Heterogeneous Catalysis with Renewed Attention: Principles,Theories, and ConceptsFranck Dumeignil,*,†,‡ Jean-Francois Paul,† and Sebastien Paul†

†Univ. Lille, CNRS, Centrale Lille, ENSCL, Univ. Artois, UMR 8181 - UCCS - Unite de Catalyse et Chimie du Solide, F-59000 Lille,France‡Institut Universitaire de France, Maison des Universites, 10 Boulevard Saint-Michel, 75005 Paris, France

ABSTRACT: With the development of a strong bioeconomy sector related tothe creation of next-generation biorefineries, heterogeneous catalysis isreceiving renewed attention. Indeed, catalysis is at the core of biorefinerydesign, and many new catalysts and catalytic processes are being developed.On the one hand, they are based on knowledge acquired during the lastcentury to efficiently upgrade fossil resources. On the other hand, they takeadvantage of the opportunity of having new substrates on which novelconversion technologies, derived from recent fundamental advances and newconcepts developed in the field of heterogeneous catalysis, can be tested andapplied. In this context, there is a global trend for establishing new coursesthroughout the world to train students and professionals in the bioeconomy/biorefinery sector. Such courses encompass many different fields in amultidisciplinary approach, including agronomy, logistics, life cycle assessment,biotechnologies, process design, economics, and, obviously, chemistry. For thelast of these, a specific important role is given to catalysis, and more especially to heterogeneous catalysis. Such courses need toexplain the basics of each scientific field involved before giving concrete examples of research/development. Among them, theaforementioned heterogeneous catalysis is an especially complex scientific field in which surface science, spectroscopy,thermodynamics, and other disciplines are intimately interwoven. While many references presenting specific aspects of catalysiscan be found, concise documents encompassing all of the angles of this very rich fieldincluding its most recent developmentsand prospects, aimed at a readership already trained in chemistryare scarce. This paper gives a global and simple“heterogeneous catalysis tour” to explain in a concise way what heterogeneous catalysis is without going into too much detail.The students who would like to know more about some specific aspects will find useful authoritative references in the text,making this paper a good introduction to all aspects of heterogeneous catalysis.

KEYWORDS: Upper-Division Undergraduate, Graduate Education/Research, Continuing Education, Physical Chemistry,Textbooks/Reference Books, Catalysis

■ GENERAL CONTEXT

We have recently seen the creation of many multidisciplinarycourses, especially in the context of biorefinery development,that involve bringing together many different scientific fields. Forexample, as of September 2016 at Lille University in France, anew M2 (master’s degree) course entitled “Biorefinery” is opento students with backgrounds in chemistry and/or biochemistry.This involves, for example, students with a biochemistrybackground being able to acquire the basics of heterogeneouscatalysis largely employed in biorefineries.1 Within 2 years, thismaster’s degree offer will evolve to integrate lectures from othernatural science disciplines (such as environmental sciences,toxicology and ecotoxicology, mathematics, computer science,etc.) as well as social sciences and humanities (philosophy ofscience, ethics, economy, laws, politics, urbanism, etc.).Wageningen University in The Netherlands will open a similarinitiative, the “Biobased Sciences” master’s degree program, inSeptember 2018 by merging the M.Sc. in Biosystems Engineer-ing, M.Sc. in Biotechnology, and M.Sc. in Plant Science and

offering skills development in the “multidisciplinary design ofproduction chains including biomass production, bioconversion,biorefining and social, logistical and economic transitionprocesses”.2 In the United States, the Bioeconomy Institute ofIowa State University proposes the interdisciplinary Biorenew-able Resources and Technology graduate program.3 Thisprogram is strongly focused on biobased chemistry withadditional courses on legal aspects and policies regardingbiorenewables.As a result, students from different backgrounds need to

quickly learn the basics of various disciplines in order to be ableto participate in such courses. While some excellent referencesdealing with important aspects of catalysis are already available,for example, on practical reaction aspects4 from a historical pointof view,5 the recent scientific context and prospects are

Received: August 11, 2016Revised: February 19, 2017Published: March 30, 2017

Article

pubs.acs.org/jchemeduc

© 2017 American Chemical Society andDivision of Chemical Education, Inc. 675 DOI: 10.1021/acs.jchemed.6b00611

J. Chem. Educ. 2017, 94, 675−689

encouraging teachers to modernize courses on chemicalcatalysis.6 There is thus a need for a document introducing allof the aspects of catalysis, including its most recent develop-ments. We present here a very comprehensive yet accessible,simple, but still sufficiently technical description of heteroge-neous catalysis that will not only be attractive to students but alsoundoubtedly to some professionals in the field of chemistryconcerning bioeconomy development.Organic as well as inorganic reactions can be conducted using

catalysts, which can themselves be synthesized through organicor/and inorganic routes. The resulting catalysts are thus organic,inorganic, or mixed organic−inorganic solids (such as metal−organic frameworks (MOFs)7). As heterogeneous catalysis,which involves a solid catalyst in contact with liquid and/or gasreactant(s), is a surface phenomenon, surface science also playsan important role, and virtually all of the possible methods forcharacterizing solids can be applied. Furthermore, recentadvances in density functional theory (DFT) (computersimulations also introduced in the present paper) and moleculardynamics open the way to a finer understanding of catalyst actionand, in the future perhaps, to predictive catalytic science.

■ FOREWORDChemical reaction is at the heart of catalytic processes. Figure 1schematically shows an example of a reactant (R), lactic acid, abioderived molecule, which is converted to a product (P), acrylicacid, over a zirconia-based catalyst.

Heterogeneous catalysis involves a space- and time-dependentmultiscale approach (Figures 2 and 3, respectively) for the designand characterization of catalytic sites at the molecular levelthrough activation of the catalysts, spectroscopic characterizationof their surfaces, and subsequently their processing at the “real”industrial scale, which involves shaping the catalyst powdersbefore use in specifically designed reactors.The development of catalytic processes involves the

application of both fundamental and applied concepts (Figure2) on different scales. It encompasses the microscopic scale(bottom blue triangle in Figure 2), the mesoscopic scale (middleorange triangle in Figure 2), and the macroscopic scale (top redtriangle in Figure 2) for designing, characterizing, andimplementing (i.e., reacting) the catalysts. At the microscopicscale (corresponding to the representation in Figure 1), thechemist’s work is to design the active phase at the molecular level(the catalytic site, represented as a yellow box in Figure 1) usingproper synthesis methods. The final molecular structure is thenchecked using characterization techniques that are able to probethe local atomic environment (e.g., NMR spectroscopy or X-rayphotoelectron spectroscopy (XPS)). Theoretical chemistry issometimes used to assess the validity of the chemical structure

deduced from these characterizations. It is also at themicroscopiclevel that the reactionmechanism is studied to understand the setof chemical events involved in, for example, the conversion oflactic acid to acrylic acid and water represented in Figure 1, whichis not a one-step reaction (Figure 4; see below). These chemicalevents involve interactions of differing complexities between thechemical structure of the surface and the chemical structure ofthe reactant (lactic acid in Figure 1). Combined with othertechniques for probing the mesoscopic scale, each techniqueprovides specific information, and gathering such informationenables the most probable surface chemical state to be depicted.The mesoscopic level is a representation of the system

typically at the micrometer scale. The mesoscopic scale is thus, inother words, the representation of the catalyst as a population ofcatalytic sites gathered on a surface and working together toconvert a given amount of substrate (reactant). To generate acatalytic surface containing a certain number of active sites, thechemist uses lab-scale preparation methods (e.g., impregnation)to generate the catalyst, as it is not formally possible to synthesizeone single catalytic site but it is possible to synthesize a surfaceover which a certain number of catalytic sites are distributed.Activation methods are sometimes needed, as the solids aresometimes not reactive in their as-synthesized form. Forexample, in the case of hydrodesulfurization catalysts, the activephases must be in a sulfide form, while the solids are firstprepared in their oxide form. An intermediate step of catalystsulfidation is then needed prior to the reaction. Similarly, becausecatalyst performance decreases with time (a phenomenon knownas “deactivation”), their regeneration must be performed underspecific conditions. For instance, for biorefinery processes, one ofthe main hurdles for catalyst implementation involves theirsensitivity to deactivation due to coking (carbonaceouscompound deposition due to uncontrolled side reactions).Regeneration by coke burning under an oxygen-containingatmosphere is one of the solutions to periodically remedydeactivation. Some of the spectroscopic techniques also provideinformation at the mesoscopic scale. For example, XRD yields

Figure 1. A heterogeneous catalytic reaction typically involves theconversion of a reactant (R) to a product (P), here exemplified by theconversion of lactic acid to acrylic acid, an important industrial chemicalintermediate. This dehydration reaction releases one water molecule.

Figure 2. Schematic spatial representation of catalytic science. Adaptedwith permission from ref 8. Copyright 1995 Springer. (Also see ref 9.)

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

676

information on the crystalline phases that are present in a givensolid. These spectroscopic techniques can be used undercontrolled conditions. In situ experiments enable the catalystto be characterized while exposed to various conditions(temperature, atmosphere), while operando experiments consistof spectroscopic data acquisition during catalysis. Operandoexperiments enable spectroscopic signatures to be correlatedwith catalytic performance, and they help in accessing structure−performance relationships. Temperature-programmed reactionmethods are also useful at the mesoscopic scale. For example, thereduction profile of the solid can be recorded as a function oftemperature in order to interpret the redox capabilities of thecatalyst. Kinetics and transport phenomena (diffusion) are alsotypical features of the mesoscopic approach, through character-ization of the active-phase reactivity and the ease of movement ofmolecules to/from the active phase. The macroscopic scale,which corresponds to direct visual observation and handling ofthe catalytic processes, is mostly linked to engineering issues,such as catalyst shaping or reactor design and engineering to runthe developed catalytic systems at a larger, practical scale.Figure 3 then shows the time and length scales of the events

occurring over a catalyst. The blue boxes correspond to the

chemical reaction occurring at the surface of the solid. Theelementary step reactions occur at the shortest time and lengthscales. A global reaction is a succession of elementary steps, eachof them being a simple chemical event, i.e., a segment of thereaction mechanism. Because of the scales involved, theseelementary steps are extremely difficult to characterize. Thereaction intermediates are partly converted molecules for whichthe reaction is still ongoing. They are formed after eachelementary step. Their concentration over the solid eventuallyreaches a constant level in the steady state. This means that theirrate of formation through an elementary step and their rate ofdisappearance through the next elementary step become equal.Then, if they are present over the solid in a sufficient quantity,they can be observed by, e.g., vibrational spectroscopytechniques. Surface restructuration is intimately linked to thetwo previous items. Indeed, the catalytic sites work as a “chemicalfactory”, and the atomic organization of the surface of the catalystcan be seen as a “chemical tool” that traps the molecules andworks on them step by step (elementary-step reactions) to yield

reaction intermediates and the final molecule, which is thenreleased. The action on the working surface of an efficientcatalyst is thus quite fast and involves phenomena operating overa relatively larger length scale. Solid-state dynamics (the greenbox in Figure 3) involves the same concept. However, it is moreof a bulk phenomenon and occurs according to parameters thatare not necessarily linked to the reaction itself (response of thesolid to temperature, pressure, etc.). The red boxes are morelinked to engineering aspects. Transport in heterogeneousstructures relates to the ease with which molecules movebetween more or less packed particles and/or within more or lessnarrow, tortuous porous networks. An observed reaction rate canbe largely influenced by transport phenomena, which can hinderaccess to the active sites and decrease the apparent reaction rate.In such a case, the real chemical reaction rate, which is usuallyhigher (see the blue boxes), is masked by transport phenomenathat hinder the reaction. External heat and mass transfers are alsophenomena to be taken into account. At time t in a gas- or liquid-phase reactor, the concentration of each species and, in mostcases, the temperature, are not uniform over the reactor volume.The reaction can be exothermic, and then heat can diffuse fromso-called “hot spots” created at the reaction sites. Furthermore, aspecies concentration gradient is usually observed near thesurface of a catalyst, as, e.g., the reaction consumes reactants.Technical reactors are thus built to make the reactions performproperly over the long term according to reactor engineeringconcepts.Nowadays, with the boom in the exploitation of biomass to

yield chemicals and energy in specifically designed “next-generation biorefineries”,11,12 catalysis is encountering whatcould be called a “second golden age”. Indeed, as an evolution ofthe initial concept of a biofuel production unit as a biorefinery, itis nowadays well-accepted that in fact a biorefinery is an industrialunit converting bioresources to a variety of products, includingenergy, heat, fuel, food, feed, chemicals, and materials. This is anabsolutely necessary condition for the economics of biorefiningwithout the need for subsidies. Biorefineries actually bridge thegap between agriculture and industry, converting multiplesources of biomass to multiple products through the develop-ment and implementation of optimized technologies. Biomassconversion routes involve a combination of biological processes,thermochemistry, and homogeneous and heterogeneous catal-ysis. In France, for example, ROQUETTE13 and the industrialcomplex built along ARD14 are emblematic success stories ofbiorefineries. Large-scale projects for designing oil-plant-based(PIVERT15) and starch-based (IFMAS16) next-generationbiorefineries are also ongoing with a strong catalysis core.The use of new substrates for conversion into value-added

compounds is a good opportunity to explore newly designedconcepts while applying existing knowledge, progressivelyrefined over the last century, to exploit natural resources moreand more efficiently. Catalysis has indeed developed rapidly forthe processing of fossil fuel resources in petrorefineries. Themolecules derived from fossil fuel resources have a relativelysimple structure but are rather difficult to activate, with hence aneed for particularly efficient catalysts to activate the chemicalbonds. The problem is quite different when dealing withbiomass-derived molecules. The advent of biomass-basedtechnologies leads to a complete rethinking of catalysis, andeverything has to be reinvented, so to speak. Biosourcedmolecules notably contain a large fraction of oxygen with manychemical functions (esters, acids, ketones, ethers, alcohols, etc.),and thus, mostly multifunctionalized substrates must be treated

Figure 3. Time-scale correlation of events and technologies inheterogeneous catalysis. Adapted from ref 10. Copyright 2003 RobertSchlogl.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

677

(reactants such as 5-hydroxymethylfurfural, glucose, lactic acid,levulinic acid, etc.).12 The use of catalysts is thus more complex.Indeed, it is necessary to find catalysts capable of targeting onlyone of the functions of the molecule in order to transform onlyone part according to the desired final molecule. Moreover, thepresence of oxygen in the biosourced molecules is a factor thatincreases their reactivity. This may appear to be an advantage, butin fact, high reactivity can be detrimental to selectivity and inaddition can lead to “runaway” phenomena with loss of control ofthe reaction and generation of carbon compounds (coke) thatcompletely block the action of the catalyst. Moreover, in thecontext of a biorefinery, the nature of the source molecules and ofthe impurities present can depend on the climatic conditions atthe moment when the biomass was grown. Thus, it is necessaryto develop robust catalysts by making them tolerant to a certainamplitude in terms of fluctuations in the composition of thecharges to be treated. Finally, the presence of water in biosourcedfeedstocks is a puzzle for the chemist, as it greatly alters theproperties of the catalysts and, hence, their chemical action. Thisdeep questioning opens the way to a new golden age of catalysisresearch.In the following, we shall explore all aspects of heterogeneous

catalysis, with a final section on the future of this scientific field,which is undergoing considerably renewed interest in the currentcontext.

■ CHEMISORPTION AND REACTION KINETICS

Basics of Chemisorption and Thermodynamics

The basis of heterogeneous catalysis is chemisorption. This is aprocess that involves chemical bonding (of varying strength) of amolecule present in the liquid or the gas phase (the adsorbate)with a surface (the adsorbent), namely, the surface of a catalyston which the active sites are located. This is exemplified in Figure4 with a reaction of which the first step is actually the adsorptionof lactic acid onto a zirconia surface. In this case, adsorptionconsists of an acid−base interaction between a Lewis acid site(vacancy on a Zr atom) and the nonbonding doublet of thealcohol function of lactic acid, with formation of a surface bond.Unlike physisorption, which involves weak interactions (van derWaals forces), chemisorption is a chemical reaction and thus

usually involves an activation energy (represented by Ea in Figure5).

The thermodynamics of adsorption can be described in simpleterms by the following equation: ΔrG = ΔrH − TΔrS (where thesubscript “r” refers to a reaction, in this case adsorption). ΔrGobviously has to be negative to favor adsorption, and ΔrS is alsonegative, as the consequence of chemisorption is to reduce thedegree of freedom of the molecule. Thus, ΔrHmust be negative,which means that the reaction is exothermic in these cases.However, the heat of adsorption is most often represented by Q,which is positive (the absolute value of ΔrH). Q is usually largerfor chemisorption compared with physisorption (see QC and QPin Figure 5). Typically, and very approximately,QP is in the rangeof 10 to a few tens of kilojoules per mole, while QC more oftenranges from a few tens to a few hundreds of kilojoules per mole.For example, the physisorption heat of n-butane over baregraphite is about 12 kJ mol−1, while it can reach about 60 kJ mol−1

for physisorption of n-heptane over a zeolite. The chemisorptionheat of lactic acid over a ZrO2 surface (Figure 2) is rather low atabout 47 kJ mol−1, while the chemisorption heat of oxygen over aW(100) surface reaches ca. 580 kJ mol−1.

Figure 4.Mechanism of the conversion of lactic acid to acrylic acid with release of a water molecule (dehydration) over a zirconia surface. These stepscorrespond to the set of chemical reactions occurring over the active catalytic sites, schematically represented by a yellow box in Figure 1. In the ball-and-stick models, the blue atoms represent C, the yellow atoms represent H, the red atoms represent O, and the orange atoms represent Zr. Adapted withpermission from ref 17. Copyright 2013 Elsevier.

Figure 5. Energetic representations of chemisorption and physisorptionas functions of the distance between the adsorbate and the adsorbent.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

678

Similarly, the equilibrium distance between the molecule andthe surface is shorter for chemisorption compared withphysisorption (see dC and dP in Figure 5). In addition,physisorption is a multilayer phenomenon, as after saturationof the whole surface by the adsorbate, other molecules canphysisorb on top of the adsorbate layer. This is not the case forchemisorption, which involves a reaction between the surfaceand the molecule and which also means that chemisorption isselective to adsorbent−adsorbate pairs, while any molecule canphysisorb onto any surface.Adsorption Isotherms and Modeling

In order to characterize chemisorption, isotherms can bedetermined experimentally. The catalyst is brought into contactwith the adsorbate under controlled conditions at a constanttemperature. This implies using small quantities of the solidcatalyst in order to avoid a significant heat rise due to theexothermicity of the reaction. The quantity of absorbateadsorbed can be determined by volumetric, gravimetric, oreven spectroscopic methods.18 In the example given in Figure 6,

the quantity of a given adsorbed molecule, q, is recorded atvarious adsorbate pressures P over a given quantity of solid at agiven temperature. q can be expressed as the number of moles(n), volume (V), or mass (m), these parameters being linkedthrough the equations PV = nRT and n = m/M, where M is themolar mass of the adsorbate. The isotherm is asymptotic, as thereare a limited number of adsorption sites on the solid, and thus, allof the sites are covered when the adsorbate pressure applied tothe solid is sufficiently high. A maximum quantity of adsorbate,qMax, can be determined, and a very useful parameter is defined asθ = q/qMax. θ represents the surface coverage and thus rangesbetween 0 and 1, with 0 meaning a bare surface and 1 meaning asurface for which all of the adsorption sites are covered.The adsorption reaction can be written simply as A(g) + *⇄

A*, where A(g) represents an adsorbate molecule in the gasphase, * a free site, and A* an adsorbed A molecule. In somecases, A can require two adsorption sites. For example, H2 candissociatively adsorb onto two sites to yield two H*. Differenttypes of adsorption can occur. Depending on the adsorbent, theadsorbate, and the experimental conditions, different models canbe applied to describe the measured isotherms, namely, theLangmuir, Temkin, and Freundlich models, which considerdifferent types of site distributions according to the heat ofadsorption profile as a function of the surface coverage (Figure7).In the Langmuir model, which is the simplest one, the heat of

adsorption is considered to be constant irrespective of the surfacecoverage. In the Temkin model, the heat of adsorption linearlydecreases with the surface coverage according to the equation Q= Q0(1 − αθ), where α is a fitting parameter used to describe the

linear decrease of the heat of adsorption with θ (the slope of thepurple line in Figure 7). Indeed, this model considers anonuniform surface, with a distribution of catalytic site strengths.The “strongest” sites, i.e., the sites having the largest affinitytoward the molecules to be adsorbed, will be naturally coveredfirst. Then the next adsorbates will interact with the remaining“stronger” free sites, and so on. The Freundlich model alsoconsiders a nonuniform surface, but with a more complexdistribution of sites, expressed by the equation =n n ei

Q Q0

/e,i e0,which is used to fit the ni site distribution according to their heatsof adsorptionQe,I. The Freundlich model simulates a surface thatcontains a certain number of strong sites (the first part of the bluecurve in Figure 7) and then an exponential decrease in the active-site strength distribution. These three models are actuallyrepresentative of the encountered types of solids, and in any case,it will be possible to represent the behavior of any catalyst usingone of them. In other words, the distribution of catalytic sitesover a solid will obey one of these laws. It is also noteworthy thatin addition to the potential active-site strength distribution, thepresence of already adsorbed molecules can have an influence onthe further adsorption of new neighbors as a result of specificinteractions.Mathematical treatment provides equations for the isotherms.

The Temkin and Freundlich cases are quite complex, and we willonly comment in detail hereafter on the equations linked to thesimple Langmuir model. For the simplest model, the Langmuirmodel for nondissociative adsorption, the equation for the

isotherm can be written as θ = λλ+

PP1or = λ

λ+q q PPMax 1, in which

P is the pressure of the adsorbate, q is the adsorbed quantity, qMaxis the maximum quantity that can be adsorbed, and λ is theequilibrium constant [it should be noted that with this definition,λ has actually dimensions of P−1; the real equilibrium constantKTis dimensionless because, in that case, the pressure is taken at thethermodynamic reference pressure P0 = 1 bar, while λ has adimension]. Linearization of the equation gives a straight linewhen plotting 1/q = f(1/P) if the system indeed obeys theLangmuir model, with the possibility of determining λ and qMax.Similarly, a straight-line dependence is observed when plotting1/q = f(1/P0.5) for the Langmuir model in the case of dual sites(dissociative adsorption), q = f(ln P) for the Temkin model, andln q = f(ln P) for the Freundlich model. The catalyst surface canbe complex, and the treatment of the isotherms can involvedifferent types of catalytic sites corresponding to different

Figure 6. Typical shape of an adsorption isotherm.

Figure 7.Models usually used to describe adsorption and correspondingdistributions of heat of adsorption or sites.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

679

models. Usually, if one wants to finely characterize catalytic sites,it is interesting to find a molecule whose structure is close to thatof the molecule to be reacted but that only adsorbs withoutreacting, which is not an easy task. In other words, if one wouldlike to characterize the catalytic sites involved in the trans-formation of lactic acid to acrylic acid (Figure 1), it would beconvenient to use lactic acid as a probe. However, if the catalysthas been designed to make lactic acid react, the surfacephenomena will be much more complex than what is desiredin that case (just adsorption). One should then find a moleculethat “looks like” lactic acid, so that it would supposedly adsorb inthe same way as lactic acid would while not being subsequentlytransformed by the catalyst. In that specific case, short carboxylicacids and alcohols that do not undergo dehydration (e.g.,isobutanol) could be pertinent choices. In a general way, for thesekinds of experiments, carefully selected probe molecules can beused to characterize acidic, basic, metallic sites, etc. Then, comingback to the example of the dehydration of lactic acid to acrylicacid, it is reasonable to first state that the catalytic sitesresponsible for dehydration are acidic sites and thus that basicprobe molecules can be used, such as pyridine or NH3. However,one must keep in mind that, depending on the selected probemolecule, the steric hindrance can be different than that of thereactant it is supposed to simulate and the number of actuallyaccessible sites might be different (e.g., as NH3 is a smallmolecule, it could reach acidic sites in narrow pores that lacticacid could not reach).Adsorption Kinetics

Usually, two different models can be used to describe adsorptionkinetics, depending on the catalytic system studied. The first oneis based on Langmuir adsorption:

+ * *X YoooA Ak

k(g)

des

ads

The reaction rate is then given by

θ θ= − −r k P k(1 )ads des

When the global adsorption rate is zero, the system is atequilibrium, and we naturally again find the Langmuir equationwith

λ θθ

= =−

kk P(1 )

ads

des

The second model is the Elovich model. In a commonlyobserved chemisorption process, a relatively slow uptake ofadsorbate follows a rapid initial uptake, which is described asactivated adsorption. The Elovich mathematical treatment givesthe following equation: q = C ln(t + t0) + D, in which q is theadsorbed quantity, C and D are constants, and t0 is a “time lag”that is usually found by iterative fitting of measured data to alinear function assuming some arbitrary initial value for t0.An interesting illustration of the usefulness of this kind of study

was recently published by Lalikoglu et al.19 These authorsactually put the focus on the separation process of lactic acid in afermentation broth in which it is highly diluted in water. Onepossible process to separate lactic acid from water is to adsorb itselectively onto a well-chosen adsorbent. In this study, layereddouble hydroxides were considered to play this role. In a firststep, experimental data were collected to plot the adsorptionisotherms, and the different models described above (Langmuir,Freundlich, and Temkin) were tested. It turned out that the bestfit was obtained for the Langmuir model. In a second part of this

study, the adsorption data were applied to various adsorptionkinetic models, among which was the Elovich model. The kineticparameters were determined from the experimental data, and thebest model was identified. This work provides the conclusionthat the layered double hydroxides are suitable adsorbents for theseparation of lactic acid from aqueous solutions.Reaction Kinetics.8,20

Kinetics is a very important but extremely complex topic incatalysis. While purely numerical models (mathematical curvefitting) can be built to predict the behavior of working catalystsover certain rather narrow ranges of pressure, temperature, andconcentration, etc., it is interesting from a scientific point of viewto understand all of the elementary reaction steps involved inbuilding amodel. In practice, it is necessary to establish equationsaccording to the possible reaction pathways, etc., and then toverify whether the outputs from the model match theexperimental results. Except for very simple reactions, it isextremely rare to get a definitive answer for the reaction ratemodel because of continuous refinement/understanding of thesequence of elementary reaction steps. Nevertheless, under-standing of the proposed reaction pathways is gradually beingimproved as research advances. Figure 4 represents, for example,the set of elementary steps involved in the dehydration of lacticacid to acrylic acid (step 1, formation of a carbanion by hydrogenabstraction; step 2, formation of acrylic acid; steps 3 and 4, watermolecule release), as validated by DFT results combined withexperimental results.20

Some of the basics of heterogeneous kinetics are presented inthe following, and more detailed information can be foundelsewhere.25 We first consider the global process reactant(s) →adsorbed reactant(s) → product(s). The steady-state assump-tion (SSA) is used when the amounts of intermediates present inthe reaction are low (i.e., when their concentrations are muchlower than those of the major species in the mechanism). Then,as the number of moles in the gas phase is much larger than the

adsorbed quantity, this formally yields = 0Nt

dd

ads (with Nads

representing the number of adsorbed molecules), and the

reaction rate is then =ν

rm

Nt

1 1 ddi

i

cata, in which νi is the algebraic

stoichiometric coefficient associated with the ith species(reactant or product), mcata is the mass of catalyst, and Ni is thenumber of moles of the ith species. Compared with purelyhomogeneous kinetics, it should be noted that the mass of solidmust be considered in the equation, which is obvious because thereaction rate is dependent on the catalyst action and thus thequantity of catalyst.We shall now consider the following simple sequence of

reactions, where A is first adsorbed onto the catalyst surface, thesurface transformation of A* to X* occurs, and X* then desorbedto give the final gas-phase product X:

+ * *

* ⇄ *

*

X Yooooo

X Yooooo

A(g) A

A X

X X(g)

k

k

k

k

k

k

des(A)

ads(A)

2

1

ads(X)

des(X)

Even in such a simple case, rigorous mathematical treatmentresults in a very complicated equation. Identifying the limitingstep, i.e., the rate-determining step (RDS), helps greatly in

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

680

obtaining more simple, directly exploitable equations. In thepresent case, if we suppose that adsorption of A is the RDS, thatthe third reaction is at equilibrium (SSA, in which

= 0t

d(Intermediate)d

), and that the surface reaction goes to

completion (k2 = zero), then

λ=

+r

k P

P1ads(A) A

X X

in which λX = kdes(X)/kads(X) (Langmuir model). It should benoted that accumulation of the product lowers the reaction rate(λXPX is in the denominator).As a last illustrative example, we consider a bimolecular surface

reaction, A + B → product(s), in which the RDS is the surfacereaction, in three different cases. The first two cases involve theLangmuir−Hinshelwood model, which implies a reactionbetween adsorbed A and adsorbed B species. In case (1) wehave competitive adsorption, which means that A and B adsorbon the same type of sites, resulting in r = kθAθB. In case (2) wehave adsorption on two types of sites, meaning that A and Badsorb independently on two distinct types of sites. This gives r =kθA1θB2. Case (3) is the Eley−Rideal model, which has thefundamental difference with the Langmuir−Hinshelwood modelthat the reaction occurs between an adsorbed A molecule and agas-phase B molecule, which gives r = kθAPB. If we now considerthe initial kinetics (where the concentration of products isnegligible), we obtain the following equations for the three cases:

(1) Langmuir−Hinshelwood competitive:

λ λλ λ

=+ +

r kP PP P(1 )0

A A B B

A A B B2

(2) Langmuir−Hinshelwood noncompetitive:

λλ

λλ

=+

×+

r kP

PP

P1 10A A

A A

B B

B B

Eley−Rideal:

λλ

=+

×r kP

PP

10A A

A AB

As a variant, in the case “A → product(s)” in which theadsorbed molecule A* reacts using a free (“vacant”) site (A* +

*), r = kθA(1-θA), which yields = λλ+

r k PP0 (1 )

A A

A A2 .

Further, the so-called Mars−van Krevelen mechanism21 is alsoa very important mechanism typically encountered in oxidationreactions. Its concept can be further applied to describe otherreactions over heterogeneous catalysts, such as hydrodesulfuriza-tion or deoxygenation.22 Figure 8 is a schematic representation ofthe Mars−van Krevelen oxidation mechanism. For example, thereactant R can be acrolein, which is oxidized to acrylic acid, theproduct P, through the action of an oxided active site, whichtransfers an oxygen species to R, leaving a reduced active site onthe surface of the catalyst. In theMars−van Krevelenmechanism,the unactivated reduced active sites are regenerated using O2added to the gas phase. The interaction of O2 with the catalystsurface can yield the desired O2− species as well as undesired O2

−

species, which are nonselective and destructive by overoxidationof the R substrate. Depending on the support, the desired O2−

species can be stored in the bulk of the catalyst, below the surface,and migrate to the reduced active sites to reoxidize them, makingthem readily available again for oxidation of R to P. Thus, the

capacity of the support to store O2− (the “oxocapacity”) and themobility of the O2− anions in the bulk (the ion conductivity) arealso very important parameters for the catalytic activity. Forexample, CeO2 is a support with an excellent oxocapacity and ahigh ion conductivity. CeO2 is thus, e.g., used in the catalyticexhaust converters of cars to oxidize unburned hydrocarbonseven under “lean” conditions, i.e., when the exhaust gas does notcontain much O2. Further, as redox reactions involve electronexchanges, the electron conductivity within the solid is also animportant parameter.The global kinetics derived from the Mars−van Krevelen

mechanism is then complex, depending on a few different rates:the rate of oxidation of R to P, the rate of reoxidation of thereduced sites to the active oxidized sites, the rate of conversion ofO2 species to O

2− species, and the O2− species mobility from/tothe bulk of the catalyst. Each of these steps can be the limitingstep, with the lowest rate, imposing its rate on the whole process.As previously mentioned in this section, complex reaction

schemes involving, e.g., parallel reactions, competing reactions,or sequential reactions can occur on a solid, which lead to verycomplex kinetic treatments. Among the approximations that canbe used to simplify the system, the most abundant reactionintermediate (MARI) approximation can be helpful: when oneintermediate is much more abundant than all of the others, thecoverage of the other intermediates can be neglected in the globalcoverage balance: 1 = ∑iθi = θv + θMARI + ∑jθj ≈ θv + θMARI.Furthermore, when needed, computer-assisted kinetics modelsnowadays enable extremely complex cases to be treatedefficiently.A very good illustration of this kind of approach can be found

in a recent paper published by Gonzalez-Borja and Resasco.24 Inthis work, the alkylation ofm-cresol with isopropanol over an HYzeolite was studied in the liquid phase. Experimental data werecollected and fitted to Langmuir−Hinshelwood and Eley−Ridealkinetic models, which allowed the adsorption and rate constantsto be estimated. A deeper understanding of the mechanism of thereaction was gained even if in that specific case, both modelsoffered good-quality fits.

Figure 8. Schematic of the Mars−van Krevelen mechanism. Adaptedwith permission from ref 23. Copyright 2011 Royal Society ofChemistry.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

681

Factors Potentially Masking the Real Catalyst Activity

A. Chemical Regime versus Diffusion Regime. Diffusionis a physical phenomenon that can detrimentally influencechemical reactivity by slowing down the transfer of moleculesbetween the liquid or gas phase and the catalytic site. Figure 9

schematically represents the path of gas or liquid reactants andproducts in a tubular reactor packed with a catalytic bedmaintained over a frit. The reactants and products must firstcirculate between the catalyst particles within the porosity of thebed [Figure 9(1)]. Too-tight packing, for example, can inducedifficulties in the flow through the catalyst bed and can generatepressure losses. In any case, a laminar film of fluid is formed at thesurface of the catalyst particles. Depending on the nature of thefluid and the operating conditions, this film can impose a strongresistance to mass transfer. This is called external diffusionlimiting. Then, after having overcome this first barrier, thechemicals must enter and circulate within the porous catalystnetwork [Figure 9(2)] in order to access the inner active sites onwhich the surface reaction proceeds [Figure 9(3)]. Too-smallpore diameters or too-tortuous internal structures can hinder thesmooth circulation of the chemicals (“internal diffusion” issue).Furthermore, depending on the porous structure (tortuosity,pores with only one opening, etc.), reactants and products caneven circulate in opposite directions to enter into or to escapefrom the inner part. Thus, when measuring reaction kinetics inorder to understand the reaction mechanism, it is necessary to besure that the system is not subject to diffusion limitations and thatthe measured performance is thus fully representative of itsintrinsic chemical reactivity. Proper design of the morphologyand shaping of the catalyst, optimization of its packing, use ofspecific flow conditions (e.g., creation of turbulences), andselection of a proper reactor type/geometry can help tackleexternal diffusion limitation issues. Internal diffusion can belimited by choosing/designing catalysts with pores sufficientlylarge compared with the size of reactants/products and withoptimized porous structures. Experience with the fluid dynamicsof the problem, together with dedicated experimentalverifications, can ensure that a purely chemical regime can beachieved.25

B. Deactivation. Even if a catalyst can be precisely designedex situ, its behavior can certainly be modified under the reactionconditions. A catalyst can be beneficially activated under the

reactant stream, but sooner or later any catalyst will encounterdeactivation, which means degradation of its performance overtime.Usually, after an induction phase (either a decrease in

performance, as the first molecules contacting the catalyst areall converted before stabilization of the reactant flow rate when asubstantial number of molecules pass through the catalyst, whichcannot treat all of them, or an increase in performance, due toactivation under the reactant stream) at the end of which a steadystate is reached, a slow decrease in performance is observed(Figure 10). We can distinguish two deactivation mechanisms:

(1) Deactivation linked with the reactivity of the catalytic sites(chemical modification). This can involve (i) anunmodified solid composition with a phase transition orsegregation or (ii) a change in the composition of the solidby reactions with the reactants, products, or impuritiesand/or by elimination of some components from thecatalyst, by the deposition of solid impurities, or evenworse, by the adsorption of some compounds thatreversibly or irreversibly alter the activity of the catalyticsites (catalyst poisoning).

(2) Deactivation linked with the catalyst morphology(physical modification), with changes in diffusion proper-ties, decrease in specific surface area, etc.

In general, the deposition of coke (carbonaceous compounds) isthe main cause of deactivation for organic reactions. The catalyticperformance can be at least partially recovered usingregeneration/rejuvenation processes such as moving bed reactortechnology for the well-known fluid catalytic cracking process orother methods, including in situ protocols, in which the catalyst iscontinuously regenerated while performing the reaction, e.g., bythe introduction of small quantities of O2 to burn the cokes.26

■ SYNTHESIS AND CHARACTERIZATION

Chemical Nature of the Catalyst

Any solid with chemical properties, including natural solids, canpotentially find catalytic applications. It is thus impossible to givehere an exhaustive list, but just a broad overview, as we can findmany different families of catalysts. Depending on the reaction,they may contain acidic or basic sites (Brønsted or Lewis), redoxsites, metallic sites, etc., or a combination of these sites whenmultifunctional properties are needed, which is the case, e.g., forthe so-called Guerbet reaction. In the Guerbet reaction, alcoholsare dimerized to long-chain alcohols (e.g., the synthesis ofbutanol from ethanol). The most generally described sequenceof reactions involved is quite complex, with a first step ofoxidation over redox/acidic−basic sites, followed by aldolcondensation of the as-formed aldehydes over basic sites to

Figure 9. Circulation of reactants and products (1) between and (2)into the catalyst particles and (3) surface reaction on a catalytic site.

Figure 10. Typical evolution of conversion as a function of time (notethat induction can also be positive in case of the activation of the catalystunder the reaction conditions).

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

682

give an aldol. The latter is further dehydrated over acidic sites andfinally hydrogenated over redox/acidic−basic sites to form thefinal higher alcohol.27

The catalyst itself can be exclusively composed of the activephase (bulk catalyst), or the active phase can be dispersed over an“inert” support (supported catalyst). Indeed, textural, structural,morphological, and even mechanical properties are veryimportant features. Usually, an optimal dispersion/distributioncan be found with a maximal number of available active sites. Theactive phase is then spread over supports, usually oxides, themostcommon one being γ-alumina, although silica, titania, zirconia,and similar materials are also used. The supports contain porousnetworks in which small layers (or particles) of the active phaseare deposited to optimize their reactive surface.Oxides themselves can be catalysts. For example, γ-alumina is

an acidic solid, catalyzing dehydration reactions. Mixed ormulticomponent oxides are also used for specific chemical and ormechanical properties. Furthermore, recent methods haveallowed oxide templates to be formed with periodic nano-designed porous networks (mesostructured solids), such as SBA-15 (honeycomb-like silica) or the KIT6 structure, and evenhierarchical materials with well-defined, multiscale, porousnetworks.28,29 In this respect, zeolites are also ordered solidsthat are useful in catalysis.It is not possible to cover all of the possibilities here, but we can

also mention (multi)metallic nanoparticles (Au, Pd, Pt, etc.),solid solutions, MOFs, nano-oxides, hydrotalcites, hydroxyapa-tites, perovskites, heteropolycompounds (also called polyox-ometalates), etc. The possible variations are almost unlimited,and it is the role of the chemist to find/design the “best”formulation for a given reaction. For example, Clerici andKholdeevea30 recently published an overview of liquid-phaseoxidation that can be performed on a wide range of suchdedicated solids.It should be noted that identifying the phenomena involved

during synthesis can be crucial when designing the active phasethat will be used in the catalytic process.

Synthesis Methods.31−33

Many methods can be used to synthesize this wide variety ofpossible heterogeneous catalysts. Basically, any method thatyields a solid can be envisaged. To prepare bulk catalysts, somecommon methods are (i) precipitation or coprecipitation, inwhich precursor solutions are typically precipitated by controlledpH variation; (ii) the sol−gel method, during which organic(alkoxides) or inorganic compounds are hydrolyzed to a gel viathe intermediate formation of a colloidal suspension (a “sol”);(iii) hydrothermal (or solvothermal) synthesis, during which asolution of precursors (nitrates, acetates, etc.) is processed byheating it to a high temperature under the condition of anautogenous increase in pressure (this method is used, e.g., tosynthesize zeolites and MOFs), and (iv) flame hydrolysis. Thefinal catalyst is usually obtained after a few steps, includingwashing and purification, and, most of the time, a drying step anda calcination step, with optionally an activation procedure(usually involving a controlled atmosphere pretreatment).In the case of supported catalysts, most often the support (or its

precursor) is first prepared (and optionally shaped), and theactive phase is then deposited by various methods:

(i) Impregnation is a very commonmethod that is based on aninteraction between a precursor(s) solution and the solidsupport. For incipient wetness impregnation, the water porevolume (or the pore volume accessible to any other

solvent) of the solid to be impregnated is first measured,and then the desired amount of active-phase precursor isdissolved in the exact quantity of solvent needed tocompletely fill the pores of the solid. In this method, thequantity of active phase that can be deposited in oneoperation can be limited by the solubility of theprecursor(s). In excess solvent impregnation, this solubilitylimit is overcome, as the support is immersed in a solutionin large excess containing the precursors. In this secondapproach, an equilibrium between the species in solutionand the species chemically bound to the support isreached. However, as a drawback of this second option,when the solvent is evaporated, some undesired “bulk”species can precipitate onto the support.

(ii) Chemical vapor deposition (CVD) is a method in which abulk catalytic phase is sublimated. The as-obtainedgaseous precursor is then brought into contact with thesupport, over which it solidifies as thin layers.

(iii) Other methods include/involve solid−liquid interfacialchemistry (oxide surfaces, surface acidity, and charge), ionexchange and equilibrium adsorption, grafting, immobili-zation of metal particles and clusters, deposition−precipitation, spreading and wetting, and heterogenizationof homogeneous catalysts (anchoring).33

It should be noted that in some more rare cases the supportand the active phase can be prepared together through suitablespecific protocols. Generally, the role of the support is not limitedto being a simple carrier, as most of the time its presence has aninfluence on the reactivity of the active phase deposited on it.In general, a thermal post-treatment (of varying complexity) is

performed to eliminate residual organic/inorganic compounds inorder to form the desired phase, to activate the catalyst, etc.Textural Characterizations34,35

N2 physisorption is conventionally used to characterize thetextures of solids. Sorption/desorption isotherms are recorded atthe temperature of liquid nitrogen (77 K), and hysteresis curvesare usually observed as a result of capillary condensationphenomena (in mesopores). The shape of the obtained curvesprovides useful information on the global type of porous network(cylindrical, ink-bottle, “V”-shaped, etc.), but more detailedinformation is obtained using mathematical treatments based onvarious models. The most popular one is the Brunauer−Emmett−Teller (BET) model, which enables the specific surfacearea (m2·g−1) to be calculated. The Barrett−Joyner−Halenda(BJH) method is useful for specifically characterizing the poresize distribution in themesopores range (2 nm< d < 50 nm). TheLangmuir method is used to characterize micropores (d < 2 nm),for which the volume can be calculated bymeans of the Harkins−Jura equation. Macropores (d > 50 nm) are usually probed usingmercury porosimetry.Many other calculation methods have been developed to

interpret the experimental results, and other possible exper-imental methods include capillary flow porometry and electro-acoustic measurements.Structural Characterizations30,36

Solid-state characterizations are of utmost importance whendeveloping catalysts, especially to identify the nature of the activephase and to establish structure−reactivity relationships. Theycan be performed in situ on the dried or calcined catalyst37 afterpretreatment (reduction, oxidation, sulfidation, etc.) in a specificchamber working under controlled conditions (temperature,atmosphere, pressure) or under in operando conditions (see

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

683

below). In addition to “conventional” spectrometers, new so-called “environmental” spectrometers are becoming more andmore popular. This concerns techniques that are traditionallyperformed under ultrahigh vacuum, such as XPS (10−10 Torr),but are now technically feasible at pressures of a few Torr,providing the possibility of observing adsorbed molecules. Forinstance, using environmental XPS during ethanol steamreforming (production of H2 from ethanol/H2O mixtures)over a NiCeO2 catalyst made it possible in particular to observe

the surface adsorption/dissociation of ethanol molecules and tounderstand the influence of water molecules on this process.Furthermore, the formation of carbonates during reactions wasalso observed. Such species were not detected when theexperiment was performed in a conventional ultrahigh-vacuumXPS spectrometer, which illustrates the huge potential ofambient-environment analytical instruments.38

In Table 1 we have listed some of the primary characterizationtechniques used in this field. It is important to recall here that

Table 1. Overview of the Main Characterization Methods Used in Heterogeneous Catalysts, with the Corresponding Acronymsand Some General Remarks

Technique Acronym Type Remarks

Laser Raman spectroscopy LRS Vibrational,essentially bulk

Resonance Raman (RR), using a tunable laser, provides a 106-fold intensityenhancement.

Surface-enhanced Raman spectroscopy (SERS) occurs in the neighborhood ofnanoparticles of Ag, Au, Cu with a 103−106-fold intensity enhancement.

Useful for characterization of crystalline and amorphous phases.(Fourier transform) infrared (FT)-IR Vibrational,

mainly bulkUsually, strong absorption of the support in the 500−1000 cm−1 range is observed.

Sum frequency generation SFG Interfacial Yields composition and structural information on interfaces between differentmedia.

Attenuated total reflectance ATR Vibrational,mainly bulk

IR spectroscopy variation allows thick or opaque media to be analyzed.

Diffuse reflectance spectroscopy/diffusereflectance infrared Fourier transformspectroscopy

DRS/DRIFTS

Vibrational,surface-sensitive

IR spectroscopy variation allows the surface of powdered samples to be probed.Very sensitive to low concentrations of vibrating species.

Ultraviolet−visible UV−vis Absorption orreflectance,bulk

Observation of transitions from the ground state to the excited state, characteristicof local environments.

Electron spin resonance/electron paramagneticresonance

ESR/EPR Bulk Characterization of local environment of paramagnetic species.Various multidimensional sequences exist.

Magic-angle spinning nuclear magnetic resonance MASNMR

Bulk Solid-state NMR spectroscopy.Possible on any isotope that has a quadrupole moment (e.g., 27Al) to characterizelocal environment of atoms.

Various multidimensional sequences exist.The recent development of the dynamic nuclear polarization (DNP) techniquebridges the gap between NMR and ESR spectroscopy.

X-ray diffraction XRD Bulk Transmission/reflection mode.Provides identification of crystallographic phases.Particle size calculations are possible.

(High-resolution) transmission electronmicroscopy

(HR)TEM

Bulk, undervacuum

Recent developments in tomography (3D imaging).Magnification up to ∼1,500,000 times.Determination of size, shape, and distribution of nanoparticles.

Scanning electron microscopy SEM Surface, undervacuum

Magnification up to 500,000 times.Basically shows the morphology of catalytic objects (grains, particles, etc.).

Atomic force microscopy AFM Surface, undervacuum

Atomic resolution.Only for flat surfaces and then limited to model catalysts.

X-ray absorption near-edge structure/extended X-ray absorption fine structure

XANES/EXAFS

Bulk Techniques using synchrotron radiation.Access to local environment of atoms and evaluation of distances betweenneighbors.

Inelastic neutron scattering INS Bulk Crystallographic information.Hydrogen species can be observed (impossible with XRD).

Low-energy ion scattering LEIS Surface, undervacuum

Provides access to the composition of the first atomic layer of a solid sample.Possibility to alternate erosion/analysis of the material to obtain progressive depthprofiling.

(Time of flight) secondary ion mass spectroscopy (ToF)-SIMS

Surface, undervacuum

Provides composition and local arrangement of atoms over a few nanometers.Erosion/analysis possible.

X-ray photoelectron spectroscopy/electronspectroscopy for chemical analysis

XPS/ESCA

Surface, undervacuum

Provides composition and information on local environment of atoms in the first∼10 nm.

Erosion/analysis possible.Information from Auger electron emission is also sometimes obtained.Variant: ultraviolet photoelectron spectroscopy (UPS).

Inductively coupled plasma atomic emissionspectroscopy

ICP-AES Bulk Used to determine bulk chemical compositions.Destructive analysis.

X-ray fluorescence XRF Bulk Nondestructive elemental analysis, but lower resolution compared with ICP-AES.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

684

some characterization techniques probe the bulk of the wholesolid, while catalysis is a surface reaction. Thus, even if thesetechniques bring valuable information, one must be prudentwhen interpreting the data obtained, as they are not necessarilyrepresentative of the active phase but of the whole solid structure.Furthermore, the active phase can be very different underworking conditions compared to ex situ or even in situ analysisresults, and only in operando analysis can provide a definitiveanswer (see below).

Thermoprogrammed Methods and Adsorption of ProbeMolecules

These methods are conventionally used to probe surfaceproperties. Thermoprogrammed reduction (TPR), usually withH2 (H2-TPR), and thermoprogrammed oxidation (TPO) areused to assess the redox properties of the solids by giving an ideaof their reduction/oxidation capacity as a function of temper-ature. Acid−base properties can also be probed. Routinetechniques to roughly assess the distributions of the quantityand strength of acidic and basic sites include NH3 and CO2

thermal programmed desorption (TPD), respectively. However,methods based on IR spectroscopy39 are useful if information onthe types of sites (Brønsted, Lewis) is also required.40 Forexample, IR observation of pyridine, lutidine, or CO providesvaluable information on acidic sites, including their nature,strength, and distribution. It should be noted that micro-calorimetry techniques are complementary, as they can also yielddetailed numerical information (in kJ mol−1) on the strength anddistribution of basic/acidic sites.CO adsorption followed by IR spectroscopy is also a very

useful technique for observing metallic sites, with muchinformation that can be deduced, such as the number of sites,the dispersion, and even the shape of the nanoparticles, etc.Slightly different concepts are differential thermal analysis

(DTA) and thermogravimetric analysis/differential scanningcalorimetry (TGA/DSC) techniques, which yield informationon phase transitions with temperature and are then largelycomplementary to XRD (which can also work in a

thermoprogrammed mode using specifically equipped diffrac-tometers).Basically, the researcher has a large choice to probe a variety of

different types of sites, including those described above, andmethods of varying sophistication can be used (or evenspecifically developed with probemolecules) to give the expectedlevel of information.

Operando Spectroscopy

Operando spectroscopy is a concept that has been developedrelatively recently. It consists of observing in a single experiment,and at the same time, the catalytic performance (conversion andselectivity) and the spectroscopic signatures of the solidfunctioning under real reaction conditions, also including theobservation of any species present in the system (reactionintermediates among others; see Figure 11). It therefore involvesthe design of specific catalytic reaction cells with walls that aretransparent to the analyzing beam to enable direct observation.The recent developments of operando spectroscopy tend tocombine different spectroscopic techniques in a single experi-ment (by, e.g., recording an XRD diffractogram simultaneouslywith a Raman spectrum during the catalytic act).Operando spectroscopy must not be confused with in situ

techniques. Indeed, the latter do not involve the simultaneousmeasurement of the catalytic performance under real catalyticconditions. Operando spectroscopy is thus a very powerful newanalysis tool, as the features of the catalyst working under realconditions in the presence of the reactants/products are oftenvery different from those traditionally observed ex situ. To date,operando spectroscopy has been mostly designed for gas-phasereactions, but is more andmore applied to the liquid phase and totrickled bed (triphasic: liquid, gas, solid) reactions, which aretaking on great importance in the context of biorefinerydevelopment. The next step is probably to design time-resolvedequipment to try to match the duration of the spectroscopicmeasurement with the reaction kinetics, which is a verychallenging task (see Figure 3). Figure 11, which was adaptedfrom a remarkable document available elsewhere,41 provides anoverview of the species that can be observed during operando

Figure 11. Schematics of the operando spectroscopy principle, outputs, and applications to predictive catalysis and expert systems. Adapted withpermission from ref 41. Copyright 2003 The PCCP Owner Societies.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

685

experiments and how they can help in understanding thequantitative structure−activity relationship (QSAR).For example, operando Raman spectroscopy was used to

observe and improve the dehydration−condensation reaction ofmethanol to form dimethyl ether over Keggin heteropolyacids.42

In that study, the authors could enhance the conversion byproperly tuning the catalyst pretreatment, i.e., the treatment ofthe catalyst prior to reaction. In this reaction, the active speciesare the protons (acids) located in the vicinity of theheteropolyanions, which also act as counteranions to ensure aglobal neutral charge. However, the heteropolyanions aresurrounded by water molecules in a semicrystalline structure,the presence of which hinders the approach of methanolmolecules to the catalytic sites. The challenge was thus to findconditions to displace these water molecules without degradingthe heteropolyanion structure in order to facilitate the access ofmethanol molecules to innerH+ active sites. In that study, in situ/operando Raman spectroscopy was a powerful tool to observethe chemical events and to identify spectroscopic signatures ofproper pretreatment procedures for boosting the catalyst activity.

Density Functional Theory and Molecular Dynamics

In the last 10 years, as a result of a significant increase incomputational power, theoretical studies of heterogeneouscatalysts have undergone considerable progress, in large partsupported by the development of density functional theory(DFT), which enables calculations to be performed on relativelylarge systems composed of a few hundred atoms. Once a modelof the active site has been defined on the basis of the availablespectroscopic and catalytic data, the interaction of the reactant(s)with the catalyst can be investigated at the atomic level. UsingDFT calculations, and taking into account the experimentalconditions, it is possible to compute the spectroscopic propertiesof the adsorbed molecules. The theoretical results are thencompared with the experimental data in order to improve theaccuracy of the assignments. This synergetic approach is veryfruitful in the case of operando spectroscopies, for which thedirect assignment of the spectroscopic data is difficult because ofthe lack of spectroscopic references for the reaction chemicalintermediates.DFT modeling of catalysts is also used to propose/refine

realistic reactionmechanisms. In such studies, the reaction path isbroken down into a sequence of possible elementary reactionsteps. The thermodynamic properties of all of the elementarysteps are then computed. If some of the proposed elementarysteps turn out to be very endothermic, this generally constitutesstrong evidence that the initially considered mechanism shouldbe modified. In a second step, the activation energies arecomputed, and the calculated values are compared withexperimental data coming from kinetic studies. Agreementbetween the theory and the experiments is a simple method forvalidating a possible reaction mechanism. Figure 4 is a goodexample of DFT results, with a complete set of elementary stepscalculated and validated among various possibilities. For step 1,which is in fact the rate-limiting step with the highest energeticbarrier of the whole sequence, Figure 4 presents ball-and-stickmodels of the initial state (adsorbed lactic acid) and thetransition state leading to a carbanion intermediate, which isfurther transformed through step 2. For each reaction step, DFThelps to determine the transition state, which is a metastableconfiguration of the system corresponding to the energeticbarrier to be overcome before progressing to the next reactionstep.

Systematic deconstruction of the active site by subtraction ofeach constituent element enables the electronic effect thatinfluences the rate-determining step to be determined, which inturn provides useful information for back-optimization of thecatalytic formulation in terms of activity and/or selectivity. Evenif the direct synthesis of the proposed ideal theoretical catalyst isnot possible, the calculations indicate the chemical properties forwhich we should look in order to optimize the available catalysts.To date, DFT calculations have been used successfully to study

gas-phase reactions. In the past year, some test calculations havebeen performed in the liquid phase (for water or organicsolvents). Because of the mobility of the molecules in the liquidphase, one needs to apply molecular dynamics calculations inorder to be able to take into account the numerousconfigurations of the solvent around the active site. Calculationsof spectroscopic properties have already been performed, but thedescription of a full catalytic cycle has still not been achieved.However, this may be possible in the next few years, thanks to theevolution of parallel computers, which will provide us with thehuge calculation power required.High-Throughput Technologies.43

High-throughput technologies (HTTs), basically developedfrom a rather long history in parallel with research in drugdiscovery, have been applied relatively recently to heterogeneouscatalysis. They must not be confused with combinatorialchemistry, which involves a totally random approach. Asheterogeneous catalysis is notor at least not yeta predictivescience, after a brainstorming phase for identifying some possibleefficient catalytic formulations to produce a given reaction, anexperimental trial-and-error phase must be carried out in the lab.This consists of synthesizing different series of catalystformulations to be further tested under various conditions, asthe final performances constitute the indicator of success orfailure. Using HTTs allows the duration (and cost) of this trial-and-error phase throughout the catalyst development chain (i.e.,synthesis, characterization, and test) to be optimized with “high-output” benefits. The REALCAT platform43 is an example of anultraintegrated laboratory for high-throughput discovery ofcatalysts. This kind of equipment can also be used to acquire ahuge quantity of data in a short time in order to, for example, feedand refine kinetics models.

■ REACTORS AND CATALYSTS SHAPING

Reactors

The reactor is the heart of every industrial catalytic process. Evenif its size is relatively small and its cost is low compared with thoseof the whole plant, its performance has a tremendous influenceon all of the other upstream and downstream operations. Forinstance, the level of reactant conversion attained has a directimpact on the downstream separation stage. The choice of thereactor type is therefore of crucial importance. For a givenreaction, the chemists and chemical engineers have to decidetogether what type and size of reactor they should use. They alsohave to choose the operating mode (batch, semibatch,continuous, steady-state, or dynamic) as well as the operatingconditions (pressure, temperature, concentrations, contact orresidence time, etc.). The kinetic and thermodynamic data forthe reaction of interest and for side reactions leading tobyproducts are the basis on which these decisions are made.The phase(s) present (liquid, gas, or both) also have an impacton the choice. If the reactant and products are in the gas phase,for example, then a continuous tubular reactor is the best

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

686

candidate. Many excellent books have already been published onchemical reaction engineering and reactor technology,44−46 butthe literature discussing this important question of the initialchoice of the type of reactor and operating conditions is scarce.Some guidelines are given below.The decision to work in batch, semibatch, or continuous mode

is generally linked to the quantity of products manufactured peryear. For a few tenths to hundreds of tons, a batch process isgenerally preferred, whereas it is not feasible from an economicalpoint of view to produce more than ten thousand tons a year of aproduct, in a batch reactor.The large majority of industrial reactors do not behave ideally,

but in order to be able to model their performances, they aregenerally considered to be close to a plug-flow reactor (PFR) orto a perfectly mixed reactor (either a batch reactor or acontinuous stirred tank reactor (CSTR)). Sometimes it isnecessary to add some correction factors to the ideal reactormodels to properly represent the real situation. In the PFR, agradient of concentration exists between the inlet and the outletof the reactor. The reactant is progressively converted toproducts as it makes its way through the reactor. In a CSTR, onthe contrary, the composition of the reactive medium is the sameirrespective of the location in the reactor. The composition,temperature, and pressure are constant over the whole reactorvolume in that case. From a kinetic point of view, most of thetime the rate of reactant consumption increases with reactantconcentration (positive kinetic order), and in that case, if only thereactant conversion is taken into account a PFR is a better choicethan a CSTR. Actually, the reactant concentration is immediatelylowered in a CSTR whereas it is more progressive in a PFR.Finally, for the same reactor volume, the mean rate of reactantconsumption is higher in a PFR than in a CSTR. This is, however,a too-simplistic conclusion. Indeed, the selectivity for a desiredproduct is generally also the target, and in that case, the choice ofthe reactor is not so straightforward.The contact time is also a criterion to take into account when

selecting the reactor. Indeed, a continuous tubular reactor mustbe reserved for relatively fast reactions for which a contact time ofa few seconds to a few minutes is sufficient. In batch reactors andCSTRs, a contact time of a few hours can be attained.Another important point to consider when selecting a reactor

is the thermicity of the desired reaction. It is of course importantto control the temperature of the reaction even if it is highlyendothermic or exothermic. In that case, the choice of the reactorhas to be made considering the possibility to add or removecalories at a high rate. In the gas phase, multitubular reactors,where there is contact with a heating or cooling fluid, are oftenused for that purpose. Fluidized-bed reactors can also be a goodchoice, as the turbulence of the medium is high and this assiststhermal transfer.Finally, the stability of the catalyst can also influence the choice

of the reactor and the operating conditions. If fast deactivationoccurs, then it is necessary to quickly regenerate the catalyst. Forinstance, this is the case in fluid catalytic cracking, for which amoving-bed reactor is used. In this type of process, the catalystcontinuously moves from the reactive part of the reactor(fluidized-bed), where coke is formed very quickly on the catalystsurface, to the regeneration zone, where the coke is burned usingair addition. Hence, this specific reactor configuration enables acontinuous catalytic cracking process to proceed.

Catalyst Shaping

Very often at the lab scale, catalyst prototypes are tested in theform of powders because this is more convenient. Indeed, thecatalysts are generally obtained in powder form following theirsynthesis. However, at the pilot and commercial scales,depending on the kind of reactor that has been selected (seeabove), most of the time it is necessary to design the shape of thecatalyst. For example, the design of a catalyst used in the liquidphase in a perfectly mixed reactor will not be the same as acatalyst used in a gas-phase reaction carried out in a fixed-bedtubular reactor. In the first case, the main problems to tackle willbe (i) the potential diffusion limitations of the reactants andproducts outside the catalyst particles (external diffusion) orwithin its porosity (internal diffusion), (ii) the capacity to easilyseparate the catalyst from the liquid after reaction, and (iii) thepossibility to efficiently stir the reactor and to create ahomogeneous suspension of the solid catalyst in the reactivemedium. In the gas phase, the objectives will mainly be to limitthe pressure drop created by the catalyst bed when the gas flowgoes through it and to maintain sufficient mechanical resistancefor the catalyst particles. The limitations due to heat transfermust also be taken into account, most particularly for veryexothermic or endothermic reactions. Of course, fine controlover the temperature, even very locally in the reactor, is of greatimportance in achieving high performance.Publications specifically on the subject of shaping of industrial

catalysts are scarce, as the highly valuable know-how is generallynot patentable.47 However, general methods used to developcommercial catalysts are well-described in the literature.31,48

In order to avoid the internal diffusion limitations, a goodstrategy is to reduce the size of the catalyst particles (50−200μm) and to increase the porosity. Too-fine particles must beavoided, however, as they are difficult to recover after thereaction (filtration issues) and can be dragged into thedownstream equipment. To prepare such catalysts, grinderscoupled with sieves are generally used.Small particles (50−250 μm) are also needed when the use of a

fluidized-bed reactor is envisioned. In that case, sphericalparticles are also highly desired, as they are more resistant tothe strong mechanical constraints encountered in such equip-ment. Spray dryers are then used to prepare these catalysts. For amoving-bed reactor, the catalyst must also be resistant toattrition, but the particles are generally bigger (up to severalmillimeters). Oil-drop systems are used for their preparation. Forfixed-bed reactors, particles (1 to 15 mm) with a lot of possibleshapes can be used to constitute the packing of the bed: spheres,cylinders, hollow and ribbed cylinders, rings, wheels, and so on.The main techniques used to produce these catalysts areextrusion and pelletization or tableting. Extrusion involves thepreparation of a paste containing the catalyst and differentadditives (lubricants, binders, etc.) that are forced through a dieto give the desired shape. The extrudates are then cut to therequired length before drying or calcination. Pelletizationconsists of pressing powders into a mold to obtain the desiredshape. Here again some additives are needed, such as lubricants.

■ FUTURE DEVELOPMENTS

Catalysis is a scientific field that is encountering a “new goldenage”, mainly due to the necessity of finding new strategies andconcepts for efficiently treating biosourced molecules andmaterials. In that context, HTTs holds an important role, andnew developments in this field are expected. Innovative reactors

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

687

are also being developed, such as microreactors49 and two-zonefluidized-bed reactors (TZFBRs).50 Process developments willinclude, for example, a better use of microwave-assisted methods,in particular to adapt them to continuous operation. New typesof catalysis are also being developed: for example, hybridcatalysis51 will bridge the gap between chemocatalysis andbiocatalysis, and interfacial catalysis assists reactions betweenhydrophobic and hydrophilic reactants via Pickering emulsions,which consist of emulsions with small particles of aheterogeneous catalyst located at the interface between theliquid droplets and the liquid medium.52 It is also possible tocombine Pickering emulsions with hybrid catalysis to produceeven more sophisticated catalytic systems. Nano-oxides couldalso reveal enhanced properties compared with their bulkcounterparts, similar to what was observed on gold, which waslargely considered as a completely catalytically inert materialuntil the idea arose of dividing it into nanoparticles, which turnedout to be exceptionally active.53

DFT and molecular dynamics are also involved, with expectednew robust methodologies combined with the evolution of thepower of parallel computers, which will enable, for example, therole of the solvent in liquid-phase catalytic systems to be betteraddressed. Furthermore, as liquid-phase (often aqueous)transformations are of specific interest in biomass upgrading,advanced spectroscopic and microcalorimetric observations ofcatalysts in this phase are also under development. As a specificpoint, characterization of basic sites is taking on a largeimportance, while to date the acid−base characterizationtechniques have beenmainly developed specifically for character-izing acidic sites (furthermore, mainly in the gas phase!).Similarly, it is now highly desired to obtain simultaneousinformation on the basic and acidic sites of catalysts, as theseantagonist functionalities always coexist in various proportionson the catalysts’ surfaces and are most likely responsible forselectivity issues, for example (however, cooperation betweenacidic and basic sites is also sometimes needed, e.g., for theGuerbet reaction54).We should also mention what we could call “photo-

constructive catalysis”, which could mimic photosynthesis inartificial systems but directly target the molecules to beproduced; could this dream be realized in 50 years? In thiscontext, photoassisted technology will probably develop rapidlythrough various channels. For example, “catalytic nanoplas-monics” based on the properties of surface plasmons created bylaser irradiation of metallic particles could lead to better energyefficiency.The possibilities are quasi-infinite, depending on researchers’

imagination and creativity, and we may most probably havesurprises in the next years with completely new developments.Indeed, catalytic science has again entered an era of globalinnovation after a long period mainly driven by incrementalimprovements of the already existing concepts, methodologies,and techniques.

■ AUTHOR INFORMATION

Corresponding Author

*E-mail: franck.dumeignil@univ-lille1.fr.

ORCID

Franck Dumeignil: 0000-0001-9727-8196Jean-Francois Paul: 0000-0003-1935-1428

Notes

The authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors warmly thank Edmond Payen (Lille University,France) for his advice and suggestions during the preparation ofthis manuscript.

■ REFERENCES(1) http://master-chimie.univ-lille1.fr/master2/Biorefinery/ (ac-cessed Feb 7, 2017).(2) http://www.wur.nl/en/Education-Programmes/master/MSc-programmes/MSc-Biobased-Sciences.htm (accessed Feb 7, 2017).(3) https://www.biorenew.iastate.edu/education/brt/ (accessed Feb7, 2017).(4) Heveling, J. Heterogeneous catalytic chemistry by example ofindustrial applications. J. Chem. Educ. 2012, 89, 1530−1536.(5) Oyama, S. T.; Somorjai, G. A. Homogeneous, heterogeneous andenzymatic catalysis. J. Chem. Educ. 1988, 65 (9), 765−769.(6) Dicks, A. P.; Batey, R. A. ConfChem Conference on Educating theNext Generation: Green and Sustainable Chemistry-Greening theOrganic Curriculum: Development of an Undergraduate CatalyticChemistry Course. J. Chem. Educ. 2013, 90, 519−520.(7) Farruseng, D. Metal−Organic Frameworks; Wiley-VCH: Wein-heim, Germany, 2011.(8) Van Santen, R. A.; Niemantsverdriet, J. W. Chemical Kinetics andCatalysis; Fundamental and Applied Catalysis Series; Plenum Press:New York, 1995.(9) Van Santen, R. A. Theoretical Heterogeneous Catalysis; WorldScientific: Singapore, 1991.(10) Schlogel, R. Presented at Nanoscience and Catalysis, a meeting atthe National Science Foundation, Washington, DC, 2003.(11) Biorefinery: From Biomass to Chemicals and Fuels; Aresta, M.,Dibenedetto, A., Dumeignil, F., Eds.; De Gruyter: Berlin, 2012.(12) Biorefineries: An Introduction; Aresta, M., Dibenedetto, A.,Dumeignil, F., Eds.; De Gruyter: Berlin, 2015.(13) https://www.roquette.com (accessed Feb 3, 2017).(14) http://www.a-r-d.fr/en/ (accessed Feb 3, 2017).(15) http://www.institut-pivert.com/?lang=en (accessed Feb 3,2017).(16) https://www.ifmas.eu/en/ (accessed Feb 3, 2017).(17) Hammaecher, C.; Paul, J.-F. Density functional theory study oflactic acid adsorption and dehydration reaction on monoclinic 011, 101,and 1 11 zirconia surfaces. J. Catal. 2013, 300, 174−182.(18) Thomas, J. M.; Thomas, W. J. Principles and Practice ofHeterogeneous Catalysis; VCH: Weinheim, Germany, 1997.(19) Lalikog lu, M.; Gok, A.; Gok, M. K.; Asci, Y. S. Investigation oflactic acid separation by layered double hydroxide: Equilibrium, kineticsand thermodynamics. J. Chem. Eng. Data 2015, 60, 3159−3165.(20) Vannice, M. A. Kinetics of Catalytic Reactions; Springer Science+Business Media: New York, 2005.(21) Mars, P.; van Krevelen, D. W. Oxidations carried out by means ofvanadium oxide catalysts. Chem. Chem. Eng. Sci. 1954, 3, 41.(22) Doornkamp, C.; Ponec, V. The universal character of the Marsand Van Krevelen mechanism. J. Mol. Catal. A: Chem. 2000, 162 (1−2),19−32.(23) Katryniok, B.; Kimura, H.; Skrzynska, E.; Girardon, J.-S.;Fongarland, P.; Capron, M.; Ducoulombier, R.; Mimura, N.; Paul, S.;Dumeignil, F. Selective catalytic oxidation of glycerol: perspectives forhigh value chemicals. Green Chem. 2011, 13, 1960−1979.(24) Gonzalez-Borja, M.; Resasco, D. E. Reaction pathways in theliquid phase alkylation of biomass-derived phenolic compounds. AIChEJ. 2015, 61 (2), 598−609.(25) Smith, J. M. Chemical Engineering Kinetics, 3rd ed.; Mc Graw-Hill:New York, 1981.(26) Doraiswamy, L. K.; Sharma, M. M. Heterogeneous Reactions:Analysis, Examples, and Reactor Design. Vol.1: Gas−Solid and Solid−SolidReactions; John Wiley and Sons: New York, 1984.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

688

(27) Silvester, L.; Lamonier, J.-F.; Faye, J.; Capron, M.; Vannier, R.-N.;Lamonier, C.; Dubois, J.-L.; Couturier, J. L.; Calais, C.; Dumeignil, F.Calais, C.; Dumeignil, F. Reactivity of Ethanol over Hydroxyapatite-Based Catalysts with Various Carbonate Contents. Catal. Sci. Technol.2015, 5, 2994−3006.(28) Regalbuto, J. Catalyst Preparation; CRC Press: Boca Raton, FL,2007.(29) Trimm, D. L. Deactivation and Regeneration. In Handbook ofHeterogeneous Catalysis; Ertl, G., Knozinger, H., Weitkamp, J., Eds.;Wiley-VCH: Weinheim, Germany, 1997.(30) Liquid Phase Oxidation via Heterogeneous Catalysis: OrganicSynthesis and Industrial Applications; Clerici, M. G., Kholdeeva, O. A.,Eds.; John Wiley & Sons, Hoboken, NJ, 2013.(31) Busca, G. Heterogeneous Catalytic Materials: Solid State Chemistry,Surface Chemistry and Catalytic Behaviour; Elsevier B.V.: Amsterdam,2014.(32) Synthesis of Solid Catalysts; De Jong, K. P., Ed.; Wiley-VCH:Weinheim, Germany, 2009.(33) http://www.fhi-berlin.mpg.de/acnew/department/pages/teaching/pages/teaching__wintersemester__2006_2007/koehler_preparation_supported_catalysts_111006.pdf (accessed Feb 5, 2017).(34) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area and Porosity;Academic Press: London, 1967.(35) Lowell, S.; Shields, J. E.; Thomas, M. A.; Thommes, M.Characterization of Porous Solids and Powders: Surface Area, Pore Size andDensity; Particle Technology Series, Vol. 16); Springer Science+Business Media: New York, 2010.(36) Characterization of Solid Materials and Heterogeneous Catalysts:From Structure to Surface Reactivity; Che, M., Vedrine, J. C., Eds.; Wiley-VCH: Weinheim, Germany, 2012; Vols. 1 and 2.(37) In-Situ Spectroscopy in Heterogeneous Catalysis; Haw, J. F., Ed.;Wiley-VCH: Weinheim, Germany, 2002.(38) Liu, Z.; Duchon, T.;Wang, H.; Grinter, D. C.;Waluyo, I.; Zhou, J.;Liu, Q.; Jeong, B.; Crumlin, E. J.; Matolin, V.; Stacchiola, D. J.;Rodriguez, J. A.; Senanayake, S. D. Ambient pressure XPS and IRRASinvestigation of ethanol steam reforming on Ni−CeO2(111) catalysts:an in situ study of C−C andO−Hbond scission. Phys. Chem. Chem. Phys.2016, 18, 16621−16628.(39) Ryczkowski, J. IR spectroscopy in catalysis. Catal. Today 2001, 68,263−381.(40) Lavalley, J. C. Infrared spectrometric studies of the surface basicityof metal oxides and zeolites using adsorbed probe molecules. Catal.Today 1996, 27, 377−401.(41) Weckhuysen, B. M. Determining the active site in a catalyticprocess: Operando spectroscopy is more than a buzzword. Phys. Chem.Chem. Phys. 2003, 5, 4351−4360.(42) Schnee, J.; Gaigneaux, E. M. Operando Raman to Enhance theMethanol-to-DME Conversion Over Non-Thermally-Pretreated Keg-gin Heteropolyacids. J. Phys. Chem. C 2017, 121, 556−566.(43) Modern Applications of High Throughput R&D in HeterogeneousCatalysis; Hagemeyer, A., Volpe, A. F., Eds.; Bentham SciencePublishers: Sharjah, United Arab Emirates, 2014.(44) Levenspiel, O. Chemical Reaction Engineering, 3rd ed.; Wiley:Hoboken, NJ, 1998.(45) Salmi, T.; Mikkola, J.-P.; Warna, J. P. Chemical ReactionEngineering and Reactor Technology; CRC Press: Boca Raton, FL, 2011.(46) Carberry, J. J. Chemical and Catalytic Chemical Engineering, 2nded.; Dover Publications: New York, 2001.(47) Pernicone, N. Scale-up of catalyst production. Catal. Today 1997,34 (3−4), 535−547.(48) Deutschmann, O.; Knozinger, H.; Kochloefl, K.; Turek, T.Heterogeneous Catalysis and Solid Catalysts, 2. Development andTypes of Solid Catalysts. In Ullmann’s Encyclopedia of IndustrialChemistry; Wiley-VCH: Weinheim, Germany, 2012.(49) Ehrfeld, W.; Hessel, V.; Lowe, H. Microreactors: New Technologyfor Modern Chemistry; Wiley-VCH: Weinheim, Germany, 2000.(50) Soler, J.; Lopez Nieto, J. M.; Herguido, J.; Menendez, M.;Santamarıa, J. Oxidative dehydrogenation of n-butane in a two-zonefluidized-bed reactor. Ind. Eng. Chem. Res. 1999, 38, 90−97.

(51) Dumeignil, F. Chemical catalysis and biotechnology: From asequential engagement to a one-pot wedding. Chem. Ing. Tech. 2014, 86(9), 1496.(52) Pera-Titus, M.; Leclercq, L.; Clacens, J. M.; De Campo, F.;Nardello-Rataj, V. Pickering interfacial catalysis for biphasic systems:From emulsion design to green reactions. Angew. Chem., Int. Ed. 2015,54, 2006−2021.(53) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Novel goldcatalysts for the oxidation of carbon monoxide at a temperature farbelow 0 °C. Chem. Lett. 1987, 16 (2), 405−408.(54) Tsuchida, T.; Kubo, J.; Yoshioka, T.; Sakuma, S.; Takeguchi, T.;Ueda,W. Reaction of ethanol over hydroxyapatite affected byCa/P ratioof catalyst. J. Catal. 2008, 259, 183−189.

Journal of Chemical Education Article

DOI: 10.1021/acs.jchemed.6b00611J. Chem. Educ. 2017, 94, 675−689

689