each temperature were calculated directly from Nishina et al.²⁶ Table III summarizes the results of the linear regression for the four mechanisms described and the number of electrons n obtained. Based on the linearity of the linear regressions, Li-K seems to change from the mechanism of simultaneous reduction of the species (SR) to the rapid equilibrium mechanism (FE) when decreasing . The value of n was found to be 1 electron for high concentrations and around 2 for lower . However, at very low concentration, the linearity is not so good and the error in n increases. For Li-Na-K at 550°C SR is the dominant mechanism with a number of electrons close to 2 for both reactions. When increasing the temperature and , the phenomena seems to be mixed between the SR and FE mechanisms, and the computed n decreases.

Table II. τ relations for electrochemical reactions including one adsorption step.

	Mechanism	Potting	Equation
SR	Simultaneous reduction of diffusing electroactive species	iτ vs τ1/2
	in solution and of adsorbed electroactive species
RA-RD	Reduction of adsorbed species	iτ vs 1/i
	previously to diffusing species
RD-RA	Reduction of diffusing species	(iτ)1/2 vs 1/i1/2
	previously to adsorbed species
FE	Equilibrium mechanism (rapid equilibrium between adsorbed and diffusing species)	Cτ1/2 vs C2/i

Zoom In Zoom Out Reset image size

Table III. Transition time analysis of the four cases proposed in Table II.

Li-K 650°C
	1 atm.		0.5 atm.			0.2 atm.
Mechanism	R2	n	R2	n		R2	n
SR	0.98	1.0	0.93	1.3		0.89	4.0
RA-RD	0.89	2.5	0.79	3.5		0.84	7.6
RD-RA	0.89	0.5	0.79	0.5		0.83	1.0
FE	0.96	1.8	0.95	2.5		0.97	5.6
Li-Na-K 550°C	0.5 atm.
	Plateau I				Plateau II
Mechanism	R2		n		R2		n
SR	0.99		1.5		0.98		2.0
RA-RD	0.94		4.3		0.69		5.0
RD-RA	0.91		4.7		0.69		4.5
FE	0.96		3.3		0.87		3.6
Li-Na-K 600°C	1 atm.
	Plateau I				Plateau II
Mechanism	R2		n		R2		n
SR	0.97		1.0		0.93		1.8
RA-RD	0.82		2.5		0.82		5.0
RD-RA	0.83		1.7		0.79		2.7
FE	0.96		1.7		0.96		3.6

In order to confirm our hypotheses on CO₂ reduction mechanisms, we performed chronoamperometric measurements. Two kind of curves were obtained with specific shapes according to the imposed potential. In the first one, displayed in Fig. 8a, we can observe at around −0.9 V vs Ag/Ag⁺ a distortion of the iτ plot with respect to a Cottrell behavior, which most probably indicates an adsorption step and is in agreement with the results obtained by chronopotentiometry. For the second one at around −1.2 V vs Ag/Ag⁺ (Fig. 8b), current decreases abruptly and then slightly until reaching the residual current. Cottrell plots (j vs 1/t^1/2) did not give strictly a straight line, which is supposed to be the case for a diffusion-controlled reaction. However, many parameters may influence the recording of the current (e.g. possible agitation by convection, electrode vibration due to convection, non-ideal stationary system), so that the hypothesis of a diffusion-controlled mechanism should not be discarded. Therefore, according to chronoamperometry, we conclude that CO₂ undergoes two reduction mechanisms. The first one at the higher potential is irreversible, as no reverse plateau (chronopotentiometry) or peak (by CV¹⁷) were observed, and is not diffusion-controlled. As this system is not always clearly observed, it might be due to the reduction of adsorbed CO_2, in specific adsorption sites, into CO⁻₂, which is unstable and, therefore, its oxidation cannot be observed. For the second at the lower potential, the reaction is quasi-reversible. Our work does not contradict previous hypothesis of a reaction involving two diffusing species, with a soluble product. This reaction corresponds, as also seen by chronopotentiometry and CV,¹⁸ to the reduction of soluble CO₂ into CO, but with different complex steps including adsorption phenomena.

Zoom In Zoom Out Reset image size