Molybdenum chemistry in molten LiCl±KCl eutectic: anelectrochemical and absorption spectroscopy study of theconcentration dependent stability of solutions of K3MoCl6

J.C. Gabriela, D. Vincentb, J. Bouteillona, J.C. Poigneta, V.A. Volkovichc,T.R. Gri�thsc,*

aLaboratoire d'Electrochimie et de Physico-Chimie des MateÂriaux et des Interfaces, UMR 5631 INPG-CNRS, ENSEEG, Domaine

Universitaire, BP 75 38402, St Martin D'Heres, FrancebUniversite Joseph Fourier, Grenoble, France

cSchool of Chemistry, The University of Leeds, Leeds LS2 WT, UK

Received 30 March 1999

Abstract

The chemical behaviour of LiCl±KCl±K3MoCl6 solutions between 400 and 5008C was investigated. They

underwent slow decomposition, according to a concentration dependent disproportionation reaction yieldingmetallic molybdenum and a volatile black higher valency molybdenum compound, which corresponded with MoCl5.Chronopotentiometric determinations showed a single 3-electron reduction, yielding Mo. The process was slow with

di�usion-controlled mass transfer. The concentration of the Mo(III) species, presumed in the form of MoCl63ÿ ions,

was determined using linear sweep voltametry and open circuit voltage measurements and the overall Mo(III)species concentration by chemical analysis. The results showed MoCl6

3ÿ to be the predominant species in the melt

and the presence of a previously proposed dinuclear species, Mo2Cl93ÿ, could not be con®rmed electrochemically.

This arose to quanti®cation errors in the determination of the electroactive species MoCl63ÿ due to inherent

problems in estimating the active surface area of the electrode. The electronic absorption spectrum of MoCl63ÿ in

LiCl±KCl eutectic at 4008C is reported for the ®rst time. Measurements were made above and below 0.045 moldmÿ3 K3MoCl6, the concentration above which the disproportionation reaction was relatively slow. The results arediscussed in detail and the spectroscopic parameters of the octahedral MoCl6

3ÿ complex were calculated as:Dq=1775 cmÿ1, B = 511 cmÿ1 and C = 1913 cmÿ1. Although an additional band, at 340 nm, appeared during

disproportionation it could not be identi®ed with any currently known polynuclear molybdenum chloro complex oroxychloro complex. It is probably associated with an intermediate in the disproportionation process. No spectralevidence was thus obtained in support of the formation of dinuclear molybdenum species. # 1999 Elsevier Science

Ltd. All rights reserved.

Keywords: Molybdenum; Molten chlorides; Electrochemistry; Spectroscopy; Potassium hexachloromolybdate(III)

Electrochimica Acta 44 (1999) 4619±4629

0013-4686/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.

PII: S0013-4686(99 )00176-0

* Corresponding author. Tel.: +44-113-233-6408; fax: +44-113-233-6565.

E-mail address: t.r.gri�ths@chem.leeds.ac.uk (T.R. Gri�ths)

1. Introduction

The present and developing technologies o�erincreasing applications for metals and alloys coatedwith niobium, tantalum, molybdenum and tungsten.

Coherent molybdenum coatings have been electrolyti-cally obtained from halide melts but more research isneeded, especially as the chemical and electrochemical

equilibria involved are not yet well established. Theaim of this paper is to compare the results of severalrelevant physico-chemical techniques, including electro-

chemical methods and electronic absorption spec-troscopy and thereby obtain a better understandingand knowledge of the chemistry of LiCl±KICl±K3MoCl6 solutions.

Chemical and electrochemical studies of molyb-denum in LiCl±KCl eutectic began over forty yearsago [1]. These solutions were found to be generally

quite stable, but low concentration molybdenum(III)solutions in this eutectic were found, somewhat sur-prisingly, to undergo greater decomposition than more

highly concentrated solutions [2,6]. This unquestion-able observation, ®rst reported by Mellors andSendero� [2], was that for low concentrations of

K3MoCl6 in LiCl±KCl mixtures, at temperatures inthe range 600±8008C, there was evolution of a volatilesubstance and precipitation of a black phase, shownby X-ray di�raction to be Mo metal. The volatile

phase was not identi®ed and the authors ascribed it toa higher valency molybdenum chloride and suggestedthat spectroscopic evidence should be helpful to ident-

ify it. Mellors and Sendero� postulated that this de-composition could be a disproportionation reaction,such as

5MoCl3ÿ6 42Mo� 3MoCl5�g� � 15Clÿ �1�

Obviously such a disproportionation reaction shouldbe favoured by high solute concentrations, which dis-agreed with the experimental ®ndings. The authors

therefore postulated that there were twomolybdenum(III) species in equilibrium

Mo2Cl3ÿ9 � 3Clÿ , 2MoCl3ÿ6 �2�

and that the reaction generating MoCl63ÿ species was

slow.The proposed cathodic deposition process has the

following chemical±electrochemical (CE-type) mechan-

ism

Mo2Cl3ÿ9 � 3Clÿ , 2MoCl3ÿ6 �3�

MoCl3ÿ6 � 3eÿ4Mo� 6Clÿ �4�

The electron transfer step was found to involve three

electrons. The observed slow overall kinetics, whichexplained the production of coherent deposits, was

attributed to the existence in the mechanism of theslow step providing the MoCl6

3ÿ electroactive species.Subsequent studies by other workers [3,4] have pro-

vided some support for this CE-type mechanism.White and Twardoch [5] criticised some of the con-clusions of Mellors and Sendero� and pointed out that

the presence of oxide contamination in the melts couldexplain certain instabilities. However, in their study ofthe cathodic reduction of solutions of K3MoCl6 in

LiCl±KCl and in LiCl±CsCl at 5008C they providedfurther support for the importance of a chemical reac-tion in the overall mechanism. They found that thekinetics of the deposition process appeared to be com-

plex and two cathodic peaks present in some voltamo-grams suggested the presence of two di�erentmolybdenum species containing Mo in the same oxi-

dation state. Thus in spite of these previous studies,the details of the solution chemistry of molybdenum(III) in molten chlorides has remained speculative.

2. Experimental

2.1. Preparation of the LiCl±KCl eutectic

The lithium±potassium chloride eutectic (41.5 mol%

KCl, m.p. 3618C) was prepared from LiCl (StremChemicals) and KCl (Prolabo Normapur). The mainimpurities in these salts are trace amounts of calcium,

magnesium, sodium and barium chlorides, which donot interfere with our electrochemical measurements.Dried powdered salts were mixed and melted under

puri®ed argon (less than 1 ppm O2 and H2O). Thesolute, K3MoCl6, was introduced into the melt duringthe experiment via a lock chamber.

2.2. Preparation of potassium hexachloromolybdate(III)

Potassium hexachloromolybdate(III) is not a com-

mon commercial product and was therefore prepared.Chilesotti [9] published the ®rst procedure, in 1903 andthis was improved in 1927 by Bucknall [10]. Later

Sendero� and Brenner [1] developed a new and morereliable process based on the following scheme:

�MoO4�2ÿ � 8H � � 3eÿ4Mo3� � 4H2O �5�

Mo3� � 6HCl4�MoCl6�3ÿ � 6H � �6�We employed this approach. The catholyte was an

aqueous solution of potassium tetraoxomolybdate(VI).Once saturated with HCl (gas), it yielded a red precipi-tate of K3MoCl6, which was dried under vacuum at

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±46294620

2008C for 20 min. The product obtained gave the fol-lowing quantitative elemental analysis (theoretical

values in parenthesis), wt%: K, 30.65 (27.54); Mo,20.51 (22.52); Cl, 48.84 (49.94). X-ray di�raction pat-terns of the product obtained corresponded closely to

those reported by Amilius et al. in 1969 [11]. Carefulanalysis of our data revealed the existence of a dis-torted perovskite structure [7] having the lattice par-

ameters: a, 1216.0 pm: b, 753.4 pm; c, 1273.1 pm; andb, 108.66.

2.3. Electrochemical cell

The melt was contained in a pyrex crucible (78 mminner diameter, 100 mm high) situated inside an air-tight pyrex tube closed by a pyrex cover ®tted with O-ring ori®ces for electrodes and tubing for gases and a

thermocouple. The pyrex cell was situated in aKanthal-wound furnace. The melt temperature wasmaintained within218C.The working electrodes used were a gold wire (1 mm

diameter) for the voltametric determinations, a passi-vated molybdenum wire for open-circuit potential de-

terminations and a molybdenum wire (1 mm diameter)freshly coated with a cathodic Mo deposit for chrono-potentiometric determinations. The reference electrode

was a silver wire dipped in a solution of AgCl in LiCl±KCl (0.75 mol kgÿ1 solvent) separated from the work-ing electrolyte by a thin pyrex membrane.

2.4. Electrochemical equipment and techniques

Open circuit voltage measurements, chronopotentio-metry and linear sweep voltametry were employed.The electrochemical signals were delivered by a PAR

173 potentiostat monitored by a PAR 175 signal pilot.Data were recorded using a digital Nicolet 2090 scopeand further processed using a HP Vectra microcom-

puter.

2.5. Analytical determination of the molybdenum speciespresent in the melts

If we consider the results of the literature review asa starting hypothesis, we can expect that the molyb-denum species involved are two Mo(III) species, suchas Mo2Cl9

3ÿ and MoCl63ÿ, plus molybdenum metal par-

ticles and possibly a Mo(V) species resulting from dis-proportionation. The total concentration ofmolybdenum species present in the molten solution

was determined by chemical analysis and the concen-tration of the electroactive species, presumed MoCl6

3ÿ,was measured using electrochemical techniques such as

linear sweep voltametry and open circuit potentiome-try. In situ spectroscopic investigations of the speciespresent in the melt were also performed.

2.6. Overall analysis of molybdenum species in solution

The total concentration of molybdenum species pre-sent in the liquid phase was determined volumetrically.Samples of the electrolyte were taken at regular time

intervals, quenched, weighed and then dissolved inwater. The solution obtained was ®ltered to eliminatethe undissolved particles (Mo plus possibly molyb-

denum oxide-covered Mo particles) and then an excessof ferric sulfate was added. The resulting Fe(II) ionswere titrated with potassium dichromate in the pre-

sence of barium diphenylaminesulfonate as redox indi-cator. The overall accuracy was within 5%.

2.7. Voltametric determination of the concentration of

the electroactive Mo(III) species

The electroactive molybdenum(III) concentration

was determined by linear sweep voltametry by measur-ing the reduction peak current. These results had anoverall accuracy of 210%, due to di�culties in esti-

mating the active surface area of the electrode.

2.8. Open circuit potential determinations of the

electroactive Mo (III) species

Open circuit voltage measurements were obtainedwith a passivated molybdenum electrode versus a silver

chloride/silver reference electrode.

2.9. Chronopotentiometric study of the mechanism of

reactions occurring at the molybdenum electrode

For these measurements, it was necessary to prepare

a molybdenum wire coated with a fresh molybdenumlayer obtained by in situ cathodic electrode position. Asilver chloride/silver electrode was used as referenceelectrode.

2.10. Spectroscopic investigations of the species presentin the melts

The spectrophotometer used at Leeds was a Cary14H, speci®cally designed for accurate measurements

of spectra at high temperatures. The necessary tech-niques, and a description of the furnace employed,have been reported previously [8].

3. Results and discussion

Once the eutectic had been melted and after additionof K3MoCl6, we observed the precipitation of a dark

phase and evolution of a phase which condensed onthe cold walls of the sealed electrochemical cell.Samples of the dark phase were identi®ed as Mo by X-

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±4629 4621

ray di�raction. The vapour phase was not identi®edbecause, once taken out of the cell, it reacted immedi-

ately with air. Quantitative chemical analysis yielded agood mass balance for the molybdenum present in thetwo solid phases and in the melt.

3.1. Results of voltametric and chemical analysis of thespecies dissolved in the melt

Fig. 1 gives a typical voltamogram, with only onereduction peak (A), which was shown to be due to anirreversible 3-electron step [7]. No evidence for molyb-

denum in solution in any oxidation state other than+3 was obtained. The concentration of the Mo(III)electroactive species was calculated from:

ip � 0:498nF�anFD=RT �1=2Cv1=2 �7�

where ip is the cathodic peak current; a the electrontransfer coe�cient; n the number of electronsexchanged; D the di�usion coe�cient of the Mo(III)

species; and v the voltage sweep rate.Fig. 2 shows the results obtained for the total Mo

and electroactive Mo species concentrations, in twosolutions containing initially 0.06 and 0.03 mol dmÿ3

K3MoCl6. The changes in concentration of the electro-active species and that of total molybdenum, deter-mined by chemical analysis, are shown as a function of

time. The initial rise of molybdenum concentrationwith time is attributed to the slow dissolution ofK3MoCl6. The total concentration of molybdenum in

the melt reached a maximum between two to threehours after K3MoCl6 had been added to the LiCl±KCleutectic and then began to decrease. Fig. 2 also shows

that decomposition was more pronounced when the in-

itial concentration of molybdenum in the melt waslow, in agreement with our previous observations [7].

From these results it is clear that the decompositionoccurring during K3MoCl6 dissolution is less pro-

nounced if the initial concentration of molybdenum inthe melt was above ca. 0.04 mol dmÿ3. These data

suggest zero order kinetics for the decomposition reac-tion.

We next investigated electrochemical measurements

in solutions with high molybdenum concentration, pre-pared 12 h in advance. The change in total and electro-

active molybdenum concentrations was found to beessentially the same (Fig. 2), upon recognising that the

magnitude of the estimated experimental errors is dueto di�culties in estimating the active surface area of

the electrode and hence the origin of some inaccuracyin the concentration of the electroactive species. This

active area problem also explains why the concen-tration of the electroactive species may appear to be

above that of total molybdenum in solution. However,for a given experiment in which concentration vari-

ations with time are investigated, the error due to thearea of the electrode vanishes since the same electrode

is used under the same conditions throughout the ex-periment. On these arguments, the electroactive species

is expected to predominate. Moreover, the 3-electronreduction reaction of the Mo(III) electroactive species

shows that this species is mononuclear: if any dinuclear

Fig. 1. Characteristic voltamogram of a 10ÿ2 mol dmÿ3 sol-

ution of K3MoCl6 in LiCl±KCl at 5008C. Sweep rate: 250 mV

sÿ1.

Fig. 2. Change with time of the total (thick line) and, electro-

active (thin line) molybdenum concentrations in LiCl±KCl±

K3MoCl6 electrolytes at 5008C for two di�erent initial con-

centrations of Mo(III). The total concentration of molyb-

denum was determined by chemical analysis and the

concentration of the electroactive species by voltametry. After

ca. 3±5 h these concentrations were essentially identical.

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±46294622

Mo(III) species is present it is behaving as electro-inac-tive.

3.2. Chronopotentiometric study of the mechanism ofreduction of the electroactive species

The voltametric study showed evidence of a singleelectroactive species. Chronopotentiometry is a con-venient method for investigating the possibility of a

chemical reaction interfering in the process. Variationsin the plot of it 1/2 (where i is the current density andt, the transition time) vs. the applied current density

depend on the way by which the electroactive speciesare supplied to the electrode surface, viz., either bysimple di�usion or by di�usion plus chemical reaction

or adsorption. Strict proportionality between it 1/2 andi is observed when the mass transfer process is purelydi�usion-controlled.A molybdenum electrode, prepared by depositing

molybdenum at low current density over a twelve hourperiod in LiCl±KCl solutions containing K3MoCl6,was used in our chronopotentiometric experiments.

This electrode preparation was necessary to preventthe interference of nucleation phenomena in the re-duction mechanism. Our results are consistent with

Sand's law (Fig. 3). Sand's equation relates the tran-sition time (t ) and the concentration (C ) of reactant inthe melt, viz.

it1=2 � 1=2p1=2nFD1=2C �8�

where D is the di�usion coe�cient of the reactant; F

the Faraday constant; and n the number of electronsparticipating in the reaction.The linear plot passing through the origin shows

that the reaction is di�usion-controlled and that nochemical reaction supplying the electroactive speciesinterferes in the mass transfer process. The Mo(III) dif-

fusion coe�cient in LiCl±KCl was calculated usingSand's law and the value obtained at 5008C was820.5 � 10ÿ6 cm2 sÿ1. This value is in agreement with

that of 1.43 � 10ÿ5 cm2 sÿ1 reported by Sendero� [2]at 6008C. The values of the di�usion coe�cients canbe related to the solvodynamic mean radii (R ) of theelectroactive species by the Stokes±Einstein equation,

R � kT=�6pZD�, where k is the Boltzmann constant; Tthe absolute temperature; Z the viscosity; and D thedi�usion coe�cient. At 5008C, this radius is close to

350 pm. Upon comparing this value with the radius ofthe Mo3+ ion (80 pm [13]) the solvation, in reality thecomplexation of this ion with the oppositely charged

nearest-neighbour chloride ions of the melt, of the elec-troactive species is substantiated. Thus MoCl6

3ÿ, withan octahedral arrangement of chloride ligands and acalculated radius ranging from 320 to a maximum of

440 pm, (assuming a chloride ion radius of 181 pm),can be assumed to be the electroactive species.

3.3. Open circuit potential determinations of theelectroactive species

The development of the open circuit potential of amolybdenum electrode should permit the change in theelectroactive molybdenum concentration to be fol-

lowed in a stable melt as a function of time. A freshlyprepared Mo electrode dipped in a concentratedmatured (12 h) K3MoCl6 solution obeyed Nernst's law

Fig. 3. Dependence of it 1/2 in stable molybdenum solutions

on the concentration of Mo(III) in the LiCl±KCl eutectic at

5008C. This plot shows that the data obtained are consistent

with Sand's law (Eq. (8)). Fig. 4. Change with time of the open circuit potential of a

passivated molybdenum electrode for several

molybdenum(III) concentrations (initial and ®nal concen-

trations are given for each curve).

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±4629 4623

for a 3-electron exchange during the ®rst three hours.For the next ®ve hours this potential changed some-

what erratically, only becoming stable after eighthours. These important changes are attributed to a

slow chemical passivation reaction of the molybdenum

electrode in the electrolyte.To overcome this problem and rapidly obtain stable

and reproducible potential measurements, the passiva-tion process was accelerated by anodic polarisation of

the electrode. This operation was performed with the

electrode in the LiCl±KCl±K3MoCl6 electrolyte, usinga current density of 120 mA cmÿ2 for 60 s.

Figure 4 shows the development of the open circuit

potential of a passivated molybdenum electrode vs.time and total molybdenum concentration. These

curves show the same trend as the pro®les observed inFig. 2.

Potentials recorded by this method showed an exactNernst dependence over a large range of concentration

with an exchange of 0.5 electrons. This was shown to

be related to the formation of a visible insoluble layerof a molybdenum compound with a valency of 2.5,

due to a slow reaction between Mo and Mo(III).Similar results have been obtained by Arkhipov et al.

in NaCl±KCl±CsCl±MoCl3 [12], who studied molyb-denum electrode position and dissolution in melts with

di�erent cation combinations and obtained compact,

columnar molybdenum coatings. They also detectedpassivation of the anode by solid electrochemical reac-

tion products in NaCl±KCl±CsCl±MoCl3 electrolytes.In NaCl±KCl±MgCl2±MoCl3 melts at high MoCl3concentrations they also noted that a lower valent salt

was deposited on the cathode together with molyb-denum.

Fig. 5. Electronic absorption spectra and their resolution into individual bands, of K3MoCl6 solutions in the LiCl±KCl eutectic at

4008C measured for high, 0.1 mol dmÿ3 (a) and low, 0.02 mol dmÿ3 (b), molybdenum concentrations.

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±46294624

3.4. Conclusions from analytical and electrochemicaldeterminations

A 3-electron irreversible reduction of a mononuclearMo(III) compound, with no kinetic complication byany preceding chemical reaction has been shown bylinear sweep voltametry and chronopotentiometry.

However, as far as the earlier [2] proposed existence ofa polynuclear species such as Mo2Cl9

3ÿ is concerned,no unambiguous conclusion from electrochemical

measurements can be reached because the determi-nation of the corresponding concentrations were notsu�ciently precise. From our experiments we can how-

ever conclude that the mononuclear species should bepredominant. However, electronic absorption spec-troscopy can be used to de®ne the structure and

explore other properties of the molybdenum speciespresent in solution. Sendero� and Mellors [2] havesuggested using spectroscopy to settle the question ofthe formation of bi- or polynuclear molybdenum

species in chloride melts but such a study has not pre-viously been reported.

3.5. Spectroscopic studies

Spectroscopic measurements were made usingsamples of the LiCl±KCl eutectic and K3MoCl6 pre-pared for the electrochemical studies, to minimise poss-

ible di�erences between the two sets of results. Tominimise or delay the onset of decomposition the low-est possible temperature of the eutectic was used,

4008C. Solutions were prepared in two ways, by add-ing K3MoCl6 to the molten eutectic and by carefulheating of the eutectic powder thoroughly mixed withthe salt. There was very little di�erence between the

two methods, but the K3MoCl6 dissolved in the latterslightly faster.Fortunately spectra could be measured between 300

and 800 nm of solutions having concentrations aboveand below the concentration established above by elec-trochemistry and con®rmed here, as that below which

disproportionation occurred. Concentrations around0.02 mol dmÿ3 could be investigated using a 10 mmpath-length cuvette and for 0.1 mol dmÿ3 solutions an

optically polished silica block of cross-section 9 � 8

mm was inserted into the cuvette to reduce the optical

path through the liquid to 1 and 2 mm respectively: it

also functioned as an e�cient stirrer.

A and B in Fig. 5 show typical spectra taken at high

and low concentrations shortly after all the K3MoCl6added had dissolved. The terms high and low refer to

concentrations above and below that at which dispro-

portionation occurs (ca. 0.04±0.05 mol dmÿ3). The

spectrum obtained at the high concentration at 4008C(Fig. 5a) exhibited maxima at 425, 552 and 685 nm

(23,530; 18,115 and 14,600 cmÿ1, respectively) and re-

sembled that reported for [MoCl6]3ÿ by She�er et al.

[14] at room temperature in a chloroaluminate melt.

This high concentration spectrum shows higher absor-

bance values than normally recommended for accurate

spectroscopic work. However, the manufacturers claim

exceedingly low stray light characteristics for the Cary

14H spectrophotometer and that spectra with absor-

bances up to 5.6 can be reliably recorded. This had

earlier been con®rmed [15] for the instrument used in

this study and thus these high absorbances are valid.

They were obtained using neutral density ®lters: the

chart recorder normally operates in the range 0±2

absorbance, but the pen can be brought back on scale

by inserting a suitable neutral density ®lter of known

absorbance into the reference beam. The true absor-

bance is then obtained by adding the chart absorbance

to that of the ®lter (which is essentially wavelength

independent).

The spectrum at low concentration in Fig. 5b has

essentially similar peak maxima but also a fourth ad-

ditional peak, at 340 nm (29,410 cmÿ1). However, the

peak maximum steadily shifted with time to lower

wavelengths, until the band maximum could no longer

be resolved (when its absorbance was in excess of 4.5).

This peak at 340 nm has not previously been reported:

the best previous study [14] only recorded spectra

down to 360 nm. Since the pro®le approaching this

wavelength was rising it is possible that the band was

present but the absorbance of the chloroaluminate sol-

vent prevented its resolution.

Table 1 includes the peak maxima and their molar

absorbance values obtained above and below the dis-

Table 1

Electronic absorption spectroscopy data for MoCl63ÿ

Solvent Temperature (8C) Peak maxima (nm) (ea (dm3 molÿ1 cmÿ1)) Ref.

44.4/55.6 mol% AlCl3±MEIC 25 439 (36), 544 (30), 685 (1.6) [14]

LiCl±KCl 400 425 (44), 552 (29), 685b (5.9) high conc.

LiCl±KCl 400 340 (86), 420 (64), 550 (39), 685b (13) low conc.

a e=molar absorbance.b Shoulder.

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±4629 4625

proportionation concentration. The latter values were

calculated from the maximum absorbance values at the

various peak maxima and employing the molybdenum

concentration corresponding to the amount of

K3MoCl6 added initially to the LiCl±KCl eutectic.

Figure 6 shows the decrease of the absorbance with

time at 550 and 420 nm, close to the wavelength max-

ima of the two most prominent peaks. The same trend

may clearly be seen as was noted in the electrochemical

studies (Figs. 2 and 4). The initial rise noted while the

K3MoCl6 continues to dissolve is also apparent. The

time taken for all the salt to dissolve could have been

reduced by operating at 5008C, at which temperature

the eutectic is much less viscous and easier to stir.

However, although initial experiments indicated as

expected a shorter time before disproportionation com-

menced, other necessary aspects of sample manipu-

lation then become technically more di�cult, leading

to unacceptable uncertainties concerning the reliability

of the resulting data.

It will be noted that a few of the later data points in

Fig. 6 do not lie exactly on a smooth curve. This is

because when disproportionation begins or is taking

place some of the molybdenum metal formed coagu-

lated before falling slowly to the bottom of the cuvette.

In consequence, small particles are seen and are pre-

sent in the light beam during the recording of some

spectra. The e�ect of this is to increase the overall

absorbance of the spectrum. A partial base-line correc-

tion can be made by subtracting any absorbance

observed at 800 nm, where essentially zero absorbance

is expected for the species present in this system and

this was undertaken. It worked well for absorbance

values below 2 but was less reliable at very high absor-

bance values. This is attributed to minor changes whenrepositioning the cuvette after removal for momentary

inspection to check for the presence of particles. Thissometimes had a noticeable e�ect upon the recordedabsorbance, but only if the absorbance was greater

than 2. Unfortunately this could not always be quanti-®ed or su�ciently corrected after these spectra wererun because the baseline could not be guaranteed to

remain linear: the contribution to the absorbance at800 nm by the particles may not be the same as that atlower wavelengths because the number and size of par-

ticles in the light beam at the time the lower wave-length region is being recorded could be di�erent.The time-scale of the stability of the higher concen-

tration solutions is also shown in Fig. 6. Once the

K3MoCl6 had all dissolved, after about 50 min, theabsorbance remained essentially constant for another200 min before slowly decreasing. The low concen-

tration solutions decreased immediately in absorbanceas soon as (and possibly before) all the molybdenumsalt was dissolved. Between 50 and 650 min the high

concentration solution has only decreased by around18% and the low by more than 50%, thereby support-ing and con®rming the electrochemical observations.

3.6. Chemistry of the [MoCl6]3ÿ ion

While there have been a number of studies of moltensalts containing molybdenum species only a few havereported the spectra of these coloured solutions. The

advent of room temperature ionic liquids has meantthat spectra can be recorded of species that are notstable at high temperatures. Further, ambient tempera-

ture conditions mean that spectral bands are subjectedto minimum spectral broadening and features pre-viously hidden may now be revealed. The neutralroom temperature melt consists of a mixture of AlCl3and MEIC, where the latter is 1-ethyl-3-methylimida-zolium chloride. Sche�er et al. [14] have recorded thespectra of molybdenum species in this melt and ident-

i®ed [MoCl6]3ÿ as the only stable Mo(III) species.

Their reported peak maxima (and molar absorbancevalues) are included in Table 1. They did not suggest

that there was any evidence for a dimer being presentin any of the various solutions investigated. Unlike inelectrochemistry, there is no equivalent to an electro-inactive species that cannot be detected by absorption

spectroscopy, provided the appropriate concentrationsand wavelength regions are investigated.Molybdenum(III) forms complexes of the type

[MoX6]3ÿ with several halogens and pseudohalogens.

Prolonged electrolytic reduction of a solution of MoO3

in concentrated hydrochloric acid gives a green sol-

ution of Mo(III) in the form of chloro complexes,from which [MoCl6]

3ÿ and [MoCl5(H2O)]2ÿ can beprecipitated with the larger alkali metal cations. The

Fig. 6. Dependence of relative Mo(III) concentration in LiCl±

KCl±K3MoCl6 melts with time for the initial high, 0.1 mol

dmÿ3 (circles) and low, 0.02 mol dmÿ3 (squares) molybdenum

concentrations. Measurements made at 550 nm (solid sym-

bols) and 420 nm (open symbols).

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±46294626

structure of the [MoCl6]3ÿ ion is well known to be

octahedral and the two main bands at 550±552 and420±425 nm are those of the transitions 4A2g4

4T2g

and 4A2g44T1g, respectively [16].

The spectra presented in Fig. 5 were resolved intoindividual bands using PeakFit computer software.

This allowed more accurate determination of bandpositions. Assigning the two bands at ca. 17,750 and23,200 cmÿ1 (563 and 431 nm, respectively) to the

above transitions it was possible to calculate the crystal®eld splitting energy, D=10Dq and the Racah par-ameter B, using the equations for energy levels for a d3

ion [17]. The values obtained for the high molybdenumconcentration (Fig. 5a) were: Dq=1775 cmÿ1 and

B = 511 cmÿ1. For a free Mo3+ ion, Bo=610 cmÿ1

and therefore the degree of `ionicity' in MoCl63ÿ can be

estimated as approx. 84%. The low intensity band

observed at ca. 683 nm (14,640 cmÿ1) is due to thespin forbidden 4A2g 4

2T2g transition and the value ofthe Racah parameter C, calculated from the energy of

this transition, is 1913 cmÿ1. The energy of the thirdspin-allowed transition, 4A2g4

4T1g(P ), calculated

using obtained values of Dq and B, is ca. 37,715 cmÿ1

and this transition is thus expected at around 265 nm.Similarly for the spectrum at the low molybdenum

concentration (Fig. 5b) values of the spectroscopic par-ameters, calculated using the best resolution the pro-gram allowed were: Dq=1746 cmÿ1, B = 623 cmÿ1

and C= 1849 cmÿ1. However, we consider this resol-ution less satisfactory, because the program was not

designed to enable us to include the evident contri-bution from the tail of the intense absorbance athigher wavenumbers. In support we note that the

resulting similarity of the Racah parameter B with B0

would now indicate the invalid conclusion that in di-lute solution the bonds in MoCl6

3ÿ are essentially ionic.

The values of the above spectroscopic parameters (Dq,B and C ) were therefore calculated using band pos-

itions obtained from the observed maxima or shouldersin the experimental spectra (Table 1). For both thehigh and low molybdenum concentration spectra these

values were now similar: Dq ca. 1460 cmÿ1; B ca. 504cmÿ1; and C ca. 1900 cmÿ1.Spectroscopic studies of the species MoCl6

3ÿ andMo2Cl9

2ÿ in the visible region, starting at 333 nm, inHCl solutions [15] gave identical spectra, except that

the dinuclear molybdenum complex showed a splittingof the low energy transition at ca. 680 nm(4A2g4

2T2g) into doublet, with a peak separation of

around 1700 cmÿ1. High temperature broadening ofbands does not allow unambiguous analysis of the

spectra obtained at high molybdenum concentration inLiCl±KCl melt, but at present it does not seem likelythat there is evidence for the formation of a Mo2Cl9

2ÿ,a molybdenum dinuclear species, in agreement with theconclusions from our electrochemical data. Attempts

to substitute a doublet for the single resolved band at683 nm (14,640 cmÿ1) in Fig. 5a, having a separation

of 1700 cmÿ1 did not result in any signi®cant improve-ment in the ®t.Carlin and Osteryoung [18] have reported spectra

measured in room temperature melts of several molyb-denum chloride dimers, including the Mo(III) andMo(II) species, Mo2Cl9

3ÿ and Mo2Cl84ÿ, respectively.

The spectrum of the molybdenum(III) dimer exhibitedfour peaks at 435, 527, 655 and 758 nm, with the mostintense peak at 527 nm and of molar extinction coe�-

cient, e, 560 l molÿ1 cmÿ1. Only one peak wasobserved in the spectrum of the molybdenum(II)dimer, at 534 nm (e=740 l molÿ1 cmÿ1). The spectrain Fig. 5 show no maxima around 520±540 nm.

Recognising the high values of the molar absorptioncoe�cients, reported by Carlin and Osteryoung [18]for these molybdenum dimers, means that it is unlikely

that either Mo2Cl93ÿ or Mo2Cl8

4ÿ were formed in ourmelt.Our spectroscopic studies have thus shown that

there is no evidence for the presence of dinuclear mol-ybdenum chloride species in equilibrium with MoCl6

3ÿ

in LiCl±KCl melt, including Mo2Cl93ÿ, postulated by

Mellors and Sendero� [2].

3.7. Disproportionation of [MoCl6]3ÿ

Our electrochemical results indicate disproportiona-tion of Mo(III) solutions at concentrations below

5 � 10ÿ2 molÿ1 dmÿ3 and the X-ray di�raction evi-dence identi®ed Mo(O) as one of the products: theother is postulated to be Mo(V). Disproportionation

reactions are known for chlorine-containing com-pounds and complexes of both molybdenum and tung-sten above 2008C, though they have not previouslybeen reported for Mo(III) species. Spectroscopic

measurements can help identify the reaction productsremaining in solution. The intermediate Mo(IV) specieswill be [MoCl6]

2ÿ and this has been reported in chlor-

oaluminate melts [14] to have two intense peaks closetogether at 366 and 391 nm, with molar absorbancesaround 4.5 � 103 dm3 molÿ1 cmÿ1. This is over a

three-fold increase in magnitude compared with thevalues for the main peaks for [MoCl6]

3ÿ and thus ifthis intermediate species was formed at any time inamounts greater than about 0.05% it would be

detected, because the absorbance below 500 nm wouldrise and not fall with time (Fig. 6). We thus concludethat there is no build-up in the concentration of any

intermediate Mo(IV) species formed and hence it im-mediately disproportionates into Mo(V) and a loweroxidation state molybdenum species. Crystals of the

tetraethylammonium salt of [MoCl6]ÿ are black but

the absorption spectrum of this octahedral complexion in fused salts has not been reported. Denisov and

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±4629 4627

Kobeleva [19] reported spectra of molybdenum(V)

chloride solutions in methanol, saturated with HCl.

These solutions showed one rather broad peak around

315 nm. The rising absorption observed in our spectra

in the 350 nm region with time is consistent with the

formation of this Mo(V) species.

The main (spectroscopic) di�erence between the

behaviour of the high and low concentration regimes is

the presence of the band at, initially, 340 nm. Fig. 7

shows and compares the absorbance at 340 nm at high

and low concentrations as a function of time. The in-

itial rise re¯ects the time taken for the eutectic to melt

and the K3MoCl6 to dissolve therein. The system was

not stirred during this period nor was the temperature

raised above 4008C to speed dissolution as this would

also a�ect the rate of disproportionation. The absor-

bance values (of the new peak) at 340 nm were

obtained by subtracting the absorbance of the rising

curve on which it appeared. The rate of fall in absor-

bance after about 30 min is slightly misleading because

the peak was also steadily shifting to 308 nm before it

could no longer be observed, as indicated earlier. This

shift arises from a steady growth of an intense band in

the ultraviolet region, the rising edge inducing the

apparent shift. Fig. 7 also indicates that this peak had

disappeared after 100 min, but again this is a conse-

quence of its shifting. It is not possible to know if it

was still present but hidden at the end of the exper-

iment.

It is tempting to associate the peak with the appear-

ance of [MoCl6]2ÿ but this species has two intense and

close but well-resolved peaks at 366 and 391 nm in

room temperature melts [14] and thus has the wrong

pro®le and peaks not near our peaks. Moreover, the

pro®le presented by She�er et al. [14] shows a mini-

mum at 340 nm, where we obtained our maximum.

MoCl6ÿ could be a candidate species. Its spectrum at

room temperature, in MeOH±HCl solutions [19], has a

main peak at 315 nm. However, to ®t with our eutectic

data at 4008C, this peak would have to have shifted to

340 nm, some 2300 cmÿ1 and a temperature shift of

this magnitude is much greater than we have ever

encountered.

There are other candidates. The compound Mo2Cl10is green±black and could be an initial decomposition

product. Alternatively, an oxychloride may be formed.

Care was taken to maintain a slight positive pressure

and ¯ow of dry argon over the mixture at all times.

However, even though the ppm levels of dioxygen are

reported as less than 10, a similar experiment in our

laboratory with lanthanide chlorides dissolved in LiCl±

KCl eutectic showed the slow formation and precipi-

tation of LnOCl (T.R. Gri�ths and H.V.St.A.

Hubbard; unpublished results). One of the oxychlor-

ides, MoOCl3 is dark brown and will co-ordinate with

chloride to form [MoOCl4]ÿ and solutions containing

[MoOCl5]2ÿ are intensely coloured. In the room tem-

perature solutions of MoOCl3 the maximum was how-

ever observed at ca. 440 nm [19].

Denisov and Kobeleva [19] have reported the spectra

of molybdenum(III) chloride complexes in methanolic

solutions. At a Mo(III) concentration of 0.04 mol

dmÿ3 there was no maximum observed in the spectra

between 280±340 nm. However, addition of HCl had a

remarkable e�ect on the spectra: with a HCl concen-

tration of 1.6 mol dmÿ3 a well pronounced maximum

appeared at ca. 310 nm. No explanation for this maxi-

mum was o�ered, but it is clear that the maximum

was dependent upon the presence of an excess of chlor-

ide ions. In our case, the maximum at 340 nm has

been observed at low molybdenum concentration and

thus in the presence of an excess of chloride ions, simi-

lar to the above situation. Thus the species associated

with this maximum cannot at present be explained and

nor were attempts to identify it using ligand ®eld the-

ory successful.

It is therefore here proposed that, since disproportio-

nation is identi®ed by both spectroscopic and electro-

chemical techniques for the low concentration

solutions, our peak at 340 nm is thus probably associ-

ated with an intermediate in the disproportionation

process or possibly the formation of another chloride

species, which at present, despite investigating all

chloro species reported to date, cannot be identi®ed

with certainty, although work is in progress.

Fig. 7. Dependence of the relative peak high at 340 nm for

the initial high, 0.1 mol dmÿ3 (.) and low, 0.02 mol dmÿ3 (w)

molybdenum concentrations. After 100 min at high Mo con-

centration the peak was completely hidden under the rising

intense band in the ultraviolet region.

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±46294628

References

[1] S. Sendero�, A. Brenner, J. Electrochem. Soc. 101

(1954) 16.

[2] S. Sendero�, G.W. Mellors, J. Electrochem. Soc. 114

(1967) 556.

[3] D. Inman, R.S. Sethi, R. Spencer, J. Electroanal. Chem.

29 (1971) 137.

[4] D. Inman, R. Spencer. Advances in Extractive

Metallurgy and Re®ning, IMM paper. 12 (1977).

[5] S.H. White, U.M. Twardoch, J. Appl. Electrochem. 17

(1987) 225.

[6] G.J. Kipouros, D. Sadoway, J. Appl. Electrochem. 18

(1988) 823.

[7] J.C. Gabriel, J. Bouteillon, J.C. Poignet, J.M. Roman,

J. Electrochem. Soc. 141 (1994) 2286.

[8] T.R. Gri�ths, P.J. Potts, J. Chem. Soc. Dalton Trans.

1975, p. 344.

[9] A. Chilesotti, Gazz. Chim. Ital. 33 (1903) 354.

[10] W.R. Bucknall, S.R. Carter, W. Wardlau, J. Chem. Soc.

1927, p. 513.

[11] Z. Amilius, B. van Laar, H.M. Rietveld, Acta. Cryst. B

25 (1969) 400.

[12] P. Arkhipov, A.N. Baraboshkin, Z.I. Valeev, Z.S.

Martem'yanova, Electrokhimiya 26 (1990) 1555.

[13] R.C. Weast (Ed.), Handbook of Chemistry and Physics,

55th ed., 1974, p. F198.

[14] T.B. Sche�er, C.L. Hussey, K.R. Seddon, C.M. Kear,

P.D. Armitage, Inorg. Chem. 22 (1983) 199.

[15] H.V.St.A. Hubbard, T.R. Gri�ths, J Chem. Soc.

Faraday Trans. 83 (1987) 1215.

[16] P.E. Kazin, N.A. Subbotina, A.I. Zhirov, V.V.

Zelentsov, M.G. Felin, Russ. J. Inorg. Chem. 34 (1989)

294.

[17] A.B.P. Lever, Inorganic Electronic Spectroscopy, in:

Elsevier Publishing Company, Amsterdam, 1968, p. 182.

[18] R.T. Carlin, R.A. Osteryoung, Inorg. Chem. 27 (1988)

1482.

[19] N.T. Denisov, S.I. Kobeleva, New J. Chem. 14 (1990)

69.

J.C. Gabriel et al. / Electrochimica Acta 44 (1999) 4619±4629 4629