A Chemical Thermodynamics Review Applied to V2O5 Chlorination

R

A

EM

a

ARRAA

KCVC

C

1

teiTgcos

(

0h

Thermochimica Acta 559 (2013) 1– 16

Contents lists available at SciVerse ScienceDirect

Thermochimica Acta

jo ur n al homepage: www.elsev ier .com/ locate / tca

eview

chemical thermodynamics review applied to V2O5 chlorination

.A. Brocchi, R.C.S. Navarro ∗, F.J. Mouraaterials Engineering Department (PUC-Rio), Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil

r t i c l e i n f o

rticle history:eceived 24 July 2012eceived in revised form 10 January 2013ccepted 25 January 2013vailable online 6 March 2013

eywords:

a b s t r a c t

This work is mainly related to the thermodynamic study of the V2O5 carbon-chlorination. In this context,three different approaches of diverse complexity level were employed. First, the formation of individualvanadium chlorides and oxychlorides were considered on the basis of the well known �G◦ × T diagrams.It is suggested that these simple constructions can help in understanding the formation sequence ofvanadium chlorinated species. Moreover, the relative stability of the most stable vanadium chloride (VCl4)and oxychloride (VOCl3) was addressed based on available thermodynamics data. Finally, a gas phase

hemical thermodynamic2O5

hlorination

speciation calculation was performed in order to obtain, simultaneously, the concentration of all possiblevanadium chlorides and oxychlorides as a function of temperature, Cl2 and O2 partial pressures. It isdemonstrated that, although diverse in the complexity level, the three methods considered have a strongrelation with each other, and converge to a better insight into the nature of the chemical equilibriumstates achievable for the reaction system under study. The same approach can be applied to any other
reaction system of technological importance.
© 2013 Elsevier B.V. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1. Reductive chlorination roasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2. The system V–O–Cl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1. Vanadium oxides and chlorides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1.1. V2O5 direct chlorination and the effect of the reducing agent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.2. Relative stability of VCl4 and VOCl3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3. Final remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

. Introduction

In general terms the present work can be considered as a con-ribution to a quantitative study review related to the chemicalquilibrium of metallic oxides chlorination. Three possible ways ofnvestigating the achievable equilibrium states will be presented.

one is defined by the minimization of the total Gibbs energy of thesystem.

The three mentioned methods are then applied to the studyof the thermodynamic viability of the reaction between V2O5 andgaseous Cl2. The importance of the presence of a reducing agent isdiscussed through the construction of proper �G◦ × T in the pres-

he first one is based entirely on the construction of �G◦ × T dia-rams. The second one applies the reactions Gibbs energies foromputing the concentration of specific chlorinated compoundsf interest and enables the study of the relative stability of thesepecies in a specific temperature range. The third and most general

∗ Corresponding author. Tel.: +55 2135271720.E-mail addresses: [email protected] (E.A. Brocchi), [email protected]

R.C.S. Navarro).

040-6031/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.tca.2013.01.025

ence as in the absence of graphite. The same method is also appliedfor the formulation of possible chlorination paths for the formationof the most stable chlorinated species produced at high P(Cl2) andlow P(O2) values, conditions usually found in most experiments.The relative stability of the mentioned chlorides is then studiedbased on the independent resolution of chemical equilibrium equa-
tions. Finally, the equilibrium states available to the system arepredicted through the minimization of its total Gibbs energy. Theeffect of T, P(Cl2) and P(O2) over the composition of the gas phase inclearly presented and discussed. Also, calculations of the enthalpy
dx.doi.org/10.1016/j.tca.2013.01.025

http://www.sciencedirect.com/science/journal/00406031

http://www.elsevier.com/locate/tca

mailto:[email protected]

mailto:[email protected]

dx.doi.org/10.1016/j.tca.2013.01.025

2 ochimica Acta 559 (2013) 1– 16

atrpi

svdsctwimwa

1

tcplpfflpcrtfpp(

midttbcmorcOtr

ipcpsmi

tabp

E.A. Brocchi et al. / Therm

ssociated with the chlorination reactions are also performed, andhe impact of the amount of Cl2 available over the heat absorbed oreleased by the reaction system is explored. Emphasis is given to theoint that the most general strategy must embrace the tendencies

ndicated by the implementation of simpler methodologies.Until recently, the approach applied for chemical equilibrium

tudies was almost exclusively based on standard free energyersus temperature and predominance diagrams. However, nowa-ays, advances in computational thermodynamics have enabled aort of software developments which can perform more complexalculations. So, it is understood that time has come for a review onhermodynamics chlorination which can combine its basic aspectsith a now available new kind of approach. Therefore, the general-

ty and depth of the produced information for this work have alsootivated the authors to publish its content in a form of a chapter,hich could well be considered as a part of a thermodynamic book

pplied to high temperature process.

.1. Reductive chlorination roasting

Chlorination roasting has proven to be a very important indus-rial route and can be applied for different purposes. Firstly, thehlorination of some important minerals is a possible industrialrocess for producing and refining metals of considerable techno-

ogical importance, such as titanium and zirconium. Also, the samerinciple is mentioned as a possible way of recovering rare earthrom concentrates [1] and metals, of considerable economic value,rom different industrial wastes, such as, tailings [2], spent cata-ysts [3], slags [4] and fly ash [5]. The chlorination processes are alsoresented as environmentally acceptable [6,7]. In general terms thehlorination can be described as reaction between a starting mate-ial (mineral concentrate or industrial waste) with chlorine in ordero produce some volatile chlorides, which can then be separated by,or example, selective condensation. The most desired chloride isurified and then used as a precursor in the production of either theure metal (reaction of the chloride with magnesium) or its oxideby oxidation of the chloride).

The chlorination reaction has been studied on respect of manyetal oxides [8–11] as this type of compound is the most common

n the mentioned starting materials. Although some basic thermo-ynamic data is enclosed in these works, most of them are relatedo kinetics aspects of the gas–solid reactions. However, it is clearhat the understanding of the equilibrium conditions, as predictedy classical thermodynamics, of a particular oxide reaction withhlorine, can give strong support for both the control and opti-ization of the process. In this context, the impact of industrial

perational variables over the chlorination efficiency, such as theeaction temperature and the reactors atmosphere composition,an be theoretically appreciated and then quantitatively predicted.n that sense, some important works have been totally devoted

o the thermodynamics of the chlorination and became classicaleferences on the subject [12–15].

As said before, the approach applied for the study of chem-cal equilibrium studies was based exclusively on �G◦ × T andredominance diagrams. Nowadays, due to the development ofomputational thermodynamics, a more detailed calculation isossible. This approach, together with the one accomplished byimpler techniques, converge to a better understanding of the inti-ate nature of the equilibrium states for the reaction system of

nterest.The present chapter focuses on the chemical equilibrium condi-

ions associated with the chlorination of V2O5 both in the presences in the absence of carbon (reducing agent). The impact of car-on over the thermodynamic viability of the reactions is clearlyresented and discussed. Also, the equilibrium conditions are

Fig. 1. Predominance diagram for the system V–O.

appreciated through graphical constructions of different complex-ity level, beginning with the well known �G◦ × T and predominancediagrams, and ending with gas phase speciation diagrams, rigorouscalculated through the minimization of the total Gibbs energy ofthe system.

2. The system V–O–Cl

Vanadium is a transition metal that can form a variety of oxides.At ambient temperature and oxygen potential, the form V2O5 isthe most stable. It is a solid stoichiometric oxide, where vanadiumoccupies the +5 oxidation state. By lowering the partial pressure ofO2, the valence of vanadium varies considerably, making it possi-ble to produce a family of stoichiometric oxides: V2O4, V3O5, V4O7,VO, VO2 and V2O3. Recently, it has been discovered that vanadiumcan also form a variety of non-stoichiometric oxygenated com-pounds [16], however, to simplify the treatment of the presentreview, these phases will not be included in the data-base used forthe following computations. Additionally, it was considered thatthe concentration of the oxides in gas phase is low enough to beneglected. Further, on what touches the computations that follows,the software Thermocalc was used in all cases, and it will alwaysbe assumed that equilibrium is achieved, or in other words, kineticeffects can be neglected.

The relative stability of the possible vanadium oxides can beaccessed through construction of a predominance diagram in thespace T–P(O2) (see Fig. 1). As thermodynamic constraints we haven(V) (number of moles of vanadium metal–it will be supposed thatn(V) = 1), T, P and P(O2). The reaction temperature will be varied inthe range between 1073 K and 1500 K and the partial pressure ofO2 in the range between 8.2 × 10−4 atm and 1 atm.

The total pressure was fixed at 1 atm. It can be seen that for thetemperature range considered and a partial pressure of O2 in theneighborhood of 1 atm, V2O5 is formed in the liquid state. Throughlowering the oxygen potential, crystalline vanadium oxides precip-itate, VO2 being formed first, followed by V2O3, VO, and finally V.The horizontal line between fields “5” and “6” indicates the melting
of VO2, which according to classical thermodynamics must occur ata fixed temperature. Next it will be considered the species formedby vanadium, chlorine and oxygen.

E.A. Brocchi et al. / Thermochimica Acta 559 (2013) 1– 16 3

Table 1Physical nature and selected data for vanadium chlorinated compounds.

Chlorinated species Physical state at 25 ◦C Equilibrium data Transition temperature (K)

VCl2 Solid Psat(T) [17]Tf = 1088 [18], 1620 [19]Ts = 1680 [17]Tb = 1803 [18]

VCl3 Solid PVCl4(T) [17]Td = 1120 [17], 914 [19]Ts = 1106 [18]

VCl4 Liquid – Tb = 398 [18], 424 [19]VOCl3 Liquid PVOCl3(T) [20] Tb = 694 [18], 400 [19], 399 [20]VO2Cl Solid PVOCl3(T) [20] Td = 450 [20]

VOCl2 Solid PVOCl3(T) [20]Td = 650 [20]Ts = 784 [21]

VOCl Solid – Ts = 1393 [21]

Psat ium VOCl3 pressure as a function of temperature; PVCl4(T), equilibrium VCl4 pressure asa temperature; Ts, sublimation temperature; Tb, ebullition temperature.

2

rVcecr

tsmVfatVio

t[tco[mb

ifVdcomsVHV

Table 2Equilibrium constant for the reaction represented by Eq. (3).

T (K) Equilibrium constant

1173 1.76257 × 10−13

(T), equilibrium vapor pressure as a function of temperature; PVOCl3(T), equilibr function of temperature; Td, dismutation/decomposition temperature; Tf , melting

.1. Vanadium oxides and chlorides

The already identified species formed between vanadium, chlo-ine and oxygen are: VCl, VCl2, VCl3, VCl4, VOCl, VOCl2, VOCl3,O2Cl. In Table 1 it was included information regarding the physi-al states at ambient conditions, some references related to phasequilibrium studies conducted on samples of specific vanadiumhlorinated compounds, as well as some available literature dataegarding specific phase transitions.

Only a few studies were published in literature in relation tohe thermodynamics of vanadium chlorinated phases. In Table 1ome references are given for earlier investigations associated witheasurements of the vapor pressure for the sublimation of VCl2 andCl3, and the boiling of VOCl3 and VCl4. There are also evidences

or the occurrence of specific thermal dismutation reactions, suchs those of VCl3 [17], VOCl2 and VO2Cl [20] (Eq. (1)). According tohe mentioned studies, at 1 atm VCl3 should decompose at 1120 K,O2Cl at 450 K and VOCl2 at 650 K. In the case of VCl3, however, it

s not clear if the two processes (sublimation and dismutation) canccur simultaneously [21].

VCl3(s) = VCl2(s) + VCl4(g)

3VO2Cl(s) = VOCl3(g) + V2O5(s)

2VOCl2(s) = VOCl3(g) + VOCl(s)

(1)

Chromatographic measurements conducted recently confirmedhe possible formation of VCl, VCl2, VCl3, and VCl4 in the gas phase22]. In this study the molar Gibbs energy models for the men-ioned chlorides were revised, and new functions proposed. In thease vanadium oxychlorides, models for the molar Gibbs energiesf gaseous VOCl, VOCl3, and VOCl2 have already been published21]. For gaseous VO2Cl, on the other hand, no thermodynamic

odel exists, indicating the low tendency of this oxychloride toe stabilized in the gaseous state.

It is worthwhile to mention that, for the calculations conductedn the present work the SSUB3 Thermocalc data-base [18] was usedor modeling the Gibbs energy of all vanadium chlorides, VCl2(s,l,g),Cl3(s,g) and VCl4(l,g). On what touches the chloride VCl(g) theata published by Hildenbrand et al. [22] was considered. In thease of the Gibbs energies of vanadium metal and the vanadiumxides V2O4(s), VO2(s,l), V2O3(s) and V2O5(s,l), the SSUB3 Ther-ocalc data-base was again employed [18], which also served as

ource for the Gibbs energy of VOCl3(l,g). In the case of the oxides4O7(s) and V3O5(s), as for VO2Cl(s), the models contained in theSC version 6.0 data-base were considered [23]. For VOCl(s,g) andOCl2(s,g), the data published by Hackert et al. [21] was employed.

1273 5.82991 × 10−11

1473 1.0397 × 10−08

2.1.1. V2O5 direct chlorination and the effect of the reducingagent

The direct chlorination of V2O5 is a process, which consists inthe reaction of a V2O5 sample with gaseous Cl2.

V2O5 + Cl2 = Chloride/Oxychloride + O2 (2)

For most pyrometallurgical processes, temperature lies usuallybetween 1173 and 1473 K. Considering the chloride VCl4, its forma-tion from V2O5 can be modeled according to Eq. (3). This reaction isassociated with very low thermodynamic driving force in the men-tioned temperature range, a fact that is demonstrated by the lowmagnitude of its equilibrium constant (Table 2).

V2O5 + 4Cl2 = 2VCl4 + 2.5O2 (3)

The very low magnitude of the equilibrium constant is an indica-tive, that at the standard conditions, the chemical equilibriumrepresented by Eq. (3) is almost entirely dislocated in the direc-tion of the decomposition of VCl4. The equilibrium could then beshifted in the direction of the formation of the mentioned chlorideif one removes O2 from the reactors atmosphere.

One possible strategy in this direction is to add to thereaction system some carbon bearing compound [24–26]. Thecompound decomposes producing graphite, which reacts with oxy-gen, thereby shifting the chlorination equilibrium in the desireddirection. A simpler route, however, would be to admit carbon asgraphite together with the oxide sample into the reactor. If graphiteis present in excess, the O2 concentration in the reactor’s atmo-sphere is maintained at very low values, which are achievablethrough the formation of carbon oxides (Eq. (4)).

2C + O2 = 2CO

C + O2 = CO2

(4)

So, for the production of VCl4 in the presence of graphite, thereaction of C with O2 can lead to the evolution of gaseous CO or
CO2 (Eq. (5)).
V2O5(s, l) + 4Cl2(g) + 2.5C(s) = 2VCl4(l, g) + 2.5CO2(g)

V2O5(sl) + 4Cl2(g) + 5C(s) = 2VCl4(l, g) + 5CO(g)(5)

4 E.A. Brocchi et al. / Thermochim

ttttmpsmahd(b

t

Fig. 2. �Gr◦ vs. T for the formation of VCl4 in the presence of graphite.

The effect of the presence of graphite over the �G◦ × T curves forhe formation of VCl4 can be seen in the diagram of Fig. 2. As a mat-er of comparison, the plot for the formation of the same species inhe absence of graphite is also shown, together with the curves forhe reactions associated with the formation of CO and CO2 for one

ole of O2 (Eq. (4)). In the temperature range considered, VCl4 isroduced in the gaseous state. It can be readily seen that graphitetrongly reduces the standard molar Gibbs energy of reaction, pro-oting in this way considerably the thermodynamic driving force

ssociated with the chlorination process. The presence of graphiteas also an impact over the standard molar reaction enthalpy. Theirect action of Cl2 is associated with an endothermic reactionpositive linear coefficient), but by adding graphite the processes
ecome considerably exothermic (negative linear coefficient).
The curves associated with the VCl4 formation in the presence ofhe reducing agent cross each other at 973 K, the same temperature

Fig. 3. �Gr◦ vs. T the formation of VCl4-melting of V2O5.

ica Acta 559 (2013) 1– 16

where the curves corresponding to the formation of CO and CO2have the same Gibbs energy value. This point is defined by thetemperature, where the Gibbs energy of the Boudouard reaction(C + CO2 = 2CO) is equal to zero. For each reaction, continuous curvesdenote the formation of the most stable species, which are associ-ated with the lowest value of the reaction Gibbs energy.

The equivalence of this point and the intersection associatedwith the curves for the formation of VCl4 can be understood, as theBoudouard reaction can be obtained through a simple linear com-bination, according to Eq. (6). So, its molar Gibbs energy is equalto the difference between the molar Gibbs energy of the VCl4 for-mation with the evolution of CO and the same quantity for thereaction associated with the CO2 production. Concomitantly, whenthe curves for the formation of VCl4 crosses each other, the differ-ence between their molar Gibbs energies is zero, and according toEq. (6) the same must happen with the molar Gibbs energy of theBoudouard reaction.

1) V2O5(s, l) + 4Cl2(g) + 5C(s)�G1−→2VCl4(g) + 5CO(g)

−2) V2O5(s, l) + 4Cl2(g) + 2.5C(s)

�G2−→2VCl4(g) + 2.5CO2(g)

=3) 2.5C(s) + 2.5CO2(g)

�G3−→5CO(g)

�G3 = �G1 − �G2

LimT→973 K

(�G3) = LimT→973 K

(�G1 − �G2) = 0

(6)

The inflexion point present on the curves of Fig. 2 is associatedwith the melting of V2O5. This inflexion is better evidenced on thegraphic of Fig. 3. As V2O5 is a reactant, the curve should experience areduction of its inclination at the melting temperature of the oxide.However, the presence of the mentioned inflexion point is muchmore evident for the reactions with the lowest variation of numberof moles of gaseous reactants, as is the case for the direct action ofCl2, which leads to the evolution of CO2 the variation of numberof moles of gaseous species (�ng) is equal to 0.5, smaller than thevalue obtained for the same reaction leading to CO (�ng = 3). Thevalue of �ng controls the molar entropy of the reaction. By loweringthe magnitude of �ng the value of the reaction entropy reduces,and the effect of melting of V2O5 over the standard molar reactionGibbs energy becomes more evident.

The presence of graphite affects in the same way the thermody-namic tendency of formation of vanadium oxychlorides. This canbe well illustrated for the synthesis of VOCl3. Its �G◦ × T curve iscompared with the one for the formation of VCl4 on the graphic ofFig. 4. As in the case of VCl4, VOCl3 is formed as a gas in the tem-perature range considered. Moreover, the inflexion around 954 Kis again associated with the melting of V2O5. Further, as the reac-tion associated with the formation of VCl4, the formation of VOCl3has a negative molar reaction enthalpy. So, if during the productionof both chlorinated compounds the gas phase is considered ideal,the system should transfer heat to its neighborhood (exothermicreaction).

On what touches the molar reaction entropy, the graphic of Fig. 4indicates, that the reaction associated with the formation of VCl4should generate more entropy (more negative angular coefficientfor the entire temperature range). This can be explained by the fact,that in the case of VCl4, the value of �ng (�ng = 3) is higher than the
same value for the formation of VOCl3 (�ng = 2). The same tenden-cies are expected for the formation reactions associated with CO2evolution, as these result from a simple linear combination betweenthe reaction equations considered in Fig. 4 and the Boudouard reac-tion.


F ◦

2assbrCpo

ig. 4. �Gr vs. T for the formation of VOCl3 and VCl4 in the presence of graphite.

.1.1.1. Multiple reaction equilibria. The �G◦ × T diagram is a valu-ble tool for suggesting possible reactions paths in the case ofystems, where multiple reactions occur simultaneously. Let’s con-ider first the formation of VCl4. In a first glance, such process coulde thought as the result of three stages. In the first one, a lower chlo-inated compound (VCl) is formed. The precursor then reacts withl2 resulting in higher chlorinated species (Eq. (7)). The �G◦ × Tlots associated with reactions paths represented by mechanismsf Eq. (7) were included in Fig. 5.

V2O5 + Cl2 + 2.5C = 2VCl + 2.5CO2

V2O5 + Cl2 + 5C = 2VCl + 5CO

VCl + 0.5Cl2 = VCl2 (7)

VCl2 + 0.5Cl2 = VCl3


Fig. 5. �Gr◦ × T for reaction paths of Eq. (7).


As said before, in each temperature range, reactions leadingto stable species are represented by continuous curves. This con-vention will be adopted in all other diagrams included in therest of the present article. On the diagram of Fig. 5, two inflex-ion points are evidenced. The first one around 1000 K is associatedwith VCl2 melting. The second more evident one, around 1100 K,is associated with the sublimation of VCl3. It can be speculatedthat only for temperatures greater than 1600 K the path describedby Eq. (7) could be realized. For lower temperatures, the molarGibbs energy of the first step is higher than the one associatedwith the second. In this situation, therefore, another mechanismmust be formulated for the VCl4 appearance. One possibility isdefined by Eq. (8). This time, VCl2 is formed directly from V2O5,
which then reacts to give VCl3 and finally VCl4. The characteristic

6 E.A. Brocchi et al. / Thermochimica Acta 559 (2013) 1– 16

�F

dVFatftiwwhttrE

0

ttsed

ptaptip

◦


G◦ × T curves for the reactions defined in Eq. (8) are presented inigs. 6 and 7.

V2O5 + 2Cl2 + 2.5C = 2VCl2 + 2.5CO2

V2O5 + 2Cl2 + 5C = 2VCl2 + 5CO



(8)

The inflexion points have the same meaning as described foriagram of Fig. 5. It can be seen that the first step (formation ofCl2) has a much higher thermodynamic tendency as the others.or temperatures lower than 953 K or higher than 1539 K the mech-nism represented by Eq. (8) describes the order of appearance ofhe chlorides as P(Cl2) gets higher. In these conditions, VCl2 is firstormed as a solid for temperatures lower than 953 K and liquid foremperatures higher than 1539 K. In the mentioned temperaturentervals, the second reaction step leads to the formation of VCl3,

hich can be produced as a solid (T < 953 K) or gas (T > 1539 K),hich then finally reacts resulting in gaseous VCl4. On the otherand, for temperatures higher than 953 K and lower than 1539 K,he step associated with the formation of VCl4 is now the one withhe second lowest standard Gibbs energy. So, in this temperatureange, VCl4 should be formed directly from VCl2, as suggested byq. (9).

.5VCl2 + 0.5Cl2 = 0.5VCl4 (9)

In order to achieve thermodynamic consistency in the tempera-ure interval between 953 K and 1539 K, the curves associated withhe formation of VCl3 and VCl4 according to Eq. (8) should be sub-tituted for the curve associated with reaction defined by Eq. (9), asvidenced on the plot of Fig. 7. The inflexion point evident in thisiagram around 1088 K is associated with VCl2 melting.

On what touches the synthesis of VOCl3, a reaction path can beroposed (Eq. (10)), in that VOCl is formed first, which then reactso give VOCl2 which transforms to VOCl3. The �G◦ × T diagramsssociated with these reactions are presented in Fig. 8. The inflexion
oint at 784 K is related to the sublimation of VOCl2, and at 1393 K tohe sublimation of VOCl. According to the �G◦ × T curves presentedn Fig. 8, it can be deduced that the reaction steps will follow theroposed order only for temperatures higher than 1053 K, gaseous

VOCl being first produced, which then reacts leading to gaseousVOCl2, which finally results in gaseous VOCl3.

V2O5 + Cl2 + 1.5C = 2VOCl + 1.5CO2

V2O5 + Cl2 + 3C = 2VOCl + 3CO

VOCl + 0.5Cl2 = VOCl2

VOCl2 + 0.5Cl2 = VOCl3

(10)

For temperatures lower than 1053 K, VOCl3 should indeed beformed directly from VOCl (Eq. (11)). Again, to attain thermody-namic consistency for temperatures lower than 1053 K, the Gibbsenergy curves associated with the formation of VOCl2 and VOCl3according to Eq. (10) must be substituted for the curve associ-ated with reaction represented by Eq. (11). It should be mentionedindeed, that the reaction equations compared must be written withthe same stoichiometric coefficient for Cl2, which was set equal to“1/2” (Fig. 9).

0.5VOCl + 0.5Cl2 = 0.5VOCl3 (11)

2.1.1.2. Predominance diagrams and �G◦ × T diagrams. Althoughreaction mechanisms lie outside the field of chemical thermo-dynamics, if a reliable data-base is available, the stability of theindividual species taking part in a specific mechanism can be judgedbased on thermodynamic equilibrium computations. In the case ofthe V–O–Cl system, for example, this means that the sequence ofappearance of chlorides and or oxychlorides as predicted by thepredominance diagram at a specific temperature must be consis-tent as the one deduced by the �G◦ × T plots analysis for eachreaction taking part in the proposed mechanism. If the formationof a compound is supported by the �G◦ × T approach and the samespecies is not present in the calculated predominance diagram,it should be classified as metastable for a short period of time,the species can be present, but as equilibrium is approached, itdecomposes resulting in the more stable species according to thepredominance diagram. Therefore, if a compound is present in thepredominance diagram, its formation must be supported by the
thermodynamically consistent �G × T plots.
Considering the formation of VCl4, the reaction path of Eq. (8) isvalid only for temperatures lower than 953 K or higher than 1539 K.In the lower temperature range this mechanism suggests that once


Fig. 10. Predominance diagram for the system V–O–Cl at 900 K.

VVg

trfd9gfdabi


included on topic (Section 2.1.2.2), where besides VCl3 and VCl4,


Cl2 is formed in the solid state, it should be converted to solidCl3 with increasing P(Cl2). Next, VCl3 reacts further resulting inaseous VCl4.

For temperatures higher than 1539 K, VCl2 must be formed inhe liquid state, being next converted to gaseous VCl3, which finallyesults in gaseous VCl4. These sequences of vanadium chloridesormation are exactly the same as observed in the predominanceiagrams calculated at 900 K (Fig. 10) and 1673 K (Fig. 11). Between53 K and 1539 K the mechanism embedded in Eqs. (8) and (9) sug-ests that VCl2 reacts directly resulting in gaseous VCl4. No VCl3 isormed in this case. This is in fact the chlorination sequence pre-icted by the predominance diagrams calculated at 1073 K (Fig. 12)
nd 1273 K (Fig. 13). In these diagrams, no VCl3 field is present, andy starting at the liquid VCl2 field, the stability area of gaseous VCl4
s finally reached as P(Cl2) achieves higher values.


A closer look on the predominance diagrams of Figs. 10–13indicates that, with the exception of gaseous VOCl3, no otheroxychloride appears (VOCl, VOCl2, or VO2Cl), suggesting that thementioned species is the only one stable vanadium oxychloridefor temperatures between 900 K and 1673 K. Therefore, consider-ing the present mechanism for the formation of the oxychlorideVOCl3 in the mentioned temperature range (Eqs. (10) and (11)),the other possible oxychlorides (VOCl and VOCl2) could be consid-ered as metastable species. Such result is not a mere consequenceof the fact that during the computations developed in the presenttopic the oxychlorides are treated as pure compounds, but are alsocontained in the results achieved through speciation calculations

VOCl3 is the only one oxycloride found in the gas phase. Anotherpossible explanation is the fact that the oxychlorides formation maynot happen by the progressive addition of chlorine atoms, but it

8 ochimica Acta 559 (2013) 1– 16

wpw(t

tcoeoiftv(

2opiotfa

CttdtrPtEdicvatrcrbeTdbVtite

meVg

Fig. 14. Partial pressure of VOCl3 as a function of P(O2) at 900 K.


ould be accomplished through reactions, where one of the alreadyresent vanadium chlorides (VCl4, VCl3 or VCl2) participate. It isorthwhile to mention that this alternative was addressed on topic

2.1.2.3), where it was considered that VOCl3 could be producedhrough the direct oxidation of one of the chlorides.

The results included on the present topic clearly illustrate howhe intimate relation between �G◦ × T and predominance diagramsan help in the reaction mechanisms discussions. The reliabilityf the computations is directly linked to the quality of the Gibbsnergy model used for describing the thermodynamic propertiesf each considered phase. If a high quality data-base is employed,t is expected that a consistent equilibrium approach would beound in both diagrams (�G◦ × T and predominance diagrams). Inhe present work this situation is exemplified for the formation ofanadium chlorides according to mechanism represented by Eqs.8) and (9) and Figs. 6, 7 and 10–13.

.1.1.3. VOCl3 formation. According to the information containedn the predominance diagrams of Figs. 10–13, depending on theartial pressure of O2 imposed, the formation of VOCl3 can be real-

zed either from the direct reaction of Cl2 with a vanadium oxider the oxidation of some vanadium chloride (VCl4, VCl3 or VCl2). Inhe first case the VOCl3 phase field is achieved as P(Cl2) gets higheror a fixed P(O2) value, and in the last case the same field is reacheds P(O2) achieves higher values – P(Cl2) fixed.

0.5V2O5 + 1.5Cl2 = VOCl3 + 0.75O2

VO2 + 1.5Cl2 + VOCl3 + 0.5O2

1/3V3O5 + 1.5Cl2 = VOCl3 + 1/3O2

1/2V2O3 + 1.5Cl2 = VOCl3 + 0.5O2

(12)

Let us take a look first on the reaction of vanadium oxides andl2. At each temperature, depending on the P(O2) value imposed,he valence of vanadium changes, and VOCl3 can be producedhrough one of the reactions represented by Eq. (12). The thermo-ynamic viability of such transformations can be studied based onhe computation of the partial pressure of VOCl3 found in equilib-ium at a specific temperature under variable P(O2). The calculated(VOCl3) values can be easily obtained through solving individuallyhe equilibrium equations associated with reactions represented byq. (12), considering in all cases P(Cl2) equal to 1 atm and the vana-ium oxides as pure compounds (unity activity). For each P(O2)

nterval, the curve associated with the most stable vanadium oxideorrespond to the lowest partial pressure of VOCl3. The mentionedanadium oxide must have the lowest molar Gibbs energy amongll possible valence states considered. As the oxide is a reactant,he reduction of its Gibbs energy results in a more positive molareaction Gibbs energy, which in the end leads to a lower chemi-al equilibrium constant. Concomitantly, the calculated P(VOCl3)educes for the same P(O2) value imposed. At 900 K, VOCl3 cane produced from one of the vanadium oxides only for ln P(O2)qual or higher than −50.22. At 1273 K, this value becomes −32.42.hese quantities limit the region of physical validity of the curvesrawn of in Figs. 14 and 15. For lower P(O2) values, VOCl3 can onlye produced through oxidation of some vanadium chloride (VCl2,Cl3 or VCl4), a fact that will be considered in more detail latter in

his section. At both temperatures studied, for ln P(O2) < −5, signif-cant partial pressures of VOCl3 are computed, which suggests thathe formation of the mentioned oxychloride is associated with anxpressive thermodynamic driving force.

The present method can also be used for comparing the ther-
odynamic stability of two chlorinated products. This fact was
xplored on topic (Section 2.1.2.1), where the stability of VOCl3 andCl4 produced through the chlorination of V2O5 in the presence ofraphite is discussed. There, the interpretation of the data is the

Fig. 15. Partial pressure of VOCl3 as a function of P(O2) at 1273 K.

opposite from the one used for the discussion of the informationcontained in Figs. 14 and 15. The most stable vanadium chlori-nated compound is associated with the highest vapor pressurevalue computed at each condition, as in this case we are speak-ing from product molecules. If the stability of one of the products ishigher than the other, its Gibbs energy must be lower, reducing themolar Gibbs energy of the reaction in that it takes part. The equilib-rium constant grows, and so the partial pressure of the chlorinatedproduct. It is interesting to see that the inflexion points contained inthe continuous curves (PVOCl3 stability locus) represented on thediagrams calculated at 900 K (Fig. 14) and 1273 K (Fig. 15) corre-spond exactly to the P(O2) values, where the equilibrium between
two vanadium oxides is established. Such fact can be readily under-stood, as the linear combination of the chlorination reactions oftwo oxides associated with neighboring stability fields, results inthe chemical equation representing the equilibrium between the


Table 3P(O2) equilibrium value under the presence of an excess of graphite.

T (K) Ln P(O2)

900 −53.041273 −42.381673 −37.17

tc

mgnUcfp(

oiVcvgeppvd

Fig. 17. Partial pressure of VOCl3 as a function of P(Cl2) at 1273 K.


wo mentioned vanadium oxides. This can be illustrated for thehlorination of V2O5 and VO2 (Eq. (13)).

(0.5V2O5 + 1.5Cl2 = VOCl3 + 0.75O2)−(VO2 + 1.5Cl2 = VOCl3 + 0.5O2) =0.5V2O5 = VO2 + 0.25O2

(13)

Now let us consider the possible formation of VOCl3 underuch more reductive conditions, as in the case where an excess of

raphite is added to the reaction system under study. If this compo-ent is present in excess, P(O2) should be limited to very low values.nder these conditions, according to the predominance diagramsonstructed on topic (Section 2.1.1.2) (Figs. 10–13), the mechanismor VOCl3 formation could be defined by two steps. First a chloride isroduced (VCl2, VCl3 or VCl4), which is then next oxidized to VOCl3Eq. (14)).

VCl2 + 0.5Cl2 + 0.5O2 = VOCl3

VCl3 + 0.5O2 = VOCl3

VCl4 + 0.5O2 = VOCl3 + 0.5Cl2

(14)

The thermodynamic tendency associated with the occurrencef the reactions represented by Eq. (14) can be studied by employ-ng an analogous strategy as used for discussing the formation ofOCl3 under variable P(O2). This time, the partial of any gaseoushloride (VCl3 or VCl4), if present, is fixed at 1 atm, and the P(O2)alue imposed is calculated according to equilibrium between aaseous phase containing CO, CO2 and O2 in the presence of anxcess of graphite (Boudouard equilibrium) at each imposed tem-
erature (Table 3). In all cases (Figs. 16–18), significant partialressures of VOCl3 are computed in comparison with the low P(O2)alue imposed, indicating that the proposed reactions are thermo-ynamically possible. It is also interesting to note that the idea of

the occurrence of oxidation reactions as those represented by Eq.(14) reappears in the discussion of topic (Section 2.1.2.2), as a plau-sible explanation of the gas phase composition variations as P(O2)is systematically varied.

The inflexion points present in the curves of Figs. 16–18 areassociated with P(Cl2) values, which are identical to the partialpressures related to the equilibrium between two successive chlo-rinated compounds, as described by the predominance diagrams.Again, the stability locus is determined by the minimum calculatedpartial pressures. The diagrams are lower limited by the horizon-tal lines, which determine the minimum vapor pressure of Cl2 forthe occurrence of the oxidation reactions considered (1673 K –ln P(Cl ) = −7.2, 1273 K – ln P(Cl ) = −10.4, 900 K – ln P(Cl ) = −15.2).
2 2 2For lower partial pressures the oxidation of the most stable chlorideresults in a vanadium oxide. Therefore, depending on the par-tial pressure of oxygen imposed, the formation of VOCl3 can be


a(bO

2

2tiTfiisfios

2VnttAsr

a

wf(Vtvnm

Fig. 19. Ln (P(VCl4)) and Ln (P(VOCl3)) as a function of Ln (P(CO)).

ccomplished either through the oxidation of a vanadium chloridelower O2 activity – Figs. 16–18) or the chlorination of the most sta-le vanadium oxide under the specified oxygen potential (higher2 activity – Figs. 14 and 15).

.1.2. Relative stability of VCl4 and VOCl3As is evident from the discussion developed on topic (Section

.1.1), the chlorinated compounds VCl4 and VOCl3 are, respectively,he most stable vanadium chloride and oxychloride to be foundn the gas phase as the atmosphere becomes concentrated in Cl2.he relative stability of these two chlorinated compounds will berst accessed on topic (Section 2.1.2.1) by applying the method

ntroduced by Kang and Zuo [27], which is indeed a very similartrategy as the one used for the discussion of possible reaction pathsor the formation of VOCl3 on topics (Sections 2.1.1.2 and 2.1.1.3), ast relies entirely on the computation of �G◦ values, and secondly,n topic (Section 2.1.2.2), through the construction of gas phasepeciation diagrams.

.1.2.1. Method of Kang and Zuo. The concentrations of VCl4 andOCl3 can be directly computed by considering that each chlori-ated compound is generated independently. It will be assumedhat the inlet gas is composed of pure Cl2 (P(Cl2) = 1 atm). Fur-her, two temperature values were investigated, 1073 K and 1373 K.t these temperatures, the presence of graphite makes the atmo-phere richer in CO, so that for the computations the followingeactions will be considered:

V2O5 + 4Cl2 + 5C = 2VCl4 + 5CO

V2O5 + 3Cl2 + 5C = 2VOCl3 + 3CO(15)

The concentrations of VOCl3 and VCl4 can then be expressed as function of P(CO) and temperature according to Eq. (16).

PVCl4 =√

P5COK1 → ln PVCl4 = ln K1

2+ 5

2ln PCO

PVOCl3 =√

P3COK2 → ln PVOCl3 = ln K2

2+ 3

2ln PCO

(16)

here K1 and K2 represent, respectively, the equilibrium constantsor the reactions associated with the formation of VCl4 and VOCl3Eq. (15)). By applying Eq. (16) the partial pressure of VCl4 andOCl3 were computed as a function of P(CO). The results were plot-

ed in Fig. 19. The significant magnitude of the partial pressurealues computed for VOCl3 and VCl4 is a consequence of the hugeegative standard Gibbs energy of reaction associated with the for-ation of these species in the temperature range considered (see

ica Acta 559 (2013) 1– 16

Fig. 4). Moreover, according to the data presented in Fig. 19, VCl4 isthe compound with the highest partial pressure for both specifiedtemperatures while, as temperature gets higher, their concentra-tions show appreciable reduction. These results will be confirmedthrough construction of speciation diagrams for the gas phase, atask that will be accomplished on topic (Section 2.1.2.2).

2.1.2.2. Gas phase speciation. The construction of speciation dia-grams for the gas phase enables the elaboration of a complementarypicture of the chlorination process in question. The word “speci-ation” means the concentration of all species in gas. This bringsanother level of complexity to the quantitative description ofequilibrium, as the species build a solution, and as so, their con-centrations must be determined at the same time. This sort ofinformation can only arises if one solves the system o equilibriumequations associated to all possible chemical reactions involvingthe species that form the gas. For the present system (V–O–Cl–C)this task becomes very tricky, as the number of possible speciespresent is pretty significant (e.g. CO, CO2, O2, VCl2, VCl3, VCl,VCl4, VOCl3, VOCl, VO2Cl and VOCl2), and so the number of pos-sible chemical reactions connecting them. So, we must think inanother route for simultaneously computing the concentration ofthe gaseous species produced by our chlorination process. The onlypossible way left consists in minimizing the total Gibbs energy ofthe system.

The equilibrium state is defined by fixing T, P, n(V2O5), P(O2) andP(Cl2). The number of moles of V2O5 is fixed at one. If graphite ispresent in excess, the partial pressure of O2 is controlled by accord-ing to the Boudouard equilibrium, so that, its presence forces P(O2)to attain very low values (typically lower than 10−2 atm). The totalpressure is fixed at 1 atm and T varies in the range between 1000 Kand 1473 K. An excess of graphite is desirable, so that the chlo-rination reactions can achieve a considerable driving force at thedesired conditions. Computationally speaking, this can be done intwo ways. One possibility is to define an amount of carbon muchgreater than the number of moles of V2O5. Other possibility, whichhas been made accessible through modern computational thermo-dynamic software, consists in defining the phase “solid graphite”as fixed with a definite amount. As carbon in graphite is consideredto be pure, the later alternative is equivalent of saying that carbonis present in the system with a chemical activity equal to one. Theequilibrium compositions (intensive variables) are not a function ofthe amount of phases present (size of the system), depending onlyof temperature and total pressure. So we are free to choose anysuitable value we desire for the fixed amount of carbon present,such for example zero. This last alternative was implemented inthe computations conducted in the present topic.

In Fig. 20, the number of moles of gas produced was plottedas a function of P(Cl2) for T equal to 1073 K, 1273 K, and 1473 K,1000 K, 1073 K, 1173 K, and 1373 K. The partial pressure of O2 wasfixed at 1.93 × 10−22 atm, and the partial P(Cl2) is varied between3.6 × 10−7 atm and 0.61 atm. Each curve represented in Fig. 20 isdefined by three stages. First, for very low values of P(Cl2), no gas isformed. At these conditions VCl2(l) is present in equilibrium with.The equilibrium ensemble does not experience any modificationuntil a critical P(Cl2) value is reached, at which a discontinuity canbe evidenced. The gas phase appears and for any P(Cl2) higher thanthe critical one, the number of moles VCl2(l) becomes equal to zero.This condition defines the second stage, where for higher P(Cl2) val-ues the gas composition changes accordingly, through forming ofchlorides and oxychlorides. Finally, a P(Cl2) value is reached, where
all capacity of the system for forming chlorinated compounds isexhausted, and the effect of adding more Cl2 is only the dilution ofthe chlorinated species formed. As a consequence, the number ofmole of gas phase experiences a significant elevation.

E.A. Brocchi et al. / Thermochim

Fig. 20. Number of moles of gas as a function of P (Cl2).

Table 4Equilibrium constants at 1073 K for the reactions represented by Eq. (17).

Chemical reaction Equilibrium constant

tsactaTaf VCl3 VCl3 VCl2

VCl2 + Cl2 → VCl4 1.95 × 105

VCl3 + 0.5Cl2 → VCl4 2.12 × 103

At 1073 K, for example, Fig. 21 describes the effect of P(Cl2) overhe gas phase composition during the second and third stages. Weee that the mol fraction of VCl4 raises (second stage) and afterchieving a maximum value starts to decrease (third stage). Theoncentration variations during the second sage can be ascribedo the occurrence of reactions represented by Eq. (17), which havet 1073 K equilibrium constants much higher than unity (Table 4).he reduction of the mol fraction of VOCl3 can be understood as

dilution effect, which is motivated by the elevation of the molractions of VCl4 and Cl2.

VCl2 + Cl2 = VCl4

VCl3 + 0.5Cl2 = VCl4(17)

Fig. 21. Concentration of vanadium chlorides and o

ica Acta 559 (2013) 1– 16 11

Besides P(Cl2), temperature should also have an effect over thecomposition of the gas phase. This was studied as follows. Six tem-perature values were chosen in the range between 1073 K and1473 K. Next, for each temperature the critical P(Cl2) value (the oneassociated with the formation of the “first” gas molecules) is iden-tified. The concentrations of the most stable gaseous species at thecomputed critical P(Cl2) value were then calculated and presentedin Table 5. During the calculations the partial pressure of O2 wasfixed at 1.93 × 10−22 atm.

As expected, the mol fraction of CO is greater than the mol frac-tion of CO2 for the entire temperature range studied. Also, thechloride VCl4 has the highest concentration at 1073 K, a phasewhich occupies a large area of the predominance diagram at thistemperature (Fig. 12). As temperature attains higher values, themol fraction of VCl4 and VOCl3 become progressive lower and theatmosphere more concentrated in VCl2 and VCl3. So, at 1473 Kthe situation is significant different from the equilibrium stateobserved at 1073 K. This behavior is again consistent with the infor-mation contained on the predominance diagrams (Figs. 10–13)where can be seen that the stability fields of VCl4(g) and VOCl3(g)shrink while the area representing the phase VCl3(g) grows, occu-pying at 1673 K a visible amount of the diagrams space (Fig. 11).

It is worthwhile to mention that a more detailed look on theresults presented in Table 5 seems to incorporate apparent incon-sistencies. (i) The minimum partial pressure of Cl2 for the formationof pure VCl4(g) at 1073 K (Fig. 12) is higher than the critical pressurefor the formation of the first gaseous species at this temperature(Fig. 20). (ii) Measurable amounts of VCl3 (greater or equal to 0.1)were detected for temperatures higher than 1100 K (Table 5) but noVCl3(g) field was observed in the predominance diagram computedat 1273 K (Fig. 13). (iii) Also, no field associated with the formationof VCl2(g) could be detected even at 1673 K (Fig. 11) but the speci-ation computation predicts its presence in measurable amounts atthe last temperature (x(VCl2) = 0.14) (Table 5). All these thermody-namic values differences are a consequence of the fact that the puremolar Gibbs energy of each component is higher than its chemicalpotential in the ideal gas solution, the former model being used forthe predominance diagrams construction while the later is appliedto the speciation calculations. Therefore, the driving force for theformation of the gaseous compounds is reduced accordingly to Eq.(18) [28].

�g − gg = RT ln xg < 0 (18)

Another possible type of computation is to study the effectof P(O2) over the composition of the gas phase. This variable isrestricted by the fact that the amount of graphite is fixed. So there

xychlorides as a function of P(Cl2) at 1073 K.

12 E.A. Brocchi et al. / Thermochimica Acta 559 (2013) 1– 16

Table 5Composition of the “first” gas formed as a function of temperature.

T (K) X(CO) X(CO2) X(VOCl3) X(VCl2) X(VCl3) X(VCl4)

1073 0.16 3.64 × 10−3 1.95 × 10−2 1.74 × 10−3 9.27 × 10−2 0.721100 0.12 1.23 × 10−3 1.02 × 10−2 2.61 × 10−3 0.12 0.751200 4.21 × 10−2 3.36 × 10−5 1.08 × 10−3 9.87 × 10−3 0.26 0.691300 1.76 × 10−2 1.60 × 10−6 1.45 × 10−4 2.98 × 10−2 0.43 0.521373 1.00 × 10−2 2.29 × 10−7 3.69 × 10−5 5.97 × 10−2 0.56 0.371400 8.26 × 10−3 1.17 × 10−7 2.27 × 10−5 7.6 × 10−2 0.59 0.321473 5.07 × 10−3 2.18 × 10−8 6.22 × 10−6 0.14 0.66 0.19

Table 6Gas phase speciation as a function of P(O2) at 1373 K.

P(O2) (atm) X(CO) X(CO2) X(VOCl3) X(VCl2) X(VCl3) X(VCl4)

−24 −4 −9 3.05 × 10−6 0.0597 0.56 0.386.03 × 10−5 0.0588 0.55 0.373.21 × 10−4 5.72 × 10−3 0.054 0.036

iteitw2bsr

orcacsbCt

eottttgoP(

ohVVo

ttV

Fig. 22. Mol fraction of VCl3 and VCl4 as a function of P (O2).

1.30 × 10 8.22 × 10 1.54 × 105.24 × 10−22 1.65 × 10−2 6.22 × 10−7

1.56 × 10−18 0.902 1.85 × 10−3

s a maximum value of P(O2) at each temperature for which thehermodynamic modeling remains consistent with the Boudouardquilibrium (Eq. (4)) and the computation can be performed. By fix-ng the temperature at 1373 K, the upper limit for P(O2) has showno be equal to 1.56 × 10−18 atm and the value of P(Cl2) associatedith the appearance of the first gaseous molecules is identified as

.05 × 10−4 atm. The composition of the gas phase is then computedy fixing P(Cl2) at 2.05 × 10−4 atm. Three different P(O2) levels weretudied, 1.3 × 10−24 atm, 5.24 × 10−22 atm, 1.56 × 10−18 atm. Theesults are presented in Table 6.

The mol fractions of CO and CO2 gets higher for higher valuesf P(O2). This is consistent with the dislocation of the equilibriumepresented by Eq. (4) in the direction of the formation of the twoarbon oxides. Also, the Boudouard’s equilibrium demands thatt the chosen temperature (1373 K) the atmosphere is more con-entrated in CO. This was indeed observed for each equilibriumtate investigated. It is interesting to observe that for P(O2) varyingetween 5.24 × 10−22 atm and 1.56 × 10−18 atm the mol fraction ofO and CO2 experience a much higher variation in comparison withhe one observed for lower P(O2) values.

In the case of the vanadium chlorides and oxychlorides an inter-sting trend is evidenced. The concentration of VOCl3 grows andf VCl2, VCl3 and VCl4 reduce appreciably for the same O2 par-ial pressure range. The concentration variations associated withhe vanadium chlorinated compounds is analogous to the varia-ions observed in the concentrations of CO and CO2. For P(O2) lowerhan 5.24 × 10−22 atm the variations are much less significant. Toet a better picture of the trend observed for the chlorides andxychlorides, their concentrations were plotted as a function of(O2), which was varied in the range spanned by the data of Table 4Figs. 22–24).

The variations depicted in Figs. 22–24 are consistent with theccurrence of reactions represented by Eq. (19). As P(O2) achievesigher values, it reacts with VCl3, VCl2 and or VCl4 resulting inOCl3. Such phenomena could explain the significant reduction ofCl3, VCl4 and VCl2 concentrations, and the concomitant elevationf the VOCl3 mol fraction.

VCl3 + 0.5O2 = VOCl3

VCl2 + O2 + VCl4 = 2VOCl3

VCl4 + 0.5O2 = VOCl3 + 0.5Cl2

(19)

The participation of VCl4 in the second reaction is supported byhe fact that its equilibrium concentration lowering is more sensibleo ln P(O2) than observed for VCl3 (Fig. 22). The consumption ofCl4 by the second reaction is also consistent with the maximum

Fig. 23. Mol fraction of VOCl3 and VCl2 as a function of P (O2).


Fig. 24. Mol fraction of VOCl3 as a function of P (O2).

Table 7Equilibrium constants at 1373 K for reactions represented by Eq. (19).

Chemical reaction Equilibrium constant

VCl + 0.5O = VOCl 8.01 × 106

oA

bpTmfc

2ikmfscali

r(aacrHtrf

TV

Fig. 25. Total enthalpy as a function of Cl2 partial pressure.

3 2 3

VCl2 + O2 + VCl4 = 2VOCl3 1.89 × 1013

VCl4 + 0.5O2 = VOCl3 + 0.5Cl2 1.01 × 105

bserved in the curve obtained for VOCl3 concentration (Fig. 24).s less VCl4 is available, less VOCl3 can be produced.

The occurrence of reactions represented by Eq. (19) is supportedy classical thermodynamics, as the equilibrium constant com-uted at 1373 K assume values appreciably greater than unity (seeable 7). It is indeed interesting to recognize that the same transfor-ations were considered as plausible reaction paths for the VOCl3

ormation based on the oxidation of one of the possible vanadiumhlorides (see Section 2.1.1.3).

.1.2.3. V2O5 chlorination enthalpy. For the implementation of anndustry process based on chemical reaction is fundamental tonow the amount of heat generated or absorbed from that. Exother-ic processes (heat is released) reach higher temperatures, and

requently demand engineering solutions for protecting the oventructure against the tremendous heat generated by the chemi-al phenomena. In this context, endothermic processes (heat isbsorbed) are easier controlled, but the energy necessary to stimu-ate the reaction must be continuously supplied, making the energynvestment larger.

The variation of the total enthalpy of the system for the chlo-ination process in question was calculated as a function of P(Cl2)Fig. 25). The partial pressure of O2 was fixed at 1.93 × 10−22 atmnd four temperature levels were studied, 1000 K, 1100 K, 1300 Knd 1700 K. It can be seen that the total enthalpy for the processonducted at 1000 K reduces with the advent of the chlorinationeactions, indicating that the chlorination process is exothermic.owever, the molar enthalpy magnitude is progressively lower up

o a certain temperature where it is zero. Above that, the molareaction enthalpy becomes positive, and Fig. 25 illustrates its value
or 1700 K.
All these facts are in agreement with the results presented inable 5. As temperature gets higher, the mol fractions of VCl4 andOCl3 reduce and that for VCl3 and VCl2 experience a significant

Fig. 26. Molar reaction enthalpy for the formation of gaseous VCl3 and VCl2.

elevation. For some temperature between 1300 K and 1373 K themol fractions of VCl4 and VCl3 assume equal values. This point isrelated to the condition where the chlorination enthalpy is zero. Forhigher temperatures, where x(VCl4) < x(VCl3) the process becomesprogressively more endothermic.

It is interesting to see that he explained behavior is consistentwith the fact that the global formation reactions of VCl3 and VCl2from liquid V2O5 in the temperature 1000–1800 K are associatedwith positive molar reaction enthalpies (Fig. 26) and that of VOCl3and VCl4 with negative molar reactions enthalpies (Fig. 27).

3. Final remarks

In this chapter three different approaches to the chlorinationequilibrium study of an oxide were presented. The first two arebased on the construction of �G◦ × T diagrams (Section 2.1.1) and
on the calculations, first introduced by Kang and Zuo [27] (Section2.1.2.1), respectively. Both of them take into consideration that eachchlorinated compound is produced independently. The third onehas its fundamental based on the total Gibbs energy minimization


F

ots

ttteasntcv

t(w�duc

sp(p9

wcfo(tcanodrosoC

where the concentrations of the two mentioned carbon oxides have

ig. 27. Molar reaction enthalpy for the formation of gaseous VOCl3 and VCl4.

f the reaction system and the gas phase equilibrium composi-ion is calculated considering that the formed species are producedimultaneously (topic Section 2.1.2.2).

The method based on the construction of �G◦ × T was applied onopic (2.1.2) for studying the thermodynamic viability of the reac-ion between gaseous Cl2 and V2O5. The discussion evidenced thathe chlorination is thermodynamically feasible only in the pres-nce of a reducing agent (graphite in the case of the present work)nd was initially focused on the production of VCl4 and VOCl3. Theame approach was employed for studying the possible mecha-isms associated with the formation of VCl4 and VOCl3. Accordingo the results (Sections 2.1.1.1–2.1.1.3), the synthesis of these twoompounds should be subdivided in different stages, which canary in nature, depending on the temperature range considered.

The sequence of formation of vanadium chlorides accordingo the predominance diagrams computed in space P(Cl2)–P(O2)Figs. 10–13) is perfectly consistent with the findings associatedith the investigation of possible chlorination paths based onG◦ × T curves, as far as these graphics are plotted with thermo-

ynamic consistence (see examples in Figs. 7 and 9). Also, it isnderstood that the interpretation of these two diagrams togetherould became a very useful didactic contribution.

According to Figs. 6 and 7, between 953 K and 1539 K, VCl4hould be formed directly from VCl2. This is indeed observed in theredominance diagrams computed at 1073 K (Fig. 12) and 1273 KFig. 13). Outside the mentioned range, VCl3 should be producedrior to VCl4. This was also confirmed in the diagrams computed at00 K (Fig. 10) and 1673 K (Fig. 11).

In the case of the vanadium oxychlorides, only the VOCl3 fieldas detected in the predominance diagrams, indicating that this

ompound is the only stable vanadium oxychloride to be formedor the conditions imposed. On the other hand, the discussion basedn �G◦ × T curves (Figs. 8 and 9) indicates that other oxychloridesVOCl and VOCl2) could be formed prior to VOCl3. Therefore, ifhe information contained in the thermodynamic data-base used isonsidered to be reliable enough, these species should be regardeds metastable, whose decomposition in the way to thermody-amic equilibrium leads exclusively to the formation of VOCl3, thenly stable oxychloride according to the calculated predominanceiagrams. Depending on the oxygen potential prevailing in theeactors atmosphere, however, some other routes for the formationf VOCl3 can be considered. For higher P(O2) values, the results pre-
ented on topic (Section 2.1.1.3) clearly indicate that the mentionedxychloride can be formed through the direct reaction betweenl2 and the most stable vanadium oxide under the specified
ica Acta 559 (2013) 1– 16

oxygen potential – V2O5, VO2, V2O3 or V3O5 (Figs. 14 and 15). Onthe other hand, if the atmosphere is made reductive enough (lowoxygen activity), the VOCl3 production should happen through theoxidation of one of the possible vanadium chlorides VCl2, VCl3 andor VCl4 (see Eq. (12)), the stimulated transformation depending onthe P(Cl2) value imposed (Figs. 16–18). In this latter case (reduc-ing atmosphere) it could also be possible an oxide partial reductionfollowed by the lower oxide chlorination.

Generally speaking, for an atmosphere rich in Cl2, VOCl3 andVCl4 are the most stable vanadium chloride and oxychloride,respectively, as their formation reactions are associated with �G◦

significant negative values (Fig. 4). However, just by looking at the�G◦ plots, it is only possible to estimate the specie in greater con-centration (VCl4) at chemical equilibrium. A quantitative approachfor the relative stability between VOCl3 and VCl4 was carried outby the implementation of Kang and Zuo method (Section 2.1.2.1).The results (Fig. 19) indicated that VCl4 should, in fact, have ahigher concentration in comparison with VOCl3 in the temperaturerange between 1073 K and 1373 K. Moreover, the data contained inFig. 19 suggest that the tendency of formation of both VCl4 andVOCl3 should reduce as temperature achieves higher values. Thisfact is also in agreement with the gas phase speciation results (seeTable 5).

It can be said that both, the method based on the �G◦ × Tdiagrams construction as well as the Kang and Zuo method, incor-porate some simplifications and are very easy to implement.However, they lead to only a superficial knowledge of the truenature of the equilibrium state achievable. Thanks to the develop-ment of computational thermodynamic software, more complexcomputations can be conducted. For example, by allowing thechlorides and oxychlorides to build a gaseous solution, the mini-mization of the total Gibbs energy of the system results in thedirect computation of the mol fraction of each chlorinated speciespresent in the gas phase (Section 2.1.2.2). This approach can beseen as a general improvement in comparison with the usuallyemployed methods, since all equilibrium equations are solvedsimultaneously, with the further advantage that there is no needto formulate a group of independent reactions that cover all pos-sible chemical interactions among the components, a task that canbecome very complex for metals, as in vanadium case, which canproduce a family of chlorides and oxychlorides.

The conclusion that carbon strongly promotes the thermody-namic driving force necessary to chlorination and that VCl4 shouldbe formed preferentially in relation to VOCl3 are perfectly consis-tent with the results based on the total Gibbs energy minimization(Tables 5 and 6). However, by the application of this last method itwas possible to go a little further, through investigation of the effectof P(Cl2) over the chlorination enthalpy (Section 2.1.2.3) and study-ing the effect of temperature, Cl2 and O2 partial pressures over theconcentrations of vanadium chlorides and oxychlorides in the gasphase (Section 2.1.2.2).

The predictions associated with the effect of temperature overthe gas phase speciation (Table 5) indicate that the mol fractionsof VCl2 and VCl3 grow significantly in the range between 1073 Kand 1473 K and, as a result, the concentrations of VOCl3, VCl4, COand CO2 exhibit a significant reduction. This finding agrees with thetendency depicted by the predominance diagrams constructed forthe system V–O–Cl, where the VCl4(g) and VOCl3(g) fields shrinkand that of VCl3(g) grows (Figs. 10–13). Also, the calculated molfraction of CO is at all temperatures much higher than the molfraction of CO2, a fact that is consistent with the establishment ofthe Boudouard’s equilibrium for temperatures higher than 973 K,

the same magnitude.The fact that the speciation computation indicates appreciable

amounts of gaseous VCl3 for temperatures higher than 1100 K and

ochim

ocgtttmtiatV

vto(Ca

pettsipimdtmdO

paesdnV(

4

pbrfcnpTttto

tsaga

[

[

[

[

[

[

[

[


f gaseous VCl2 for temperatures higher than 1473 K, apparentlyontradicting the information contained in the predominance dia-rams of Figs. 11 and 13, is a mere consequence of the fact that onopic (Section 2.1.2.2) the gaseous chlorides build an ideal gas solu-ion. The chemical potentials of VCl3 and also of VCl2 are lower thanheir pure molar Gibbs energies. As a result, the species become

ore stable in the gaseous solution in relation to the pure state, andheir mol fractions assume higher values for the same temperaturemposed. The same idea explains why VCl4 is formed in significantmounts at 1073 K for a P(Cl2) value lower than the one observed inhe predominance diagram of Fig. 12 for the equilibrium betweenCl2(l) and VCl4(g).

The effect of adding more Cl2 after all vanadium has been con-erted to gaseous chlorinated compounds is also consistent withhe expectations. At 1073 K the results indicate that the mol fractionf VCl4 grows while all other relevant chlorinated species reducesFig. 21). This can be explained by the reaction of VCl2 and VCl3 withl2 resulting in VCl4, which have a significant negative driving forcet 1073 K (Table 4).

The study of the impact of varying P(O2) over the gas phase com-osition at 1373 K indicated that the mol fractions of CO and CO2xperience significant elevation as P(O2) becomes higher, a facthat is also observed in the case of VOCl3 (Table 6). The concen-ration of all other chlorinated compounds reduces for the sametudied range of P(O2). The influence of the oxygen chemical activ-ty over the gas phase speciation can be explained by a group ofroposed reactions between VCl4, VCl3 and VCl2 with O2 resulting

n VOCl3 (Eq. (19)). All these reactions have equilibrium constantsuch higher than one, indicating an expressive thermodynamic

riving force at the standard conditions for the specified tempera-ure (Table 7). The same reactions were put forward as a possible

ethod for the production of VOCl3 from one of the stable vana-ium chlorides (VCl2, VCl3 or VCl4) when low partial pressures of2 are imposed (Section 2.1.1.3).

The conclusions about the exothermic nature of the chlorinationrocess (Section 2.1.2.3) in the temperature range between 1000 Knd 1300 K and the observation that it becomes progressively morendothermic as 1700 K is approached (Fig. 25), are perfectly con-istent with the fact that the atmosphere becomes progressivelyiluted in VCl4 and VOCl3, whose formations are associated withegative molar enthalpies (Fig. 27) and becomes richer in VCl2 andCl3, whose molar enthalpy of formation are considerably positive

Fig. 26).

. Conclusions

According to the obtained results, the chlorination of V2O5 in theresence of carbon is a thermodynamic feasible process in the rangeetween 900 K and 1600 K. During the process, VCl4 and VOCl3 are,espectively, the most stable vanadium chloride and oxychlorideormed. In the case of VCl4 it has been observed that its formationan be explained as a progressive path starting from VCl2, which isext converted to VCl3 or directly to VCl4, depending on the tem-erature. In relation to VOCl3 two possibilities seems to be present.he first one, starts with VOCl, which, depending on the reactionemperature, can be converted to VOCl3 either directly or passinghrough VOCl2. In the second one, VOCl3 is formed from one ofhe vanadium oxides (higher oxygen activity) or chlorides (lowerxygen activity).

The speciation calculations indicate that the concentration ofhe gas phase generated during the chlorination process is very
ensitive to variation in temperature, P(Cl2) and P(O2). As temper-ture achieves higher values, the mol fractions of VCl3 and VCl2row while VCl4 and VOCl3 reduce. Regarding the effect of P(Cl2),s this parameter achieves higher values, the formation of VCl4 is
[

[

ica Acta 559 (2013) 1– 16 15

stimulated as VCl2 and VCl3 reacts with the Cl2 available. The effectof increasing the oxygen potential P(O2) can be seen as a way ofpromoting the formation of VOCl3 through the oxidation of one ofthe vanadium chlorides already present (VCl2, VCl4 or VCl3). More-over, the chlorination process is considerably exothermic at 1000 K,but becomes progressively less exothermic at higher temperatures.This fact is correlated with the mol fraction reductions of VCl4 andVOCl3 and the concomitant increasing of those for VCl3 and VCl2.

Finally, we can conclude that the study of the equilibrium statesachievable through the reaction between a transition metal oxideand gaseous Cl2, can be now approached through the implemen-tation of methods of different complexity levels. The most generalone, in which the total Gibbs energy of the reaction system is min-imized, enables the construction of a more detailed (quantitative)picture of the equilibrium states involved. However, as it is evidentfrom the comparisons explained above, the most general methodmust be consistent with the tendencies predicted by simpler cal-culation procedures (qualitative picture), as the one defined by theconstruction of �G◦ × T diagrams, or the simple solution of individ-ual chemical equilibrium equations. Although the discussion in thiswork focuses exclusively on the V2O5 chlorination, the followedapproach can be applied to any reaction system.

Acknowledgments

The authors are especially grateful for the financial supportand scholarships provided by CAPES, CNPQ and FAPERJ during thedevelopment of the present work.

References

[1] L. Zhang, et al., Rare earth extraction from bastnaesite concentrate by stepwisecarbochlorination–chemical vapor transport-oxidation, Metall. Mater. Trans. B35 (2) (2004) 217–221.

[2] E. Cecchi, et al., A feasibility study of carbochlorination of chrysotile tailings,Int. J. Miner. Process. 93 (3/4) (2009) 278–283.

[3] I. Gaballah, M. Djona, Recovery of Co, Ni, Mo, and V from unroasted spenthydrorefining catalysts by selective chlorination, Metall. Mater. Trans. B 26 (1)(1995) 41–50.

[4] E.A. Brocchi, F.J. Moura, Chlorination methods applied to recover refractorymetals from tin slags, Miner. Eng. 21 (2) (2008) 150–156.

[5] K. Murase, et al., Recovery of vanadium, nickel and magnesium from a fly ash ofbitumen-in-water emulsion by chlorination and chemical transport, J. AlloysCompd. 264 (1/2) (1998) 151–156.

[6] D. Neff, Environmentally acceptable chlorination processes, in: Aluminum CastHouse Technology: Theory & Practice, Australasian, Asian, Pacific Conference,1995, pp. 211–225.

[7] D. Mackay, Is chlorine the evil element? Environ. Sci. Eng. (1992) 49–52.[8] G. Micco, A.E. Bohe, H.Y. Sohn, Intrinsic kinetics of chlorination of WO3 particles

with Cl2 gas between 973 K and 1223 K (700 ◦C and 950 ◦C), Metall. Mater. Trans.B 42 (2) (2011) 316–323.

[9] J.P. Gaviria, A.E. Bohe, Carbochlorination of Yttrium oxide, Thermochim. Acta509 (1/2) (2010) 100–110.

10] M.R. Esquivel, A.E. Bohé, D.M. Pasquevich, Carbochlorination of samariumsesquioxide, Thermochim. Acta 403 (2003) 207–218.

11] M.W. Ojeda, J.B. Rivarola, O.D. Quiroga, Study of the chlorination of molybde-num trioxide mixed with carbon black, Miner. Eng. 15 (2002) 585–591.

12] H.H. Kellog, Thermodynamic relationships in chlorine metallurgy, J. Met. 188(1950) 862–872.

13] C.C. Patel, G.V. Jere, Some thermodynamical considerations in the chlorinationof ilmetite, Trans. Metall. Soc. AIME 218 (1960) 219–225.

14] R.F. Pilgrim, T.R. Ingraham, Thermodynamics of chlorination of iron, cobalt,nickel and copper sulphides, Can. Metall. Q. 6 (4) (1967) 333–346.

15] N. Sano, G. Belton, The thermodynamics of chlorination of vanadium pentoxide,Trans. Jpn. Inst. Met. 21 (9) (1980) 597–600.

16] L. Brewer, B.B. Ebinghaus, The thermodynamics of the solid oxides of vanadium,Thermochim. Acta 129 (1988) 49–55.

17] R.E. McCarley, J.W. Roddy, The vapor pressures of vanadium(II) chloride,vanadium(III) chloride, vanadium(II) bromide, and vanadium(III) bromide byKnudsen effusion, Inorg. Chem. 3 (1) (1964) 60–63.

18] N. Saunders, A.P. Miodownik, Calphad: Calculation of Phase Diagrams—AComprehensive Guide, Elsevier Science Ltd., The Boulevard, Langford Lane,Kidlington, Oxford, 1998.

19] O. Knacke, O. Kubashcewski, K. Hesselmann, Thermochemical Properties ofInorganic Substances, Springer-Verlag, Berlin, 1991.

1 ochim

[

[

[

[[

[

[

6 E.A. Brocchi et al. / Therm

20] H. Oppermann, Gleichgewichte mit VOCl3, VO2Cl, und VOCl2, Z. Anorg Allg.Chem. 331 (3/4) (1967) 113–224.

21] A. Hackert, Plies, R. Gruehn, Nachweis und thermochemische Charakterisierungdes Gasphasenmolekuls VOCl, Z. Anorg. Allg. Chem. 622 (1996) 1651–1657.

22] D.L. Hildenbrand, et al., Thermochemistry of the gaseous vanadium chlorides
VCl, VCl2, VCl3, and VCl4, J. Phys. Chem. A 112 (2008) 9978–9982.
23] HSC version 6.0 data-base.24] E. Allain, M. Djona, I. Gaballah, Kinetics of chlorination and carbochlorination

of pure tantalum and niobium pentoxides, Metall. Mater. Trans. B 28 (1997)223–232.

[

[

ica Acta 559 (2013) 1– 16

25] J. Gonzallez, et al., �-Ta2O5 carbochlorination with different types of carbocanadian, Metall. Q. 41 (1) (2002) 29–40.

26] P.K. Jena, E.A. Brocchi, J. Gonzalez, Kinetics of low-temperature chlorination ofvanadium pentoxide by carbon tetrachloride vapor, Metall. Mater. Trans. B 36(2) (2005) 195–199.

27] S.X. Kang, Y.Z. Zuo, Chloridizing roasting of complex material containing lowtin and high iron at high temperature, Kunming Metall. Res. Inst. Rep. 89 (3)(1989).

28] D. Robert, Thermodynamics in Materials Science, Second ed., Taylor and Fran-cys, NY, USA, 2006.

A Chemical Thermodynamics Review Applied to V2O5 Chlorination

Documents

Transcript of A Chemical Thermodynamics Review Applied to V2O5 Chlorination