Ut

Fa

Vb

a

ARR1AA

KRTERA

1

sg(aviit

aasmmprp

q

h0

Applied Catalysis A: General 504 (2015) 220–227

Contents lists available at ScienceDirect

Applied Catalysis A: General

jou rn al hom epage: www.elsev ier .com/ locate /apcata

nraveling the mechanism of chemical reactions throughhermodynamic analyses: A short review�

.C. Meuniera,∗, J. Scalbertb, F. Thibault-Starzykb

Université Lyon 1, CNRS, UMR 5256, IRCELYON, Institut de recherches sur la catalyse et l’environnement de Lyon, 2 avenue Albert Einstein, 69626illeurbanne, FranceLaboratoire Catalyse et Spectrochimie, UMR 6506, CNRS-ENSICAEN, Université de Caen, 6 Bd. Marechal Juin, 14050 Caen, France

r t i c l e i n f o

rticle history:eceived 21 September 2014eceived in revised form0 December 2014ccepted 14 December 2014vailable online 23 December 2014

a b s t r a c t

Basic thermodynamic analyses can provide in-depth knowledge of the mechanism of chemical reactions.This short review recalls first the definitions of the approach-to-equilibrium �, the reaction quotient Qand the thermodynamic equilibrium constant K. Thereafter, four case studies specifically dealing withgas-phase heterogeneously catalyzed reactions are reviewed: (i) alkane hydroisomerization, (ii) NO oxi-dation during the selective catalytic reduction of NOx with propene, (iii) the steam reforming of methanoland (iv) ethanol condensation to butanol. These examples illustrate in different manners how a reaction

eywords:eaction mechanismhermodynamicsquilibrium constant

mechanism can be supported or rejected based on rather simple analyses of the concentrations of reac-tants and products. While the examples used here are all referring to catalyzed reactions, it must beemphasized that the method can be applied to non-catalytic systems.

© 2014 Elsevier B.V. All rights reserved.

eaction quotientpproach to equilibrium

. Introduction

Chemical processes can be made more efficient by better under-tanding the various reaction steps involved in the conversion of aiven feedstock into the desired products so that each step can bei) promoted by a suitable catalyst and/or (ii) carried out underppropriate operating conditions as to push away the limiting con-ersion associated with the thermodynamics of the system. Insightsnto reaction mechanisms can be obtained, for instance, via study-ng the nature (e.g. acid/base [1], redox [2]) and structure [3,4] ofhe catalytic sites or by in situ and operando spectroscopy [5–8].

Thermodynamics is routinely used as a tool to determine favor-ble reaction conditions so that high yields of products can bechieved. However, it is less common to use thermodynamics toupport or rule out a particular reaction step. The power of thisethod is recalled here, showing how details of complex reactionechanisms can be unraveled. The method is based on the com-

arison of the proportions of reactants and products present in theeactor effluent to various thermodynamic equilibrium constants

ertaining to the system.

This short review recalls first the definitions of the reactionuotient Q and the thermodynamic equilibrium constant K. The

� Part of the special issue of Applied Catalysis A: General dedicated to Prof J. Vedrine.∗ Corresponding author. Tel.: +33 472445468; fax: +33 472445365.

E-mail address: fcm@ircelyon.univ-lyon1.fr (F.C. Meunier).

ttp://dx.doi.org/10.1016/j.apcata.2014.12.028926-860X/© 2014 Elsevier B.V. All rights reserved.

calculation of these numbers will be detailed over a few exam-ples, taken primarily from published work of the authors. The directcomparison of these two numbers (or the value of the ratio Q/K = �,so-called “approach-to-equilibrium” parameter) can be used tosupport or reject the relevance of a reaction step in a complexscheme.

It is important to stress that the method described here does notinvolve the thermodynamics of adsorbed species and therefore noinference can be made as to the nature and concentration of surfaceintermediates. The examples used in the present paper are dealingwith gas-phase compounds and heterogeneous catalysis, but it iscrucial to realize that the method can be applied to non-catalyticreactions and condensed systems, as long as activity coefficientsare available.

The insights obtained in each of the examples enabled to derivebetter catalytic formulations or guide future catalyst developmentand improved reaction conditions. Numerical examples of calculusare given so that a reader not familiar with this topic should fullyunderstand the whole procedure. Most calculations can actuallybe done on user-friendly commercially available thermodynamicsoftwares such as HSC Chemistry®.

2. Definitions

The reaction quotient Q for a given reaction is the ratio betweenthe product of the (dimensionless) chemical activity of each of the

lysis A: General 504 (2015) 220–227 221

pciamtb

n

n

2

sptvettcs

Q

aruoG

�

�

eceTas

shTatat–u

btdprt

F.C. Meunier et al. / Applied Cata

roducts to that of the reactants, each raised to the power of theorresponding stoichiometric coefficient [9]. For reactions involv-ng only gaseous compounds at moderate pressure, the chemicalctivities can be approximated to the pressures of reactants nor-alized to the standard pressure P◦ = 1 bar (i.e. ideal case for which

he gas fugacities are unity). Examples of expressions for Q are givenelow for some of the reactions discussed here [Eqs. (1)– (3)].

− butane � isobutane Q1 = Pisobutane

Pn–butane(1)

-butane � H2 + 1-butene Q2 =(

PH2 × P1-butene

)(Pn-butane × P◦)

(2)

acetaldehyde + 2 H2 � butanol + H2O Q3

=(

PH2O × Pbutanol × P◦2)

(Pacetaldehyde × PH2

)2(3)

If no reaction product is present in the system (typically at thetart of the reaction in a batch reactor or at the front of the bed in alug flow reactor), the reaction quotient Q is equal to zero. If the sys-em under study has reached the thermodynamic equilibrium thealue of the reaction quotient is equal, by definition, to the reactionquilibrium constant K [Eq. (4)]. The value of K can be determinedhrough the knowledge of the standard Gibbs free energy of reac-ion �rG◦ at the temperature considered, which is tabulated forommon reactions and can be calculated in most thermodynamicoftwares.

equilibrium = K = exp(−�rG◦

RT

)(4)

The approach-to-equilibrium (noted “�”) is the ratio between Qnd K [Eq. (5)] and takes a maximum value of one as the systemeaches equilibrium. The approach-to-equilibrium is commonlysed as a means to obtain forward rates of reaction when the extentf backward reaction becomes significant [10]. The variation ofibbs free energy of reaction �rG can be written as follows [Eq. (6)]:

= Q

K(5)

rG = �rG◦ + RT ln Q = RT lnQ

K= RT ln � (6)

The change of the Gibbs free energy of reaction must remainqual to or lower than zero at all stages of a spontaneous chemicalonversion of the system. Therefore, the value of the approach-to-quilibrium � must remain lower or equal to unity, i.e. � = Q/K ≤ 1.his is schematically represented in Fig. 1, in which the open circlesre associated with allowed evolutions of the system, while thetates in the red shaded zone are forbidden transformations.

The above paragraph recalls that the Gibbs free energy of theystem at any point of a spontaneous transformation cannot beigher than its initial value. This is the realm of thermodynamics.his statement is not contradictory with the fact that at any time

minute fraction (determined by statistical thermodynamics) ofhe molecules will have a sufficiently high energy to overcome anctivation barrier to form the products. This fraction of moleculeshat can react will be increased – and so will be the reaction rate

if the activation barrier of the transformation can be lowered bysing a catalysts or a suitable solvent. This is the realm of kinetics.

The level of kinetic analysis used in the following examples isasic, mostly consisting in measuring the concentrations of reac-ants and products. Detailed kinetic analyses (e.g. based on the

etermination of reaction orders, primary and secondary reactionroducts, microkinetic models) are also a powerful tool to unraveleaction mechanism, but are not the point of the present contribu-ion.

Fig. 1. Gibbs free energy of reaction as a function of the approach-to-equilibrium �.Only transformations associated with � ≤ 1 are allowed.

3. Case studies

The calculation of � requires the determination of the reactionquotient Q, which implies measuring the partial pressure (or con-centration) of each of the species appearing in the stoichiometricequation. If a reactant or product is in large excess, then the evo-lution of the reaction quotient can be satisfactorily monitored byonly considering the variation of the concentration of the othercompounds.

This is exemplified in the first two case studies described belowdealing with alkane hydroisomerization and NOx oxidation, forwhich the H2 and O2 concentrations, respectively, were consid-ered to be constant. The thermodynamic analysis was there merelyreduced to comparing two-term ratios (i.e. [A]/[B]) of pairs of con-centrations of reactant and/or products.

The third case study dealing with methanol steam reformingshows how the thermodynamically relevant reaction pathway canbe readily determined from the comparison between the measuredmolar fractions and those calculated at the thermodynamic equi-librium.

The final case study dealing with ethanol condensation involvesthe analysis of an intermediate reaction step based on the exactdetermination of the reaction quotient Q. The utilization of equi-librium diagrams is shown to be a useful guide in the search ofpossible reaction pathways and the rejection of forbidden routes.

3.1. Alkane hydroisomerization

Solids derived from tungsten and molybdenum carbides andoxides can be alternative to noble metals to catalyze reactionssuch as the reverse water–gas shift and alkane hydroisomeriza-tion [11–14]. The nature of the reaction mechanism of alkanehydroisomerization over molybdenum oxide-based catalysts hadbeen a matter of debate. Katrib et al. [15] and Matsuda et al.[16] had proposed that a traditional bifunctional mechanismoperated, involving alkenes as reaction intermediates. On thecontrary, Bouchy et al. [17] had proposed a metallacyclobutaneintermediate-based mechanism.

The hydroisomerization of n-butane to isobutane [Eq. (1)] overreduced MoO3 provided a case study in which thermodynamicsenabled to positively determine both the reaction mechanism and

the rate-determining step (RDS) [18]. MoO3 is not active for thisreaction and the catalytically active phase (thought to be an oxy-hydroxide of molybdenum [19]) is gradually formed under thereaction stream at 350 ◦C (Fig. 2). Note that the insertion of carbon

222 F.C. Meunier et al. / Applied Catalysis A: General 504 (2015) 220–227

F3i

it

iftorac

titm1a

(dtdbmMe

iatposIh

rmtc

[

Several other catalysts based on alumina led to similar observations[26,27].

Another mechanistic route enabling higher NO2/NO ratios musttherefore exist. A possible pathway involves the formation and

ig. 2. Isobutane and butene yields during the reductive activation of MoO3 at50 ◦C. (a) isobutane, (b) trans-2-butene, (c) cis-2-butene, (d) 1-butene and (e)

sobutene. Feed: 7% n-butane in H2, at ambient pressure. (Adapted from Ref. [18].)

n the catalyst structure (to form an oxicarbide) is detrimental tohe selectivity, leading to butane hydrogenolysis to methane [20].

The concentration of the various butenes, potential reactionntermediates produced by dehydrogenation of n-butane [Eq. (2)or the case of 1-butene], could be accurately measured at the reac-or exit, despite being present at levels more than two and threerders of magnitude lower than that of isobutane and n-butane,espectively (Fig. 2). The catalytic experiments were carried out in

large excess of dihydrogen, the concentration of which could beonsidered as constant.

For the sake of brevity only the ratios of the hydrocarbon concen-rations were considered and reported at various time on streamn Table 1, since the concentration of H2 was essentially constanthroughout the experiment. The corresponding ratios at the ther-

odynamic equilibrium are reported in the last column of Table 1.-butene, cis-butene and trans-butene were quickly equilibratednd are not discussed any further [18].

Steady-state experimental ratios (n-butane/1-butene) andisobutane/isobutene) were equal to those associated with theehydrogenation thermodynamic equilibria within the experimen-al error (Table 1). This indicates that the hydrogenation andehydrogenation reactions were fast and not rate-limiting, if theifunctional mechanism applied. The phase responsible for theseetal-like reactions has been proposed to be molybdenum dioxideoO2, which exhibits a metallic band structure with delocalized �

lectrons [21].On the contrary, the skeletal isomerization between butenes and

sobutene had not reached equilibrium (Table 1) and was proposeds being the RDS. This assumption was subsequently confirmed byhe fact that the addition of an acidic microporous aluminophos-hate (CoAlPO-11), a known catalyst for the selective isomerizationf alkenes [22], led to a large enhancement of the butane hydroi-omerization activity when mechanically mixed with MoO3 [23].t must be noted that CoAlPO-11 itself was not active for butaneydroisomerization.

In conclusion, our approach enabled determining the maineaction mechanism of n-butane hydroisomerization over reducedolybdena, which is summarized in Scheme 1. It must be stressed

hat no labeled compounds were used in this study, as is often thease for mechanistic investigations on similar systems [24].

The cases of n-pentane and n-hexane were also investigated23]. The same bifunctional mechanism operated, with the main

Scheme 1. Reaction equilibria (�) and rate-determining step (RDS) relevant tobutane hydroisomerization over reduced MoO3 at 350 ◦C. All the C4 molecules drawnin the scheme refer to gas-phase species. (Adapted from Ref. [18].)

difference being that the RDS was no longer the skeletal isomeriza-tion, but instead the dehydrogenation/hydrogenation steps weredetermining the overall rate of the reaction [23].

3.2. Oxidation of NO to NO2 during the selective reduction of NOxwith propene

The selective catalytic reduction of NO with propene over alu-mina provided an example in which the long-thought-crucial directoxidation of NO with O2 to give NO2 [Eq. (7)] was actually found tobe irrelevant [25].

NO + 0.5 O2 � NO2 (7)

The thermodynamic analysis was focused on the NO2/NO ratio,as O2 was in great excess and its concentration was essentially con-stant throughout the experiment. The NO2/NO ratio above a certainvalue of pseudo-contact time W/F (weight to flow ratio) was fargreater than that associated with the thermodynamics of the reac-tion described by [Eq. (7)] (Fig. 3). Such high NO2/NO ratios arethermodynamically forbidden, if the relevant route for NO2 for-mation were to be that described by the direct oxidation [Eq. (7)].

Fig. 3. Selective catalytic reduction of NO using propene over alumina at 540 ◦Cas a function of the W/F: NO2 to NO ratio (a) experimental and (b) at the ther-modynamic equilibrium of the reaction NO + 1/2 O2 � NO2 (—). Feed: 500 ppmNO + 500 ppm C3H6 + 2.5% O2 in Ar. (Reprinted from Ref. [25].)

F.C. Meunier et al. / Applied Catalysis A: General 504 (2015) 220–227 223

Table 1Ratios of the concentrations of various C4–compounds observed during the activation of MoO3 under 7% n-butane + 93% H2 at 350 ◦C and ambient pressure. (adapted fromRef. [18]).

Molar ratios 1 h 6 h Steady-state at 350 ◦Ca Thermodynamic equilibrium ratio at 350 ◦Cb

Dehydrogenationn-butane/1-butene 16000 7000 6000 ± 800 6600Isobutane/isobutene 0.045 17 500 ± 80 470IsomerizationIsobutene/1-butene 0.27 3.3 1.2 ± 0.2 8.12n-butane/isobutane 1.6 × 106 120 9.3 ± 1 1.73

a The average value for the three data points after 20 h is reported.b Calculated using the HSC software (version 4.1, © Outokumpu Research Oy, Pori, Finland, A. Roine).

Fe0

oscwpe

C J mo

C 4 kJ m

botipat

omt(o

fv(mirofsd

Fig. 5. The influence of W/F on the product compositions at T = 300 ◦C over

ig. 4. Effect of the W/F on the C3H6-SCR of NO over Ag/Al2O3 at 590 ◦C: NO2/NOxperimental ratio (o) and theoretical thermodynamic ratio (dotted line). Feed:.05% NO + 0.05% C3H6 + 2.5% O2/He. (Reprinted from Ref. [26].)

xidation of organonitrogen compounds, which have often beenuggested as potential reaction intermediates [28]. The nature andoncentration of such compounds are yet unclear. Nitromethaneas selected as a model compound, since its thermodynamicotentials are readily available to calculate the standard Gibbs freenergy of the reactions of interest [Eqs. (8) and (9)].

3H6 + 1.5O2 + 2NO � 2CH3NO2 + CO �rG◦(540 ◦C) = −298 k

H3NO2 + 1.75O2 � NO2 + CO2 + 1.5H2O �rG◦(540 ◦C) = −73

The thermodynamics of formation of nitromethane and its com-ustion to NO2 are largely favorable at the temperature of reactionf 540 ◦C, while the standard Gibbs free energy of the direct oxida-ion [Eq. (7)] at the same temperature is about + 4 kJ mol−1. The truentermediates and reaction pathways are likely to be more com-lex, but these calculi show that organonitrogen-based routes canfford high concentration of NO2 at high temperatures, contrary tohe case of the direct oxidation.

Some rational bases behind the thermodynamic relevance ofrganonitrogen routes are that (i) the resulting intermediate isuch more stable than the highly energetic NO radicals [Eq. (8)], (ii)

he formation of highly stable water and CO2 provided an energyenthalpy) sink during oxidation [Eq. (9)] and (iii) the larger numberf molecules on the product side increased entropy [Eq. (9)].

Similar data collected over a silver-promoted alumina wereurthermore interesting because the level of conversion could bearied to a larger extent when changing the pseudo-contact timeFig. 4). The NO2/NO ratio rapidly and largely exceeded the ther-

odynamic limit associated with direct oxidation [Eq. (7)] withncreasing W/F. The NO2/NO ratio then decreased and eventuallyeached the value associated with the thermodynamic equilibrium

f the direct oxidation [Eq. (7)]. It was proposed that the NO2 rapidlyormed via the organonitrogen pathway [Eqs. (8) and (9)] and wasubsequently slowly decomposed to NO and O2 according to theirect route [Eq. (7)], once propene had been fully consumed [26].

l−1 (8)

ol−1 (9)

a Cu–Zn–Zr–Al–O catalyst, m = equilibrium excluding CO from calculations,n = equilibrium including CO in calculations (H2O/CH3OH = 1.3, P inert dilu-ent = 55 kPa, Ptotal = 101 kPa). (Adapted from Ref. [31].)

The case studies described in Section 3.1 and 3.2 have underlinedhow detailed mechanistic insights could be obtained by simplyconsidering the two-term ratios of the concentrations of a couple

of reactant and/or products. The latter example even showedthat NO2 formation/decomposition occurred via two very differentroutes depending on whether propene was present or not. It alsoproved that a catalytic function able to oxidize directly NO to NO2was not a pre-requisite to obtain an effective catalyst to reduce NOwith propene. On the contrary, such strong oxidizing centers leadto unselective propene combustion.

3.3. Methanol steam reforming

The selectivity to CO2 is crucial during the steam reforming ofmethanol to maximize the yield of H2 and limit the formation of CO,which can poison fuel cell electrodes [29,30]. The steam reformingof methanol [Eq. (10)] over Cu–Zn–Zr–Al–O catalyst provided anexample of a system in which thermodynamics enabled discard-ing a reaction pathway [31]. A critical question was whether ornot CO was a primary reaction product, which would be formedprior to CO2 [Eq. (11)] and would then be converted to CO2 via thewater–gas shift reaction [Eq. (12)].

CH OH + H O → CO + 3H (10)

3 2 2 2

CH3OH → CO + 2H2 (11)

CO + H2O → H2 + CO2 (12)

224 F.C. Meunier et al. / Applied Catalysis A: General 504 (2015) 220–227

Scheme 2. Ethanol condensation mechanism based on the sel

Fig. 6. Product composition at the thermodynamic equilibrium calculated atT = 300 ◦C in a system noted “m” containing methanol, H2, CO2 and water and a sys-twH

maWor

ta“cpmrt

tet(cs

em noted “n” containing the same components and CO. Calculus initial conditions:ater/methanol = 1.3, P inert diluent = 55 kPa, Ptotal = 101 kPa. Calculated using theSC software (version 4.1, © Outokumpu Research Oy).

Fig. 5 represents the product compositions (left scale) and theethanol conversion (right scale) as a function of the W/F. CO only

ppeared after the other reaction products, H2 and CO2, as the/F was increased and methanol conversion was complete. This

bservation already strongly suggested that CO was not a primaryeaction product.

The expected proportions of methanol, water, H2 and CO2 athe thermodynamic equilibrium of the direct reforming [Eq. (10)]t 300 ◦C were calculated for a system free of CO, referred to asm” (Fig. 6). The calculation shows that essentially full methanolonversion can be expected at the equilibrium and that the molarroportions of H2 and CO2 reach ca. 69% and 23%, respectively. Theolar proportions of H2 and CO2 decreased to ca. 66% and 20%,

espectively, when CO was included in the system, which is referredo as “n” (Fig. 6).

The thermodynamic limits for each product associated withhe systems m and n are reported in Fig. 5 as dotted lines. Thexperimental proportions of H2 and CO2 first clearly matched the

hermodynamic composition associated with a CO-free systemline m), being in excess of the proportion associated with a systemontaining CO (line n). Therefore, these data unambiguously, andurprisingly, proved that CO was not the precursor of CO2. Instead,

f-aldolization of acetaldehyde. (Adapted from Ref. [37].)

CO was probably formed from CO2 as a result of a reverse water–gasshift reaction (reverse of Eq. (12)). This observation stressed that thereverse water–gas reaction should be suppressed to maximize H2production.

3.4. Ethanol condensation to butanol

Ethanol can be converted to higher molecular weight alcohols,so-called Guerbet alcohols [32,33]. The traditional (partly-homogeneous) synthesis of Guerbet alcohols is thought toproceed via several consecutive steps [34,35]. Aldehyde forma-tion by alcohol dehydrogenation and its self-aldolization is awell-accepted reaction pathway in the case of the reaction cat-alyzed by alkali metal hydroxides in the presence of a metalliccatalyst.

Scheme 2 illustrates this pathway in the case of the ethanolconversion to butanol. The metal carries out the ethanol dehydro-genation to acetaldehyde, as well as the hydrogenation of reactionintermediates to butanol (Steps 1, 3 and 4 in Scheme 2). Thehomogeneous base is responsible for acetaldehyde self-aldolizationto 3-hydroxybutyraldehyde, which is then dehydrated to cro-tonaldehyde. Crotonaldehyde hydrogenation eventually leads tobutanol.

Interestingly, metal-free basic oxides were shown to be activeand selective, albeit at higher temperatures, typically between300 and 450 ◦C [36–39]. The nature of the reaction mechanismof the metal-free systems has been widely debated and alter-native mechanisms have been proposed. Yang and Meng [36]proposed a direct route with no identifiable reaction intermedi-ates (Scheme 3.A) and a semi-direct pathway involving ethanolcondensation with acetaldehyde, which was formed from ethanoldehydrogenation (Scheme 3.B). Other authors have proposed thatthe direct pathway (Scheme 3.A) could be taking place simulta-neously with the acetaldehyde self-aldolization route (Scheme 2)[38].

We have recently reported thermodynamic-based investiga-tions of the high temperature condensation of ethanol to butanol

over a hydroxyapatite Ca3(PO4)2Ca(OH)2, noted “HAP”, one of thebest known catalysts for this reaction [40,41]. The formation ofacetaldehyde was of particular interest, since this compound wasoften proposed as being a crucial reaction intermediate, through

F.C. Meunier et al. / Applied Catalysis A: General 504 (2015) 220–227 225

Scheme 3. Alternative ethanol condensation mechanisms: (A) direct dimerizationand (B) semi-direct dimerization. (Adapted from Ref. [36].)

FtW

i(

awraw

teo1owta

tiBc3tc

Fig. 8. Composition at the thermodynamic equilibrium of a system containingethanol, butanol, acetaldehyde, crotonaldehyde, butanal, 2-buten-1-ol, H2 andwater as a function of temperature. Total pressure is 1 bar and the initial state ofthe system corresponds to an equivalent 11 mol.% of ethanol (balanced by Ar). Themolar fractions of ethanol and 2-buten-1-ol were always lower than ca. 0.01 moland cannot be seen on the graph (Reprinted from Ref. [41].)

Table 2Ratio between the reaction quotient Q and the equilibrium constant K for Eq. (3),obtained under various reaction conditions.

Temperature (◦C) Ethanol inlet concentration (%) WHSV (h−1) � = Q/K

350 15.2 28 623400 15.2 28 2430440 15.2 28 2730400 7.6 1.4 8440

ples of calculus are detailed elsewhere [40,41].

ig. 7. Yields of the main reaction products during ethanol reaction overhe hydroxyapatite as a function of temperature. Feed: Ethanol = 15.2% in Ar,

HSV = 14 h−1. (Reprinted from Ref. [40].)

ts self-aldolization (Scheme 2) and condensation with ethanolScheme 3.B) [38,39,42–44].

1-Butanol (noted here “butanol”) was the main product andcetaldehyde the main by-product (Fig. 7). Butadiene and ethyleneere also formed, the latter probably by ethanol dehydration on

esidual acid sites [39]. Minute concentrations of butenol, butanalnd crotonaldehyde were also observed. The selectivity to butanolas highest at around 400 ◦C.

Equilibrium composition diagrams were drawn as a meanso comprehend the global thermodynamics of systems of inter-st. The thermodynamic equilibrium of a system containing mostf the species present in Scheme 2 was calculated over the00–500 ◦C temperature range (Fig. 8). The thermodynamic dataf 3-hydroxybutyraldehyde were not available and this moleculeas therefore omitted from the system. Its equilibrium concentra-

ion is yet expected to be very low and should not change the valuesssociated with other species.

The striking feature of the data reported in Fig. 8 is thathe equilibrium concentration of butanol drops dramatically withncreasing temperature and becomes negligible above 300 ◦C.utanol is replaced by butanal (and H2) and then acetaldehyde androtonaldehyde. The equilibrium distribution of products above

50 ◦C (Fig. 8) is clearly at odds with the catalytic data collected inhe 350–440 ◦C temperature range reported in Fig. 7. Butanol waslearly the main reaction product obtained over our HAP catalyst.

400 7.6 7.0 4560400 7.6 14 4430

This combination of kinetic (Fig. 7) and thermodynamic (Fig. 8)data suggests that the reaction scheme as proposed above(Scheme 2) is likely to be irrelevant. In simple terms, if butanaland acetaldehyde were to be reaction intermediates in the forma-tion of butanol, then the concentration of these compounds shouldremain higher than that of butanol when operating above ca. 350 ◦C,because those are more favored thermodynamically.

A precise quantitative comparison of reaction quotient (Q) andequilibrium constant (K) is yet required to be able to draw adefinitive conclusion on the relevance of the pathway describedin Scheme 2, since there are many other by-products (e.g., buta-diene, ethene, H2) which affect various reaction equilibria. Thesimplest system based on acetaldehyde self-aldolization that can becomputed (i.e. having intermediates of accurately measurable con-centrations and available thermochemical data) is given by reaction[Eq. (3)]:

2 acetaldehyde + 2 H2 � butanol + H2O Q3

=(

PH2O × Pbutanol × P◦2)

(Pacetaldehyde × PH2

)2(3)

It must be stressed that the water and the H2 concentrationsneeded to quantitatively measured, alongside those of butanol andacetaldehyde. The ratios between the reaction quotient Q and theequilibrium constant K were calculated at various temperaturesand for different experimental conditions (Table 2). Practical exam-

Since the reaction quotient would be higher than the equilib-rium constant, in fact, Q � K (Table 2), the reaction given by Eq.(3) is irrelevant to describe the mechanism of butanol formation

226 F.C. Meunier et al. / Applied Catalysis A: General 504 (2015) 220–227

S over

m

tov

aw[tr

prlh

ow(tos

4

otstdutt(Td

A

cU

[[

[[[

[

[

[

[[

[

[[[

[[[

[

[

[

[

cheme 4. The ethanol condensation mechanism prevailing at high temperaturesetal-containing systems (right).

aking place over our metal-free HAP between 350 and 440 ◦C. Inther words, the condensation of ethanol to butanol did not proceedia acetaldehyde self-aldolization in the present case.

This observation is in contrast with the low temperature cat-lytic systems, which typically include a metallic phase, and whichere shown to operate through aldol self-condensation (Scheme 4)

34]. The metallic phase is crucial for the low-temperature systemo enable (i) alcohol dehydrogenation and (ii) the hydrogenation ofeaction intermediates.

Alcohol dehydrogenation is facile on basic oxides and iso-ropanol dehydrogenation is actually routinely used as a probeeaction to measure oxide basicity. Yet, the absence of a metal-ic phase would make very difficult the activation of molecularydrogen needed to hydrogenate unsaturated intermediates.

Other reaction pathways must be considered in the case ofur HAP at temperatures above 350 ◦C, which are discussed else-here [40]. In brief, the main pathway would be the direct route

Scheme 3.A) and a minor parallel pathway would be similar tohe indirect route (Scheme 3.B), but in which the hydrogenationf unsaturated intermediates proceeds via hydrogen-transfer fromacrificial ethanol molecules and not from molecular hydrogen [40].

. Conclusions

The approach-to-equilibrium ratio should always remain lowerr equal to one for a thermodynamically feasible reaction. Reac-ion pathways that breach this rule can be ruled out andeveral examples have been detailed here. An advantage of thishermodynamic-based method is that it is uses steady-state kineticata obtained under operating conditions and does not require these of transient methods or labeled compounds. It is yet necessaryo have the thermochemical data of the compounds appearing inhe stoichiometric equation and the corresponding concentrationsand possibly the activity coefficient, if not under ideal conditions).he method may unambiguously discard a reaction pathway or mayetermine the relevance of others routes.

cknowledgements

F.C.M. thanks Prof J.R.H. Ross and Dr J.P. Breen for fruitful dis-ussions on some of the methods described here while at theniversity of Limerick.

[[[

[

metal-free catalysts (left) is different from that occurring at low temperatures on

References

[1] E.G. Derouane, J.C. Védrine, R. Ramos Pinto, P.M. Borges, L. Costa, M.A.N.D.A.Lemos, F. Lemos, F. Ramôa Ribeiro, Catal. Rev. Sci. Eng. 55 (2013) 454–515.

[2] N. Dimitratos, J.C. Vedrine, Appl. Catal., A: Gen. 256 (2003) 251–263.[3] J.C. Vedrine, Appl. Catal., A: Gen. 474 (2014) 40–50.[4] F.C. Meunier, L. Domokos, K. Seshan, J.A. Lercher, J. Catal. 211 (2002) 366–378.[5] M.D. Rhodes, K.A. Pokrovski, A.T. Bell, J. Catal. 233 (2005) 210–220.[6] L. Piccolo, H. Daly, A. Valcarcel, F.C. Meunier, Appl. Catal., B: Environ. 86 (2009)

190–195.[7] S. Chansai, R. Burch, C. Hardacre, J.P. Breen, F.C. Meunier, J. Catal. 276 (2010)

49–55.[8] S. Chansai, R. Burch, C. Hardacre, J.P. Breen, F.C. Meunier, J. Catal. 281 (2011)

98–105.[9] P. Atkins, J. de Paula, “Atkin’s Physical Chemistry 8th Edition”, Oxford University

Press, Oxford, page 202.10] J. Wei, E. Iglesia, J. Catal. 224 (2004) 370–383.11] M.D. Porosoff, X.F. Yang, J.A. Boscoboinik, J.G.G. Chen, Angew. Chem. Int. Ed. 53

(2014) 6705–6709.12] R.B. Levy, M. Boudart, Science 181 (1973) 547–549.13] R. Burch, J. Chem Soc, Faraday Trans. I 74 (1978) 2982–2990.14] A.P.E. York, C. Pham-Huu, P. Delgallo, E.A. Blekkan, M.J. Ledoux, Ind. Eng. Chem.

Res. 35 (1996) 672–682.15] A. Katrib, V. Logie, N. Saurel, P. Wehrer, H. Leflaive, G. Maire, Surf. Sci. 377 (1997)

754–758.16] T. Matsuda, K. Watanabe, H. Sakagami, N. Takahashi, Appl. Catal., A: Gen. 242

(2003) 267–274.17] C. Bouchy, C. Pham-Huu, B. Heinrich, C. Chaumont, M.J. Ledoux, Appl. Catal., A:

Gen. 215 (2001) 175–184.18] F.C. Meunier, Chem. Commun. 15 (2003) 1954–1955.19] A. Goguet, S. Shekhtman, F. Cavallaro, C. Hardacre, F.C. Meunier, Appl. Catal., A:

Gen. 244 (2008) 30–35.20] M.J. Ledoux, F. Meunier, B. Heinrich, C. Pham-Huu, M.E. Harlin, A.O.I. Krause,

Appl. Catal., A: Gen. 181 (1999) 157–170.21] A. Benadda, A. Katrib, A. Barama, Appl. Catal., A: Gen. 251 (2003) 93–105.22] J. Houzvicka, V. Ponec, Catal. Rev. 39 (1997) 319–344.23] F.C. Meunier, F. Cavallaro, T. Le Goaziou, A. Goguet, C. Rioche, Catal. Today 112

(2006) 64–67.24] E. Iglesia, J.E. Baumgatner, F.H. Ribeiro, M. Boudart, J. Catal. 131 (1991) 523–544.25] F.C. Meunier, J.P. Breen, J.R.H. Ross, Chem. Commun. 3 (1999) 259–260.26] F.C. Meunier, J.P. Breen, V. Zuzaniuk, M. Olsson, J.R.H. Ross, J. Catal. 187 (1999)

493–505.27] F.C. Meunier, V. Zuzaniuk, J.P. Breen, M. Olsson, J.R.H. Ross, Catal. Today 59

(2000) 287–304.28] J.A. Martens, A. Cauvel, A. Francis, C. Hermans, F. Jayat, M. Remy, M. Keung, J.

Lievens, P.A. Jacobs, Angew. Chem. Int. Ed. 37 (1998) 1901–1903.29] M. Kusche, F. Enzenberger, S. Bajus, H. Niedermeyer, A. Boesmann, A. Kaftan, M.

Laurin, J. Libuda, P. Wasserscheid, Angew. Chem. Int. Ed. 52 (2013) 5028–5032.30] M. Friedrich, S. Penner, M. Heggen, M. Armbruester, Angew. Chem. Int. Ed. 52

(2013) 4389–4392.

31] J.P. Breen, F.C. Meunier, J.R.H. Ross, Chem. Commun. 22 (1999) 2247–2248.32] W. Birky, J.T. Kozlowski, R.J. Davis, J. Catal. 298 (2013) 130–137.33] G.R.M. Dowson, M.F. Haddow, J. Lee, R.L. Wingad, D.F. Wass, Angew. Chem. Int.

Ed. 52 (2013) 9005–9008.34] S. Veibel, J.I. Nielsen, Tetrahedron 23 (1967) 1723–1733.

lysis A

[[[[

[

[

F.C. Meunier et al. / Applied Cata

35] R.E. Miller, G.E. Bennett, Ind. Eng. Chem. 53 (1961) 33–36.36] C. Yang, Z. Meng, J. Catal. 142 (1993) 37–44.

37] A.S. Ndou, N. Plint, N.J. Coville, Appl. Catal., A: Gen. 251 (2003) 337–345.38] J.I. Di Cosimo, C.R. Apesteguia, M.J.L. Gines, E. Iglesia, J. Catal. 190 (2000)

261–275.39] T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J. Catal. 259

(2008) 183–189.

[

[[[

: General 504 (2015) 220–227 227

40] J. Scalbert, F. Thibault-Starzyk, R. Jacquot, D. Morvan, F.C. Meunier, J. Catal. 311(2014) 28–32.

41] J. Scalbert, F. Thibault-Starzyk, F.C. Meunier, C. R. Chim. (2014),http://dx.doi.org/10.1016/j.crci.2014.07.002 (accepted).

42] S. Ogo, A. Onda, K. Yanagisawa, Appl. Catal., A: Gen. 402 (2011) 188–195.43] K.W. Yang, X.Z. Jiang, W.C. Zhang, Chin. Chem. Lett. 15 (2004) 1497–1500.44] I.-C. Marcu, D. Tichit, F. Fajula, N. Tanchoux, Catal. Today 147 (2009) 231–238.