Citation
Redox mechanism of sulfide ion in molten lithium chloride-potassium chloride eutectic

Material Information

Title:
Redox mechanism of sulfide ion in molten lithium chloride-potassium chloride eutectic
Creator:
Waggoner, James Richard, 1952-
Publication Date:
Language:
English
Physical Description:
v, 108 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Dimerization ( jstor )
Electric current ( jstor )
Electrodes ( jstor )
Electrons ( jstor )
Glassy carbon ( jstor )
Ions ( jstor )
Molten salts ( jstor )
Oxidation ( jstor )
Sulfides ( jstor )
Sulfur ( jstor )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Lithium chloride ( lcsh )
Oxidation-reduction reaction ( lcsh )
Potassium chloride ( lcsh )
Potentiometer ( lcsh )
Spectrum analysis ( lcsh )
Sulfides ( lcsh )
Voltameters ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis(Ph. D.)--University of Florida, 1982.
Bibliography:
Bibliography: leaves 104-107.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by James Richard Waggoner.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
029895422 ( ALEPH )
09941117 ( OCLC )

Downloads

This item has the following downloads:


Full Text












REDOX MECHANISM OF SULFIDE ION IN
MOLTEN LITHIUM CHLORIDE-POTASSIUM CHLORIDE EUTECTIC

















By

JAMES RICHARD WAGGONER


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA 1982

















ACKNOWLEDGEIENS


The author wishes to express his gratitude to Professor H.A,

Laitinen f or encouragement and guidance throughout his graduate tenure. No matter what problems arose, Professor Laitinen was always concerned

with the author's professional and personal welfare, and this attitude will always be an inspiration.

Many thanks are also due to Professor R.G. Bates, who so kindly lent his services and advice, especially during the writing of this manuscript.

The assistance and understanding of the author's parents, to whom this thesis is dedicated, are most gratefully appreciated.


ii

















TABLE OF CONTENTS


Page


ACKNOWLEDGEMENTS---------------------------------------ABSTRACT-----------------------------------------------CHAPTER

1. INTRODUCTION-------------------------------------Historical------------------------------------Cyclic Voltainmetry------------------------Chronopotentiometry-----------------------Chronoamperometry-------------------------Previous Work--------------------------------2. EXPERIMENTAL-------------------------------------Cell Materials and Design--------------------Electrode Design-----------------------------Working Electrodes------------------------Reference Electrode-----------------------Counter Electrode-------------------------Sulfide Generating Electrode--------------Equipment-------------------------------------Temperature Control-----------------------Electrochemical Equipment-----------------Cell Cleanup and Salt Purification----------Experimental Procedure -----------


iii


ii


v


2
7
10 11 17 17

21 21 25 25 28 29 29 29 30


33











3. RESULTS --------------------- 35

Determination of Diffusion Coefficient of S
Ton--------------------------------------------- 35

Cyclic Voltammietry------------------------------- 38

Variation of Scan Rate----------------------- 38
Effects of Change in Concentration ---- 64 Temperature Effects-------------------------- 71
Current-Voltage Response of S on Pt
Electrode------------------------------------ 72

Chronopotentiometric Response of Sulfide --- 81 ESR Results-------------------------------------- 87

4. DISCUSSION AND CONCLUSION-------------------------- 92

REFERENCES------------------------------------------------ 104

BIOGRAPHICAL SKETCH-------------------------------------- 108


iv


















Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partical Fulfillment of the
Requirements for the Degree of Doctor of Philosophy REDOX MECHANISM OF SULFIDE ION IN MOLTEN LITHIUM CHLORIDE-POTASSIUM CHLORIDE EUTECTIC By

James Richard Waggoner

December, 1982

Chairman: Herbert A. Laitinen Major Department: Chemistry

The redox kinetics and mechanism of solutions of sulfide ion were studied in molten LiCl/KCl eutectic using cyclic voltainmetry, chronopotentiometry and ESR spectroscopy. The effects of scan rate, current density, and electrode material were investigated.

Temperatures were varied from 375'C to 4750C. Sulfide concentrations ranging from lxlO- M to lXlO- M were generated coulometrically from a Ni/NiS eutectic electrode.

Teh first oxidation process of sulfide ion involves a one'electron transfer to form the radical anion. An irreversible dimerization of the electrogenerated radical ensues, and the second order rate constant for the dimerization was estimated.


v
















CHAPTER 1
INTRODUCTION

Recently, there has been much interest shown in high temperature secondary energy sources (1-4). Examples of such high temperature systems currently being investigated are the sodium-sulfur cell, the aluminum-sulfur cell, and the lithium-iron sulfide cell. Due to their relative abundance and low equivalent weight, sulfur and sulfur compounds are popular as electrode materials in high temperature systems. However, lack of knowledge of the electrochemical reaction mechanisms of sulfur and its compounds in molten salt media has hindered a more complete understanding of the discharge-charge properties of these battery systems. Due to the attention being given to high temperature secondary cells of the type

Li/LiCl,KC1/MS x x = 1 or 2

it was felt that a study of the redox properties of sulfide ions in the LiCl/KCl eutectic would be beneficial to the further development of molten salt batteries.

Historical

The details of the physical structure and properties of molten salts has been dealt with extensively in several reviews (5-9). Therefore this presentation will be restricted to a discussion of two topics which are relevent to the present work.


1




2


First, the electrochemical methods which have been employed in this work will be reviewed. The theories as they apply to molten salt system.Ts will be analyzed and any criticisms or corrections will be made.

Second, past work dealing specifically with the electrochemistry

of sulfur and sulfide in molten salts, particularly the LiCl/KCl eutectic system,will be reviewed.

During the course of this work, three electrochemical methods were employed to help determine the redox mechanism of sulfide ion in LiCl/KCl. These methods are cyclic voltanretry, chronopotentiometry, and chronoamperometry.

Cyclic Voltammetry

The theory of cyclic voltammetry has been well developed by several noted workers (10-16). Briefly, the theoretical derivations involve the solution of Fick's equations for semi-infinite linear diffusion.

The solution is obtained by transforming Fic
i = nFA C' /rD-a X(ar)

for totally irreversible charge transfer,

i = nFA C' V5Thb X(bT)

to obtain the current at a particular potential.










Eqjuat ion


Preference


o + ne -~R

(Reversible charge transfer)


X(aT) [ [2V'y exp (-.U)


Tanh (u-y+z) d cosh 2(u-y+z)


k
0 + ne R


(Totally irreversible charge transfer)


Ajl


(1+l(vTr)~x exp [) ic~F(E-EO+
/(-j+ -1)RT


RT / o C(NaF Z k
s


o + ne R 2R -* z


(Reversible charge transfer
with irreversible dimerization)


X (aT) exp (aT) 1--1/ a


Case


Table 1


14.


14


x (T) d-


3/2


18


+ i / -Z
2 foy




4


The application of these equations to electrode processes in molten salts at other than room temperature is straightforward, involving only RT RT
a change in the factor -or -N F Figure 1 illustrates the effect
a
of temperature change on the value of X(ar) for a reversible one electron process. The curves were calculated by computer through the use of the first equation in Table 1. The integration was accomplished using a modified algorithm of Simpson's rule, and the values are accurate to + 0.5%.

It is necessary to introduce the equations for the theoretical cyclic voltammograms, since these equations will be used to construct baselines from which values of the current in multi-step processes can be obtained.

From the practical viewpoint, cyclic voltammetry is an excellent qualitative tool for the study of electrode mechanisms. However, its

use as a quantitative method should be applied with caution. The theory is derived by assuming semi-infinite linear diffusion. This condition does not apply at low scan rates where density gradients may cause significant convection, or where thermal gradients give rise to convective currents near or at the electrode surface. Techniques are available to determine if convection is significant within the time frame of the experiment (19). one of the better methods used, and the method of choice for this work, is the determination of the constancy of the quantity iT 12as i decays with time. Upon application of a constant potential to an electrode, a current decay is observed with time. If linear diffusion is the only mode of mass transport, the product of the current and the square root of time should be constant according to the Cottrell equation 1/2 F D1 2C

iT
I1/2























Figure 1. Variation in X(at) as a Funcation of Temperature






0.4




0.3

X (at)

0.2 7990K 291%2K


0.1




-0.45 -035 -0.25 -0.15 -0.05 005 0.15 0.25 0.35
E-Ey,9/n




7


where n = number of electrons, F = the Faraday (96500 coul/eq.), A = the area of the electrode surface, D = the diffusion coefficient, and C b= the concentration of the electroactive species in the bulk of the solution.

The constancy of the electrode area is another factor which can influence qualitative and quantitative results in cyclic voltammetry as well as other electrochemical techniques. Several phenomena can cause a change in electrode area. First, faulty insulating seals around the electrode deteriorate rapidly in molten salts, which can increase the electrode area and produce spurious currents. Second, dendrite growth or porous film formation during the deposition of many metals (20,21) has been found to cause significant changes in electrode area. The appearance of dendritic growths on a surface can also disrupt the diffusion layer, causing spurious currents and worsening the results. Third, alloy formation or chemical reaction of the deposited species with the electrode substrate can alter the surface area considerably. This effect is especially prevalent in high temperature molten salt systems (20, 22) .

Taking into consideration the above effects, one should use care when employing-cyclic voltammetry and for that matter other electrochemical techniques for quantitative work.

Chronopotentiometry

The essence of the method of chronopotentiometry is the measurement of the time interval between the start of constant current electrolysis and the point at which complete concentration polarization is achieved. This time is known as the transition time and is a function of concentration of the electroactive species. As with cyclic voltammetry,




8


most of the theoretical development has assumed semi-infinite linear diffusion as the sole mechanism of mass transport.

The theory of chronopotentiometry was first developed by Weber (23) and later verified by Sand (24). Sand showed the dependence of E VS T

to be applicable only to linear diffusion controlled processes. Delahay et al. (25,26) developed theoretical treatments for reactions that were not necessarily reversible or diffusion controlled. The extension of the method to systems involving chemical kinetics was achieved by Testa and Reinmuth (27-29).

The theoretical development of chronopotentiometry involves the

solution of Fick's second law with the appropriate boundary conditions. The resulting equation, known as Sand's equation, which releases the current density and transition time to the concentration of the electroactive species is

V2 F~r1/2 D1/2 b



where i 0 current density and T= transition time, the other parameters having been defined previously. This equation is valid for either reversible or irreversible diffusion controlled processes, since no assumptions concerning the kinetics of charge transfer were made in the derivation (30). By the addition of the appropriate boundary conditions, chronopotentiometric equations can be derived for a number of kinetic situations. Examples of these equations are given in Table 2.

Excellent reviews of chronopotentiometry in kinetic systems can be found in the literature (J9, 30-34) and the reader is urged to refer to these for a more complete treatment of the subject.











Table 2


Case

o + ne R o + ne R


Equation

E = E RT 9nIT 1/2 7t1/2
1/2 + nFT 1/2





E E +RT 9,[T 1/2_ 1/2 RT Z[1+ 1/2 nF 1/2 nF 1+K


K
R~ z


Reference

30


1/21/ 1/21/
2 (1+K) (k 1+k 2) 1/t/


31


o + ne R

k
2R z


RT 1/2 -t1/2 RT 8D klCo E =E' k + kn [I
nF 1/6 3 nF 3 Tr


32




10


The extension of the theory of chronopotentiometry to molten


salt systems is trivial, again involving only the RTfactor. However, as in cyclic voltainmetry, the method requires maintenance of linear diffusion conditions, a not always easy matter with molten salts. Previous work indicates (35-38) that precision on the order of 2-4% is possible with chronopotentiornetry in molten salts. In each case, care was taken to thermostat the system to at least + 31C. The position of the electrode with reaards to the direction of diffusion was also found to vary the results somewhat, probably due to the enhanced thermal and convective currents.

The problem of maintaining a constant electrode area is also present in chronopotentiometry. The deposition of solid products on the electrode surface frequently results in denciritic growths, which cause the surface roughness and area to change with time. This effect is especially prominent at high current densities. However at low current densities,

the electrolysis time may be so slow that convective mass transport and spherical diffusion become increasingly important, leading to inconsistent results.

Chronoamp ero m etry

This particular technique involves the application of a controlled potential pulse to the working electrode. The current is monitored as a function of time and varies according to the Cottrell equation mentioned previously. In this study, chronoamperometry was used for an accurate determination of the diffusion coefficient of sulfide ions. The method was also used to corroborate the results of the kinetic studies from cyclic voltamnmetry and chronopotentiometry.




11


Previous Work

Since the late 1960's several workers have studied the electrochemistry of sulfur and sulfide ion in molten LiCl/KCl (3,39-41.). Kennedy and Adanuo(39) found from cyclic voltammetry and controlled potential electrolysis of a sulfur solution that the reduction of sulfur in LiCl/KCl occurs in 2 steps, which they ascribed to

S + e &~
n n
S +-~-S2n n

According to the authors, the electrochemical reactions were not strictly reversible at 4201C, since the peak separation of the anodic and cathodic processes was greater than the theoretical 132/n mV, where

n is the number of electrons transferred. However, the authors reported that the peak separation did not vary with scan rate from 40 mV/sec to 200 mV/sec. Tischer and Ludwig (4) explained these two contradictory facts by assuming a catalytic mechanism

S 2+ e S2


S2 +S4 S2 +S4

A plot of square root of scan rate vs. peak current would readily confirm or deny this mechanism, but no mention of such a diagnosis was made.

Bodewig and Plairteck (40) studied the potentiometric behavior of the sulfur-sulfide couple in LiCl/KCl at 400-4501C. They observed Nernstian response of the sulfur electrode with changes in concentration of coulometrically generated sulfide ions. The E' of the sulfur-sulfide couple was calculated to be -1.008V vs the standard molar Platinum electrode (SMPE).




12


During coulometric generation of sulfide from a sulfur pool, they

noted the appearance of a blue color, which they ascribed to a polysulfide species S x Upon applications of a vacuum to the cell, or by applying a potential anodic of the equilibrium. potential for sulfide generation, the blue color was found to disappear. If sulfur was then added to the cell, the blue color returned, which supports their hypothesis of a

colored polysulfide. Other workers (42-45) have determined spectroscopically that the blue color is due to a supersulfide, S 2or S species. Both findings are consistent if an equilibrium of the type

S + 2e S
x x

x Sx 4_ 2 x/2

where x > 6 is assumed.

These values for x are consistent with the gas phase measurements by Beckowitz (46) of sulfur vapor composition. From vapor pressure vs temperature plots, he found that sulfur vapor consists mainly of S 7 units at 425'C, with a slightly smaller amount of S 8and S 6*The proportion of S 7 and S increased with increasing temperature.

Bodewig and Planibeck (40) found from chronopotentiometric experiments that the oxidation of sulfide obeyed the Sand equation with rhenium and gold electrodes. Their value for the diffusion coefficient of sulfide is 3.12x10- cm 2/sec. They make mention of the fact that this value of the diffusion coefficient is almost an order of magnitude smaller than 2+ 5 2
the lowest value for a divalent ion (D for Pb 1 .3x10 cm /sec), indicating that the diffusing species may be a large polysulfide ion, Sx All the evidence taken into consideration seems to indicate that the

potential and chronopotentiometric data obtained in this work was for the sulfur-polysulfide system, instead of the sulfur-sulfide couple.







Birk and Steunenberg (41) performed a more comprehensive examination of the sulfur-sulfide system in LiCl/KCl involving the temperature and scan rate dependence of the cyclic voltammograms of sulfide ion.

Figure 2 shows a typically cyclic voltammogram obtained in their study. From peak separation data, the authors conclude that the f irst electrochemical reaction is reversible at all scan ratEs investigated. This contradicts the f indings of Kenredy and Adamo (39) mentionled earl ier. They also claimed that this first electrochemical reaction is probably an unsymmetrical reaction, involving different numbers of electrons on charge and discharge.

Based on their literature review and the results of their experiments, Birk and Steunenberg hypothesized the following reaction scheme:

anod ic

1) 2S S 2+2e 2) S 2- 2S +2e


3) 3S+S ~2S 2 S4


cathodic

4) 2S + 2e S~ S2 5) s2 + e -S2


6) s2 + 2e -~2S


The existence of the supersulfide ion, S 2 in reactions (3) and

(5) has previously been proposed (47-50) to account for the deep blue color observed when both sulfur and sulfide are present in solution. E.S.R., T.R., and Raman data indicate that an S species may also be present in the LiCl/KCl melt (51-53). The contribution of the























Figure 2. Cyclic Voltammogran of a Solution of Li2S in LiCl/KCl According to Birk and Steunenberg











10





E0
F
Z:
LU
-5



-10


1.2 1.6 2.0 2.4 2.8




POTENTIAL (VOLTS VS. Li)




16


tetrasulfide ion, S 4 in reaction (3) is not expected to be significant, since higher polysulfide species in LiCl/KCl have been found to be unstable with respect to formation of sulfur and lower polysulfides (54,55).

Cleaver, et al. (54) examined the cyclic voltammetric behavior of Na2Sand Na 2S 22in the LiCl/KCl eutectic. They found Na2 Sto be practically insoluble in the melt at 420'C, while solutions of Na 2S 2.2 give cyclic voltammograms very similar to those of Li 2S. The authors found that the solutions of Na 2 S2.2 gradually lost sulfur, indicating that the disulfide ion may be unstable toward decomposition to sulfur and sulfide.
















CHAPTER 2
EXPERIMENTAL

Cell Materials and Design

A photograph of an assembled and dissassernbled cell

and cell housing is shown in Figure 3. Both the outer and inner cell housings were made of Pyrex. The inner Pyrex container was used to prevent the molten salt from inadvertently coming in contact with the outer Pyrex container. When this did happen, diffusion of Li+ and K +into the hot glass changed the composition and thermal properties of the Pyrex, causing weakening of the walls and cracking of the entire container on cooling.

Pyrex or any other silica containing substance was not used for any part of the cell which would be in contact with the molten salt. This is due to the highly basic and corrosive properties of sulfide ion.

Immediately upon contact with a sulf ide containing melt, silica becomes etched due, to attack by sulfide and introduces silicon containing compounds into the melt (56) Therefore all cell components were made of high pi~rity aluminum oxide (Norton Co., Worcester, Mass.). The melt container was made of high density Al 20 3(AH999), while the reference electrode and counter electrode compartments were made from a slightly porous Al 20 3(A1788). The porosity allowed electrical contact to be maintained between the different compartments while minimizing intermixing of the solutes.


17






































Figure 3. Photographs of Cell and Cell Housing





pjj




























- 20




21


The top portion of the outer container was fitted with 5 threaded pyrex openings (Ace Glass, Inc., Vineland, N.J.) by the Department of

Chemistry glass shop. These openings allowed electrode access to the cell. By inserting the electrodes through the threaded Teflon bushing and 0-ring assembly and tightening the bushings into the threaded pyrex openings, a good vacuum seal could be maintained.

Electrode Design

Working Electrodes

The construction of working electrodes with acceptable performance characteristics has always been a challenge in molten salt systems. For the study of non-corrosive solutes and solvents, the most common method of fabricating working electrodes is the glass to metal seal. In

instances where the wetting properties of the glass preclude the use of glass to metal seal, most workers have relied on a seal-free electrode. This is simply a metal foil connected by a thin contact wire known as a "flag type" electrode.

The glass seal construction was unacceptable due to the corrosive

nature of the sulfide ion mentioned previously. The flag type construction was found to be adequate for qualitative measurements, but gave

irreproducible results when quantitative measurements were attempted. Therefore, it was necessary to devise an electrode system which had a

definite working area, the desired geometry, and was inert with respect to attack by sulfide.

Figure 4 is a diagram of the electrode system which was found to give excellent results.





































Figure 4. Diagram of Working Electrode System




23


GLASSY CARBON (CENTER) A[2034 -iBORON NITRIDE (BN)


END VIEW


PYRE*X TUBE


GLASSY


CARBO!


SIDE VIEW


I BN


MY




24


The outer portion of the electrode consists of a cylinder of hot pressed boron nitride (Union Carbide, New York) A hole with a diameter slightly larger than the diameter of the electrode is bored through the center of the boron nitride cylinder. The electrode material consisting either of glassy carbon or Pt rod previously cleaned in boiling, concentrated HCl is then coated with a layer of Ultra Bond 552 high temperature ceramic adhesive (Aremco Products, Inc., Ossining, N. Y.). The electrode is then fitted into the boron nitride cylinder and the adhesive is allowed to dry at room temperature. After drying, the electrode is heated to 500'C for 2

hours in a muffle furnace in order to evaporate the organic portion of the ceramic adhesive. The temperature should be brought to 500'C slowly to prevent the formation of bubbles and voids between the electrode and the boron nitride.

Electrical connection to the instrumentation for glassy carbon

electrodes was made by wrapping Pt wire around the exposed glassy carbon at the top of the electrode. Gold paste was applied over the Pt wire and glassy carbon for mechanical strength and to improve conductivity. Nickel wire was then spot welded to the Pt wire and run through a length of 6 mm diameter Pyrex tubing. The Pyrex tubing was then vacuum collapsed around the end of the glassy carbon-platinum wire arrangement to complete

the electrode.

Platinum electrodes were spot welded directly to a nickel wire and run through a 6 mm diameter pyrex tubing. The tubing was then vacuum collapsed around the Pt rod-Ni wire junction. Working electrodes orePared in this manner were resistant to attack by sulfide ion. The ceramic adhesive prevented the melt from creeping between the electrode and the




25


baron nitride. No corrosive effects of sulfide on the ceramic adhesive were noted, and the electrodes usually remained intact throughout the entire experiment.

Reference Electrode

The reference electrode used in this study was the Pt/Pt 2+system. This reference electrode system was chosen because of the convenience of construction and good stability (20,21,57,58).

The Pt 2+for the reference compartment was generated coulometrically from a large Pt foil. The potential was allowed to stabilize for

approximately 30 minutes before the experiment was begun. The drift of the Pt/Pt reference electrode was checked against the following reference electrode of the second kind:




This reference electrode has a total drift of < 0.0005V over a period of several days (59), and was thus suitable for checking the performance of the Pt/Pt 2+system. In this manner, the Pt/Pt reference was found to be stable to within 0.002V over a period of 24 hrs. Counter Electrode

The counter electrode consisted of a graphite rod (Spectroscopic

Grade, National Carbon Co., Cleveland, Ohio) inserted into the counter electrode compartment. Figure 5 shows a diagram of the counter electrode in the compartment.

The Pyrex cap inserted over the top of the counter electrode

compartment prevented any chlorine gas generated at this electrode from escaping into the melt compartment. A continuous flow or argon or helium under positive pressure outside of the compartment swept all of the generated chlorine gas to the outside atmosphere.





































Figure 5. Counter Electrode Assembly




27


PYREX CAP

(-A[2 0 3 TUBE

___GRAPHITE ROD


4-1 MELT




28


Sulfid2 Generating Electrode

Previously, all studies of the electrochemistry of sulfide ion in molten LiCl/KCl have been accomplished by adding a weighed amount of Li 2S or Na 2S to the melt. This method resulted in the introduction of impurities such as sulfur and H 20 by virtue of the hygroscopic and air sensitive nature of the alkali metal sulfides. A much more convenient and accurate method of sulfide addition to the melt was developed by Liu et al. (60). This method involves the coulometric generation of sulifde ion in the melt from a Ni/NiS eutectic electrode. This eutectic electrode is easily prepared from the elements in a 2:1 mole

ratio of Ni:S. The nickel and sulfur are placed in a quartz tube under 200 mm Hg argon pressure and heated to > 700'C. The silver melt produced is the Ni/NiS eutectic. The metal can then be cast into any convenient shape for use as an electrode. In this work, 5 mm diameter rods of the Ni/NiS were used. Electrical contact with the sulfide electrode was made by welding a 2 mm diameter Pt rod to the electrode. A nickel wire was then spot welded to the platinum for connection to the outside of the cell container.

The coulometric generation of sulfide was found to be essentially

100% current efficient as long as the temperature was greater than approximately 4700C. For studies below this temperature, the sulfide was first generated at 4750C, then the temperature was lowered to the appropriate value. In each experiment the sulfide ion concentration 2+
was cross checked by titration with coulometrically generated Ni The concentration of sulfide was monitored potentiometrically by means of a Ni/NiS/S= electrode according to Liu et al. (60).




29


Equipment

Temperature Control

Temperature monitoring and control was accomplished by a Wheelco Panelmount Capacitrol coupled to a chromel-alunel thermocouple Inserted into the outer cell compartment. The heating element was a "Nevi-Duty" multiple unit furnace (Hevi-Duty Heating Equipment Div., Watertown, Wis.) with a water cooled heat dissipator located above the furnace. The temperature was calibrated against the melting point of zinc, and was controllable to within + 3'C.

Electrochemical Equipment

Cyclic voltammograms were obtained with a Princeton Applied Research (PAR) model 174A Polarographic Analyzer coupled with a PAR model 175 Universal Programmer. The three electrode system was used in all experiments.

The constant current source for the chronopotentiometric experiments was a Buchler Instruments D.C. Power Supply in series with a 5000Q precision variable resistor. This arrangement allowed current control to + 0.5%.

All potential measurements were made with a Hewlett Packard 34703A digital D.C.V., D.C.A., Q meter.

A PAR model RE0074 X-Y recorder was used to record all slow scan

(lO-200mv/sec) cyclic voltammograms. Rapid scan cyclic voltammograms and chronopotentiograms were recorded on a Tektronix model 549 storage oscilloscope fitted with a model 53/54C fast rise plug-in preamplifier.




30


cell Cleanup and Salt Purification

Before each experiment, the alumina melt container, electrodes, and electrode compartments were first boiled in 12M HCl for 2 hrs. This was followed by a 24 hr. cleaning in fresh 12M HCl in an ultrasonic bath. The final step involved boiling for 2 hrs in triple distilled and deionized water. The crucibles and electrodes were then dried in an oven at 140'C until ready for use.

The LiCl,'KCl eutectic used in this study was supplied by Anderson

Physics Laboratories, Inc., Champaign, Illinois. The salt was introduced into the cell compartment in powdered form and placed under vacuum at 300'C for 24 hrs. Electronic grade HCl gas (Matheson Co., East Rutherford, N.J.) dried over anhydrous magnesium perchlorate and a dry iceacetone trap was then passed over the salt as the temperature was raised. This treatment reversed the hydrolysis of LiCl:

LiCl + H 0 ~-LiGH + HCl
2

After the salt was molten, dry 0 2 gas pretreated in the same manner as the HCl gas was bubbled into the melt for 2 hrs. This oxidized any organic contaminants which may have been present. Dry HCl was again introduced to react with any oxide ions which may have formed from the

introduction of 0 .Finally, dry Helium or argon, passed over anhydrous
2'

magnesium perchlorate, hot copper turnings and a dry ice-acetone traD

was bubbled through the melt for 2 hrs. The melt purity was then checked voltaminetrically as described by Laitinen and Gaur (61). Figure 6 shows the experimental set-up, with a completely assembled cell, gas purification train and instrumentation.
























Figure 6. Photograph of Experimental System














oil




33


Experimental Procedure

The procedure used during each experiment was essentially the

same. After the blank current was determined using both the Pt and glassy carbon electrodes, sulfide was coulometrically added to the melt until the desired concentration was reached. The temperature was then lowered to 3751C and the system was allowed to equilibrate for 30 minutes.

Cyclic voltanimograms were obtained at scan rates from 0.01 V/sec to 75 V/sec with both types of electrodes. The switching potential was also varied to observe the effect on the reverse scan.

Chronopotentiograns were acquired at several different current

densities at both electrodes to check the constancy of the iT product.

In order to determine an acceptable value for the diffusion coefficient of sulfide ion at the various temperatures, the method suggested by Adams (62) was used. This method has proved extremely reliable in aqueous solvents, as evidenced by the work of von Stackelberg et al. (63), and should apply equally well in molten salts. It involves the application of a controlled potential pulse to the electrode, while monitoring the current-time decay curve. By plotting current vs. time and extrapolating to t=0, radial diffusion and convective effects at

unshielded electrodes can be neglected, and a true value of D can be calculated from the Cottrell equation.

At the highest temperature studied, a portion of the sulfide in

the melt was electrolyzed at constant potential and a sample drawn out for analysis by ESR spectroscopy. The sampling technique used was to place quartz ESR tubes just above the melt. A vacuumi was then applied




34




and the tubes were lowered below the melt line. The cell was then vented to helium pressure and the sample entered the tubes. A blank sulfide solution was also obtained in this manner.
















CHAPTER 3
RESULTS

Determination of Diffusion Coefficient of S Ion

As mentioned in the last section, the diffusion coefficient of sulfide ion was determined by observing the decay of current with time after application of a controlled potential pulse.

Since the oxidation of sulfide ion is followed by at least one chemical reaction as evidenced by this work as well as the work of

previous authors, it was necessary to measure the current at times short enough to neglect the chemical reaction. These times can be determined from a graphical plot of it 12vs t, which should be constant

for a system uncomplicated by preceeing or subsequent chemical processes.

Figure 7 shows a plot of it 12vs t for the oxidation of sulfide at three different temperatures. As indicated by the data, the i / factor becomes essentially constant for times less than about 10 m sec. Extrapolation to T-0 gives the appropriate value of it 1/2 for determining the diffusion coefficient. These values were obtained at the highest concentration of S =studied (9.8xlo- M) to minimize errors due to contaminants. Glassy carbon electrodes were used to avoid interference due to surface processes such as adsorption and metal sulfide formation.

The extrapolated values of it 12were substituted into the Cottrell e q u a t i o n i / F D 2


T1/2


35






















Figure 7. itl/2 vs t as a Function of Temperature










































30

t( SEC.x


(-~J
-I


0
0
A


2











1


60


9
A


0


0


o 4750C o 4250C A375 C


0
a
A


0


0
a
A


0


10


20


40


50


I I I


I




38


and solved for D. Table 3 gives values of D for sulfide ion at 375, 425, and 475'C.

Table 3

3750 4250 4750 D cm. /sec 2.6x10- +0.lxlO- 2.9x10- +0.lxlO- 3.lxlO- +0.2x10The values of D given in Table 3 were corrected for the slight change in concentration with temperature.

Cyclic Voltammetry

The cyclic voltammetric response of S ion in LiCl-KCi eutectic was studied as a function of temperature, scan rate, concentration, and electrode material. Scan rate was varied from 0.01 V/sec to 75 V/sec and the temperature varied from 375'C to 4750C in steps of 50'C. All experimental potentials were corrected for the increase in temperature and for the small IR drop between the working and reference electrodes. Concentrations ranging from hOxiG- Mj to lOxlO -2M in sulfide were examined at these different temperatures and scan rates. Finally, the effect of

electrode material was investigated to determine if adsorption played a significant role in the redox mechanism, and to lend support to the

proposed mechanism. Glassy carbon and platinum were the principal materials used as working electrodes in this work. Variation of Scan Rate

Since the largest variation in the electrochemical behavior of

sulfide occurs when the scan rate is varied, these results will be Presented first.

A typical cyclic voltammogram of a solution of Li 2 in molten LiCl-KCl eutectic at 3750C and [S I] 9.8x10 -3m is shown in Figure 8.


























Figure 8.





0


200





I-1o
LU


200




400


-1.55


- 1.35


-1.15


-0.95


-0.75


-0,55


-0.35


E ( VOLTS VS. 1M Pta"/Pt)


A NOD IC


SCAN RATE20mVS'


400


200 mV S-'




41


All potential scans in this work were made from negative to positive potentials, so that the first charge transfer process is the oxidation of S ion. This charge transfer can be represented by the following reaction

S Sn-2 + neBy examining the current voltage curve for this process, the value of "n" can be calculated. Several relations derivable from theory (17) are used for this purpose. The first of these relations is the equation describing the dependence of the Peak potential, E pand half-peak potential, E ID2on the value of "n" for a reversible charge transfer


E E =2.20-R p p/2 nF

The se cond relation involves the peak potential of the reverse cathodic process as well

E -B E 2.22 R
p(anodic) p(cathodic) nF

Table 4 gives the values of E E E and p(anodic)' p/2(anodic), p (cathodic) "n" for the assumed reversible oxidation of S

From the experimentally determined potential values, the number of electrons involved in the first anodic process is found to be n=1.3. one can see in Table 4 that the calculated value of "n" at polarization rates greater than 1 volt/sec. decreases with increasing scan rate, approaching the theoretical value of n=1.0. The data suggest that a different mechanism may be operating under conditions of rapid scan. By comparing cyclic voltammograms obtained at various scan rates, it is possible to derive information concerning a consistent overall mechanism for the first charge transfer. Figure 9 illustrates six current-voltage curves of 9.8x10- M S= at scan rates of 0.1, 0.2, 0.5, 1, 5, and






















Figure 9. Current-Voltage Curves of 9.8xlO- M Sulfide as a Function of Scan Rate











0.1) 0. 2 V /SEC


-1.35


-0.95


-0.55


0.5)2) 5)10 V/SEC

















-1.75 -1l45 -1,15 -0.85 -0,55 -0-25


E


E


w


-1.75




44


10 volts/sec. at 3751C. Several salient features can be seen in this Figure. The first aspect is that the peak potential of the first oxidation process (Peak I) stays constant at scan rates below approximately 0.5 V/sec. The peak potential then shifts anodically until a scan rate of 5 V/sec. beyond which it again becomes constant. Figure 10 plots the variation of the peak potential of peak I with scan rate.

The behavior of the peak potential with scan rate is indicative

of an irreversible dimerization of the electrogenerated product according to the following reaction scheme (18,64):

R- 0 + ne


20~ X

where in this case R is the S ion, 0 is the sulfur anion radical S and X is the disulfide ionS2

The determination of the homogeneous rate constant k 1can be made

by means of the following equation first derived by Sav~ant and Vianello


(65) .E p = E + 0.90 RT RT Zn 2RT T c'n k1R
p 1/2 + 2nF 3nF 3nF 3nF V

where ES/ is the half wave potential of the charge transfer step when the dimerization reaction is negligible, C Ris the concentration of substance R in the bulk solution, and V is the scan rate. The evaluation of E 12is the difficult step in the use of this equation. In this work, the value of E / was obtained from cyclic voltammograms at scan rates rapid enough so that the dimerization could be neglected. At the rapid polarization rates used, the current-voltaqe curve is described by the theoretical voltammogram for the uncomplicated reversible charge transfer. In this work EB and E / were found to be constant from 5 V/sec to 50 V/sec























Figure 10.








1.27


-E MI 1.20

1.1 B 1.16

1.14


1


o 3750 C A4250C o4750C


0


0


310


A1


0


a


0


2


3


4


SCAN RATE (VSC


A
0


5


6


7


V / SEC)




47


after correction for the slight IR drop. Using the values of the peak current and peak potential at rapid scan rates, E 12can be easily estimated (17). It must be stated, however, that since E 12is an exponential function of k1, slight errors in the measurement ofE1/ cause rather large variations in the rate constant. Therefore, care should be taken to obtain a value of the half wave potential which is as precise as possible. In the present work, the half wave potential could not be estimated to better than +0.005V and thus the error in k 1is approximately + 40%. The value of k 1for the irreversible dimerization

3 -1-_1
of S at 375'C is calculated to be 3x10 Z M S

As a check on the consistency of the proposed irreversible dimerization, a plot of peak height vs. square root of scan rate was made as shown in Figure 11. The equation describing the variation of peak height with /V for a reversible charge transfer followed by irreversible dimerization is given by (18)


i =0.27n 32F 32D1/2 ACV1/2
p R1/2 T1/2 b


where the various terms have been previously defined. The slope of the i vs /V line was calculated and solved for the diffusion coefficient.
p
Since the slope varies as the square of the diffusion coefficient, this is a rather sensitive test for conformity. The diffusion coefficient calculated in this manner at 3751C and 9.8x1- 3 ms is 3.4x10- cm 2/sec + 1.OXlO 5, in good agreement with the more accurate value of

2.6x10- cm 2/sec + 0.lxlO- determined by chronoamperometry.

The second feature that can be seen in Figure 9 occurs in the potential range between the first oxidation process and what appears to be







































Figure.




49


200













15






-20 iAx1O




10






50




00

50
0
0

0
0
0
0
-0





0 1 1/2 2




50


the second oxidation process. The current in the intermediate region is nearly constant, rather than decaying as would be expected for a simple, diffusion controlled charge transfer process. Observation

suggests that an intermediate electrochemical reaction is occurring between the two more obvious charge transfers. In order to discern the intermediate process moreclearly, a computer assisted subtraction procedure was applied to each of the current voltage curves. Briefly, the subtraction was accomplished by programming the computer to generate current-voltage curves using the series and integral equations given in the first chapter. The experimental current was entered into the computer and the generated current was subtracted from it. Allowance was also made for the residual charging current. The result was a set of voltammograms without interference from the current due to the first peak, allowing minor features to become more prominent. Figure 12 shows the results-of this procedure for [S I=9.8x10 M at 375'C on glassy carbon. It is readily obvious from the peak morphologies that there are at least two electrochemical oxidations (peak II and peak III) occurring after the initial oxidation of S ion to S

A graphical analysis of the effect of scan rate on the current and current ratio of the two peaks is given in Figure 13-15. These graphs show that the variation of the peak height with the square root of scan rate is not linear. Peak II increases exponentially with /V while peak III increases logarithmically with /V. In effect, peak II appears to increase with scan rate at the expense of peak III. The behavior of the peak currents indicates that the kinetics of the two oxidations are coupled in some way.






































Figure 12.




52


960- SCAN RATE(V/SEC)

5
840720600

480360- 1

0.2
240
0.1
120 -0.05


-1.10 -0.95 -0.80 -0.65


E




































Figure 13. Variation of Current for Peak II as a Function of (Scan Rate)1/'2




54


03


1200







900 600


a


0i Al


0


-4

0


* 375*
*~ 425"
* 475*


1 2


0


300







0


2





































Figure 14. Variation of Current for Peak III as a Function of (Scan Rate) 1/2




56


0


0


4


A


A3750

c3 4250


0 4750






-I A I I


1


2


1 /2


300


200

-A) 100


0





































Figure 15. Ratio of Currents for Peaks II and III as a Function of (Scan Rate)1!2



















6









4 ip(II)






2


o 375"


a


L3475*











a


a 425


a


0


2
vl'2 (SEC1/2)


58


0


0


0


0 0


I I I


1


I I


I I I





59


The features of the cathodic half of the cyclic voltammograms

appear to behave normally with increasing scan rate. Cathodic peak IV is a stripping peak, involving the reduction and subsequent cathodic dissolution of a sulfur film deposited on the electrode surface during the anodic scan. The number of electrons involved in the dissolution of the sulfur film is difficult to determine. However, the ratio o-f the

number of electrons involved in the deposition step to the number involved in the stripping step can be determined. This was accomplished by integrating the current-voltage, i.e. current-time, curves for the deposition and stripping. The values of the total charge Passed for each process was compared and the ratio found to be 1:2 for deposition and stripping, respectively. Therefore, if the number of electrons involved in the sulfur deposition can be determined, the value for the cathodic

dissolution of sulfur will be known. The total charge data were obtained at the lowest temperature studied (3750) in order to minimize errors due to vaporization and subsequent loss of material.

Cathodic peak V appears to involve the reverse of anodic peak I. The peak current varies linearly with /V and "n" was calculated to be

1.4+0.1 for this process. However, at scan rates on the order of 1-10 V/sec the peak potential shifts in the cathodic direction, indicating the possible presence of a preceding chemical reaction. It is more probable that other phenomena such as the reduction of two species with similar E01s cause this Particular effect. This is not difficult to imagine, since the sulfur cathodically stripped from the electrode surface in the previous step can exist in several polymeric forms at the temperatures involved.




60


Since the average number of electrons transferred in the striuOpinq step is most likely two, the formula for the product can be represented as S 2 < X =8, depending on the temperature and concentration at the electrode. The following cathodic process, designated as peak V, would then involve the reduction of several polysulfide species, each with EOts differing by a small value depending on the value of X. Evidence for this particular explanation can be seen by comparing the cyclic voltammograms at various temperatures (Figure 16). As the temperature is increased, peak V broadens by an amount greater than predicted for the reduction of a single species. A second peak begins to appear at a potential which is close to the peak potential of the oriqinal process. The appearance and growth of this second peak would seem to indicate the increasing presence of a reactant which is more stable at higher temperatures, and which possesses a similar E'. These facts are easily explained if the presence of several polysulfide species is postulated. The assumption of similar E'1s for each of the polysulfide species is reasonable, because the reduction potential for a species such as S5 would not be expected to be much different from species like S 6or S7 other evidence pointing to the similarity of various _polysulfide species is seen in aqueous polarography. The oxidation of sulfide and polysulfides occurs by means of similar mechanisms on a mercury drop electrode:

S 2-_ HgS + 2e 1< X < 4

In this case, E 12for this reaction is the same for all possible values of X.

Further support for this particular reduction mechanism is found in the value for the number of electrons transferred in the last cathodic step. The value of "n", as previously mentioned was calculated to be 1.4+0.1.





































Figure 16. Cyclic Voltanunograms of Sulfide Ton as a Function of Temperature.










600

400 200

0

200 400


-1.7


-1.35


375 0


-0.95


-0.55


62


5


I'


E


I


I








400 200 (Im A) 0

200 400




800

400 (a A) 0

400 800


-1.77


63


425


.1


-1.36


-0.96


E


-0.56


47


-1.37


-0.97


E


-0.57


a


I


I


I


I


I


6




64


According to theoretical considerations put forth by Polcyn and Shamn

(66), electron numbers intermediate between whole numbers are indicative of either a multistep reduction with similar E0 s or the reduction of more than one component, again with similar E~s.

The appearance of a second peak with increasing temperature can be thought of as due to the reduction of lower polysulf ides formed from the thermal decomposition of the higher polysulfide species. This behavior is consistent with the fact that the higher polysulf ides (X > 3) are known to be unstable in LiCl/KCl at temperatures above 4000C. No data on the stability of these compounds below 400'C have been published. Effects of Change in Concentration

Figure 17 shows the current voltage response of S ion on glassy carbon as a function of concentration at a scan rate of 0.1 V/sec.

The striking aspect of these curves is the rapid change with concentration in the ratio of the currents of the first and second anodic processes, and the lower response of the third process at the lower concentrations. Table 5 gives the ratios of the peak currents of anodic processes I and II as a function of concentration of S .The values of i A(11)/i A(I) are plotted in Figure 18, and are the average of two separate experiments with three determinations for each experiment.

The peak potentials of the anodic reactions at the lower concentrations could not be accurately measured since the standard potentials of the three separate processes are close enough to cause significant overlap of each wave. At concentrations greater than about 6x10_ M S= the value of -1.200+0.003V vs SMPE at 4250C for the potential of peak I was measured. This value was measured at 100 mV/sec on glassy carbon and found to be invariant at these higher concentrations.





































Figure 17. Cyclic Voltanmograns of Sulfide Ion as a Function of Concentration of Sulfide




66


6.1 x104 m


-1.35


E


-0.95


-0.55


- 3.0 x10-' M


-175 -1.35 E-09-05


80


40


-1.75


6001 300


0


300


&


i


-1.75


-0.55


-0.95




67


80

40

0

40


-1.35


-0.95


-0.55


E


3.4x 10 -4M





L


-1.35


-0.95


-0.55


E


1.7 x10 -4N


-1,75


80


40


-1.75


i


I


I


I


a


I


I


I


a


i) (,uA )




68


Table 5


IS]1



1.xo-4
3.4 xl 04 4. 3x10 4 5. 0x10 4 6.1lx10 4 6. 6x10 4 9. 3x10 4 2. 5x10-3

4.2x10- 3M 9.8x10- 3M


A (II) /i CT()


2.5 + 0.3 1.9 + 0.3 1.5 + 0.2 1.4 + 0.2 0.9 + 0.2 1.0 + 0.2 o0.85 + 0 .2

0.58+ 0.15 0.47+ 0.15 0.32+ 0.10





































Figure 18. The Ratio of the Currents Due to Peak I and II as a Function of Concentration of Sulfide




70


0





0







0




0








0

0

0

0
0

0





0




I I I I I I I

) 1 2




71


No significant variations in peak structure on the cathodic side of the cyclic voltammogram could be detected. Temperature Effects

-3
Cyclic voltammograms of 9.8x10 M S at 100 mV/sec on a glassy carbon electrode aie shown in Figure 16 for 3750, 4251, and 4750C. By comparing the anodic portions of each curve, the decrease in the height of peak III with increasing temperature can be observed. A corresponding increase in the current due to the second anodic process accompanies the decrease of peak III. This phenomenon is also seen at both the lower and h:Igher scan rates studied. Again, a kinetic coupling between the two proc, ;esses is suggested.

For the same electrode and sulfide concentration, Figure 15 illustrates the variation of the ratio of peaks TI and ITT as a function of temperature and scanl rate. The results of these experiments provide more evidence for kinetic coupling, as will be discussed in more detail

in the "Conciuslions" section.

The dimerization rate constant, ki, was also determined as a function of temperature.' Table 6 displays the results from 3750 to 475'C.

Table 6

[S ]=9.8x10 M 3750 4250 4750

-1 3 3 3
k ,~M S 3.OxlO l.7x10 l.Ox10 +0.9x10 +0.7x10 +0.6xl0




A possible explanation for the anomolous behavior'of the rate constant with increasing temperature will be reserved for the next chapter.

--4m




72


Current-Voltage Response of S on Pt Electrode

This particular portion of the section on cyclic voltammetrv is presented as a separate subject since good quantitative data were difficult to obtain on Pt. The cause was the occurrence of several surface processes during the redox cycle of S .These surface processes

caused skewing of the cyclic voltammograms, especially at high scan rates and temperatures. However, it was felt that the electrochemical and chemical surface reactions could be used to enhance the knowledge of the redox mechanism of sulfide.

Figure 19 gives cyclic voltammograms of 9.8x10- M S= on a platinum disk electrode at 3750, 4250, and 475'C. Scan rates of 20 and 200 mV/sec are shown to illustrate the effect of polarization rate. Several surface peaks not found with glassy carbon are seen on the cathodic scan of the cyclic voltammogram. These peaks are labeled I sand IT sfor convenience.

At 3750C, both peak I sand ITI are seen at scan rates from 20-200 mV/sec. For scan rates above 500 mV/sec only peak Is is found.

A slightly different situation occurs at 425'C. Peak Is does not appear until 90 mV/sec while Peak IIS is present up to 200 mV/sec. Peak I~s dissappears above 1 V/sec, while the reduction process causing. peak Is continues to occur.

At 4751C, peak Is does not appear at any scan rate, while peak I~s exists at all scan rates.

This behavior suggests that peak I sis due to an adsorption process, which becomes less significant at higher temperatures. Peak II on the

other hand is probably due to a chemical reaction between a sulfur species and the platinum surface, which becomes more prominent with increasing temperature.





































Figure 19. Cyclic Voltainmograms of Sulfide Solution on a Platinum Disk Electrode




74


ES 375" 100

0

10020 0
-1.75 -1,35 -0.95 -0.55




2004250 100

0

100

200F

-1.76/ -1.36 -0.96 -0.56
E















475




I2000


A.A) 2OmVf


A) 20VS


-0.57


75


0


1


-1.77


-1.37


-0.97


E


i


I




76


An attempt was made to determine the potential at which these surface peaks formed. This was done by initiating the scan at a potential more negative than the first anodic peak and then reversing the direction of scan at different potentials. Figure 20 gives the results of this experiment at 425'C. Peak I occurred if the potential scan was reversed at the first anodic wave. At this point there was no evidence of peak II At the beginning of the second anodic wave, peak II began to appear. This implies that the product of the first oxidation peak is responsible for peak I while the oroduct of the
5
second oxidation wave is responsible for peak II

The surface peaks were integrated to obtain the total charge passed in the reduction process. The number of coulombs as a function of time is plotted for peaks I sand II sin Figure 21. T=O was taken at the foot of anodic peak I. From this graph, the rate of adsorption for peak I sis seen to decrease with time, indicating a simple adsorption process. The coverage of the electrode surface at all times shown is less than one monolayer. At the maximum time of 5 sec the coverage was estimated to be approximately 0.6 of a monolayer. Peak II s, however, behaves irregularly, providing more evidence for something other

than simple adsorption.

Finally, calculations were made to determine the number of electrons involved in the reduction of the adsorbed species in peak T However, the theory from which the equations were developed does not apply in this case, due to the following chemical reaction during the anodic scan. The value of "n" is therefore unknown for either of these processes.





































Figure 20. The Effect of Reversal Potential on Cyclic Voltanmogram of Sulfide on Platinum




78


200 100 (,"A) 0


100 200r
-1.76 -1.36 -0.96 -0.56
E





































Figure 21.




80


8 0





6


COUL.









2

0

~Is 123 4 5
t (S EfC




81


Chronopotentiometric Response of Sulfide

In this work three concentrations of sulfide were studied with chronopotentiometry and current reversal chronopotentiometry. The current was varied over an order of magnitude so that any trends in the response could be easily seen. A typical chronopotentiogran for the oxidation of S on glassy carbon is shown in Figure 22.

No attempt to calculate kinetic parameters was made since the

theoretical chronopotentiometric response for multistep charge transfers coupled by dimerization of the primary product cannot be solved analytically. However, trends in the value of iT 1/2 as a function of temperature, concentration, and current density can be used to support or rule out certain mechanisms.

Table 7 gives the results at 425'C for the controlled current

oxidation of sulfide as a function of current and concentration. The working electrode in this case is glassy carbon. The transition times were measured graphically between lines drawn tangent to the lines representing the potential shift.

Figure 23 graphically represents the current vs iT 1/2 data for the three concentrations.

The data show that at low current densities, iTl/2 is constant.

At higher current densities, the value of iT 1/2 increases, approaching a new constant value. At high concetrations, large current densities caused the chronopotentiograns to become noisy in the potential region of sulfur deposition. The exact cause of this noise could not be determined.






































Figure 22.

















- 0.080


E


-1.280


1.0 t (SEC)


83


0.2


1.8


I


I


I


I












[SI

5.6x10- 3M


7x10- 3M


332

433 549 650 759

854 961 1060 2050 3270


*1/2 1/2 4 -5
iE AmD.Sec xlO E2x10


i.Arnperes x 106

110 267 358 533 707

1051 1327

2734


-3


9.8x10- 3M 240 372 570 625 770 950 1236 1862


84


Table 7


4.3 4.2 4.4 4.3 4.3 5.1 5.5

6.4


4.6 4.7

4.6 4.8 5.0 5.2

5.3 5.3

7.4

8.3


8.2

8.4 8.3

8.4 8.4 8.4 9.1 10. 3





































Figure 23.




86


o 5.6x1O3M A 7x1OF3M


3





2





1


0

0
0


I i


6
T21/2(A


7


-SEC1/2 x


13


9.8x1OF M S













0


4


5


i


8


9


10


10 4


2


i


I


I




87


Temperature variation caused no significant differences in the chronopotentiometric results. The current density at which i / becomes non-linear is about the same differing by less than the error in measuring T This behavior appears to contradict the results obtained for the determination of the dimerization rate constant from cyclic voltammetry. This will be discussed in detail in the next section.

ESR Results

The purpose of this experiment was to determine if a chemical

equilibrium was established between S and electrogenerated S radical.

Samples of the sulfide containing melt were potentiostatically oxidized at -1.225 V vs SMPE on a large glassy carbon rod at 475'C. This potential occurs approximately 0.06 volts before the first anodic peak, and was chosen to insure that no anodic process other than oxidation of S to S took place. The total charge used during the sample preparation was calculated by integrating the current-time curve obtained by monitoring the current on a high impedance X-t recorder. The concentration of S_ generated was l.5x10- M assuming 100% current efficiency for the one electron transfer. The samples were drawn into several quartz ESP. tubes and quickly cooled, usually in less than 20 seconds. The frozen samples were then analyzed by ESR to determine if a radical species was present.

Figure 24 shows the results obtained at two sensitivities. These results show that down to the limit of detection of the instrument, i.e. approximately lxlO 1 M, no radical species existed in these samples.






































Figure 24.




89


GAIN- 4x 105 ABS







-0.2 -0.1 0 0.1 0.2 SCAN [RANGE (GAUSSx1 4)




90


GAIN -1.25 x106 ABS






-0.2 -0.1 0 0.1 0j 3 .2 SCAN RANGE (GAUSS i)




91




The large absorptions indicated bv the arrows were due to

cavity background, and the small signals seen at the higher sensitivity were also present in the blank.
















CHAPTER 4
DISCUSSION AND CONCLUSION

The results presented in the last chapter have shown that the

redox mechanism of sulifde ion in molten LiCl/KCl is more complicated

than previously thought. These results contradict some of the Previous work done on the sulfide system in LiC l/KCl. This situation is not

surprising, considering the various ways in which some of the studies were carried out. Most early workers used Pyrex or quartz containers or frits for work in saturated sulfide melts. All of these workers noted severe etching of the glass components even after an exposure of only a few seconds to the sulfide solution. Raleigh et al. (56), were the first to note altered electrode kinetics of metals in sulfide solutions exposed to Pyrex or quartz. They found that the electrode surfaces had been coated with a layer of silicon-containing compounds such as Na 2Sio2 S. Since the etching process is rapid, and presumably does not stop at the surface layers of the glass, variations in the concentration of sulfide in the molten salt will occur, especially if less than saturated solutions are used. Soluble silicon-containing compounds might also be expected to contaminate the melt.

Birk and Steunenberg (41) used quartz frit compartment separators in their study of the sulfide system. They also chose to use graphite as the working electrode material. This is not considered to be the best choice, due to the rather high porosity of the graphite and the


92




93


tendency for carbonyl and other functional groups to be present initially on the surface. The existence of electroactive surface

groups could lead to false assumptions concerning the redox mechanism of sulfide, since spurious currents due to the changing electrode area and oxidation or reduction of the surface layer would be encountered.

other workers also chose to use electrodes with areas that were

not well defined. For example, Raleigh et al. (56) used graphite rods and platinum and molybdenum strips dipping into the solution as working electrodes. Such a procedure is inadvisable in kinetic studies

since the LiCl/KCl melt has a tendency to creep along surfaces which are easily wetted by the melt, resulting in changes in the electrode area. Altered electrode kinetics could also result, due to thin layer effects produced by such solvent creeping.

For these reasons, no Pyrex or quartz was employed in this work. The wotking electrode system of glassy carbon and platinum sealed in

boron nitride gave a non-porous surface with a seal which was essentially leak proof during the time of the experiment. The cell and cell comnpartments were constructed of high purity A12 0 3 which showed no sign of attack by the sulfide.

As mentioned previously, the results show that the first stepD in the oxidation of sulfide ion in LiCl/KCl at the temperatures studied is the one electron transfer to form the sulfur anion radical, S .If a following chemical reaction is assumed, several possibilities exist for the product of such a reaction. The first is disproportionation. This mode of radical combination is well known in several aqueous and organic reactions such as the disproportionation of 2Cu(T) to Cu + Cu(TT) and 2C 2H 5*to C 2H 6+ C 2H 4




94


In order to determine if disproportionation occurs to any significant extent, the cyclic voltammograms and chronopot ent io grams were analyzed assuming this particular mechanism. The theoretical considerations used

in this analysis were due to Saveant and Nadjo (67) and Saveant and Mastragostino (68). The results showed that disproportionation could not be a significant factor in the electrochemical mechanism..

The second Possibility is radical recom~bination to form a dimer.

Again the results given in the last section were analyzed and found to conform to the theoretical descriptions given in the literature. The data obtained from chronopotentiometry, cyclic voltammetry, and chronoamperometry were all self consistent with regards to the proposed dimerization step. For an oxidation process in linear sweep voltamnietry, the shift in peak potential from a constant value at low scan rates to a more positive constant value at rapid scan rates indicates the presence of a following irreversible chemical reaction. If any long lived compounds formed between the electrogenerated S and the solvent ions, i.e. Li+ or K +, are considered unlikely then the dimerization of S_ is the most probable result.,

other evidence for the dimerization mechanism comes from the values of the peak potentials and half peak potentials of the first oxidation process. According to theory, E E =1.51 ELfor a totally irp p/2 nF

reversible dimerization. As can be seen in Table 4, the values of E P- E p2fit the equation to within + 8%. This accuracy of agreement cannot be obtained with similar equations describing other mechanisms.

Comparison of the results for the determination of the diffusion coefficient of sulfide ion also supports the dimerization scheme, as shown in the last section. However, the large errors due to the




95


square root dependence of the diffusion coefficient make it necessary but not sufficient evidence for the proposed reaction scheme.

The decrease in-the dimerization rate constant with increasing temperature is difficult to explain by any mechanism. The fact that

only three temperatures were studied coupled with the large standard deviations in the values of the constant casts doubt on whether or not the observed trend is significant. If indeed the trend is significant, then an explanation should be made which deals satisfactorily with the anomolous behavior of the rate constant. One possible reason for this behavior is the following. Free radicals are known to dimerize with little or no activation energy and thus show no increase in the rate constant with increasing temperature. The radicals in this work, however, are anion radicals and it can be assumed as a first approximation that any activation energy present is purely electrostatic in nature. If the trend in the value of the rate constant is significant, then the electrostatic activation energy must increase, resulting in a decrease in the effective collision rate with increasing temperature.

The decrease in the number of collisions can be thought of as due to the decreasing effect of ion pairing between the solvent cations, mainly Li +, and the anion radicals. At lower temperatures, the ion

pairing between Li + and S lowers the electrostatic repulsion between two S ions, and allows the close approach of another S or Li +S At higher temperatures, the ion pairs probably exist to a lesser extent and farther apart, Providing less electrostatic shielding and therefore increasing the electrostatic barrier to collisions. It must be reemphasized that the preceding is only one possible explanation for the




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EHF545IBR_JXNCQT INGEST_TIME 2017-07-11T21:10:08Z PACKAGE AA00040009_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

REDOX MECHANISM OF SULFIDE ION IN MOLTEN LITHIUM CHLORIDE-POTASSIUM CHLORIDE EUTECTIC By JAMES RICHARD WAGGONER A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLHENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1982

PAGE 2

ACKNm-vLEDGEr1ENTS The author wishes to express his gratitude to Professor H.A. Laitinen for encouragement and guidance throughout his graduate tenure. No matter what problems arose, Professor Laitinen was always concerned with the author's professional and personal welfare, and this attitude will always be an inspiration. Many thanks are also due to Professor R.G. Bates, who so kindly lent his services and advice, especially during the writing of this manuscript. The assistance and understanding of the author's parents, to whom this thesis is dedicated, are most gratefully appreciated. ii

PAGE 3

TABLE OF CONTENTS Page ACKNOWLEDGEMENTS -------------------------------------ii ABSTRACT ---------------------------------------------v CHAPTER 1. INTRODUCTION -----------------------------------1 Historical ----------------------------------1 Cyclic Voltammetry -----------------------2 Chronopotentiometry ----------------------7 Chronoamperometry -----------------------10 Previous Work -------------------------------11 2. EXPERIMENTAL -----------------------------------17 Cell Materials and Design -------------------17 Electrode Design ----------------------------21 Working Electrodes -----------------------21 Reference Electrode ----------------------25 Counter Electrode ------------------------25 Sulfide Generating Electrode -------------28 Equipment -----------------------------------29 Temperature Control ----------------------29 Electrochemical Equipment ----------------29 Cell Cleanup and Salt Purification ---------30 Experimental Procedure ----------------------33 iii

PAGE 4

3. RESULTS ----------------------------------------35 Determination of Diffusion Coefficient of S Ion ----------------------------------------35 Cyclic Voltammetry --------------------------38 Variation of Scan Rate -------------------38 Effects of Change in Concentration -------64 Temperature Effects ---------=------------71 Current-Voltage Response of Son Pt Electrode -------------------------------72 Chronopotentiometric Response of Sulfide ----81 ESR Results ---------------------------------87 4. DISCUSSION AND CONCLUSION ----------------------92 REFERENCES ------------------------------------------104 BIOGRAPHICAL SKETCH ---------------------------------108 iv

PAGE 5

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partical Fulfillment of the Requirements for the Degree of Doctor of Philosophy REDOX MECHANISM OF SULFIDE ION IN MOLTEN LITHIUM CHLORIDE-POTASSIUM CHLORIDE EUTECTIC By James Richard Waggoner December, 1982 Chairman: Herbert A. Laitinen Major Department: Chemistry The redox kinetics and mechanism of solutions of sulfide ion were studied in molten LiCl/KCl eutectic using cyclic voltarnmetry, chronopotentiometry and ESR spectroscopy. The effects of scan rate, current density, and electrode material were investigated. Temperatures were varied from 375c to 475C. Sulfide concentrations ranging from lxlO-4 M to lxlO-2 M were generated coulo-metrically from a Ni/NiS eutectic electrode. Teh first oxidation process of sulfide ion involves a one' electron transfer to form the radical anion. An irreversible dimerization of the electrogenerated radical ensues, and the second order rate constant for the dimerization was estimated. v

PAGE 6

CHAPTER 1 INTRODUCTION Recently, there has been much interest shown in high temperature secondary energy sources (1-4). Examples of such high temperature systems currently being investigated are the sodium-sulfur cell, the aluminum-sulfur cell, and the lithium-iron sulfide cell. Due to their relative abundance and low equivalent weight, sulfur and sulfur compounds are popular as electrode materials in high temperature systems. However, lack of knowledge of the electrochemical reaction mechanisms of sulfur and its compounds in molten salt media has hindered a more complete understanding of the discharge-charge properties of these battery systems. Due to the attention being given to high temperature secondary cells of the type Li/LiCI,KCI/MS x x = 1 or 2 it was felt that a study of the redox properties of sulfide ions in the LiCI/KCI eutectic would be beneficial to the further development of molten salt batteries. Historical The details of the physical structure and properties of molten salts has been dealt with extensively in several reviews (5-9). Therefore this presentation will be restricted to a discussion of two topics which are relevent to the present work. I

PAGE 7

2 First, the electrochemical methods which have been employed in this work .will be reviewed. The theories as they apply to molten salt will be analyzed and any criticisms or corrections will be made. Second, past work dealing specifically with the electrochemistry of sulfur and sulfide in molten salts, particularly the LiCl/KCi eutectic system,will be reviewed. During the course of this work, three electrochemical methods were employed to help determine the redox mechanism of sulfide ion in LiCl/KC1. These methods are cyclic voltammetry, chronopotentiometry, and chronoamperometry. Cyclic Voltammetry The theory of cyclic voltammetry has been well developed by several noted workers (10-16). Briefly, the theoretical derivations involve the solution of Fick's equations for semi-infinite linear diffusion. The solution is by transforming Fic:k's equations into non-linear integral equations and applying numerical techniques (17) to obtain values of the current function, x(aT), as a function of potential. Examples of the equations,used to calculate values of x (aT) for various mechanistic situations, are shown in Table 1. To derive the current, the values of x (aT) or X(bT), calculated from these equations, are substituted into one of the following two equations (17): for reversible charge transfer, i = nFA Co lIT Da X (aT) for totally irreversible charge transfer, i = nFA Co mi) X (bT ) to obtain the current at a particular potential.

PAGE 8

Case o + ne -+ R + (Reversible charge transfer) o + ne R (Totally irreversible charge transfer) o + ne -+ R + 2 R z (Reversible charge transfer with irreversible dimerization) Table 1 E quation x(aT) = + ITI 1 J:V rTanh (u-y+z) d vz 2 z 2 0 cosh (u-y+z) X (br ) X (aT) 1 00 -L: ( / :rr ) j /ITDob (_ly+l x exp [(_ ] aNaF) (E-Eo+ RT Q,n v---=:":':'" RT aNaF k In j=l exp (aT) s 1IT -1/2 / cf[ ,/-u x (n) d n (aT -n) 1/2 3/2 Reference 1 4 w 1 8

PAGE 9

4 The application of these equations to electrode processes in molten salts at other than room temperature is straightforward, involving only RT RT a change in the factor nF or Figure 1 illustrates the effect aN . F a of temperature change on the value of for a reversible one electron process. The curves were calculated by computer through the use of the first equation in Table 1. The integration was accomplished using a modified algorithm of Simpson's rule, and the values are accurate to + 0.5%. It is necessary to introduce the equations for the theoretical cyclic voltammograms, since these equations will be used to construct baselines from which values of the current in multi-step processes can be obtained. From the practical viewpoint, cyclic voltammetry is an excellent qualitative tool for the study of electrode mechanisms. However, its use as a quantitative method should be applied with caution. The theory is derived by assuming semi-infinite linear diffusion. This condition does not apply at low scan rates where density gradients may cause significant convection, or where thermal gradients give rise to convective currents near or at the electrode surface. Techniques are available to determine if convection is significant within the time frame of the ex-periment (19). One of the better methods used, and the method of choice for this work, is the determination of the constancy of the quantity iT 1/2 as i decays with time. Upon application of a constant potential to an electrode, a current decay is observed with time. If linear dif-fusion is the .only mode of mass transport, the product of the current and the square root of time should be constant according to the Cottrell equation 1/2 lT 1/2 7T

PAGE 10

Figure 1. Variation in X(at) as a Funcation of Temperature

PAGE 11

0.4 0.3 X (at) 0.2 0.1 0 .45 -035 -0.25 -0.15 -0.05 005 0.15 0.25 0.35

PAGE 12

7 where n = number of electrons, F = the Faraday(96500 coul/eq.) A = the area of the electrode surface, D = the diffusion coefficient, and C = the concentration of the electroactive species in thebulk of b -the solution. The constancy of the electrode area is another factor which can influence qualitative and quantitative results in cyclic voltammetry as well as other electrochemical techniques. Several phenomena can cause a change in electrode area. First, faulty insulating seals around the electrode deteriorate rapidly in molten salts, which can increase the electrode area and produce spurious currents. Second, dendrite growth or porous film formation during the deposition of many metals (20,21) has been found to cause significant changes in electrode area. The appearance of dendritic growths on a surface can also disrupt the diffusion layer, causing spurious currents and worsening the results. Third, alloy formation or chemical reaction of the deposited species with the electrode substrate can alter the surface area considerably. This effect is especially prevalent in high temperature molten salt systems (20,22). Taking into consideration the above effects, one should use care when employingcyclic voltammetry and for that matter other electrochemical techniques for quantitative work. Chronopotentiometry The essence of the method of chronopotentiometry is the measurement of the time interval between the start of constant current electrolysis and the point at which complete concentration polarization is achieved. This time is known as the transition time and is a function of concen-tration of the electroactive species. As with cyclic voltammetry,

PAGE 13

8 most of the theoretical development has assumed semi-infinite linear diffusion as the sole mechanism of mass transport. The theory of chronopotentiometry was first developed by Weber (23) and later verified by Sand (24). Sand showed the dependence of E vs T to be applicable only to linear diffusion controlled processes. Delahay et ale (25,26) developed theoretical treatments for reactions that were not necessarily reversible or diffusion controlled. The extension of the method to systems involving chemical kinetics was achieved by Testa and Reinrnuth (27-29). The theoretical development of chronopotentiometry involves the solution of Fick's second law with the appropriate boundary conditions. The resulting equation, known as Sand's equation, which releases the density and transition time to the concentration of the electro-active species is 1/2 T = o F 1/2 1/2c n TI D b 2 where i = current density and T = transition time, the other parameters o having been defined previously. This equation is valid for either re-versible or irreversible diffusion controlled processes, since no assump-tions concerning the kinetics of charge transfer were made in the deriva-tion (30). By the addition of the appropriate boundary conditions, chrono-potentiometric equations can be derived for a number of kinetic situations. Examples of these equations are given in Table 2. Excellent reviews of chronopotentiometry in kinetic systems can be found in the literature (]9, 30-34) and the reader is urged to refer to these for a more complete treatment of the subject.

PAGE 14

Case o + ne o + ne K R -+-Z + -+R + -+R + o + ne -+R + E E E Table 2 Equation 1/2 1/2 T -t [ 1/2 T 1/2 1/2 RT n [T -t ]_ El/2 + nF Nn t1 / 2 1/2 f[k +k )1/2 1/2] RT 1 'IT K er 1 2 t nF Q,n[l+K + 1/2 1/2 2(1+K) (k1+k2 ) t 1/2 1/2 BD k Co EO + RT Q, [T -t ] RT Q,n [ R 1 nF n 1/6 + .3TiF 3 'IT T Reference 30 31 32

PAGE 15

10 The extension of the theory of chronopotentiometry to molten salt systems is trivial, again involving only the RT nF factor. How-ever, as in cyclic voltammetry, the method requires maintenance of linear diffusion conditions, a not always easy matter with molten salts. Previous work indicates (35-38) that precision on the order of 2-4% is possible with chronopotentiometry in molten salts. In each case, care was taken to thermostat the system to at least 3C. The position of the electrode with regards to the direction of diffusion was also found to vary the results somewhat, probably due to the enhanced thermal and convective currents. The problem of maintaining a constant electrode area is also present in chronopotentiometry. The deposition of solid products on the electrode surface frequently results in dendritic growths, which cause the surface roughness and area to change with time. This effect is especially prominent at high current densities. However at low current densities, the electrolysis time may be so slow that convective mass transport and spherical diffusion become increasingly important, leading to inconsistent results. Chronoamperom e try This particular technique involves the application of a controlled potential pulse to the working electrode. The current is monitored as a function of time and varies according to the Cottrell equation mentioned previously. In this study, chronoamperometry was used for an accurate determination of the diffusion coefficient of sulfide ions. The method was also used to corroborate the results of the kinetic studies from cyclic voltammetry and chronopotentiometry.

PAGE 16

11 Previous Work Since the late 1960's several workers have studied the electro-chemistry of sulfur and sulfide ion in molten LiCl/KCl (3,39-41). Kennedy and Adamo (3Y) found from cyclic voltamrnetry and controlled potential electrolysis of a sulfur solution that the reduction of sulfur in LiCl/KCl occurs in 2 steps, which they ascribed to S + e + S n n 2S + e + S n n According to the authors, the electrochemical reactions were not strictly reversible at 420C, since the peak separation of the anodic and cathodic processes was greater than the theoretical l32/n mV, where n is the number of electrons transferred. However, the authors reported that the peak separation did not vary with scan rate from 40 mV/sec to 200 mV/sec. Tischer and Ludwig (4) explained these two contradictory facts by assuming a catalytic mechanism = A plot of square root of scan rate vs. peak current would readily confirm or deny this mechanism, but no mention of such a diagnosis was made. Bodewig and Plambeck (40) studied the potentiometric behavior of the sulfur-sulfide couple in LiCl/KCl at 400-450C. They observed Nernstian response of the sulfur electrode with changes in concentration of coul-ometrically generated sulfide ions. The EO of the sulfur-sulfide couple was calculated to be -l.OOSV vs the standard molar platinum electrode (SMPE)

PAGE 17

12 During coulometric generation of sulfide from a sulfur pool, they noted the appearance of a blue color, which they ascribed to a polysulfide species Sx' Upon applications of a vacuum to the cell, or by applying a potential anodic of the equilibrium potential for sulfide generation, the blue color was found to disappear. If sulfur was then added to the cell, the blue color returned, which supports their hypothesis of a colored polysulfide. Other workers (42-45) have determined spectroscopi-cally that the blue color is due to a supersulfide, S2 or S3 species. Both findings are consistent if an equilibrium of the type S + 2e + S x + x = S + x + where x > 6 is assumed. These values for x are consistent with the gas phase measurements by Beckowitz (46) of sulfur vapor composition. From vapor pressure vs temperature plots, he found that sulfur vapor consists mainly of S7 units at 425C, with a slightly smaller amount of S8 and S6' The proportion of S7 and S6 increased with increasing temperature. Bodewig and Plambeck (40) found from chronopotentiometric experiments that the oxidation of sulfide obeyed the Sand equation with rhenium and gold electrodes. Their value for the diffusion coefficient of sulfide -6 2 is 3.l2xlO cm /sec. They make mention of the fact that this value of the diffusion coefficient is almost an order of magnitUde smaller than 2+ -5 2 the lowest value for a divalent ion (D for Pb = 1.3xlO cm /sec), in-dicating that the diffusing species may be a large polysulfide ion, Sx All the evidence taken into consideration seems to indicate that the potential and chronopotentiometric data obtained in this work was for the sulfur-polysulfide system, instead of the sulfur-sulfide couple.

PAGE 18

Birk and 8teunenberg (41) performed a more comprehensive examina-tion of the sulfur-sulfide system in LiCl/KCl involving the temperature and scan rate dependence of the cyclic voltammograms of sulfide ion. Figure 2 sholtls a t yp:ically cyclic voltarmnogram obtained in their study. From peak separation data, the authors conclude that the first electro-chemical reaction is reversible at all scan rates investigated. This contradicts the findings of Kenredy and Adamo (39) mention. e.d earlier. They also claimed that this first electrochemical reaction is probably an unsyrmnetrical reaction, involving different numbers of electrons on charge and discharge. Based on their literature review and the results of their experi-ments, Birk and 8teunenberg hypothesized the following reaction scheme: anodic 1) 28 = + 8 2 + 2e 2) 8 2 + 28 + 2e 3) cathodic 4) 5} = 38+8 + 28 2 + 8 +4 = 28 + 2e + 8 2 + e + 8 2 = 6) 8 2 + 2e + 28 The existence of the supersulfide ion, 8 2 in reactions (3) and (5) has previously been proposed (47-50) to account for the deep blue color observed when both sulfur and sulfide are present in solution. E.8.R., I.R., and Raman data indicate that an 8 3 species may also be present in the LiCl/KCl melt (51-53). The contribution of the

PAGE 19

Figure 2. Cyclic Voltammogram of a Solution of Li2 S in LiCl/KCl According to Birk and Steunenberg

PAGE 20

10 5 --. ro E I-' t-0 U1 Z w e:::: e:::: -5 ::::> LJ -10 1.2 1.6 2.0 2.4 2.8 POTENTIAL (VOLTS Vs. Li)

PAGE 21

16 tetrasulfide ion, S4-' in reaction (3) is not expected to be significant, since higher polysulfide species in LiCl/KCl have been found to be unstable with respect to formation of sulfur and lower polysulfides (54,55) Cleaver, et al. (54) examined the cyclic voltammetric behavior of Na 2 S and Na 2s2 2 in the LiCl/KCl eutectic. They found Na 2 S to be practically insoluble in the melt at 420C, while solutions of Na 2s2 2 give cyclic voltammograms very similar to those of Li2S. The authors found that the solutions of Na 2s2 2 gradually lost sulfur, indicating that the disulfide ion may be unstable toward decomposition to sulfur and sulfide.

PAGE 22

CHAPTER 2 EXPERIMENTAL Cell Haterials and Design A photograph of an assembled and dissassembled cell and cell housing is shown in Figure 3. Both the outer and inner cell housings were made of Pyrex. The inner Pyrex container was used to prevent the molten salt from inadvertently coming in contact with the outer Pyrex container. When this did happen, diffusion of .+ + Ll and K into the hot glass changed the composition and thermal properties of the Pyrex, causing weakening of the walls and cracking of the entire container on cooling. Pyrex or any other silica containing substance was not used for any part of the cell which would be in contact with the molten salt. This is due to the highly basic and corrosive properties of sulfide ion. Immediately upon contact with a sulfide containing melt, silica becomes etched due. to attack by sulfide and introduces silicon containing compounds into the melt (56). Therefore all cell components were made of high pqrity aluminum oxide (Norton Co., Worcester, Mass.). The melt container was made of high density A12 0 3 (AH999), while the reference electrode and counter electrode compartments were made from a slightly porous A1 2 0 3 (A1788). The porosity allowed electrical contact to be maintained between the different compartments while minimizing inter-mixing of the solutes. 17

PAGE 23

Figure 3. Photographs of Cell and Cell Housing

PAGE 24

19

PAGE 25

20

PAGE 26

21 The top portion of the outer container was fitted with 5 threaded pyrex openings (Ace Glass, Inc., Vineland, N.J.) by the Department of Chemistry glass shop. These openings allowed electrode access to the cell. By inserting the electrodes through the threaded Teflon bushing and o-ring assembly and tightening the bushings into the threaded pyrex openings, a good vacuum seal could be maintained. Electrode Design Working Electrodes The construction of working electrodes with acceptable performance characteristics has always been a challenge in molten salt systems. For the stuay of non-corrosive solutes and solvents, the most common method of fabricating working electrodes is the glass to metal seal. In instances where the wetting properties of the glass preclude the use of glass to metal seal, most workers have relied on a seal-free electrode. This is simply a metal foil connected by a thin contact wire known as a "flag type" electrode. The glass seal construction was unacceptable due to the corrosive nature of the sulfide ion mentioned previously. The flag type construction was found to be adequate for qualitative measurements, but gave irreproducible results when quantitative measurements were attempted. Therefore, it was necessary to devise an electrode system which had a definite working area, the desired geometry, and was inert with respect to attack by sulfide. Figure 4 is a diagram of the electrode system which was found to give excellent results.

PAGE 27

Figure 4. Diagram of Working Electrode System

PAGE 28

23 GLASSY CARBON (CENTER) f-iBORON NITRIDE (BN) END. VIEW PYR[X TUBE GLASSY CARBON SIDE VIEW

PAGE 29

24 The outer portion of the electrode consists of a cylinder of hot pressed boron nitride (Union Carbide, New York) A hole with a diameter slightly larger than the diameter of the electrode is bored through the center of the boron nitride cylinder. The electrode material consisting either of glassy carbon or Pt rod previously cleaned in boiling, concentrated HCl is then coated with a layer of Ultra Bond 552 high temperature ceramic adhesive (Aremco Products, Inc., Ossining, N. Y.). The electrode is then fitted into the boron nitride cylinder and the adhesive is allowed to dry at room temperature. After drying, the electrode is heated to 500C for 2 hours in a muffle furnace in order to evaporate the organic portion of the ceramic adhesive. The temperature should be brought to 5000C slowly to prevent the formation of bubbles and voids between the electrode and the boron nitride. Electrical connection to the instrumentation for glassy carbon electrodes was made by wrapping Pt wire around the exposed glassy carbon at the top of the electrode. Gold paste was applied over the Pt wire and glassy carbon for mechanical strength and to improve conductivity. Nickel wire was then spot welded to the Pt wire and run through a length of 6 rom diameter Pyrex tubing. The Pyrex tubing was then vacuum collapsed around the end of the glassy carbon-platinum wire arrangement to complete the electrode. Platinum electrodes were spot welded directly to a nickel wire and run through a 6 rom diameter pyrex tubing. The tubing was then vacuum collapsed around the Pt rod-Ni wire junction. Working electrodes prepared in this manner were resistant to attack by sulfide ion. The ceramic adhesive prevented the melt from creeping between the electrode and the

PAGE 30

25 boron nitride. No corrosive effects of sulfide on the ceramic adhesive were noted, and the electrodes usually remained intact throughout the entire experiment. Reference Electrode 2+ The reference electrode used in this study was the Pt/Pt system. This reference electrode system was chosen because of the convenience of construction and good stability (20,2J,57,58). 2+ The Pt for the reference compartment was generated coulometrically from a large Pt foil. The potential was allowed to stabilize for approximately 30 minutes before the experiment was begun. The drift 2+ of the Pt/Pt reference electrode was checked against the following reference electrode of the second kind: This reference electrode has a total drift of < 0.0005V over a period of several days (59), and was thus suitable for checking the 2+ / 2+ f performance of the Pt/Pt system. In thls manner, the Pt Pt re erence was found to be stable to within 0.002V over a period of 24 hrs. Counter Electrode The counter electrode consisted of a graphite rod (Spectroscopic Grade, National Carbon Co., Cleveland, Ohio) inserted into the counter electrode compartment. Figure 5 shows a diagram of the counter electrode in the compartment. The Pyrex cap inserted over the top of the counter electrode compartment prevented any chlorine gas generated at this electrode from escaping into the melt compartment. A continuous flow or argon or helium under positive pressure outside of the compartment swept all of the generated chlorine gas to the outside atmosphere.

PAGE 31

Figure 5. Counter Electrode Assembly

PAGE 32

27 PYREX CAP Al2 0 3 TUBE ROD MELT

PAGE 33

28 Sulfid2 Generating Electrode Previously, all studies of the electrochemistry of sulfide ion in molten LiCl/KCl have been accomplished by adding a weighed amount of Li2 S or Na 2 S to the melt. This method resulted in the introduction of impurities such as sulfur and H 2 0 by virtue of the hygroscopic and air sensitive nature of the alkali metal sulfides. A much more con-venient and accurate method of sulfide addition to the melt was developed by Liu et al. (60). This method involves the coulometric generation of sulifde ion in the melt from a Ni/NiS eutectic electrode. This eutectic electrode is easily prepared from the elements in a 2:1 mole ratio of Ni:S. The nickel and sulfur are placed in a quartz tube under 200 mm Hg argon pressure and heated to > 700C. The silver melt pro-duced is the Ni/NiS eutectic. The metal can then be cast into any convenient shape for use as an electrode. In this work, 5 mm diameter rods of the Ni/NiS were used. Electrical contact with the sulfide electrode was made by welding a 2 mm diameter Pt rod to the electrode. A nickel wire was then spot welded to the platinum for connection to the outside of the cell container. The coulometric generation of sulfide was found to be essentially 100% current efficient as long as the temperature was greater than approximately 470C. For studies below this temperature, the sulfide was first generated at 475C, then the temperature was lowered to the appropriate value. In each experiment the sulfide ion concentration .2+ was cross checked by titration with coulometrically generated Nl The concentration of sulfide was monitored potentiometrically by means of a Ni/NiS/S electrode according to Liu et al. (60).

PAGE 34

29 Equipment Temperature Control Temperature monitoring and control was accomplished by a Wheelco Panelmount Capacitrol coupled to a chromel-alumel thermocouple inserted into the outer cell compartment. The heating element was a "Hevi-Duty" multiple unit furnace (Hevi-Duty Heating Equipment Div., Watertown, Wis.) with a water cooled heat dissipator located above the furnace. The temperature was calibrated against the melting point of zinc, and was controllable to within + 3C. Electrochemical Equipment Cyclic voltammograms were obtained with a Princeton Applied Research (PAR) model 174A Polarographic Analyzer coupled with a PAR model 175 Universal Programmer. The three electrode system was used in all experiments. The constant current source for the chronopotentiometric experiments was a Buchler Instruments D.C. Power Supply in series with a cis ion variable resistor. This arrangement allowed current control to + 0.5%. All potential measurements were made with a Hewlett P .ackard 34703A digital D.C.V., D.C.A., meter. A PAR model RE0074 X-Y recorder was used to record all slow scan (IO-200mv/sec) cyclic voltammograms. Rapid scan cyclic voltammograms and chronopotentiograms were recorded on a Tektronix model 549 Storage Oscilloscope fitted with a model 53/54C fast rise plug-in preamplifier.

PAGE 35

30 Cell Cleanup and Salt Purification Before each experiment, the alumina melt container, electrodes, and electrode compartments were first boiled in 12M HCl for 2 hrs. This was followed by a 24 hr. cleaning in fresh 12M HCl in an ultra-sonic bath. The final step involved boiling for 2 hrs in triple dis-tilled and deionized water. The crucibles and electrodes were then dried in an oven at 140C until ready for use. The eutectic used in this study was supplied by Anderson Physics Laboratories, Inc., Champaign, Illinois. The salt was introduced into the cell compartment in powdered form and placed under vacuum at 300C for 24 hrs. Electronic grade HCl gas (Matheson Co., East Ruther-ford, N.J.) dried over anhydrous magnesium perchlorate and a dry ice-acetone trap was then passed over the salt as the temperature was raised. This treatment reversed the hydrolysis of LiCl: LiOH + HCl After the salt was molten, dry 02 gas pretreated in the same manner as the HCl gas was bubbled into the melt for 2 hrs. This oxidized any organic contaminants which may have been present. Dry HCl was again in-troduced to react with any oxide ions which may have formed from the introduction of 02 Finally, dry Helium or argon, passed over anhydrous magnesium perchlorate, hot copper turnings and a dry ice-acetone trap was bubbled through the melt for 2 hrs. The melt purity was then checked voltammetrically as described by Laitinen and Gaur (61). Figure 6 shows the experimental set-up, with a completely assembled cell, gas purification train and instrumentation.

PAGE 36

Figure 6. Photograph of Experimental System

PAGE 37

. ...... '" '''' .... 'l'AI : r ,' 04'.1. -0: II:! . ::: W N

PAGE 38

33 Experimental Procedure The procedure used during each experiment was essentially the same. After the blank current was determined using both the Pt and glassy carbon electrodes, sulfide was coulometrically added to the melt until the desired concentration was reached. The temperature was then lowered to 375C and the system was allowed to equilibrate for 30 minutes. Cyclic voltamrnograms were obtained at scan rates from 0.01 V/sec to 75 V/sec with both types of electrodes. The sw itching potential was also varied to observe the effect on the reverse scan. Chronopotentiograms were acquired at several different current d t t b th 1 d t h k h f h 1/2 enSl les a 0 e ectro es 0 c e c t e constancy 0 t e IT product. In order to determine an acceptable value for the diffusion co-efficient of sulfide ion at the various temperatures, the method suggested by Adams (62) was used. This method has proved extremely reliable in aqueous solvents, as evidenced by the work of von Stackelberg et al. (63), and should apply equally well in molten salts. It involves the application of a controlled potential pulse to the electrode, whle monitoring the current-time decay curve. By plotting current vs. time and extrapolating to t=O, radial diffusion and convective effects at unshie.lded electrodes can be neglected, and a true value of D can be calculated from the Cottrell equation. At the highest temperature studied, a portion of the sulfide in the melt was electrolyzed at constant potential and a sample drawn out for analysis by ESR spectroscopy. The sampling technique used was to place quartz ESR tubes just above the melt. A vacuum was then applied

PAGE 39

34 and the tubes were lowered oelow the melt line. The cell was then vented to helium pressure and the sample entered the tubes. A blank sulfide solution was also obtained in this manner.

PAGE 40

CHAPTER 3 RESULTS Determination of Diffusion Coefficient of S Ion As mentioned in the last section, the diffusion coefficient of sulfide ion was determined by observing the decay of current with time after application of a controlled potential pulse. Since the oxidation of sulfide ion is followed by at least one chemical reaction as evidenced by this work as well as the work of previous authors, it was necessary to measure the current at times short enough to neglect the chemical reaction. These times can be determined from a graphical plot of itl/2 vs t, which should be constant for a system uncomplicated by preceeing or subsequent chemical processes. Figure 7 shows a plot of itl/2 vs t for the oxidation of sulfide at three different temperatures. As indicated by the data, the itl/2 factor becomes essentially constant for times less than about 10 m sec. E t 1 h 1 f' 1/2 f d t ., x rapo atlon to T-O glves t e approprlate va ue 0 lt or e ermlnlng the diffusion coefficient. These values were obtained at the highest concentration of S-studied (9.8xlO-3M) to minimize errors due to con-taminants. Glassy carbon electrodes were used to avoid interference due to surface processes such as adsorption and metal sulfide formation. The extrapolated values of itl/2 were substituted into the cottrell equation 7T 1/2 3 5

PAGE 41

Figure 7. it1 / 2 vs t as a Function of Temperature

PAGE 42

2 9 0 0 A a c 6 0 X 0 A 0 LJ D 0 w W l:l c -...J lJ) A 0 a 0-2: A 1 47SoC 0 t--' --' C 42SDC b 375 etc o 10 20 30 40 50 60 t( SEC.x 103 )

PAGE 43

38 and solved for D. Table 3 gives values of D for sulfide ion at 375, 425, and 475C. Table 3 2 D cm /sec -5 -5 2.6x10 +O.lxIO -5 -5 +O.lxIO -5 -5 3.lx10 +0.2xIO The values of D given in Table 3 were corrected for the slight change in concentration with temperature, Cyclic Voltarnrnetry The cyclic voltarnmetric response of S ion in LiCl-KCl eutectic was studied as a function of temperature, scan rate, concentration, and electrode material. Scan rate was varied from 0.01 V/sec to 75 V/sec and the temperature varied from 375C to 475C in steps of 50C. All experimental potentials were corrected for the increase in temperature and for the small IR drop between the working and reference electrodes. Concentrations ranging from to I.OxlO-2 M in sulfide were examined at these different temperatures and scan rates. Finally, the effect of electrode material was investigated to determine if adsorption played a significant role in the redox mechanism, and to lend support to the proposed mechanism. Glassy carbon and platinum were the principal materials used as working electrodes in this work. Variation of Scan Rate Since the largest variation in the electrochemical behavior of sulfide occurs when the scan rate is varied, these results will be pre-sented first. A typical cyclic voltarnrnogram of a solution of Li2 S in molten LiCl-KCl eutectic at 375C and [S ] = 9. 8xl0 is shown in Figure 8.

PAGE 44

Figure 8.

PAGE 45

400 SCAN RATE 200 mV S-I 200 --. c.... L < r-:z w c:: e::: ::::> LJ 200 ANODIC 400 -1.55 -1.35 -1.15 0.95 -0.75 -0, 55 -0. 35 E (VOLTS VS. 1 M pta; Pt)

PAGE 46

41 All potential scans in this work were made from negative to positive potentials, so that the first charge transfer process is the oxidation of S ion. This charge transfer can be represented by the following reaction S n-2 -+ S + ne +By examining the current voltage curve for this process, the value of "n" can be calculated. Several relations derivable from theory (17) are used for this purpose. The first of these relations is the equation describing the dependence of the peak potential, E and half-peak pot enp tial,Ep / 2 on the value of E -E = P p/2 "nil for a reversible charge transfer RI' 2.20 nF The second relation involves the peak potential of the reverse cathodic process as well E -E ,= 2.22 RI' p(anodic) p(cathodlc) nF Table 4 gives the values of E ( d')' E /2( d') E ( h d' ) and p ano lC p ano lC, P cat 0 lC "n" for the assumed reversible oxidation of S From the experimentally determined potential values, the number of electrons involved in the first anodic process is found to be n=1.3. One can see in Table 4 that the calculated value of "n" at po;I.arization rates greater than 1 volt/sec. decreases with increasing scan rate, approaching the theoretical value of n=l.O. The data suggest that a different mechanism may be operating under conditions of rapid scan. By comparing cyclic voltarnrnograms obtained at various scan rates, it is possible to derive information concerning a consistent overall mechanism for the first charge transfer. Figure 9 illustrates six current-voltage -3 = curves of 9.8xlO M S at scan rates of 0.1, 0.2, 0.5, 1, 5, and

PAGE 47

Figure 9. Current-Voltage Curves of 9.8xlO-3 M Sulfide as a Function of Scan Rate

PAGE 48

0.1) 0.2 V I SEC -1.75 -1.35 -0.95 -0.55 -1.75 -1.45 -1.15 -0.85 -0.55 -0.25 E E

PAGE 49

44 10 volts/sec. at 375C. Several salient features can be seen in this Figure. The first aspect is that the peak potential of the first oxidation process (Peak I) stays constant at scan rates below approxi-mately 0.5 V/sec. The peak potential then shifts anodically until a scan rate of 5 V/sec.beyond which it again becomes constant. Figure 10 plots the variation of the peak potential of peak I with scan rate. The behavior of the peak potential with scan rate is indicative of an irreversible dimerization of the electrogenerated product according to the following reaction scheme (18,64): R-+ +o + ne x -where in this case R is the S ion, 0 is the sulfur anion radical S and X is the disulfide ion S2 The determination of the homogeneous rate constant kl can be made by means of the following equation first derived by Saveant and Vianello (65) RT Q ., n 3nF k C 1 R V where El/2 is the half wave potential of the charge transfer step when the dimerization reaction is negligible, C is the concentration of subR stance R in the bulk solution, and V is the scan rate. The evaluation of El/2 is the difficult step in the use of this equation. In this work, the value of El/2 was obtained from cyclic voltammograms at scan rates rapid enough so that the dimerization could be neglected. At the rapid polarization rates used, the current-voltage curve is described by the theoretical voltammogram for the uncomplicated reversible charge transfer. In this work Ep and El/2 were found to be constant from 5 V/sec to 50 v/sec

PAGE 50

Figure 10.

PAGE 51

1.22 -E (I) 1.20 A P 1.18 c 1.16 1.14 c A c 375C A 42S0C o 475C o 1 2 3 4 5 SC A N RAT E (V I SEC) 6 7

PAGE 52

47 after correction for the slight XR drop. Using the values of the peak current and peak potential at rapid scan rates, El/2 can be easily estimated (17). It must be stated, however, that since El/2 is an exponential function of k l slight errors in the measurement of El/2 cause rather large variations in the rate constant. Therefore, care should be taken to obtain a value of the half wave potential which is as precise as possible. In the present work, the half wave potential could not be estimated to better than +0.005V and thus the error in kl is approximately 40%. The value of kl for the irreversible dimerization f 375' lId b 3xl03 n M-lS-l o S at C ca cu ate to e N As a check on the consistency of the proposed irreversible dimeriza-tion, a plot of peak height vs. square root of scan rate was made as shown in Figure 11. The equation describing the variation of peak height with ;vfor a reversible charge transfer followed by irreversible dimeri-zation is given by (18) i p 3/2 3/2 n F 0.527 1/2 1/2 R T where the various terms have been previously defined. The slope of the i vs IVline was calculated and solved for the diffusion coefficient. p Since the slope varies as the square of the diffusion coefficient, this is a rather sensitive test for conformity. The diffusion coefficient -3 = -5 2 calculated in this manner at 375C and 9.8xlO M S is 3.4xlO cm /sec 1.OxlO-5 in good agreement with the more accurate value of -5 2 -5 2.6xlO cm /sec O.lxlO determined by chronoamperometry. The second feature that can be seen in Figure 9 occurs in the tial range between the first oxidation process and what appears to be

PAGE 53

Figure 11.

PAGE 54

20 15 10 8 o o o 49 o o o o

PAGE 55

50 the second oxidation process. The current in the intermediate region is nearly constant, rather than decaying as would be expected for a simple, diffusion controlled charge transfer process. Observation suggests that an intermediate electrochemical reaction is occurring between the two more obvious charge transfers. In order to discern the intermediate process more clearly, a computer assisted subtraction procedure was applied to each of the current voltage curves. Briefly, the subtraction was accomplished by programming the computer to generate current-voltage curves using the series and integral equations given in the first chapter. The experimental current was entered into the computer and the generated'current was subtracted from it. Allowance was also made for the residual charging current. The result was a set of voltammograms without interference from the current due to the first peak, allowing minor features to become more prominent. Figure 12 -3 shows the results, 'of this procedure for [S ] = 9. 8xlO H at 375C on glassy carbon. It is readily obvious from the peak morphologies that there are at least two electrochemical oxidations (peak II and peak III) = occurring after the initial oxidation of S ion to S A graphical analysis of the effect of scan rate on the current and current ratio of the two peaks is given in Figure 13-15. These graphs show that the variation of the' peak height with the square root of scan rate is not linear. Peak II increases exponentially with IVwhile peak III increases logarithmically with IV-. In effect, peak II appears to increase with scan rate at the expense of peak III. The behavior of the peak currents indicates that the kinetics of the two oxidations are coupled in some way.

PAGE 56

Figure 12.

PAGE 57

960 840 720 600 i,fA 480 360 240 120 -1.10 -0,95 -0.80 E 52 SCAN RATE(V/SEC) -0,65 5 2 1 0,2 0.1 0.05

PAGE 58

Figure 13. Variation of Current for Peak II as a Function of (Scan Rate) 1/2

PAGE 59

1200 900 600 300 c a A A 0 A 0 0 0 0 0 A 0 1 54 A tJ 0 y1/2 o o o 375 0 A 425 o 475

PAGE 60

Figure 14. Variation of Current for Peak III as a Function of (Scan Rate)1/2

PAGE 61

56 o 300 o c A 200 c 4 20 A 3750 0 100 CJ 4250 o 4750 1 2

PAGE 62

Figure 15. Ratio of Currents for Peaks II and III as a Function of (Scan Rate) 1/2

PAGE 63

58 6 o 3750 4250 c 475 0 4 IpOI) 0 Ip(III) Q A 2 cC c 0 D 0 0 0 0 1

PAGE 64

59 The features of the cathodic half of the cyclic voltammograms appear to behave normally with increasing scan rate. Cathodic peak IV is a stripping_peak, involving the reduction and subsequent cathodic dissolution of a sulfur film deposited on the electrode surface during the anodic scan. The number of electrons involved in the dissolution of the sulfur film is difficult to determine. However, the ratio the number of electrons involved in the deposition step to the number in-volved in the stripping step can be determined. This was accomplished by integrating the current-voltage, i.e. current-time, curves for the deposition and stripping. The values of the total charge passed for each process was compared and the ratio found to be 1:2 for deposition and stripping, respectively. Therefore, if the number of electrons involved in the sulfur deposition can be determined, the value for the cathodic dissolution of sulfur will be kna.vn. The total charge data were obtained at the lowest temperature studied (375) in order to minimize errors due to vaporization and subsequent loss of material. Cathodic peak V appears to involve the reverse of anodic peak I. The peak current varies linearly with I v and "n" was calculated to be 1.4+0.1 for this process. However, at scan rates on the order of 1-10 V/sec the peak potential shifts in the cathodic direction, indicating the possible presence of a preceding chemical reaction. It is more probable that other phenomena such as the reduction of two species with similar EOs cause this particular effect. This is not difficult to imagine, since the sulfur cathodically stripped from the electrode surface in the previous step can exist in several polymeric forms at the temperatures involved.

PAGE 65

60 Since the average number of electrons transferred in the stripping step is most likely two, the formula for the product can be represented 2-as Sx 2 X = 8, depending on the temperature and concentration at the electrode. The following cathodic process, designated as peak V, would then involve the reduction of several polysulfide species, each with EOs differing by a small value depending on the value of X. Evidence for this particular explanation can be seen by comparing the cyclic voltammograms at various temperatures (Figure 16). As the tem-perature is increased, peak V broadens by an amount greater than pre-dicted for the reduction of a single species. A second peak begins to appear at a potential which is close to the peak potential of the original process. The appearance and growth of this second peak would seem to indicate the increasing presence of a reactant which is more stable at higher temperatures, and which possesses a similar EO. These facts are easily explained if the presence of several polysulfide species is postu-lated. The assumption of similar EO's for each of the polysulfide species is reasonable, because the reduction potential for a species such as S5 would not be expected to be much different from species like S6 or S7 Other evidence pointing to the similarity of various polysulfide species is seen in aqueous polarography. The oxidation of sulfide and polysulfides occurs by means of similar mechanisms on a mercury drop electrode: 2S -+ HgSX + 2e X 1 < X < 4 In this case, El/ 2 for this reaction is the same for all possible values of X. Further support for this particular reduction mechanism is found in the value for the number of electrons transferred in the last cathodic step. The value of "n", as previously mentioned was calculated to be 1.4+0.1.

PAGE 66

Figure 16. Cyclic VoltamrnograIIls of Sulfide Ion as a Function of Temperature.

PAGE 67

62 600 4QO I ()LA) 0 200 4(00 -1.75 -1.35 0.95 -0.55 E

PAGE 68

63 400 4250 200 400 -1.36 -0.96 -0.56 E 800 400, __ I 0 400 800 -1.77 -1.37 -0.97 -0.57 E

PAGE 69

64 According to theoretical considerations put forth by Polcyn and Shain (66), electron numbers intermediate between whole numbers are indicative of either a multisbep reduction with similar EO's or the reduction of more than one component, again with similar EO's. The appearance of a second peak with increasing temperature can be thought of as due to the reduction of lower polysulfides formed from the thermal decomposition of higher polysulfide species. This behavior is consistent with the fact that the higher polysulfides (X > 3) are known to be unstable in LiCl/KCl at temperatures above 400C. No data on the stability of these compounds below 400C have been published. of Change in Concentration Figure 17 shows the current voltage response of S ion on glassy carbon as a function of concentration at a scan rate of 0.1 V/sec. 'The striking aspect of these curves is the rapid change with con-centration in the ratio of the currents of the first and second anodic processes, and the lower response of the third process at the lower concentrations. Table 5 gives the ratios of the peak currents of anodic processes I and II as a function of concentration of S The values of iA(II)/iA(I) are plotted in Figure 18, and are the average of two separate experiments with three determinations for each experiment. The peak potentials of the anodic reactions at the lower concentra-tions could not be accurately measured since the standard potentials of the three separate processes are close enough to cause significant -4 = overlap of each wave. At concentrations greater than about 6xlO M S the value of -1.200+0.003V vs SMPE at 425C for the potential of peak I was measured. This value was measured at 100 mV/sec on glassy carbon and found to be invariant at ,these higher concent:rrations.

PAGE 70

Figure 17. Cyclic Voltammograms of Sulfide Ion as a Function of Concentration of Sulfide

PAGE 71

66 f 80 40 o 40 -1.75 -1.35 -0.95 E -0.55 600 300 o 300 -1.75 -1.35 E -0.95 -0.55

PAGE 72

80 40 o 40 -1.75 i i) ()l-A ) 1 80 4U o 40 .. 1.75 67 -1.35 -0.95 -0.55 E 3.4x 1 04 M -s--1.35 -0.95 -0.55 E

PAGE 73

68 Table 5 = [S ] i (II)/i (I) A A 1.8x10 -4 2.5 + 0.3 3 .4xlO-4 1.9 + 0.3 4.3x10 -4 1.5 + 0.2 5.0x10 -4 1.4 + 0.2 6.1x10 -4 0.9 + 0.2 -4 6.6x10 1.0 + 0.2 9.3x10 -4 0.85 + 0.2 2.5x10 -3 0.58+ 0.15 -3 4.2x10 M 0.47+ 0.15 -3 9.8x10 M 0.32+ 0.10

PAGE 74

Figure 18. The Ratio of the Currents Due to Peak I and II as a Function of Concentration of Sulfide

PAGE 75

70 o 0 o o -3 o o o o o o

PAGE 76

71 No significant variations in peak structure on the cathodic side of the cyclic voltammogram could be detected. Temperature Effectsl I -3 Cyclic voltarnrnograms of 9.8xlO M 8 at 100 mV/sec on a glassy carbon electrode ar"e shown in Figure 16 for 375, 425, and 475C. -By comparing the anodic portions of each curve, the decrease in the j height of peak III increasing temperature can be observed. A corresponding in the current due to the second anodic process accompanies the decr!ease of peak III. This phenomenon is also seen at both the lower and higher scan rates studied. Again, a kinetic coupling i between the two prodesses is suggested. i For the 'same el:ectrode and sulfide concentration, Figure 15 illus, trates the of the ratio of peaks II and III as a function of temperature and scari rate. The results of these experiments provide I more evidence for kinetic coupling, as will be discussed in more detail in the "Conclus"ions" section. The dimerizatio'n rate constant, k l was also determined as a func-tion of temperature.: Table 6 displays the results from 375 to 475C. 1 Table 6 [8 ]= -3 9.8xlO M 375 425 475 -1 -1 3 3 3 k l,9, M 8 3.0xlO 1. 7xlO 1.OxlO 3 3 3 +0.9xlO +0.7xlO +0.6xlO A possible explanation for the anomolous behaviorof the rate constant with increasing temperature will be reserved for the next chapter.

PAGE 77

72 Current-Voltage Response of S on Pt Electrode This particular portion of the section on cyclic voltammetry is presented as a separate subject since good quantitative data were difficult to obtain on Pt. The cause was the occurrence of several surface processes during the redox cycle of S. These surface processes caused ske\ving of the cyclic voltammograms, especially at high scan rates and temperatures. However, it was felt that the electrochemical and chemical surface reactions could be used to enhance the knowledge of the redox mechanism of sulfide. -3 Figure 19 gives cyclic voltammograms of 9.8xlO r.1 S on a platinum disk electrode at 375, 425, and 475C. Scan rates of 20 and 200 mV/sec are shown to illustrate the effect of polarization rate. Several surface peaks not found with glassy carbon are seen on the cathodic scan of the cyclic voltammogram. These peaks are labeled I and II for convenience. s s At 375C, both peak I and II are seen at scan rates from 20-200 s s mV/sec. For scan rates above 500 mV/sec only peak Is is found. A slightly different situation occurs at 425C. Peak Is does not appear until 90 mV/sec while peak lIs is present up to 200 mV/sec. Peak lIs dissappears above 1 V/sec, while the reduction process causing peak Is continues to occur. At 4 75
PAGE 78

Figure 19. Cyclic Voltammograms of Sulfide Solution on a Platinum Disk Electrode

PAGE 79

. ITs Is 200 1 .L 100 o 100 200 i -1.75 l (?A) 1 I 200 100 o 100 200 -1.76 74 -135 -0.95 E -0.55 -1.36 -0.96 -0.56 E

PAGE 80

75 I -1 1 o Of-A ) 20mVS I -1 100fLA) 200 mV S o -1.77 -1.37 -0.97 -0.57 E

PAGE 81

76 An attempt was made to determine the potential at which these surface peaks formed. This was done by initiating the scan at a potential more negative than the first anodic peak and then reversing the direction of scan at different potentials. Figure 2 0 gives the results of this experiment at 425C. Peak I occurred if the potential s scan was reversed at the first anodic wave. At this point there was no evidence of peak II. At the beginning of the second anodic wave, s peak II began to appear. This implies that the product of the first s oxidation peak is responsible for peak I, while the product of the s second oxidation wave is responsible for peak II s The surface peaks were integrated to obtain the total charge passed in the reduction process. The number of coulombs as a function of time is plotted for peaks I and II in Figure 21 T=O was taken at the s s foot of anodic peak I. From this graph, the rate of adsorption for peak I is seen to decrease with time, indicating a simple adsorption s process. The coverage of the electrode surface at all times shown is less than one monolayer. At the maximum time of 5 sec the coverage was estimated to be approximately 0.6 of a monolayer. Peak II hows ever, behaves irregularly providing more evidence for something other than simple adsorption. Finally, calculations were made to determine the number of electrons involved in the reduction of the adsorbed species in peak Is' However, the theory from which the equations were developed dOES not apply in this case, due to the following chemical reaction during the anodic scan. The value of "n" is therefore unknown for either of these processes.

PAGE 82

Figure 20. The Effect of Reversal Potential on Cyclic Voltammogram of Sulfide on Platinum

PAGE 83

78 200 100 __ __ __ -1.76 -1.36 -0.96 -0.56 E

PAGE 84

Figure 21.

PAGE 85

80 o o 8 o A o 6 o o A 4 2 o o Is A TIs 1 2 3 4 5 t (SE[)

PAGE 86

81 Chronopotentiometric Response of Sulfide In this work three concentrations of sulfide were studied with chronopotentiometry and current reversal chronopotentiometry. The current was varied over an order of magnitude so that any trends in the response could be easily seen. A typical chronopotentiogram for the oxidation of S on glassy carbon is shown in Figure 22. No attempt to calculate kinetic parameters was made since the theoretical chronopotentiometric response for multistep charge transfers coupled by dimerization of the primary product cannot be solved analyt 11 H t d th 1 f' 1/2 f f lca y. owever, ren s ln e va ue 0 IT as a unctlon 0 tem-perature, concentration, and current density can be used to support or rule out certain mechanisms. Table 7 gives the results at 425C for the controlled current oxidation of sulfide as a function of current and concentration. The working electrode in this case is glassy carbon. The transition times were measured graphically between lines drawn tangent to the lines re-presenting the potential shift. Figure 23 graphically represents the current vs iTl/2 data for the three concentrations. The data show that at low current densities, iTl/2 is constant. h h d h 1 f 1/2 h At 19 er current ensltles, t e va ue 0 IT lncreases, approac lng a new constant value. At high concetrations, large current densities caused the chronopotentiograms to become noisy in the potential region of sulfur deposition. The exact cause of this noise could not be determined.

PAGE 87

Figure 22.

PAGE 88

83 0.080 E -1.280 0.2 1.0 1.8 t (SEC)

PAGE 89

84 Table 7 = x 1 0 6 1/2 1/2 4 5 [ S ] i, A mperes. IT A mn.Sec x l 0 2xl 0 -3 5 .6xl 0 M llO 4.3 267 4.2 3 5 8 4.4 53 3 4 3 707 4.3 1 0 51 5.1 1327 5.5 2734 6.4 -3 7xl 0 M 332 4.6 433 4.7 549 4.6 650 4.8 75 9 5.0 8 5 4 5 2 961 5 3 1 0 6 0 5.3 2050 7 4 3270 8 3 3 9.8xl0 M 240 8 2 372 8 4 570 8 3 625 8.4 770 8 4 950 8 4 1236 9.1 1862 1 0 3

PAGE 90

Figure 23.

PAGE 91

In '0 3 X 2
PAGE 92

87 Temperature variation caused no significant differences in the chronopotentiometric results. The current density at which iTl/2 becomes non-linear is about the same differing by less than the error in measuring T. This behavior appears to contradict the results obtained for the determination of the dimerization rate constant from cyclic voltarnrnetry. This will be discussed in detail in the next section. ESR Results The purpose of this experiment was to determine if a cherni: c al equilibrium was established between Sand electrogenerated S radical. Samples of the sulfide containing melt were potentiostatically oxidized at -1.225 V vs SMPE on a large glassy carbon rod at 475C. This potential occurs approximately 0.06 volts before the first anodic peak, and was chosen to insure that no anodic process other than oxida-tion of S to S took place. The total charge used during the sample preparation was calculated by integrating the current-time curve obtained by monitoring the current on a high impedance x-t recorder. The concentration of S generated was 1. 5xlO -\1 assuming 100% current efficiency for the one electron transfer. The samples were drawn into several quartz ESR tubes and quickly cooled, usually in less than 20 seconds. The frozen samples were then analyzed by ESR to determine if a radical species was present. Figure 24 shows the results obtained at two sensitivities. These results show that down to the limit of detection of the instrument, -10 i.e. approximately lxlO M, no radical species existed in these samples.

PAGE 93

Figure 24.

PAGE 94

89 GAIN-4x105 ASS -0.2 -0.1 o 0.1 0.2 SCAN RANGE (G AUSS x 10-4 )

PAGE 95

90 GAIN -1.25 x106 ASS -0.2 -0.1 0 0.1 0.2 SCAN nANGE (GAUSS x10-3 )

PAGE 96

91 The large absorptions indicated by the arrows were dueto cavity background, and the small signals seen at the higher sensitivity were also present in the blank.

PAGE 97

CHAPTER 4 DISCUSSION AND CONCLUSION The results presented in the last chapter have shown that the redox mechanism of sulifde ion in molten LiCl/KCl is more complicated than previously thought. These results contradict some of the previous work done on the sulfide system in LiCl/KC1. This situation is not surprisDlg, considering the various ways in which some of the studies were carried out. Most early workers used Pyrex or quartz containers or frits for work in saturated sulfide melts. All of these workers noted severe etching of the glass components even after an exposure of only a few seconds to the sulfide solution. Raleigh et al. (56), were the first to note altered electrode kinetics of metals in sulfide solutions exposed to Pyrex or quartz. They found that the electrode surfaces had been coated with a layer of silicon-containing compounds such as Na 2Si02S. Since the etching process is rapid, and presumably does not stop at the surface layers of the glass, variations in the concentration of sulfide in the molten salt will occur, especially if less than saturated solutions ,are used. Soluble silicon-containing compounds might also be expected to contaminate the melt. Birk and Steunenberg (41) used quartz frit compartment separators in their study of the sulfide system. They also chose to use graphite as the working electrode material. This is not considered to be the best choice, due to the rather high porosity of the graphite and the 92

PAGE 98

93 tendency for carbonyl and other functional groups to be present initially on the surface. The existence of electroactive surface groups could lead to false assumptions concerning the redox mechanism of sulfide, since spurious currents due to the changing electrode area and oxidation or reduction of the surface layer would be encountered. other workers also chose to use electrodes with areas that were not well defined. For example, Raleigh et aL (56) used graphite rods and platinum and molybdenum strips dipping into the solution as working electrodes. Such a procedure is inadvisable in kinetic studies since the LiCl/KCl melt has a tendency to creep along surfaces which are easily ,Jetted by the melt, resulting in changes in the electrode area. Altered electrode kinetics could also result, due to thin layer effects produced by such solvent creeping. For these reasons, no Pyrex or quartz was employed in this work. The wot.king electrode system of glassy carbon and platinum sealed in boron nitride gave a non-porous surface with a seal which was essentially leak proof during the time of the experiment. The cell and cell compartments were constructed of high purity A12 0 3 which showed no sign of attack by the sulfide. As mentioned previously, results show that the first step in the oxidation of sulfide ion in LiCl/KCl at the temperatures studied is the one electron transfer to form the sulfur anion radical, S. If a following chemical reaction is assumed, several possibilities exist for the product of such a reaction. The first is disproportionation. This mode of radical combination is well known in several aqueous and organic reactions such as the disproportionation of 2Cu(I) to Cu + Cu(II) and 2C2 H 5 to C 2 H 6 + C 2 H 4

PAGE 99

94 In order to determine if disproportionation occurs to any signifi-cant extent, the cyclic voltammograms and chr0nopotentiograms were analyzed assuming this particular mechanism. The theoretical considerations used in this analysis \"ere due to Saveant and Nadjo (67) and Saveant and Mastra-gostino (68). The results showed that disproportionation could not be a significant factor in the e"lectrochemical mechi'l .nism. The second possibility is radical recombination to form a dimer. Again the results given in the last section were analyzed and found to conform to the theoretical descriptions given in the liberature. The data obtained from chronopotentiometry, cyclic voltammetry, and chronoamperometry were all self consistent with regards to the proposed dimerization step. For an oxidation process in linear sweep voltammetry, the shift in peak potential from a constant value at low scan rates to a more positive constant value at rapid scan rates indicates the presence of a following irreversible chemical reaction. If any long lived com-pounds formed between the electrogenerated S and the solvent ions, i.e. Li+ or K+, are considered unlikely then the dimerization of S is the most probable result .. Other evidence for dimerization mechanism comes from the values of the peak potentials and half peak potentials of the first oxidation R1' According to theory, E -E = 1.51 --for a totally ir-p p / 2 nF process. reversible dimerization. As can be seen in Table 4, the values of Ep -Ep/2 fit the equation to within 8%. This accuracy of agreement cannot be obtained with similar equations describing other mechanisms. Comparison of the results for the determination of the diffusion coefficient of sulfide ion also supports the dimerization scheme, as shown in the last section. However, the large errors due to the

PAGE 100

95 square root dependence of the diffusion coefficient make it necessary but not sufficient evidence for the proposed reaction scheme. The decrease in_ the dimerization rate constant with increasing temperature is difficult to explain by any mechanism. The fact that only three temperatures were studied coupled with the large standard deviations in the values of the constant casts doubt on whether or not the observed trend is significant. If indeed the trend is significant, then an explanation should be made which deals satisfactorily with the anomolous behavior of the rate constant. One possible reason for this behavior is the following. Free radicals are known to dimerize with little or no activation energy and thus show no increase in the rate constant with increasing temperature. The radicals in this work, however, are anion radicals and it can be assumed as a first approxima-tion that any activation energy present is purely electrostatic in nature. If the trend in the value of the rate constant is significant, then the electrostatic activation energy must increase, resulting in a decrease in the effective collision rate with increasing temperature. The decrease in the number of collisions can be thought of as due to the decreasing effect of ion pairing between the solvent cations, mainly Li+ and the anion radicals. At lower temperatures, the ion pairing between Li+ and S lowers the electrostatic repulsion between -+ -two S ions, and allows the close approach of another S or Li S At higher temper.atures, the ion pairs probably exist to a lesser extent and farther apart, providing less electrostatic shielding and therefore increasing the electrostatic barrier to collisions. It must be re-emphasized that the preceding is only one possible explanation for the

PAGE 101

96 anomolous rate constant effect, and is relevant only if the effect is significant. Another possible .explanation is that in approaching the higher temperatures, a point is reached where equilibrium is beginning to occur in the formation of the dimer. This would explain the apparent decrease in the rate constant. However, other evidence such as peak potential shifts in cyclic voltammetry indicates that the equilibrium, if it exists at all, is sluggish. Obviously, more work is needed in this area to determine the cause of the observed phenomenon. The second oxidation step established by the subtraction procedure previously outlined appears to involve the oxidation of two species. The relative amounts of the two species, and therefore the oxidation mechanism, depend on the concentration of S and the time involved from the onset of the first oxidation to the beginning of the second oxidation. = If the time is short or the concentration of S is low, the radical anion, S generate.d in the first oxidation step dimerizes only to a small extent before the beginning of the next oxidation. The sulfide oxidation mechanism thus proceeds in a simple two step process, with one electron transferred in each step. The cyclic voltammograms at low sulfide concentration in Figure 16 appear to show that as the concentra-tion of sulfide is reduced, the oxidation mechanism approaches the simple two step two electron process. On the other hand, if the duration of the experiment is relatively long or the concentration of sulfide is high, the extent of dimerization increases rapidly. As a result, the dimer concentration at the beginning of the second oxidation step is high, while the radical monomer concentration is low. The mechanism in this case proceeds in what appears to be three one electron steps. In cyclic

PAGE 102

97 voltammetry at relatively slow scan rates, the current due to the second oxidation, that is the oxidation of the 5; dimer, would be expected to be much less than the current due to the oxidation of sulfide ion. The reason for the attenuated current is that the con-centration of the dimer at the electrode surface is approximately one half that of the sulfide ion. The diffusion coefficient of the dimer also would be somewhat smaller, although the current depends only on the square root of the diffusion coefficient. At rapid scan rates, the current of the second oxidation process should be larger, since the principal reactant is now s-, which would have a concentration about equal to the sulfide ion concentration. The diffusion coefficient of 5 would be expected to be larger than 5 thereby giving rise to a larger current. The argument supports the proposed mechanism as can be seen in Figure 25. The ratio of the peak currents'of oxidation processes II and I respectively as a function of the square root of scan rate is seen to increase as the scan rate is increased; i.e. oxidation of 5 is contributing more to the current as the scan rate is increased. The final oxidation step appears to be the one electron oxidation of S;. Since the concentration of 5; at the electrode is entirely de-pendent on the concentration of S2 and thus on the dimerization rate of 5 and the time duration (i.e. scan rate) of the experiment, the current -due to 52 oxidation would be expected to decrease as the scan rate in-creased. This is the result that is observed and the effect is shown as a plot of the ratio of the peak currents for anodic peaks II and III in Figure ]5. The ratio of peak II to peak III is seen to increase with increasing scan rate, due as much to a decrease in peak III as an increase in peak II.

PAGE 103

Figure 25. Current Ratio of Oxidation Processes I and II as a Function of Scan Rate

PAGE 104

-,,...... ...... ,o o 99 o o o o 1 2

PAGE 105

100 The results of the above analysis can be summarized in the following equations, indicating an oxidation mechanism which fits the data. (1) 8 -+ 8 + e -+(2 ) 28 kl -+ 8 2 (3) 8 -+ -+-8 + e (4 ) 28 -+ 8 2 (5) 8 2 -+ -+-8 2 + e -(6) 8 2 -+ -+-8 2 + e Reactions (1), (3), and (4) would be the primary reaction path at rapid scan rates, (1) through (6) at intermediate scan rates, and (1), (2), (5), and (6) at slow scan rates. The indication in reactions (3) and (6) is that the sulfur is deposited as a monomer and a dimer. Actually, there is no evidence from this or other work which points to the sulfur being deposited in any particular form in LiCl/KC1. Gas phase studies have shown that at the temperatures investigated in this work, elemental sulfur occurs in several linear polymeric forms, from 8 2 to 8 8 Any particular length cannot be unambiguously assigned in this case, due to the obviously large differences between the gas phase and the molten salt medium. Therefore the sulfur is arbitrarily assumed to exist as the dimer 8 2 on the electrode surface. The chronopotentiometric results also support the above mechanism, although as mentioned in the last chapter, the rate constant change with

PAGE 106

101 temperature is not consistent. Figure 23 showed that the product 1 / 2 IT was a constant at low current densities. At intermediate d . 1/2 d' eel current ensltles IT lsplay a rapid change with increasing current. At the highest current densities studied, iT 1 / 2 again approached a constant, indicating that a new mechanism was operating. If the dimerization rate constant varied with temperature in the same way that was found for the cyclic voltammetric experiments, then th d hi h 1/2 d e current enslty at w c IT became non-llnear woul be lower at higher temperatures. This would be a direct result of the dimeriza-tion being slower at higher temperatures. The facts seem to favor the alternative explanation of the establishment of a slow equilibrium between the dimer S; and the monomer S This would also help explain the results of the ESR experiment. Since approximately 20-30 seconds were required to solidify the molten samples, there ma y have been time enough spent at lower temperatures in the molten state to allow the radical anion to dimerize. Thus even if an equilibrium or quasi-equilibrium state was reached at the higher temperatures, the experiment may not have been able to show it. The lack of knowledge in this respect is a direct result of the experimental procedure, rather than a funda-mental physical limitation. Mare elaborate equipment such as a spectroelectrochemical cell or a stop-flow ESR cell would provide con-clusive evidence for the establishment of an equilibrium state. Turning to the cathodic portion of the cyclic voltammograms, one finds that at most uemperatures and scan rates studied, there were only two major reduction processes, peaks IV and V, occurring at -0.95V and -1.33V, respectively, vs. SMPE. The exception to this is at 475C, where a third reduction begins to appear at -1.38V.

PAGE 107

102 The first cathodic process, peak IV, is a stripping peak due to the reduction of the sulfur layer deposited during the anodic scan. The product of the first reduction appears to be S2' since two electrons are involved per S2 unit. This was shown in the previous chapter by comparing the charge transferred in the last anodic step with the charge transferred in the first reduction peak. Due to the fact that there is only one reduction process left, peak V, this would mean that the final cathodic wave involves the two electron reduction of S;. However, the calculated value for n, the number of electrons involved in the reduction, is found to be about ].4 at slow scan rates and approaches 1.0 at rapid scan rates. As previously men-tioned, the presence of other sulfur species such as S-with EOs similar to S; would explain these values of n. At rapid scan rates, anodic reaction step (4) may not have time to occur to any significant degree, causing the two reduction processes to occur as one electron transfers. The proposed reduction mechanism can be represented by the following reaction scheme: (7 ) S + e -+ S = (8 ) S2 + 2e -+ S2 (9) S + e -+ S (10) S2 + 2e -+ 2S The relative extent to which each of these reactions occurs would depend on the concentration of S-and the scan rate. Thus, the redox chemistry of S ion is seen to be a complex cornbina-tion of interdependencies. The influence of concentrations, temperature, and experimental time has been examined and a mechanism proposed.

PAGE 108

103 The cathodic reaction scheme in particular is probably not the only one that can be envisioned to fit the experimental data. However, the proposed mechanism appears to he the simplest one to explain the results. Future work involving more sophisticated methods may necessitate changes in the way we view the redox reactions of sulfide and sulfur. The hope is that this work in some way has brought us closer to understanding the electrochemical behavior of sulfide and in molten salts.

PAGE 109

REFERENCES 1. "Proceedings of the Symposium on Electrode Materials and Processes for Energy Conversion and Storage," The Electrochemical Society, Inc., Vol. 77.6,1977. 2. "High Performance Batteries for Electric Vehicle Propulsion and Stationary Energy Storage," Progress Report 79-94, Argonne National Laboratory (1978-1979). 3. B. Cleaver, A. J. Davies, and M. D. Hames, Electrochimica Acta, 18, 719 (1973). 4. R. P. Tischer and F. A. Ludwig, "The Sulfur Electrode in Non-Aqueous Media," Review Article, Ford Motor Company, Dearborn, Mich. 5. T . Porland, in "Fused Salts," edited by B. R. Sundheim, McGraw-Hill,New York, 1964. 6. H. Bloom, and J. O'H. Bockris, ibid. 7. B. R. Sundheim, ibid. 8. M. Blander, in "Molten Salt Chemistry," edited by M. Blander, Wiley Interscience, New York, 1964. 9. I. K. Delimarskii and B. F. Markov, "Electrochemistry of Fused Salt," Sigma Press, Washington, D. C. 1961. 10. P. Delahay, "New Instrumental Methods in Electrochemistry," Interscience Publishers, New York, 1954. 11. J. E. B. Randles, Trans. Faraday Soc., 44, 327 (1948). 12. A. Sevcik, Coll. Czech. Chern. Cornrn., 13, 349 (1948). 13. R. S. Nicholson and I. Shain, Anal. Chern., 36, 706 (1964). 14. W. H. Reinrnuth, J. Am. Chern. Soc., 79, 6358 (1957). 15. H. Marsuda and Y. Agabe, Z. Elektrochern. 59, 494 (1955). 16. Y. P. Gokhshtein, Dokl. Akad. Nauk. SSSR, 126, 598 (1959). 17. R. S. Nicholson, Anal. Chern., 37, 667 (1965). 18. C. P. Andrieux, L. Nadjo, and J. M. Saveant, J. Electroanal. Chern., 26,147 (1970). 19. R. N. Adams "Electrochemistry at Solid E1ectrodes",Marcel Dekker, Inc., New York (1969) pp. 43-66. 104

PAGE 110

105 20. H. A. Laitinen, C H. Liu, and W. S. Ferguson, Anal. Chem., 30, 1266 (1958). 21. H. A. Laitinen and J. W. Pankey, J. Am. Chem. Soc., 81, 1053 (1959). 22. I. D. Pachenko, Ukr. Khim. Zh., 468 (1955); 22 1953 (1956). 23. H. F. Weber, Wied. Ann., 536 (1879). 24. H. J. S. Sand, Phil. Mag., !, 45 (1901). 25. P. Delahay and G. Mamantov, Anal. Chem., 478 (1955). 26. T. Berzins and P. Delahay, J. Am. Chem. Soc. 22, 4205 (1953). 27. W. H. Reinmuth, Anal. Chem., 1515 (1960). 28. A. C. Testa and W. H. Reinmuth, Anal. Chem., 1320, 1324 0-961). 29. A. C. Testa and W. H. Reinmuth, J. Am. Chem. Soc., 784 30. Z. Galus, "Fundamentals of Electrochemical Analysis," pg. 242, Ellis Horwood Ltd., New York, 1976. 31. P. Delahay, C. C. Mattax, and T. Berzins, J. Am. Chem. Soc., 5319 (1954). 32. J. Koutecky and J. Cizek, ColI. Czechoslov. Chem. Corom., 22, 914 (1957) 33. D. B. McDonald "'rransient Techniques in Electrochemistry," Plenum Press, New York, 1977. 34. A. J. Bard and L. R. Faulkner "Electrochemical Method," John Wiley & Sons, New York 1980. 35. H. A. Laitinen and W. S. Ferguson, Anal. Chem., 29, 4 (1957). 36. D. Inman and J. O'M Bockris, J. Electroanal. Chem. 2, 126 (1962). 37. H. A. Laitinen and D. R. Rhodes, J. Electrochem. Soc., 109, 413 (1962) 38. C. H. Liu, Anal. Chem., 1477 (1961). 39. J. H. Kennedy and F. Adamo J. Electrochem. Soc., 119, 1518 (1972). 40. F. G. Bodewig and J. A. Plambeck J Electrochem. Soc. 116, 607 (1969). 41. J. R. Birk and R. K. Steunenberg "Chemical Investigations of Lithium/ Sulfide Cells," Paper presented at Spring ACS Meeting, Los Angeles (1974)

PAGE 111

106 42. R. Bonnaterre and G. Cauquis, J. Chern. Soc., Chern. Commun., !, 293 (1972). 43. T. Chivers and I. Drummond, Inorg. Chern. !!, 2525 (1972). 44. F. Seel and H. J. GuttIer, Angew. Chern., 416 (1973). 45. W. Giggenback, J. Inorg. Nucl. Chern., lQ, 3189 (1968). 46. J. Berkowitz "Elemental Sulfur" Wiley-Interscience, New York, 1965, pp. 148-149. 47. G. Delarue, Bull. Soc., Chim. France, 906 (1960). 48. G. Delarue, ibid, 1654 (1960). 49. G. Delarue, Chernie Analytique, 44, 91 (1962). 50. J. Greenberg, B. R. Bundheim, and D. M. Gruen, J. Chern. Phys., 461 (1958). 51. W. Giggenback, Inorg. Chern., 10; 1308 (1971). 52. Y. Matsunaga, Can. J. Chern., 38, 310 (1960). 53. J. Schneider, B. Dischler, and A. Rauber, Phys. Status Solidi, !i, 141 (1966). 54. B. Cleaver, A. J. Davies, and D. J. Schiffrin, Electrochimica Acta, 18, 747 (1973). 55. J. R. Waggoner, Unpublished Data, University of Florida, (1980). 56. D. O. Raleigh, J. ':1:'. White, and C. A. Ogden, J. Electrochern. Soc., 126, 1087 (1979). 57. H. A. Laitinen and C. H. Liu, J. Am. Chern. Soc., 80, 1015 (1958). 58. H. A. Laitinen and R. A. Osteryoung, J. Electrochern. Soc., 102, 598 (1955). 59. Argonne National Laboratory, Progress Report Oct. 1978 Sept 1979, ANL-79-94, PCJ. 121. 60. C. H. Liu, A. J. Zie1en, and D. M. Gruen, J. E1ectrochern. Soc., 120, 67 (1973). 61. H. A. Laitinen and H. C. Gaur, J. Electrochern. Soc., 104, 730 (1957). 62. R. W. Adams "Electrochemistry at Solid Electrodes" Marcel Dekker, Inc., New York, 1969, pp. 214-231.

PAGE 112

107 63. von Stackelberg, M. Pilgram, and V. T-orne, Z. Elektrochem., 342 (1953). 64. M. L. Olrnste'ad, R. G. Hamilton, and R. S. Nicholson, Anal. Chern., 41, 260 (1969). 65. J. M. Save ant and E. Vianello, Cornpt. Rend., 256, 2597 (1963). 66. D. S. Polcyn and I. Shain, Anal. Chern., 370 (1966). 67. J. H. Saveant and L. Nadj.o, J. Electroanal. Chern., 419 (1971). 68. J. M. Saveant and M. Mastragostino, Electrochirnica Acta, .!1., 751 (1968).

PAGE 113

BIOGRAPHICAL SKETCH James R. Waggoner was born in Salem, on November 11, 1952. He moved to west Palm Beach in 1962. After graduation from Cardinal Newman High School, he entered the University of Florida, where he was granted a Bachelor of Science degree in chemistry in 1978. Mr. Waggoner then entered the graduate program at the University of Florida. During the course of his graduate work, he was the recipient of the Electrochemical Society Battery Fellowship. He is a member of the American Chemical Society and the Electrochemical Society. 108

PAGE 114

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. f / / ____ __ Herbert A. Laitinen, Chairman Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Roger G. Bates Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. j/ / ,,"--..-.-/ Richard A. Yost / Assistant Professpr' of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Q G. Dorsey tant Professor I certify that I have read this study and that in my opinion it conforms to acceptsble standards of scholorly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Gar B. Hoflund Associate Professor of Chemical Engineering

PAGE 115

This d issertation was submitted to the Graduate Faculty of the Department of Chemistry in the College of Liberal Arts and Sciences and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December, 1982 Dean for Graduate Studies and Research