Zinc Ions Reduction on Solid Metal Electrodes in Chloride Melts

ZINC IONS REDUCTION ON SOLID METAL ELECTRODES IN CHLORIDE MELTS

Alex Lugovskoy *1a, Zeev Unger 1,2b, Michael Zinigrad 1c,

Doron Aurbach 2d

1Material and Chemical Engineering Department, Ariel University Center of Samaria, Ariel, 40700, Israel

2Department of Chemistry, Bar-Ilan University, Ramat-Gan, 52900, Israel

a, b, с, d

keywords: electrodeposition, chloride melts, cyclic voltammetry, high-temperature electrochemistry

Abstract

The reduction of zinc ions on solid tungsten and platinum electrodes in chloride melts at the temperatures 700 – 750 °C was studied by cyclic voltammetry, chronoamperometry and energy dispersion spectroscopy. It was established that no zinc is reduced on platinum electrodes. As for the reduction of zinc ions on tungsten electrodes, the process has a complex character: it starts as an irreversible two-electron zinc ion reduction and, after the new phase is formed, the process of saturation of the electrode surface with lithium or sodium begins. As the second process develops, the alkaline metal becomes essentially the only constituent on the electrode surface.

General

Since zinc is industrially recovered from sulfate solutions rather than from melts and because its melting temperature (419.5 °C) is lower than the temperatures of most molten chloride compositions, the reduction of zinc ions on solid electrodes in chloride melts has been investigated relatively poorly. There are quite a few papers devoted to the electrolysis of zinc containing chloride melts (1, 2) and these cover only some details of the electrochemistry of this metal. However, zinc is not only an engineering metal. It can often be a component of molten chloride systems, in which various processes of synthesis or purification are performed. Therefore, the detailed electrochemical behavior of zinc can be of great importance.

The study of electro-reduction processes of zinc ions on solid tungsten and platinum electrodes in eutectic NaCl – KCl and LiCl – KCl melts in the temperature range of 700 – 750 °C is presented in this work. These temperatures are somewhat higher than the eutectic points of NaCl – KCl (646 °C ) and LiCl – KCl (628 °C) and the melts are therefore "liquid enough" to be used in technologically important processes of lanthanides and actinides separation, reduction and rectification. On the other hand, these temperatures are significantly lower than the boiling point of zinc (907 °C) and there is essentially no loss of the metal due to evaporation.

Experimental

The electrochemical experiments were performed using a three-electrode cell, made of sintered alumina, placed in an alumina crucible under nitrogen atmosphere. Tungsten (99.95%, 1 mm diameter) and platinum wires (99.95%, 0.5mm diameter) were used as the working electrodes and their surface area was controlled by immersion depth (typically, 6–12mm) and by measuring their diameter before and after each experiment. A 1mm tungsten wire served as a pseudo-reference electrode, and a flat spiral tungsten wire, set perpendicular to the

working and reference electrodes close to the bottom of the cell, served as the counter electrode. The area of the counter electrode was ~ 20 fold as large as that of the working electrode. ZnCl2, LiCl, NaCl and KCl (99.0% +ACS grade, Alfa Aesar) were used for the preparation melts without further purification.

Zinc chloride was mixed with alkaline metals chlorides using mortar and pestle in a glove-bag in dry nitrogen atmosphere. The mixture was then placed into a crucible, the electrode cell was mounted and transferred into the furnace (single-zone Carbolite 1600 °C STF tube furnace). In the furnace the mixture was first dried under vacuum at 40–50 °C for an hour. After completing the drying, dry nitrogen was bubbled through the electrolyte during its heating up to the temperature of the experiments (700–750 ◦C) for another hour. The temperature was controlled by a type S thermocouple, placed next to the cell and protected by an alumina capillary, thus maintaining a precision of ±1 °C in measuring and controlling the temperature. Dry nitrogen atmosphere (1 bar) was maintained in the furnace during the measurements and the post-experimental cooling. The electrochemical measurements were carried out using an Autolab PGStat-12 potentiostat. SEM images and element analysis by EDS were performed with a SEM system from JEOL Inc., Model JSM 7000F.

Results and discussion

Deposition of zinc on a tungsten electrode

Some typical voltammograms for the electrochemical reduction of Zn(II) are shown in Fig. 1.

Fig. 1. Cyclic voltammograms related to the electrochemistry of Zn2+ ions (0.163 mol / L) in equimolar NaCl-KCl melt on a W electrode at 700°C. Scan rates are 50 mV / sec (solid line), 200 mV /sec (slashed line) and 500 mV / sec (dotted line). Each charge density was calculated as the sum of areas limited by the baseline and the appropriate current density curves for the forward and backward semi-cycles.

As follows from Fig. 1, a single cathodic peak C corresponds to one anodic peak A. The potential, shape and behavior of the cathodic peak are typical for the metal deposition on a solid electrode (2-4). No difference is observed between the reduction of zinc ions in NaCl – KCl and in LiCl – KCl melts. Peak A is assigned to the reoxidation of zinc. Both peaks are clearly not independent on the scan rate. Rather, peak C is shifted to more negative potentials and peak A moves to more positive potentials as the scan rate increases. The dependence of the cathodic peak potential on the scan rate is shown in Fig. 2. Such voltammetric response is typical for irreversible processes.

Fig. 2. Dependence of the cathodic peak potential on the scan rate for the reduction of Zn2+ (0.163 mol / L) at 710°C on a W electrode.

The cathodic peak C appears at about -0.6 V vs. tungsten electrode for the scan rate of 50 mV/sec and at -0.7 V for 500 mV/sec. Such a significant shift is a clear indication that the process is irreversible. The cathodic peak not only is shifted as the scan rate grows, but it becomes broader so that the difference |Ep – Ep/2| grows from 0.1 V for 0.05 V/sec to 0.15 V for 0.5 V/sec. Values of an calculated by equation 23, are in the range of 1.56 for low scan rates to 1.04 for high scan rates. The most logical interpretation of this finding is that the charge-transfer is of two-electrons, which is not surprising in the case of Zn2+ ions reduction. The value of a is then 0.78 for 0.05 V/sec and 0.52 for 0.5 V/sec. This is evident that the rate determining step is the Faradaic process

Zn2+ + 2e-à Zn,

when the system is close to the steady state. Note that at low enough potential scanning rates diffusion limitations may be less influencing while at higher scan rates the diffusion limitations are more important. Randles-Sevcik dependencies for the zinc (II) ions reduction demonstrate linearity, but their intercepts are apparently non-zero (Fig. 3).

Fig. 3. Randles-Sevcik plots for Zn2+ ions reduction on W in a NaCl-KCl melt at 700 °C, different concentration of the ions (peak C in Figure 39). ○ 9.00x10-5 mol/mL Zn2+; r 1.63x10-4 mol/mL Zn2+; ▲ 1.77x10-4 mol/mL Zn2+.

It is evident that the process Zn2+ + 2e-à Zn is complicated by something else. Despite the irreversible character of the deposition process, it is still reasonable to roughly evaluate the diffusion coefficient of Zn2+ according equation 1:

ip = 0.6105 ×(nF)3/2/(RT)1/2×D1/2×C*×n1/2 (11)

where ip is the peal current density (A / cm2), n is the number of electrons, F is Faraday constant (96500 C), R is the gas constant (8.314 J/mol∙K), T is the absolute temperature (K), D is the diffusion coefficient (cm2 / sec), C* is the bulk concentration of a Red (Ox) species (mol / cm3) and n is the scan rate (V / sec).

Thus calculated diffusion coefficients are shown in Table 1.

Table 1. Diffusion coefficients of Zn2+ to a tungsten electrode in NaCl-KCl melt.

C*×105, mol / L / D× 105, cm2 / sec
9.00 / 9.55/ana
16.3 / 10.20/ana
17.7 / 13.64/ana

Given that the value of n for the reduction of Zn2+ cannot exceed 2 and 0 ≤ a ≤ 1 (a ≈ 0.5 for most cases), reasonable values of ana must be close to 1-2. Therefore, the values of the diffusion coefficients from Table 2 lie in the range of 1-6∙10-4 cm2/sec. Available literature data for the diffusion coefficients of most metal ions lie in the range 10-5-10-4 cm2/sec. Particularly, T. Støre, G. M. Haarberg and R. Tunold found that the values of the diffusion coefficients for Zn2+ in KCl-LiCl melts at 400°C lie in the range 0.6 – 1.06∙10-5 cm2/sec (2). Delimarski provides the value of the diffusion coefficient of Zn2+ in NaCl-KCl at 710°C, which is 2.3∙10-5 cm2/sec (5). The deviation of our results from the literature data can hint that that the process cannot be treated as simple zinc ion reduction on the surface of tungsten.

It is worth to mention that the fact that the diffusion coefficient for zinc ions in the chloride melt lay in the range 10-4 – 10-5 cm2/sec may serve as an indirect argument in the discussion about the existence of complex species described by the general formula [ZnxCly]z+ in chloride melts. While some authors argue in favor of the formation of complex ions (6 – 10), other studies give evidence for the existence of individual zinc ions as the key reacting species (11 – 12). The relatively high values of the diffusion coefficients found in our experiments hint that the charge is transferred by individual ions rather than by more massive complex moieties.

Fig. 4. Randles-Sevcik plots for Zn2+ reduction on W in a NaCl-KCl melt for different temperatures. [Zn2+] = 9.00x10-5 mol/mL

Another intriguing aspect of the zinc ions deposition process on a tungsten electrode can be seen in the temperature dependence of Randles-Sevcik plots (Fig. 4). As seen from Fig. 4, Randles-Sevcik plots do not change (to the accuracy of the experiment) as the temperature rises from 700°C to 750°C.

The lack of dependence of Randles-Sevcik plots on the temperature is really surprising. A plausible explanation to this could be an additional process in the system, which occurs simultaneously with the observed process, but does not involve charge-transfer and cannot be detected electrochemically. Such a process could compensate for the expected increase of the slope of Randles-Sevcik plots as the temperature grows and thus distort the temperature dependence.

The most probable candidates for such competing processes are a coupled chemical (not charge-transfer) reaction or a process of phase-formation. However, cyclic voltammetry alone cannot discriminate between these two possibilities.

Fig. 5. A chronoamperometric plot for the deposition of Zn2+ on a tungsten

electrode. Temperature 725°C, [Zn2+] = 9.00x10-5 mol/mL. The potential was stepped from OCV to -0.55 V.

A further insight on the nature of the deposition process can be provided by chronoamperometry. As seen from Fig. 5, the current falls in the course of the first 11 seconds of the experiment and then rises, reaches a peak and gradually declines as expected with time until the end of the experiment (300 seconds).

The initial falling and rising of the current can be attributed to the nucleation of the deposits, fluctuations of current for more advanced reaction times as seen in Fig. 5, may indicate to a very active charge-transfer process, which cannot be explained by a simple zinc deposition process.

Even more surprising information is provided by EDS analysis of the working electrode after a 3000 second deposition experiment at – 0.55 V (Fig. 6, Table 2). The most striking result of the analysis is the unexpectedly high content of sodium on the electrode surface. This amount of sodium cannot be accounted for melt adhesion or penetration, because the percentage of potassium and chlorine is much smaller. In fact, the working electrode looks as it was made of sodium with moderate inclusions of tungsten and zinc, rather of tungsten.

Fig. 6. An EDS spectrum of tungsten working electrode after 3000 second

deposition at – 0.55 V. Temperature 725°C, [Zn2+] = 1.38x10-4 mol/mL.

Table 2. Element composition of the tungsten working electrode surface calculated from the EDS spectrum after 3000 second deposition at – 0.55 V. Temperature 725°C, [Zn2+] = 1.38x10-4 mol/mL.

Element / Na / K / Cl / W / Zn
At. % / 60.84 / 5.80 / 28.61 / 2.24 / 1.91

A somewhat similar phenomenon was reported by Thus T. Støre, G. M. Haarberg and R. Tunold for the deposition of Zn2+ on a glassy carbon electrode in KCl-LiCl melts at 400°C (2). They observed a “substantial residual current observed prior to the Zn(II) reduction peak”. This current was attributed by them to lithium intercalation into the lattice of the glassy carbon electrode.

Unfortunately, the data about standard reduction potentials of many important ions in molten chlorides are lacking. The only source, in which suitable potentials were found, is the book of Yu. Delimarski “Electrochemistry of Ionic Melts” (5). The values of standard potentials tabulated in this book, were calculated on the base a few assumptions and are far from being strictly thermodynamical. However, they are helpful from the practical point of view. The potentials relevant for this discussion are summarized in Table 3.

Table 3. Standard reduction potentials in molten chlorides (adopted from ref. [5]).

As seen from Table 3, the standard potentials of lithium and sodium are very close to each other. Therefore, it is not surprising that the interference from sodium in the deposition of zinc ions is similar to that of lithium, as reported by T. Støre, G. M. Haarberg and R. Tunold. Of course, it is not intercalation that serves as the moving force of the process of sodium penetration into the surface layers of zinc deposit on the tungsten electrode.

The large amounts of sodium in the deposits obtained in the study of the Zn2+ ions reduction on tungsten electrodes cannot be explained as the formation of a W-Na alloy, because such a process is not observed by the cyclic voltammograms of NaCl-KCl on tungsten electrodes in the absence of zinc ions (3). Therefore, it is zinc, which triggers the deposition of sodium. Moreover, the data obtained by chronoamperometry at E = – 0.55 V vs. W (Fig. 5) indicate that there are two sequential faradaic processes. The first of them is relatively weak and is completed after ~ 11 seconds. Then the second process starts and its current only grows with time. The first process can be related to the reduction of zinc ions and the formation of zinc deposits. As the electrode surface is covered by a layer of zinc, the interaction of this layer with Na+ ions begins. Apparently sodium ions are absorbed by the liquid zinc (Tm = 419 °C) and this facilitates their reduction at the potential so much more positive than the sodium reduction potential in the absence of zinc ( - 1.1 V vs. W). Both lithium and sodium are liquid at the temperature of the experiment and these two metals form on the electrode surface a liquid solution with zinc, which continues to absorb new portions of the lithium or sodium ions.