Published: October 04, 2011

Copyright r 2011 American Chemical Society andDivision of Chemical Education, Inc. 1711 dx.doi.org/10.1021/ed101010x | J. Chem. Educ. 2011, 88, 1711–1715

LABORATORY EXPERIMENT

pubs.acs.org/jchemeduc

The Sabatier Principle Illustrated by Catalytic H2O2 Decomposition onMetal SurfacesAnders B. Laursen,*,† Isabela Costinela Man,‡ Ole L. Trinhammer,§ Jan Rossmeisl,‡ and Søren Dahl†

†Center for Individual Nanoparticle Functionality (CINF) at the Department of Physics, ‡Center for Atomic-Scale MaterialsDesign (CAMD) at the Department of Physics, and §Department of Physics, The Technical University of Denmark,2800 Kgs. Lyngby, Denmark

bS Supporting Information

Heterogeneous catalysis is important in today’s industry. Asstudents venture into the field of chemistry, it is imperative

to introduce them to the possibilities and principles of hetero-geneous catalysis as early as possible to motivate them to pursuethis field. This work presents a possible approach to do so at anintroductory level. Students are introduced to the advantages ofcombining a theoretical model with experiments to gain anunderstanding of the differences in catalytic activity of relevantmaterials via the Sabatier principle. Furthermore, this approachgives the students hands-on lab experience in a simple and funexperiment, which can be interpreted qualitatively by use of theperiodic table and quantitatively by computational simulations.

By definition, a catalyst is a reagent that is not consumedduring the reaction. A catalyst is either homogeneous or hetero-geneous, depending on whether the catalyst exists in the same ordifferent phase, respectively, as the reactant(s). In the following,only heterogeneous catalysis is examined. The active part of anindustrial heterogeneous catalyst is most commonly a solidsurface of a metal or a metal oxide. The surface offers a low-barrier energy path from reactants to products by bindingreactants and reaction intermediates and also avoids the problemof the unlikely three-body interactions in the gas phase. The bestcatalytic surfaces for a given reaction will have a reactivity that is acompromise with respect to reactivity. Therefore, the bindingenergy of the reactants to the surface must be strong enough thatthey produce reaction intermediates, but has to be weak enoughto allow the products to leave the surface thus allowing morereactions to occur at that site. This duality is the essence of the

Sabatier principle.1 The optimal catalyst surface will be a trade-offbetween binding the reagents to the surface and not binding anyof the reaction intermediates too strong relative to the product.The principle explains why a so-called volcano curve is mostoften obtained when plotting catalytic activity versus surfacereactivity. It is important to note the distinction between activityand reactivity. The activity is a measure for how fast the catalyticreaction occurs on the surface and therefore is measured persurface area (also called the specific rate). If not measured thisway, the activity of a foil and a powder catalyst would differ for thesame material, which they clearly should not. The reactivity, onthe contrary, is a measure of the inherent reactivity of the surfacetoward adsorbing molecules. The reactivity can be quantified bythe binding energy between the surface and a reaction inter-mediate that is central to the catalytic reaction. The obtainedvolcano curves can be helpful in the design of optimized catalysts.If similar materials are compared, a qualitative relative surfacereactivity can be obtained from the periodic table of the elements.Pure transition metals are, for instance, often used as catalysts,and for these, the reactivity decreases when going from the left tothe right in the periods. However, if different surfaces are takeninto account, for example, both metals and metal oxides,computational chemistry is of paramount importance as it offersa fast way of calculating the binding energy of conceivable

ABSTRACT: Heterogeneous catalysis is important in today’s industry.Hence, it is imperative to introduce students to this field and its tools. Anew way of introducing one of these tools, the Sabatier principle, via alaboratory exercise is presented. A volcano plot is constructed for the well-known heterogeneous H2O2 catalytic decomposition reaction on variousmetal foils. The activity per catalyst surface area versus the computationallycalculated binding energy of OH groups on the catalysts is plotted. The OHgroup is identified as the only surface intermediate in an intuitive reactionmechanism, and hence, it is the relevant reactivity parameter. From thecalculated binding energies and the reaction mechanism, the volcano peakposition is inferred. This work is relevant to introductory levels of chemistry inadvanced high school classes and initial levels of university.

KEYWORDS: First-Year Undergraduate/General, High School/Introduc-tory Chemistry, Upper-Division Undergraduate, Analytical Chemistry, Laboratory Instruction, Physical Chemistry, Hands-OnLearning/Manipulatives, Catalysis, Kinetics, Rate Law

1712 dx.doi.org/10.1021/ed101010x |J. Chem. Educ. 2011, 88, 1711–1715

Journal of Chemical Education LABORATORY EXPERIMENT

reaction intermediates, relevant to the reaction, on a variety ofsurfaces.

These principles will be used to get a volcano curve for thefamiliar classroom experiment of catalytic hydrogen peroxidedecomposition reaction using MnO2 and potato or liverenzymes:2

2H2O2ðaqÞ f 2H2OðlÞ þ O2ðgÞ ð1Þ

A laboratory experiment examining the reaction kinetics of thisreaction has been published in this Journal3 and could be com-bined with the experiment described here. To apply theoreticalcalculations to this reaction, a possible reaction path must bedevised, which will be discussed later.

’THEORETICAL AND EXPERIMENTAL WORK

The theoretical binding energy calculations and the experi-mental procedure for measuring the activity of H2O2 decom-position for different surfaces are described below.

Theoretical CalculationsComputer simulations offer a way to calculate binding en-

ergies of molecules on surfaces. The ground-state energy andelectronic structure, for a given surface or molecule, can beobtained using density functional theory (DFT).1,4 This is aquantum mechanical method where the interactions betweenelectrons and nuclei are accurately accounted for. The input is theatomic structure of the system, that is, the atomic numbers andthe position of the atoms in space. The output is the ground-stateenergy. Binding energies can be calculated by comparing en-ergies, for example, of a system containing a molecule bound to asurface with the energy of the same system where the moleculeis separated from the surface. The binding energies of OH(a relevant reaction intermediate in H2O2 decomposition) ondifferent surfaces, obtained by DFT calculations, are given inTable 1.5 The obtained free energies are converted to conditionsat 298 K by taking entropy into account, the reference point ofliquid water, oxygen gas, and the bare catalyst surface at standardconditions.6

Experimental SetupTomeasure the catalytic activity of some of the surfaces where

the binding energy of OH was calculated, representative foilsof metals were chosen. A titanium metal foil (Sigma-Aldrich)that spontaneously formed surface oxide layer, gold (Goodfellow),platinum(Goodfellow),nickel (Goodfellow), andcopper(Goodfellow)foils and a silver wire (Sigma-Aldrich) were tested. All foilswere of >99% purity and cut into pieces approximately 1 cm�2 cm(1 mm diameter silver wire of 5 cm length was used). One pieceof each metal was used, except for platinum where two smallpieces making 0.7 cm2 were used. Expenses of the materials may

be decreased by only testing nickel and titanium foils and, forinstance, a piece of gold or silver jewelry (as regular as possible to easethe surface area estimation).7 Powders of the materials can also beused, but the specific surface area of the powder is required.Preparation time for this experiment is 30�45 min and eachexperiment takes 20�30 min. Data processing takes about 10�20min per sample. It is not necessary for the students to measure theleveling off of the data (observed in the inset of Figure 2).

A reaction solution of 250mL of 0.1MKOH and themetal foilto be tested were placed in a B€uchner flask fitted with a rubberstopper and containing a magnetic stir bar. One end of the hose,attached to the side arm of the flask, was placed in an upturned10 mL graduated cylinder, which was filled with water and placedin a water bath (Figure 1). The graduated cylinder is filledcompletely to the brim with water and a hand is placed over it toseal it, and then it is turned upside down. The hand is gentlyremoved once the opening of the cylinder is completely im-mersed in the water bath. A few bubbles in the cylinder will notaffect the measurements. The experiment is initiated by adding10 mL of 30% H2O2 to the B€uchner flask and immediatelystoppering the flask. The timer is started as the bubbles fill thecylinder to the first volume mark; this is done to eliminate theeffect of possible bubbles already present in the cylinder. Onlythe volume of evolved oxygen is measured in the cylinder becausethe gaseous water from the reaction (eq 1) is cooled andcondenses when traveling through the water bath.

To obtain the catalytic activity, the evolved oxygen inmole perarea of foil is plotted as a function of time to give data similar tothose shown in Figure 2 (see the spreadsheet in the SupportingInformation for an example of the calculations). As definedearlier, the true activity is the evolved oxygen per surface areaper time, which is equal to the initial slope of the curve (Figure 2).Here, it is implicitly assumed that the roughness of the foils andwire are the same for all the metals; that is, the geometric ormacroscopic area represent the true surface area. The initial slopeof the reaction is the activity used to compare the catalysts.Initially, the curve is nearly linear and then drops off slightly asthe H2O2 is used up (see inset). This is the reason that theactivity is onlymeasured initially where there is plenty of H2O2 sothat the only limit to reaction is the properties of the surfaces.The reduced rate for copper after approximately 30 s is not due tolack of reagent but most likely the formation of an ill-definedsurface oxide. This oxide, though not easily modeled to obtainOH binding energies, is expected to have a higher bindingenergy. This is common for all metals when oxidized. Limitingthe amount of gas, which is collected to the first linear part of thereaction, removes the need to identify the linear part and makesthe data treatment easier.

Table 1. Binding Energies of OHGroups on Relevant Metalsand Metal Oxides As Obtained form DFT Calculations

Catalyst Binding Energy/ (kJ/mol) Measured Activity of mol O2/(m2 s)

Ni �42.45 0.00457

Cu �7.72 0.00271

Pt 36.66 0.0319

Ag 63.68 0.0123

Au 162.09 0.0104

TiO2 324.52 0.00191

Figure 1. (A) Schematic illustration of the experimental setup and (B)photograph of experimental setup. The reaction flask on the stirrercontains KOH, H2O2, a magnetic stir bar, and the catalyst. As thereaction proceeds, oxygen gas is evolved and measured in the graduatedcylinder as a function of time.

1713 dx.doi.org/10.1021/ed101010x |J. Chem. Educ. 2011, 88, 1711–1715

Journal of Chemical Education LABORATORY EXPERIMENT

’HAZARDS

The H2O2 solution is a strong oxidizer and is corrosive,causing burns to the eyes, skin, and respiratory tract. KOH iscaustic; it causes burns to any area of contact. Standard labora-tory practice such as the use of gloves and goggles is necessary.

’RESULTS AND INTERPRETATION

A reaction mechanism is needed to find a relevant bindingenergy to calculate. A simple, intuitive reaction mechanism isshown in Figure 3 (an animated movie is available in theSupporting Information). H2O2 adsorbs on the catalytic surfaceto make two OH groups, which next react with an additional

H2O2, transferring two H atoms from the second reactant to thetwo adsorbedOH species leavingO2 and twowater molecules onthe surface. These species subsequently desorb and diffuse intothe surrounding solvent and the surface is ready for anotherreaction. This is called a catalytic cycle. The calculated bindingenergies of the OH group on various surfaces tabulated earliermay then be used with the measured activity to plot the volcanoin Figure 4. The activity in the form of moles of oxygen evolvedper second per square meter of catalyst is used in the plot, whichallows for the direct comparison of powders and foils. Theseunits are highlighted to the students and a discussion on how tooptimize surface area ensues.

To illustrate the importance of surface area in heterogeneouscatalysis, a small quantity of silver powder, with a known surfacearea, was tested to compare its activity to that of the silver wire.After normalizing the activity to surface area, both results agreewithin the first two decimals (as shown in the spreadsheet in theSupporting Information). If the specific surface area of a silverpowder is available from the manufacturer or is obtained using anitrogen adsorption instrument (BET analysis), the students areencouraged to run the experiment with both silver samples. Notethat a measuring cylinder of 250 mL is necessary for 30 mg silverpowder as the rate is so fast. The powder (Sigma-Aldrich cat#327085) was measured by BET to have a surface area of 0.61 m2/g.Comparison of the two silver catalysts offers a memorable op-portunity to visualize the effect of increased surface area.

A crucial point must be considered before contemplating thevolcano plot in detail. If the free energy of each of the steps of thereaction pathway (Figure 3) is plotted, a three-step curve isobtained. If the energy of the reaction is considered figuratively asa ball rolling down from the initial high energy to the low-energyproduct state, then it follows that the ball will only roll downhill.This scenario will prevent the ball from rolling up the TiO2 pathand also down the copper path because it would then have to passan uphill stretch (Figure 5). In terms of the Sabatier principle,this is interpreted as the binding of the intermediate OH beingtoo weak on the TiO2 and too strong on the copper. One maythen state that the energy of OH on the surface must lie betweenthe energy of the bare surface plus two free H2O2 molecules andthe bare surface plus a free oxygen and two free water molecules.In Figure 4, these limits are illustrated as shaded regions atwhich the energy of the bound OH groups become unfavorable.

Figure 3. (Top) Schematic interpretation of the reaction modelproposed for the H2O2 decomposition: Ti is in blue, O in red, and Hin white. The model shows that a H2O2 molecule is adsorbed onto aclean surface (A) and split in two OH groups bound to the surface (B).Subsequently, another H2O2 molecule comes in and donates two Hatoms to the OH on the surface (C). This results in two water moleculeson the surface and an oxygen molecule (D). These molecules desorb toleave the clean surface for a new reaction (E). (Bottom) Equationsassociated with the reaction steps; 2* represents two reaction sites on thecatalyst.

Figure 4. Volcano curve of reactivity, in millimole oxygen evolvedper second and square meter catalyst surface area (on a logarithmicscale), as a function of OH binding energy.

Figure 2. Amount of oxygen evolved per area of metal foil as a functionof time for titanium dioxide, copper, and gold. The initial slope of thereaction is the activity used to compare the catalysts. Initially, the curve isnearly linear and then drops off slightly as the H2O2 is used up (see insetwith the gold data).

1714 dx.doi.org/10.1021/ed101010x |J. Chem. Educ. 2011, 88, 1711–1715

Journal of Chemical Education LABORATORY EXPERIMENT

It is worth noting that faded boundaries imply that this is not arigid rule for activity, but more as a constraint for the optimalcatalysts.

From the volcano plot in Figure 4, one can see that the relativeactivities predicted by our model corresponds well with theobserved activities. That is, the metals that bind OH too stronglyare poor catalysts and thus lie on the left side of the volcano.These metals are inherently unstable in an oxidizing environ-ment; indeed, these metals will form oxides after prolongedreaction and very quickly in the case of copper. Hence, OHadsorbs and reacts irreversibly with the metals; this will alter themetallic surface during the reaction. This leads to another crucialpoint: If the surface changes significantly, so will the bindingenergy and false conclusions may be drawn. At the same time,one notices that the right side of the volcano, that is, the weakhydroxide binding surfaces are stable. Students are encouraged tosee the parallel to everyday life; a suitable question relates to thebehavior of gold jewelry in contrast to a copper pot. Noting thatthe frequency that these items needing polishing is inverselyrelated to their stability should encourage discussion. At thetop of the volcano, that is, within the window of opportunity,platinum, silver, and gold are located. These are the best catalystsfor the decomposition reaction. From the classic classroomexperiments,2 manganese dioxide, with a binding energy of106.1 kJ/mol also falls in the window of opportunity.

Although the model seems to correlate well with the actualreaction pathway, this is not sufficient proof that the reactionhappens as our models states. However, it is hard to imagine anyreaction mechanism where the reaction intermediates do notbind to the surface via oxygen such as OH, which means that the

OHbinding energy is a goodmeasure of the surface reactivity. Animportant aspect of the volcano curve is that it is a good tool toscreen for other possible catalysts.

Typical data obtained by students are shown in Figure 6A. Thedata are not corrected for differences in surface area. Studentswere instructed to obtain measurements up to a given volume, asdetermined by the instructor for the amount of catalyst, so thatonly linear data are obtained. Student data deviated about 40%from the data presented in Figure 4; this must be attributed to anonidentified constant error such as faulty equipment (a hosewas found to be slightly leak) or possibly chemical preparation(concentration of KOH stock solution). The latter is suspected,as the student volcano curve seen in Figure 6B is a verticaldisplacement of the one in Figure 4. Copper is seen to fall with alarge scatter (up to 90% of the data presented here) in the studentexercise so it might be advantageous to omit this for consistencypurposes. These issues with reproducing the data demonstratethe strength of the exercise. Even an unfortunate error does notaffect the idea of these experiments as it does not affect the trendin the Sabatier plot.

’CONCLUSION

This lab experiment is instructive because (i) the students gethands-on laboratory experience. (ii) The concept of surface areais nicely illustrated when using metal foils. (iii) The processshows how powerful a tool DFT-calculations may be for pre-dicting reactivity for a given reaction. (iv) It shows that a modelsolely guided by chemical common sense may predict reactivity.

Figure 6. (A) Typical data collected by students as volume of oxygenversus time; note that the volume of gas is not normalized to surface area.(B) The Sabatier plot as obtained from two different students; theexcellent agreement for most points is clearly seen. The deviation of thecopper data is probably due to the formation of a surface oxide.

Figure 5. Energy diagram with balls rolling. For TiO2, the surface bindsOH too weakly and nothing happens as the ball cannot roll uphill. In thecase of platinum, the OH energy is just right and the ball rollsspontaneously to the product energy level. For copper, the OH bindingenergy is too strong and the OH sticks to the surface, which issymbolized by the ball getting trapped in the valley. These three casesillustrate the window of opportunity, as any surface with an OH bindingenergy within the initial and final energy will catalyze the reaction.Underneath each step of the energy diagram, the situation on the surfaceis illustrated with the TiO2 surface.

1715 dx.doi.org/10.1021/ed101010x |J. Chem. Educ. 2011, 88, 1711–1715

Journal of Chemical Education LABORATORY EXPERIMENT

(v) Procedures such as the one demonstrated here are usedwidely in academia as well as in the industry to predict whycatalysts deactivate or to find new catalyst candidates.8 (vi) High-school students are capable of performing the experiments andthe level of introduction of the calculated quantities may bevaried to fit the student’s level. Furthermore, more experiencedstudents may be introduced in greater extent to the calculationsto give them better challenges. Here, we have chosen anintermediate level.

’ASSOCIATED CONTENT

bS Supporting InformationInstructions for the students; notes for the instructor;

movie; spreadsheet. This material is available via the Internet athttp://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*E-mail: a.b.laursen@fysik.dtu.dk.

’ACKNOWLEDGMENT

We gratefully acknowledge the Catalysis for SustainableEnergy (CASE) initiative funding by the Danish Ministry ofScience, Technology and Innovation, and the European Com-mission (Marie Curie Research Training Network MRTN-CT-2006-032474.

’REFERENCES

(1) Chorkendorff, I.; Niemantsverdriet, J. W. Concepts of ModernCatalysis and Kinetics, 2nd ed.; Wiley-VCH Verlag GmbH & Co KGaA:Weinheim, Germany, 2007.(2) Lister, T. 58. Catalysts for the decomposition of hydrogen

peroxide. In Classic Chemistry Demonstrations: One Hundred Tried andTested Experiments; O’Driscoll, C., Reed, N., Eds.; The EducationDivision, The Royal Society of Chemistry; London, 1995; pp 145�146.(3) Barlag, R.; Nyasulu, F. J. Chem. Educ. 2010, 87, 78–80.(4) David, S.; Sholl, J. A. S. Density Functional Theory, A Practical

Introduction; John Wiley & Sons, Inc.: Hoboken, NJ, 2009.(5) The description of the computational method may be omitted if

the students are at an introductory level. This is especially true for highschool students who may gain just as much from the exercise if thecalculations are viewed as mere tabulated values.(6) Rossmeisl, J.; Logadottir, A.; Nørskov, J. K. Chem. Phys. 2005,

319 (1), 78–184.(7) The silver jewelry should not be left in the solution for extended

periods of time as the surface may slowly dissolve.(8) Nørskov, J. K.; Bligaard, T.; Rossmeisl, J.; Christensen, C.H.Nat.

Chem. 2009, 1, 37–46.