b919543m

Upload: vazzoleralex6884

Post on 03-Jun-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/12/2019 b919543m

    1/24

    This article was published as part of the

    In-situcharacterization of heterogeneous

    catalysts themed issue

    Guest editor Bert M. Weckhuysen

    Please take a look at the issue 12 2010table of contentsto

    access other reviews in this themed issue

    View Online

    http://pubs.rsc.org/en/Journals/JournalIssues/CS#/issueID=CS039012&Type=Current&issnprint=0306-0012http://pubs.rsc.org/en/Journals/JournalIssues/CS#/issueID=CS039012&Type=Current&issnprint=0306-0012http://dx.doi.org/10.1039/b919543mhttp://pubs.rsc.org/en/Journals/JournalIssues/CS#/issueID=CS039012&Type=Current&issnprint=0306-0012http://pubs.rsc.org/en/Journals/JournalIssues/CS
  • 8/12/2019 b919543m

    2/24

    4928 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    Analysing and understanding the active site by IR spectroscopyw

    Alexandre Vimont, Fre de ric Thibault-Starzyk and Marco Daturi*

    Received 7th June 2010

    DOI: 10.1039/b919543m

    IR spectroscopy is a technique particularly adapted for understanding the mechanism of catalytic

    reactions, being able to probe the surface mechanisms at the molecular level. In this critical review

    the main advances in the field are presented, both under the aspects of the in situ and operando

    approaches. A broad view of the most authoritative literature of the domain is given, based

    largely on the experience built up at the LCS laboratory in the last decades. After having presented

    the general methodology to observe a potential active site directly or by probe molecule adsorption,

    several examples illustrate the qualitative and quantitative analysis of the physicalchemical

    properties of the surface entities. The last part of the review is dedicated to the discrimination

    of the role of the active site and its links with the catalytic steps; the hot problem of the reaction

    intermediates and their visibility via spectroscopic techniques is critically addressed (138 references).

    Introduction

    Heterogeneous catalysis is finding an increasing importance in

    everyday life. The industrial heterogeneous catalysts are

    shaped multifunctional devices intended to offer to the reacting

    agents a multitude of active sites, distributed on the external

    surface or inside the porosity of the materials, in order to

    optimise the contact between the reacting molecule and the

    transformation centre in the sense of the Sabatiers principle.

    On these sites the reacting molecules are adsorbed, the inter-

    mediates are generated and the products formed. Physically,

    the active centres are the sites where the reaction crosses the

    energy barrier, as classically represented in Fig. 1.Therefore, the active sites are intrinsically linked with the

    intermediate species and they can be affected by the poisons

    which can be formed during the reaction. An accurate design

    of the active sites (in terms of quality, strength, position, . . .) is

    obviously the key to obtain the optimum catalyst. On the way

    to catalyst optimisation and rational design, we have to

    thoroughly characterize the active sites, particularly when in

    action, considering that they are not static entities, but they

    undergo modifications depending on the reaction conditions

    and surface restructuration phenomena. For this purpose, IR

    spectroscopy is one of the most adapted tools, being extremely

    sensitive to the molecular vibrations and able to discriminate

    the different geometrical distortions of the molecules accordingto the adsorption state on a site.

    In this contribution, we will present how it is possible to

    evidence the adsorption sites on a catalyst, particularly using

    Laboratoire Catalyse et Spectrochimie, ENSICAEN,

    Universitede Caen, CNRS, 6 Bd Marechal Juin, F-14050 Caen,France. E-mail: [email protected]; Fax: +33-231452822;Tel: +33-231452730w Part of the themed issue covering recent advances in the in-situcharacterization of heterogeneous catalysts.

    Alexandre Vimont

    Alexandre Vimont (born 1972,

    France) received his PhD in

    2000 from the University of

    Caen in the field of in situ

    and operando infrared spectro-

    scopy applied to catalysis,

    under the supervision of Jean-

    Claude Lavalley and FredericThibaut-Starzyk. He then

    joined the laboratoire Catalyse

    et Spectrochimie as a permanent

    CNRS research engineer. His

    current research interests focus

    on the comprehension at the

    molecular scale of the adsorp-

    tion sites in acid materials and

    in particular in metal organic frameworks, using in situ and

    operando Infrared spectroscopy.

    Fre de ric Thibault-Starzyk

    Frederic Thibault-Starzyk was

    born in Saint-Lo (France) in

    1965. He received his PhD

    in synthetic organic chemistry

    in 1992 from the University of

    Caen. After post-doctoral

    work with Pierre Jacobs at

    the University of Leuven, hebecame Chargede Recherche

    (1995) and Director of Research

    in the CNRS (2009) at the

    Catalysis and Spectrochemistry

    Laboratory. In 20034, he was

    an overseas fellow, Churchill

    College, working with David

    King at the Chemistry Depart-

    ment, University of Cambridge. His research interests are in

    heterogeneous catalysis and infrared spectroscopy, including

    operando spectroscopy, time resolved measurements, and new

    spectroscopic approaches for the characterisation of solids and

    zeolites.

    CRITICAL REVIEW www.rsc.org/csr | Chemical Society Reviews

  • 8/12/2019 b919543m

    3/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4929

    probe molecules. We will describe the most accurate methodo-

    logies to obtain information on physicalchemical properties

    of the sites, as well as on their concentration and strength. We

    will show then how to differentiate an active site from a bare

    adsorption centre by coupling FT-IR with catalytic evidences.

    We will finally discuss briefly the problem of the detection

    limit of intermediate species (intrinsically linked to active sites)

    using infrared spectroscopy.

    Surface adsorption site identification on a catalyst

    The best identification of the potential active sites by IR

    spectroscopy is the direct detection of their IR fingerprint.The best examples are probably those given by the zeolites, for

    which the strong Brnsted-acid sites are often identified

    through the n(OH) band of hydroxyl groups. If this is not

    possible, specific methods must be employed to obtain a

    spectroscopic response for the sites. The most common is the

    adsorption of probe molecules, which provides IR spectra

    specific to the interaction with a single site. From the spectro-

    scopic point of view, several reviews have given criteria about

    the choice of the right molecule,35 However, without any

    further considerations, it is worthwhile noticing that whatever

    the probe molecule used, it generates a perturbation of the

    surface of the catalysts: indirect perturbations such as electron

    withdrawing effect, or direct perturbations like chemical reactions

    (protonation, electron transfer, decomposition, . . .). These

    phenomena can be considered as invasive, but according to

    the probe used and the chemical reaction considered, this can

    also give information on the dynamic response of a surface to

    the molecule adsorption. Therefore, it can be concluded that

    whatever the probe molecule used, the best one is the reactant

    itself, providing the same perturbation as during the chemical

    catalytic reaction. In this view, the best conditions for a study

    appear to be those during the reaction, i.e. the operando

    conditions (real in situ);6 we will develop this point towards

    the end of the review.

    The spectroscopic study of the adsorption sites on a catalyst

    surface requires a good knowledge of the surface state itself, as

    well as an excellent reproducibility of the characterisation, to

    fix the parameters and the limits of the surface description. In

    this view, it is necessary to be aware of the presence and nature

    of surface impurities, to establish how to clean the surface and

    make the sites accessible.

    Impurities and surface activation

    IR is a very sensitive technique for detecting surface impurities

    such as water, carbonates (formed by catalyst contact with

    ambient atmosphere), organics and other residual species after

    synthesis, such as nitrates, sulfates, templates, etc., because

    these species present characteristic bands in the IR spectra.

    Although many of these species do not represent a real

    problem for catalytic reactions, being often eliminated at the

    temperature at which the process takes place, or even taking a

    beneficial role in the reaction itself, their presence may inhibit

    (at least partially) probe adsorption, or simply disturb the

    correct spectral interpretation by band overlapping. Band

    position and intensity can vary depending on the sample

    nature; an abundant literature exists describing these spectra

    (see for example ref. 5). Thermal stability of these species

    strongly depends on acidbase properties of the material: on

    relatively acidic compounds (alumina, zirconia, . . .) carbonate

    and nitrate species can be removed at mild temperatures. On

    basic oxides, such as MgO, carbonate impurities can still be

    observed after a thermal treatment at 750 1C. In the case of

    sulfates, a reducing treatment at higher temperature is even

    necessary to clean the surface, but traces of sulfur can remain.7

    Nevertheless, we should consider the fact that these residual

    species are essentially bulk moieties, almost inert during the

    catalytic process; therefore it is better to leave them in

    place rather than to heat the sample at very high temperature

    (with the aim to clean all the impurities), so producing a

    drastic sintering of its surface.

    It is much more difficult to detect the presence of alien

    anionic entities such as sulfur, Cl and F, or cationic species

    Fig. 1 reaction profile and intermediates in the simple case of a

    process going through two elemental reaction steps A - I- B: a is

    the profile for a non catalytic thermal reaction; b is a catalytic reaction

    using a good catalyst; c is a catalytic reaction using a bad catalyst.

    If I is an intermediate with a low stability, it is hard to detect; if I 0 is a

    stable intermediate it will easily be detected. The transition state A* or

    I* is not detectable.1,2

    Marco Daturi

    Marco Daturi was born in

    Genoa (Italy) in 1964. After

    having obtained a Master in

    Physics, he studied Chemistry,

    receiving a PhD in Chemical

    Engineering in 1996 from the

    University of Genoa (director:Prof. G. Busca). After post-

    doctoral work with Dr J.-C.

    Lavalley at the LCS Catalysis

    and Spectrochemistry Labo-

    ratory, he became Lecturer

    (1998) then Professor (2002)

    at the University of Caen,

    where he teaches thermo-

    dynamics and spectroscopy.

    His research topics deal with in situ and operando IR spectro-

    scopy, heterogeneous catalysis, material surface investigations

    and design. He applies fundamental studies to the domains of

    pollutants abatement and environmental protection.

  • 8/12/2019 b919543m

    4/24

    4930 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    such as alkaline cations in the zeolites, due to the absence of

    specific bands. An indirect method consists in the observation

    of the n(OH) range of the catalyst spectrum: OH/Cl, OH/F

    substitutions and OH/ONa+ alkali exchange strongly modify

    the intensity and the position of the n(OH) bands. These

    impurities and contaminants strongly influence the acidbase

    properties of the samples. As an example, Cl, F and sulfate

    species can increase the strength of the acid sites on alumina,8

    whereas sodium (even if present in as small quantities as

    300 ppm) poisons the strongest Lewis-acid sites,9,10 Traces of

    chloride also hinder oxygen mobility on cerium-based com-

    pounds11 and inhibit the oxidation properties of the supported

    metals.12

    Effect of the temperature of activation

    The increase of the temperature of activation for metal oxides

    not only removes the impurities and molecularly adsorbed

    water but also leads to surface reconstruction, due to the

    elimination of hydroxyls, according to the mechanism involving

    two MOH groups: 2 MOH - MOM + H2O (with the

    possible creation of a vacancy). In extreme conditions, if the

    activation temperature exceeds the calcination conditions

    during sample synthesis, sintering phenomena can also occur.13

    The thermal treatment leads to surface reconstruction, and

    the amounts of exposed OH groups, cationic, anionic and

    defective sites, which govern their acidbase and redox properties,

    are not predictable, but can be estimated by IR absorption

    experiments of probe molecules at various temperatures. The

    most complete IR studies about the effect of the temperature

    of activation on the surface modification spectroscopy probably

    deal with MgO and alumina.1417

    Direct observation of the active sites

    Case of Brnsted-acid sites.Direct observation of the poten-

    tial active sites is possible when Brnsted-acid sites are con-

    sidered and this possibility is well illustrated by the zeolites or

    silica, for which the hydroxyl groups can be directly observed

    through their n(OH) bands. The spectrum of hydroxyl groups

    of steamed HY zeolite (Fig. 2) allows one to observe the acidic

    OH group located in different sites: bridged hydroxy groups in

    supercages (3625 cm1), or in sodalite cages (3545 cm1).

    Perturbation of bridged OH groups by extraframework

    entities generates highly acidic species with specific n(OH)

    band (3602 cm1 for those in the supercages, 3520 in the

    sodalite cages). For this zeolite, the presence of the 3602 cm1

    band is often related to the activity of the catalyst in strongly

    demanding reactions like n-hexane cracking,18,19 The OH

    bands in the dealuminated Y zeolites have been studied in

    detail and a new model has been proposed recently.20

    Such a

    sensitivity to the environment is well illustrated with the IR

    spectra of the mordenite, which presents three nOH bands,

    depending on the location (Fig. 2b).21 The spectra of the

    sample during the conversion of xylene show that only the

    hydroxy groups present inside the main channels act as active

    sites towards the isomerisation reaction. Sites located in the

    narrow side pockets are not catalytically active, but have a

    strong influence on selectivity. This case illustrates nicely the

    fact that the direct observation of the potential sites can give

    valuable indications on the reaction protagonists.22

    Thermally activated divided metal oxides present residual

    OH groups having IR stretching frequencies related to the

    nature of surface cations. The multiplicity of the n(OH) bands

    of oxides such as alumina has been mainly explained invoking:

    (i) the multifold coordination of the hydroxyls themselves

    (linear species giving rise to n(OH) bands at a higher frequency

    than those characterising bridged species), (ii) the coordination

    number of the cation to which OH groups are bound, (iii) the

    presence of morphologic defects on the surface (edges, corners,

    etc.). In the case of many oxides such as Al2O3, Ga2O3, CeO2,

    ZrO2, the higher n(OH) wavenumber component presents

    a basic rather than acidic character, as shown by CO2experiments.

    2325Their relation to defect crystallographic

    sites has also been invoked considering the sensitivity of this

    component to the presence of coordinated species in the

    neighborhood.24 This shows that the co-existence of several

    types of sites on the surface complicates the relation between a

    n(OH) band and a well defined Brnsted surface sites in the

    case of metal oxides.

    Case of Lewis-acid sites. Direct IR observation of a

    Lewis-acid site itself is not possible since a coordinatively

    unsaturated site is not a vibrator. Considering the Lewis site

    and its first coordination sphere, the vibrations (metaloxygen

    vibrations for metal oxides) are generally coupled with the

    very intense bands from the bulk vibration modes and only

    studied by mixing the samples with KBr; therefore only

    hydrated solids are generally studied. However in few cases,

    defect sites are detected in the IR spectrum of activated

    samples by specific IR bands, mainly in the low frequency

    range: bands situated at 3782/880 cm1

    on beta zeolite26

    and

    around 960 cm1 on Ti-silicalite27 can be considered as a

    fingerprint of Lewis-acid sites and have been partly related

    to the enhanced activity of these Lewis-acid catalysts in the

    catalyzed MeerweinPondorfVerley (MPV) reaction.28

    Surface AlO species (at ca. 1050 cm1) on alumina29 and

    strained siloxane bridges on highly dehydroxylated silica

    (880940 cm1) are dissociation sites for water, alcohols, H2S.30

    However, the number of cases in which direct characterisation

    Fig. 2 IR spectra of the hydroxy groups of a steamed Y zeolites

    (a) and of a mordenite zeolite (b) after activation at 450 1C.

  • 8/12/2019 b919543m

    5/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4931

    of Lewis acidity is possible is limited, and complementary IR

    adsorption experiments using probe molecules are often

    necessary.

    Probe molecule use.The direct detection of adsorption sites

    by IR spectroscopy is often not possible, therefore specific

    methods must be employed to obtain a spectroscopic response

    of the sites. The most common is the adsorption of probe

    molecules which gives IR spectra specific to the interaction

    with the site, as mentioned above. Concerning the choice of

    the appropriate probe molecule, Lercher et al.3 gave special

    emphasis to the criteria that have to be met to arrive at a

    characterization of materials that are useful for catalytic

    application, selecting the right molecule for the right site. The

    choice of the adapted probe will depend on many parameters:

    the chemical function providing the interaction with the site

    under study; the size of the molecule, depending on the site

    accessibility; the optimum of the interaction (sufficient to

    furnish valuable information, not excessive so as to limit

    surface modifications); the spectral response, producing a signal

    intense enough, with band positions sensitive to the interaction;

    stability on the surface catalyst to avoid decomposition; and

    sufficient vapour pressure to be easily introduced in an IR cell.

    The investigated sites are very often cations, acting as Lewis

    centres, but infrared spectroscopy has also been widely used

    in order to characterize the metallic centres in oxides and

    deposited metal complex. Even if direct investigations on

    metaloxygen vibrations are reported3133 most of the studies

    related with catalysis are dealing with the adsorption of probe

    molecules. Among these molecules, N2, methanol, NO and CO

    are frequently used and the latter two are also by far the most

    common in the literature. In 2002 a very complete review

    concerning the infrared spectra of chemisorbed carbon

    monoxide as a characterization tool for the cationic sites of

    oxides34 was published. A specific property of CO is that the

    slightly antibonding HOMO 5s orbital is occupied. This

    orbital is very important for the electron-donating properties

    of CO, because a decrease of electron occupation on this

    orbital leads to stabilization of the entire molecule and thus

    to an increase of the n(CO) wavenumber (compared to the

    n(CO) for the gas phase at 2143.5 cm1

    ). On the contrary, the

    addition of electron density from a metal d orbital to one of

    the 2p* LUMO orbitals (so called p back donation) leads to a

    substantial decrease of the vibrational frequency of CO, i.e.to

    a weakening of the CO bond. As far as the red-ox properties

    are concerned and in the ideal case, the expected information

    with CO as a probe are the following:

    oxidation state of the cations on the surface,

    coordination state of these cations,

    location of the cations on flat planes or other surface

    structures,

    location of the active phase on the support,

    surface phase analysis,

    existence of strong oxidizing agents on the surface.

    The characterization of the various sites on mixed oxides

    can be advantageously carried out by CO adsorption at

    various equilibrium pressures at low temperature, followed

    by evacuation at increasing temperatures to obtain infor-

    mation about the stabilities of the various species. Although

    the CO stretching frequency is the most informative parameter,

    the data determining the stabilities of the various species can

    be decisive for the assignment of the bands. Multiple carbonyls

    adsorbed on the same metal cation are possible and in order to

    identify them, isotopic mixtures should be used. This was the

    case, for example, of a PtNamordenite sample, where the

    use of a 12CO13CO isotopic mixture combined with analysis

    of the second derivatives of the spectra was very useful

    for proving the polycarbonyl structures.35 Again, using iso-

    topic labeled 13CO and 15NO molecules mixed with their

    most abundant analogues, it was possible to describe the

    multiplicity and symmetry of CO and NO ligands on a

    complex coordinated with Rh2+ in a Rh-ZSM-5 sample;

    geometrical structures and band assignments were supported

    by DFT computational results.36 However, sometimes the

    polycarbonyls are very stable and in this case, if 12CO is

    adsorbed first and then 13CO introduced, mixed species may

    not form at ambient temperature.

    Concerning NO, Hadjiivanov37 reported that the coordi-

    nation of the NO molecule to a cationic site via the nitrogen

    atom is accompanied by a partial charge transfer from the

    5sorbital together with an increase in the bond order, just as

    in the case of CO. Formation of a p-back-bond, although not

    as easy as with CO, is also possible, and this results in a

    decrease in the NO stretching modes. The different surface

    mononitrosyls absorb in a wide spectral range: 19661710 cm1.

    When only a s bond is formed, a frequency above that of

    gaseous NO (1876 cm1

    ) is expected, whereas with low-valent

    cations, rich in d-electrons, p-back donation is possible and

    the NO stretching modes can fall below 1876 cm1. Cations

    having no d-electrons produce mononitrosyls only; on the

    contrary, dimeric molecules are very often the principal

    adspecies on transition metal cations. This is the case of NO

    adsorption on V-, Cr-, Mo-, W-, Fe-, Cu- and Co-containing

    oxide systems where the metal cations are not in their highest

    oxidation state. Thus, it is clear that a p-back donation

    stabilizes dinitrosyls. Examples of complementary information

    obtained by CO and NO coadsorption are also available.38

    Acid sites. A review by Busca39 describes the bases of IR

    spectroscopic methods for the characterization of the surface

    acidity (both of Lewis and Brnsted type), on different mixed

    oxides. A systematization is proposed associating the surface

    acidity with the ionicity/covalency of the elementoxygen

    bond, mainly affected by the size and charge of the cation.

    In a parallel work, the results obtained for the characterization

    of the Lewis-acid strength of more than 30 binary and ternary

    mixed oxides are interpreted on the basis of the different

    polarizing powers of the involved cations.40

    Recent progress

    on acidity characterization is reviewed elsewhere and

    described to be related to the broadening of the spectral range

    (investigation of overtones, combination bands, and low-

    frequency modes) and to the adsorption of new non-traditional

    probe molecules for identification of acid sites.41,42

    Some examples on oxide and zeolitic compounds.The choice

    of the probe molecule is crucial to obtain an overall view of the

  • 8/12/2019 b919543m

    6/24

    4932 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    acidity. Its size has to be small enough to interact with all

    available sites and to avoid confinement effects43 but its basic

    strength has to be strong enough to interact even with the

    weakest acidic sites. Ammonia seems to be a good candidate

    for this but due to the high polarity of the NH bonds,

    hydrogen bonding with basic entities governs the coordination

    of adsorbed species and direct conclusions about acidic

    strength are not straightforward.44 That is why most often

    the adsorption of several probe molecules is required. For

    example, the FAU and the MOR structures are made of big

    and small cavities in which some acidic hydroxyls are out of

    reach of basic molecules such as pyridine. Co-adsorbing the

    strongly basic trimethylamine (TMA) and NH3(see Fig. 3), we

    were recently able to give for the first time an infrared evidence

    of three distinct acidic hydroxyls in defect-free HY,45 to give

    an assignment for the corresponding wavenumbers and to

    characterize their respective acidic strength.46 Moreover,

    TMA desorption associated with the recovery of hydroxyls

    at 3656 and 3638 cm1

    (and two corresponding n(NH) bands

    reveal the presence of at least two distinct acidic strengths for

    the hydroxyls located inside the supercages. For the same site

    location, the local chemical factor should then play a role: the

    aluminium distribution in the framework is not necessarily

    homogeneous, and the number of Al next-nearest neighbours

    influences the acidic strength of a given site. Another explana-

    tion for the unusual 3656 cm1 component could be that part

    of the O4

    crystallographic sites is a proton holder for this low

    Si/Al ratio HY sample; in such a case, all the four theoretically

    forecasted sites in the zeolitic FAU structure would have been

    observed by IR spectroscopy.46

    The combined use of these

    two molecules moreover helped us to better characterize the

    various coordinated NH4+ and determine the activity ranking

    between the ammonium species and coordinated ammonia

    over Lewis sites during NOx

    SCR.47

    In the zeolites, some strong Lewis-acid sites can be obtained

    during steaming leading to the formation of extra-framework

    aluminium species. Mild Lewis sites may be naturally present

    (in the alkaline form) or generated upon ionic exchange with

    transition metals which are necessary for NOx

    SCR with

    hydrocarbons. For over-exchange level, some Lewis species

    may remain on the external surface and coadsorption of the

    bulky ortho-toluonitrile and CO was recently reported to

    identify the different Con+ species and their location in a Co

    H-MFI zeolite.48 In this respect, the use of the nitrile probe

    provided more precise evidences about the distribution of Co

    species in active CH4-SCR Co-HMFI than those arising

    from previous UVvis, EXAFS and XRD data.4953 The

    oTN (ortho-toluonitrile) and NO co-adsorption allowed to

    determine that a significant amount of cobalt species is at the

    external surface, mostly in the form of divalent cobalt. On

    the other hand, in the internal surface part of Co species are

    trivalent, together with predominant divalent Co ions. These

    observations coupled with those coming from the operando

    study were valuable for reactivity explanation, inferring that

    the active sites for CH4-SCR in Co-MFI are Co3+ species

    (presence of a nitrosyl n(NO) band at 1930 cm1) located in

    the cavities, but likely in non-classical cation positions, which

    are able to convert NO to an adsorbed bridging nitrate species,

    that can be later decomposed to give gas phase NO2.54

    Moreover, the cavity may contribute to the stabilization of

    aggregates containing trivalent cobalt. At the same time, the

    presence of Co-isocyanates involved in the SCR suggests that

    a possible route for the reaction implies the reduction of

    nitrate-like species by methane, forming H2O and isocyanates,

    which could later react with NO producing N2 and CO2. On

    the contrary, it seemed that substitutional Co2+ ions did

    not play a key role in the reaction, being very likely almost

    redox-inactive. Co2+-dinitrosyls formed on them being

    decomposed well below the reaction temperature, they did

    not seem to be involved in the reaction.54

    These considerations

    linking the active site with the reactivity of the species

    coordinated on it and with the possible intermediates can be

    generalized to all the reactions and will be discussed later in

    this review.

    Metalorganic framework: the ideal case for spectroscopic

    identification of adsorption sites. In the case of the metal

    oxides, surface relaxation and reconstruction phenomena

    might not allow to accurately identify the nature of the

    Lewis-acid sites (coordination, oxidation degree, unsaturation

    degree) via the study of spectra after the adsorption of probe

    molecules. On the contrary, this is not the case for the majority

    of hybrid organicinorganic solids such as MOFs (metalorganic

    frameworks), which are crystalline nanoporous solids, with a

    framework of inorganic units (clusters, chains or planes) and

    organic linkers (phosphonates, carboxylates, sulfonates). The

    majority of MOFs present framework metal sites that can

    exhibit coordinative vacancies upon solvent removal; such

    sites represent Lewis-acid centers of well defined symmetry

    and oxidation degree. As has been shown in the case of

    MIL-10055 and H-KUST56,57 it is possible to assign in a clear

    and indisputable way the bands due to the adsorbed species,

    like those (still debated nowadays) for carbonyls or nitrosyls

    coordinated on cations such as chrome, copper and iron. CO

    adsorption on MIL-100/101(Cr) is an example of the con-

    tribution of IR to the identification of the potential active sites

    in MOF.55,58,59 Three n(CO) bands are observed at 2207,

    2200 and 2193 cm1 (Fig. 4), showing that Cr3+ sites are

    Fig. 3 Infrared spectra of the HY sample upon TMA adsorption and

    NH3saturation evidencing OH in the supercages at 3637 cm1, OH in

    the sodalite units at 3548 cm1 and OH in the hexagonal prism at

    3501 cm1 (from ref. 45). Reproduced by permission of the American

    Chemical Society.

  • 8/12/2019 b919543m

    7/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4933

    not equivalent. This heterogeneity is attributed to the possible

    presence of fluoride ions (synthesis being performed in the

    presence of hydrofluoric acid) on the metallic trimers, with 2, 1

    or no fluorine ion, respectively, in the neighborhood of the

    coordinatively unsaturated (CUS) Cr3+ considered site.55

    Quantitatively, the number of free Cr3+ sites in the activated

    compound is exactly that expected, considering that onecorner over the three octahedra is occupied by one anion.

    On MIL-101(Cr), this methodology allows identifying these

    sites as the grafting centres of catalytically active sites.59

    Relationship between probe molecules and activity: the case of

    acetonitrile. Two possible aspects of acidity are generally

    considered. The first one is the hydrogen bond that can be

    established between a Brnsted site and the basic probe

    molecule (for example when carbon monoxide is interacting

    with acidic zeolites at liquid-nitrogen temperature). The second

    one is the extent of proton transfer, or more exactly the

    amount of protonated probe molecule (e.g. pyridine) on the

    surface of the catalyst at room temperature. These two facets

    of acidity, however, can not reliably be used for explaining the

    catalytic activity in acid-catalyzed reactions. The H-bond

    between a basic molecule and the various acid sites in a solid

    is strongly influenced by temperature. The linear relation-

    ship between the strength of a H-bond with CO at 100 K

    (often used for comparing solid catalysts) and the activation of

    proton transfer to a hydrocarbon in a chemical reactor at

    700 K is far from being established. Acetonitrile has been

    increasingly used as an infrared probe molecule for solid

    catalysts, and might well provide an integrated approach of

    acidity. Both H-bond and protonation can be observed, and it

    can be used to probe the actual proton transfer in reaction

    conditions.

    In the absence of water, heating acetonitrile on an acidic

    zeolite leads to the reversible protonation of the probe

    molecule.60

    Protonation of acetonitrile on zeolite Brnsted

    sites at high temperature has been used to build an acidity

    scale agreeing with catalytic activities in the conversion of

    saturated hydrocarbons: acidity is determined under conditions

    near to that of catalytic reactions (high temperature), leading

    to a more relevant parameter for the prediction of catalytic

    activity than H-bonds. Acid catalysis involves protonation

    of the reactant, and a scale built on actual protona-

    tion, preferably in reaction conditions, is more interesting.

    The protonation temperature was measured on a series of

    zeolites, and depends very much on the pore size.61 The nearer

    the pore size is to the size of the adsorbed molecule, the lower

    the protonation temperature. For example, in mordenite, two

    main locations exist for the Brnsted site: in the main channels

    and in the side pockets. From the point of view of CO at

    low temperature, a stronger H-bond is created in the main

    channels, and the stronger site would therefore be in the main

    channels. However, protonation has only been observed in the

    small side pockets, where maximum confinement takes place.

    The measurement of the protonation temperature is a way to

    know how easy the proton transfer is from the acid catalyst

    to the basic adsorbed molecule. It is therefore a new spectro-

    scopic measurement for the acidity of the solid, one that

    does not only involve the interactions strength between the

    adsorbed molecule and the surface, but the actual proton

    transfer, the real nature of acidity. The probe is here not

    the adsorbed molecule itself, but rather the actual catalytic

    protonation reaction. The new scale obtained between various

    zeolites was compared to the activity in catalytic conversion of

    saturated hydrocarbons at high temperature, a reaction where

    activity is linked to the acid strength of very strong sites.

    Contrarily to what was observed with H-bonding of CO or

    with pyridine protonation at room temperature, the scale of

    acetonitrile protonation temperature was perfectly linked with

    the catalytic activity.62

    Modification of the basicity of the probe molecule by the acid

    site.Acetonitrile has also shown that the basicity of the probe

    is influenced by the adsorption on the solid, especially on

    zeolites where confinement is important. During molecular

    dynamics simulations of acetonitrile on mordenites,63 the

    electric dipole on the molecule was significantly modified in

    the move from the main channel to the inside of the small

    lateral cavity. In the purely siliceous mordenite used for the

    simulation, the molecular dipole of acetonitrile was enhanced

    from 3 D (in the large channels or in the gas phase) to 4 D in

    the small cavities. Such enhanced dipole could also imply that

    the basicity of acetonitrile would be increased by a third on

    entering the small cavities. Moreover, the probe molecule is

    locked in the cavity, and does not escape easily. It is thus

    kept in close distance to the OH group, thus enhancing the

    probability for the proton transfer (Fig. 5).

    When the size of the molecule compares with that of the

    cavity, electron densities in the basic molecule are modified,

    the basicity of the probe can be enhanced, and the proton

    transfer can happen more easily. All these parameters show

    that the basicity of the probe molecule can not be considered

    without the solid, and that it is always a pair at work.

    Modification of the acidity of the site by the probe molecule.

    In some cases, the probe molecule itself modifies the surface

    acid sites: during comparative study of the Lewis acidity

    between SiO2B2O3 and alumina by adsorption of pyridine

    (py), CD3CN (MeCN) and CO, no coordination of carbon

    monoxide has been observed on silicaboria even at low

    temperature whereas CO strongly coordinates on alumina.64

    Similarly, coordinated pyridine and acetonitrile show a

    much lower thermal stability on SiO2B2O3 than on Al2O3,

    Fig. 4 Left: CO adsorption sites of the trimers of chromium

    octahedra in MIL-100(Cr). Right: n(CO) bands of CO adsorbed on

    MIL-100(Cr).55

  • 8/12/2019 b919543m

    8/24

    4934 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    indicating that silicaboria presents much weaker Lewis-acid

    sites than g-Al2O3. On the other hand, coordinated pyridine

    and acetonitrile species show that infrared frequency shifts

    (n8a, n19b and n(CN), respectively) are larger on B2O3SiO2than on Al2O3, suggesting that charge transfer from these

    probe molecules is more important on the B3+ than on the

    Al3+ Lewis acid (Fig. 6). DFT calculations of the interaction

    of these probe molecules with models representing Al3+ and

    B3+ Lewis-acid sites adequately reproduce these experimental

    observations (Fig. 7). The weakness of the B3+

    Lewis-acid

    sites is ascribed to the p-character of BO bonds, which

    disfavours the conversion of boron from a trigonal planar

    conformation to a tetrahedral conformation upon adsorption

    of probe molecules and decreases the adsorption energy of

    pyridine and acetonitrile despite a strong charge transfer.

    The absence of interaction noted during the adsorption of

    CO on SiO2B2O3 has been explained by its basicity, not

    strong enough to compensate for the energy required for the

    conformational change of the B3+ Lewis-acid centre.

    Transformation LewisBrnsted sites. Brnsted acidity can

    be generated by water on the surface of crystalline or amorphous

    solids. On metal oxides, it is well known that water can

    transform Lewis into Brnsted acidity by dissociative water

    adsorption on the Md+Od acidbase pairs, with the con-

    sequent creation of MOH acidic groups. On aluminium

    fluorides, water addition transforms Lewis-acid sites into

    Brnsted sites on activated compounds.65

    Due to the strong

    Lewis acidity, water is strongly coordinated on fluorides and

    generates strong Brnsted sites, as evidenced by CO adsorp-

    tion. The absence of sufficiently strong basic sites on fluorides,

    as shown by an unpublished study of propyne adsorption,

    inhibits dissociative adsorption of water on such materials.

    The consequences of the adsorption of protic molecules

    (water and alcohols) on the acidity of MOFs, have been well

    identified by IR spectroscopy:55,58 adsorption of water on

    activated MIL-100(Cr) leads to the formation of coordinated

    species well characterized by two narrow n(OH) bands at

    about 3700 and 3580 cm1. CO adsorption at 100 K shows

    that coordinated water induces the creation of Brnsted-acid

    sites with a strength close to that reported in the case of

    phosphated silica. The successive addition of CO molecules on

    hydrated MIL-100(Cr) shows that each coordinated water

    molecules creates two Brnsted-acid sites (see Scheme 1).

    This LewisBrnsted acid conversion was used to modulate

    the strength of the created Brnsted-acid sites according to the

    Fig. 5 Molecular dynamic simulation of acetonitrile in a purely

    siliceous mordenite. The Ycoordinate corresponds to the distance of

    the probe molecule from the centre of the main channel. The Z

    coordinate corresponds to the distance along the main channel.

    (A) Positions of the centre of gravity of acetonitrile during the

    simulation, showing it can go from the main channels to the side

    pockets. (B) Molecular dipole and (C) location of the molecule during

    the simulation vs. distance from the main channel in the mordenite

    structure along the axis of the side pocket (ZB 2 A ) (adapted from

    ref. 63). Reproduced by permission of the American Chemical Society.

    Fig. 6 IR spectra of activated Al2O3(top) and boriasilica (bottom)

    after introduction of pyridine at room temperature (left) and CO at

    100 K (right-spectra at various coverage).64

  • 8/12/2019 b919543m

    9/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4935

    nature of the adsorbate used: the comparison of the results

    deduced from the grafting of CH3OH, H2O, CF3CH2OH and

    (CF3)2CHOH shows that the stronger the acidity of the

    adsorbate, the higher the acidity of the generated Brnstedsites. Their strength is directly related to the nature of the

    grafted molecules and can reach that of the zeolitic Brnsted-

    acid sites when fluorinated alcohols are used (see Fig. 8).58

    Basic sites. Heterogeneous catalysis using basic solids has

    been much less studied than acidic catalysis. It seems even that

    a concern exists about the definition of basicity itself. It may be

    useful to recall here that, in spite of the common use, there is

    not a physical difference between the so-called Brnsted and

    Lewis basicity, because for both it is due to electrons on

    oxygen atoms. In fact, for Brnsted, an acid site is a proton

    donor (i.e. an hydroxyl on heterogeneous catalysts), whereas

    Fig. 7 Evolution of the total interaction energy (DEtot, ), deformation energy of the Lewis-acid center (DEdef(A), ), deformation energy of the

    probe molecule (DEdef, --), and interaction energies of the probe with the Lewis center at their geometry in the complex (DEint, -- -) for six acidbase

    complexes as a function of the ML distance (M = B or Al; L = C or N). 64 The deformation energy is the difference between the energies of the

    isolated species at their equilibrium geometries and the energy of the isolated species at their geometry in the complex.

    Scheme 1 Interaction of CO with coordinated water molecules in

    MIL-100(Cr).58 Reproduced by permission of the American Chemical

    Society.

  • 8/12/2019 b919543m

    10/24

    4936 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    for Lewis it is an electron pair acceptor (i.e. a cation); but

    looking at basicity definitions, in the sense of Brnsted a basic

    site is a proton acceptor, while for Lewis it is an electron pair

    donor, therefore an oxygen site in both cases, or an halogen

    atom for halogen based catalysts. Some simple concepts

    addressing this problem have also been mentioned by

    Zecchinaet al.66

    Concerning the difficulties in studying surface basicity, we

    might explain it with the fact that FT-IR is a technique

    sensitive to molecular bonds: in the case of acidity, when using

    a probe molecule to characterize Lewis sites, the molecular

    deformations of the molecule induced by the cationic polarizing

    effect are measured. In the case of a basic site, the surface is

    donating electrons and the back-donation effect on a molecule

    is a more complicated phenomenon to quantify. When studying

    proton donation or acceptance, the spectra are always com-

    plicated by broad and badly resolved features. From a general

    point of view, anyway, protonic molecules seem particularly

    adapted for probing basic sites, taking into account the

    basicity definition itself (and impacting them only by weak

    interactions).

    As an acid site is always associated to its conjugated basic

    site, reviews dealing with characterization of acidity by infrared

    also report interesting data about basicity (see for example

    ref. 41) and even describe some typical probe molecule inter-

    actions: CH-acids such as chloroform (Cl3CH(D)), acetylene

    (C2H2) and methylacetylene (CH3C2H) are shown to be

    potentially suitable probe molecules for basic properties using

    the H-bonding method.67 All three molecules undergo

    Oz2 HC hydrogen bonding and the induced red-shift of

    the CH stretching frequency permits a ranking of the base

    strength of a given series of materials. Many other probe

    molecules were tested for the specific study of the surface

    basicity of divided metal oxides, and Lavalley reviewed some

    years ago the infrared spectrometric studies of the surface

    basicity of metal oxides and zeolites using adsorbed probe

    molecules.4 Results obtained from carbon monoxide (CO),

    carbon dioxide (CO2), sulfur dioxide (SO2), pyrrole (C4H5N),

    chloroform (CHCl3), acetonitrile (CH3CN), alkanes, thiols,

    boric acid tri-Me-ether, ammonia (NH3), and pyridine

    (C5H5N) were discussed in that well cited review. As we

    already noticed in the case of the acidity study, the author

    reminds us that no probe can be used universally. CO2 for

    weakly basic metal oxides and for basic OH groups, CO for

    the characterization of highly basic structural defects on metal

    oxides activated at high temperature and pyrrole in the case of

    alkaline zeolites, appear to be quite suitable probes. Moreover,

    both NH3 and pyridine (generally used as probes for the

    measure of the acidity of catalysts) are also described to

    adsorb on basic oxides via dissociative chemisorption. In this

    respect we should stress that a strongly interacting probe (such

    as CO2 and NO2, for example) highly modifies the surface

    behaviour, often leading to false interpretations on the

    strength of the basic site. Moreover, a probe which dissociates

    is an unfriendly tool for site concentration characterisation

    when using volumetric methods. Again, a non-dissociating,

    weakly interacting protonic probe would appear as a most

    adapted tool for surface basic site characterization.

    A pioneering review on the basic properties of zeolites was

    proposed by Barthomeuf.68 More recently, NO2disproportio-

    nation on alkaline zeolites was used to generate nitrosonium

    (NO+) and nitrate ions whose infrared vibrations are shown

    to be very sensitive to the cation chemical hardness and to the

    basicity of zeolitic oxygen atoms.69

    Recently, Michalska et al.70 have pointed out that propyne

    is an excellent probe for the study of oxygen basicity. Addi-

    tionally, they have verified that probe dissociation does not

    depend on the site strength, but can be due to the presence

    of Lewis-acid sites coupled with the basic ones: the formation

    of the hydrogen bond weakens the bond betweenRC and H.

    Fig. 8 Brnsted-acid strength of OH groups from various grafted species on MIL-100(Cr) measured by CO adsorption: correlation between the

    n(OH) shifts, the H0 values and the n(CO) position.58

    The y axis label should be read as |Dn(OH)|. Reproduced by permission of the AmericanChemical Society.

  • 8/12/2019 b919543m

    11/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4937

    In the presence of an acid site, which can host the CH3CRC

    moiety, the surface protonation is therefore easily achieved.

    Therefore, propyne dissociation is a good probe for the

    presence of acidbase pairs on a surface.70

    Basicity case study. Among the catalysts having basic

    properties, ceria is probably the most common owing to its

    properties in the domains of oxidation and hydrogen produc-

    tion. It is the base material for car exhaust control devices,

    notably for Three Way Catalyst (TWC) applications. For

    these reasons an extensive characterisation of its properties

    has been undertaken for at least two decades. Pyrrole adsorp-

    tion on CeO2leads to dissociative adsorption characterized by

    stretching ring vibrations at 1444 and 1367 cm1 typical of

    pyrrolate ions and n(OH) vibration at 3628 cm1 typical

    of surface hydroxyls formed upon proton transfer.71

    This

    complete dissociation of C4H5N is indicative of the high

    basicity of CeO2 surface O2 ions but does not allow an

    investigation of its variation upon reducing ceria. CO2 was

    further adsorbed as it acts as a Lewis acid toward either O2

    surface ions (with the production of carbonates) or residual

    basic OH surface species (with the production of hydrogen

    carbonates (HC)). The study extended to CeZr mixed oxides72

    indicates that hydrogen-carbonates (indicative for basic OH

    groups) are mainly observed for rich ceria compounds and

    that the intensity of carbonate species is directly proportional

    to the cerium content, as expected according to the basic

    properties of this element. Identification of the spectral

    features typical of each species arising from CO2 adsorption

    was clarified by studying the splitting of the n3 band of

    carbonates and their thermal stability.

    Nowadays, basicity is becoming a more and more important

    parameter, and basic materials are involved in numerous

    industrial processes, such as fine chemical productionviagreen

    chemistry routes (replacing homogeneous by heterogeneous

    procedures, for example in esterification reactions in the

    absence of any solvent), or environmental catalysis. One of

    the most investigated methods for nitrogen oxide removal is

    NOx

    -trapping. In this process, NOx

    are stored and concen-

    trated in highly basic compounds, before being submitted to a

    reduction. The materials used in this respect are typically

    barium, alkaline metals, alkaline-earths and rare earth

    oxides.73 In such a case CO2 and NO2 are the most adopted

    probe molecules, giving rise to carbonates and nitrates,

    respectively. The storing materials are obviously submitted

    to severe surface and bulk restructuration, but in the same way

    they will be under service, exposed to the reacting lean and

    reach flows, containing NOx

    and high concentrations of CO2.

    General indications on nitrate characterization and their

    coordination on a number of solids can be found in the

    excellent review by Hadjiivanov.37

    A comparative study of

    barium and potassium based formulation74 indicated upon

    NO2adsorption the formation of both ionic and covalent-like

    NO3 species over PtRh/Ba/Al2O3, whereas only very stable

    ionic potassium nitrates (sharp peak at 1373 cm1) were

    detected over Pt/K/Mn/CeAl2O3. This is due to the higher

    basicity of the potassium sites which furthermore enlarges

    the adsorbing temperature window and delays the nitrate

    release during the rich step impeaching NO sudden outlet.

    A comparison of the latter composition with a great number

    of other NSR catalysts was reported in ref. 75, where the exact

    chemistry of nitrite and nitrate formation was investigated,

    and their coordination on specific structural sites of the oxide

    determined thanks to the parallel use of TEM analyses.

    Another interesting example is the synthesis of phyto-

    sterol esters from transesterification of a fatty methyl ester

    (dodecanoate) with b-sitosterol carried out in the presence of

    basic solid catalysts, such as lanthanum oxides.76 The

    acidbase properties of La2O3 were characterized by IR

    spectroscopy, which revealed the presence of residual unidentate,

    bidentate, polydentate, and mineral carbonate species inside

    all of the solids even after activation, suggesting different basic

    and catalytic characteristics of the samples. The carbonate

    strengths were determined by propyne adsorption, using the

    shift of the n(CRC) stretching mode for the adsorbed species:

    the lower the position of that vibration, the greater the basicity

    of the corresponding site. A ranking of the basic strength of

    the surface carbonates species of the lanthanum oxycarbonate

    samples was thus possible and it was correlated to the catalytic

    activity: the lower the basicity of carbonates, the higher the

    phytosterol ester yield. Moreover, a thorough spectral

    characterization of KBr-diluted samples indicated that the

    higher the intensity of unidentate carbonate bands (1499 and

    1382 cm1), the higher the sterol ester yield, suggesting those

    moieties of medium basic strength play a central role in the

    catalytic mechanism.

    It was shown in our laboratory that COS hydrolysis77 and

    CS2 hydrolysis78 could be used as test-reactions for metal

    oxide hydroxyl basicity. However, the use of IR spectroscopy

    should add value to this form of characterisation: adsorption

    of CS2 on a series of metal oxides (Al2O3, ZrO2, ZnO and

    CeO2) gave rise to a specific interaction with O2 sites, leading

    to the formation of xanthate (COS2

    )2 species characterized

    by bands in the 12001000 cm1 range whose intensities

    correlate well with the relative basicity of the analyzed oxides.

    Co-adsorption experiments of CS2with either CO2or pyridine

    showed that its adsorption sites are mainly those giving rise to

    bidentate carbonates, showing that this new probe is more

    specific than CO2. Moreover, the transformation of xanthate

    into carbonate species at low temperature could account also

    for the surface oxygen mobility.79

    Metallic and redox sites. Transition metal oxides, rare-earth

    oxides and various metal complexes deposited on their surface

    are typical catalytic phases leading to redox properties. For

    each of these phases, complementary tools exist for an appro-

    priate characterization of the metal coordination number,

    oxidation state or nuclearity. Among all the techniques, IR

    provides information by characterizing the characteristic

    vibrations of intrinsic (hydroxyls) or extrinsic (methanol,

    CO, . . .) probes.

    We have already discussed the principles of the use of probe

    molecules for the characterisation of surface species in the

    section concerning probe molecule use. CO and NO can

    provide highly valuable information on the supported metal

    dispersion and coordination, as well as on the oxidation degree

    of such moieties. Typical examples can be found in the

    works of Binet8083 and Bazin.84,85 This methodology appears

  • 8/12/2019 b919543m

    12/24

    4938 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    particularly adapted to the case of redox supports, where the

    alternative hydrogen chemisorption will lead to imprecise

    results. Another positive point for the use of such probes

    is their small size, allowing site accessibility even in small

    cavities. Moreover, the fair interaction energy between the

    probe and the site is a guarantee for a minor perturbation

    of the surface state upon interaction. A useful volumetric/

    spectroscopic CO adsorption combined method allows

    metal dispersion calculation in a simple and reliable way, by

    integrating the bands relative to CO adsorption on the metal

    sites vs. the molar amount of the introduced probe, as shown

    in Fig. 9.85,86 In such a way CO adsorption mode, sites and

    quantity are continuously monitored by IR spectra upon

    calibrated doses introduction.

    Additionally, carbon monoxide and nitrogen monoxide

    adsorption can provide very useful information on the coordi-

    nation mode of such molecules (of primary importance for

    environmental issues) and on the complexes formed with

    the metal particles, potential sites for pollutant abatement.

    For example, it was found that the adsorption of CO on a non-

    reduced Pt/TiO2sample reveals the existence of Pt3+ and Pt2+

    cations as well as some amount of metallic platinum. However,

    Pt4+ species are also present on the sample but, being

    coordinatively saturated, cannot adsorb CO, while NO forms

    nitrosyl species with bare Pt2+ and Pt0 sites, but it is not

    coordinated to Pt3+ ions. Therefore, it appears that NO is a

    more sensitive probe than CO for testing the state of Pt2+

    cations. Moreover it must be underlined that probe adsorption

    is not totally innocent: adsorbed CO slowly reduces the

    platinum cations, whereas NO oxidizes metallic platinum

    even at ambient temperature.87

    Similar results were found on

    Rh-ZSM-5, where a new kind of rhodium gem-dicarbonyls

    was discovered. The shift of the n(CO) vibration allowed under-

    standing Rh position in the porous structure, its oxidation

    state and the capacity to host different chemical species having

    different stability, especially in the presence of water. These

    data are fundamental for understanding the mechanism of

    different catalytic reactions.88 According to this methodology,

    CO and NO adsorption on zeolithe supported Rh nano-

    particles containing different promoter elements permitted to

    both characterise the effect of the additive and the catalytic

    activity of the noble metal.89

    Many studies were carried out on copper oxidation state,

    due to its importance when inserted in ZSM-5 zeolites for NO

    reduction. Its characteristic carbonyl band at 2158 cm1

    provides quantitative results on integrating its molar extinc-

    tion coefficient.90 Hadjiivanov et al.91 described the water

    effect (which is always present during deNOx real conditions)

    and reported that bands at 2158 and 2134 cm1 may charac-

    terize CO bound to dry and wet Cu+ centres, respectively,

    the latter being also possibly assigned to a CuO-like phase.

    However a comparative study using both Cu-ZSM-5 and

    CuO/Al2O3 allowed Praliaud et al. to propose the n(CO)

    at 21232133 cm1 to be due to non-isolated Cu+ species

    (Cu+

    surrounded by Cu2+

    ions) arising from the partial

    reduction of bulk CuO. Whereas, the band at 21522157 cm1

    would characterize isolated Cu+ ions, which are described to

    be responsible for the high activity in NO reduction into N2.92

    Concerning NO, its adsorption even at room temperature may

    lead to Cu+ oxidation to Cu2+ and therefore its use for the

    determination of copper oxidation state distribution is rather

    problematical. However the formation of Cu+ mono-

    and dinitrosyls is observable for high temperature (770 K)

    Cu-ZSM-5 activated under vacuum and Datka et al.93 even

    reported the possible existence of two distinct Cu+

    NO

    species at 1812 and 1825 cm1

    , associated to two distinct

    Cu+ sites differing in the density of oxygen packing. A

    discussion around Cu+ carbonyls can be also found in

    the Strauss94,95 and Zecchina96 reports. Further detailing the

    characterization of Cu+ by NO, it has been shown that the

    zeolitic structure also influences the symmetry of the dinitrosyl

    species and for a CuMOR sample the Cu+NO (1813 cm1)

    transformation into Cu+(NO)2 leads to different spectral

    feature depending on the Cu+ location: a doublet with

    ns and nas at 1828 and 1730 cm1 in the main channels

    and another doublet with ns and nas at 1870 and 1785 cm1

    in the constrained side-pockets.97

    The use of both probes can also be interesting when each one

    is sensitive to an oxidation state of an element, as in the case of

    copper, probed by NO in the Cu2+ state, whereas CO adsorp-

    tion would be more specific to the Cu+ state (since CO com-

    plexes with Cu2+ are only stable at very low temperature).98

    Another example concerns the adsorption of both CO and

    NO as informative for the determination of the oxidation state

    of vanadium on vanadiatitania catalysts. Although the

    carbonyl bands for V4+

    CO, V3+

    CO and Ti4+

    CO species

    almost coincide,99 the fact that NO forms dinitrosyls with Vn+

    but not with Ti4+ allows the effective use of NO as a probe

    molecule.100

    Iron-containing ZSM-5 catalysts were also studied for their

    potential application in deNOx by hydrocarbons. Interesting

    results identifying different Fe2+ species as active sites for

    NOx

    SCR with propene can be found in ref. 101 and 102. Fe3+

    characterization in the Y structure was also previously studied

    Fig. 9 Integrated intensity of the n(CO) bands bound to platinum as

    a function of introduced CO amount (reproduced from ref. 85). The

    abscissa of the line intersection corresponds to a CO monolayer on

    the exposed metal surface. This allows calculating the number ofaccessible Pt atoms,i.e.the metal dispersion. Reproduced by permission

    of Elsevier.

  • 8/12/2019 b919543m

    13/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4939

    using CO adsorption,103 even if IR spectroscopy of probe

    molecules is more suitable for characterization of Fe2+ than

    Fe3+ cations. More recently, we tried to resolve this inter-

    pretation conflict present in the specialized literature, reporting

    NO adsorption followed by infrared spectroscopy to charac-

    terize iron cations in Fe-ferrierite.104 Different iron sites

    forming mononitrosyl species were identified. The comple-

    mentary use of Mo ssbauer spectroscopy enabled us to deter-

    mine that the iron oxidation state is essentially +2. For low Fe

    loading (by ion exchange), the main fraction of Fe2+ cations is

    suggested to be located in highly accessible positions of the

    ferrierite, where ionic exchange takes place in the easiest

    way. When the amount of Fe is increased, a second site in a

    less accessible position is detected. When an oxidative

    pretreatment is applied, only the iron cations in the confined

    positions lead to the formation of Fe3+OH species. More-

    over, NO appears to be able to form polynitrosyl species with

    these confined Fe2+ cations. It thus appears that both the

    oxidation and the coordination states of confined Fe2+ may

    change easily, which makes them excellent candidates for

    active redox sites.104 Combining CO and NO as molecular

    probes, we were able to go into very fine detail for site

    characterisation in Fe-FER. It was ascertained that type I

    sites are the most populated and the Fe2+ ions located in the

    so-called G-positions are the most symmetric ones. Their

    unique geometry allows two guest molecules approaching

    the site from different cages, resulting in the formation of

    monocarbonyls (2195 cm1

    ), converted then, at low tempera-

    tures, into dicarbonyls (2188 cm1). Type I Fe2+ cations are

    hardly sensitive to oxidizing treatment and show little tendency

    to yield Fe3+ cations. Type II sites are less populated and less

    symmetric: Fe2+ ions in these B-positions form, with CO,

    monocarbonyls (2189 cm1) and, with NO, mononitrosyls

    (1880 cm1) practically coinciding in wavenumber with the

    nitrosyls formed with type I Fe2+ ions. These cations are

    sensitive to oxidizing treatment and are easily oxidized to

    Fe3+ ions most probably associated to the formation of

    a-oxygen species. When the iron concentration in the samples

    increases, a third site (F-site) is occupied. Iron ions in this

    position change easily and reversibly their oxidation state from

    Fe2+ to Fe3+ thus forming Fe3+OH or Fe3+O species.

    When in the Fe2+ state, iron ions form the most stable

    carbonyl species (2196 cm1) which can be converted, at

    low temperature, into di- (B2188 cm1) and tricarbonyls

    (B2180 cm1). With NO these Fe2+ ions form nitrosyls

    absorbing at 1895 cm1. With time, in NO atmosphere, the

    Fe2+ cations are displaced from their original positions in

    order to form tetranitrosyl species.105

    The specific structure of MOF compounds gives particular

    adsorption, separation and catalytic properties to these

    materials. For example, the controlled reduction of a large-pore

    iron(III) trimesate with unsaturated iron sites, MIL-100(Fe),

    strongly increases the strength of interaction with unsaturated

    gas molecules, such as propylene and CO, that have either a

    double or triple bond. Therefore, this property leads to a

    dramatic improvement of not only preferential gas sorption

    but also separation performance for the investigated hybrid

    compound: it could be used for the removal of CO impurity to

    protect the deactivation of Pt electrodes in low temperature

    fuel cells, for example, or the removal of CO from CO2

    -rich

    mixtures arising from the production of hydrogen from

    biomass. The use of mixed-valent MIL-100(Fe) may be also

    considered for further applications involving the selective

    separation or purification of olefins and acetylenes from

    hydrocarbon mixtures. The reducibility of MIL-100(Fe) to

    form FeII CUS has been unambiguously shown by in situ IR

    spectroscopic analysis using CO as a probe, which also

    allowed to quantify the concentration of Fe2+/Fe3+ sites as

    a function of the activation treatment (see Scheme 2 and

    Fig. 10). Moreover it provided evidences that unsaturated

    sites can be created not only through the removal of water

    but also to a much lesser extent from the departure of other

    Scheme 2 Representations of MIL-100(Fe): (a) one unit cell, (b) two types of mesoporous cages shown as polyhedra, (c) formation of

    FeIII CUS and FeII CUS in an octahedral iron trimer of MIL-100(Fe) by dehydration and partial reduction from the departure of anionic ligands

    (X

    = F

    or OH

    ) (from ref. 106). Reproduced by permission of Wiley-VCH.

  • 8/12/2019 b919543m

    14/24

    4940 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    molecules, such as trimesic acid or anionic ligands (F and OH)

    coordinated on terminal sites of FeIII.106

    As already mentioned, ceria and ceriazirconia are catalyst

    components of primary importance, notably for their red-ox

    properties. This arises from the ability of the Ce cation

    oxidation number in ceria and ceria-zirconia to easily change

    between 3+ and 4+. The surface state (reduced or oxidized)

    and composition (ratio between cerium and zirconium cations)

    is available using methanol (CH3OH) adsorption.107 Its dissocia-

    tion leads to cation coordinated methoxy and hydroxyl

    formation. Both then(CO) wavenumbers associated to methoxy

    species108,109 and the n(OH) associated to surface hydroxyls110

    depend on the cerium oxidation state. The Ce4+/Ce3+ surface

    ratio is thus available from the quantitative study of the

    corresponding methoxy intensities. Fig. 11 shows the con-

    secutive adsorption of oxygen (O2) calibrated doses after

    methanol dissociation over pre-reduced ceriumzirconium

    mixed oxide which enable the determination of the oxygen

    storage capacity of the sample.

    Moreover, the methoxy species is very sensitive to the local

    coordination site and it allows discriminating between the

    different cations on a surface. Concerning ceriazirconia solid

    solutions, for example, methoxys on Ce4+ and Zr4+ surface

    sites present specific IR fingerprints. Therefore, the surface

    cationic composition can easily been obtained upon methanol

    adsorption at room temperature.25 Rare-earth compounds

    also present electronic transitions, arising for ions in internal

    structural defects; for ceria, at temperatures above 523 K

    a new band appears at 2120 cm1 and is attributed to the2

    F5/2-2

    F7/2electronic transition of Ce3+

    , thus indicating the

    beginning of bulk oxide reduction.111

    Accessibility of sites. To be active in a catalytic reaction, a

    surface site must be accessible to reactants and products.Isotope labelling is here again a very useful tool. Deutera-

    tion of the surface OH groups is only possible if they are

    accessible to the deuterated molecule used for the exchange. In

    silica, for example, internal silanol groups are distinguished

    from external ones based on accessibility to deuterated water

    molecules.112

    An accessibility index (ACI) was derived from infrared

    spectroscopy of substituted alkylpyridines with different sizes

    (pyridine: 0.57 nm, 2,6-lutidine: 0.67 nm, collidine: 0.74 nm)

    over hierarchical ZSM-5 crystals. The samples were preparedby selective silicon extraction of a parent commercial sample in

    alkaline medium (desilication) and contained different degrees

    of intracrystalline mesoporosity. The enhanced accessibility of

    acid sites in the hierarchical zeolites was shown. A relatively

    bulky molecule such as collidine, which probes practically no

    acid sites of the parent medium-pore MFI structure, can access

    up to 40% of the Brnsted sites in the mesoporous sample.

    The ACI is a powerful tool to standardize acid site accessibility

    in zeolites and can be used to rank the effectiveness of

    synthetic strategies towards hierarchical zeolites (mesoporous

    crystals, nanocrystals, and composites).113

    Time resolved studies of probe molecules: 2D-pressure jumpspectroscopy. Time-resolved IR spectroscopy can be used in

    probe molecule studies. Microsecond infrared spectroscopy

    can be used to monitor adsorbed probe molecules after a

    pressure jump on the surface of a porous catalyst. The analysis

    of the time behavior of the adsorbed molecule can be obtained

    by a Fourier Transform of the IR spectra over time. A 2D map

    is obtained, showing frequency response vs. IR spectra for the

    adsorbed probe molecule on the catalyst. This technique has

    Fig. 10 (a) IR spectra of MIL-100 under a stream of 10% CO at 25 1C after activation under a helium flux at various temperatures over 3 or 12 h.

    (b) Amount of FeIII CUS and FeII CUS detected by IR analysis upon CO adsorption at 173 1C on MIL-100(Fe) activated under high vacuum at

    different temperatures (from ref. 106). Reproduced by permission of Wiley-VCH.

    Fig. 11 Progressive reoxidation of a mixed Ce0.80Zr0.20O2compound

    by introduction of small doses of O2 after reduction under hydrogen

    at 673 K.

  • 8/12/2019 b919543m

    15/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4941

    been used to study platinum catalysts supported on MFI

    zeolites. The 2D map can be read as a map showing loca-

    tion vs. size for the metal particles. Small particles can be

    distinguished from large ones, and their location can be

    determined in the pores or on the outer surface of the zeolite.

    The role of the pores was demonstrated for the protection

    of small particles during ageing of the catalyst, and sintering

    was limited by the pore diameter where the particle was

    located.86

    Quantitative analysis by coupling IR with gravimetry.Thermo-

    gravimetric analysis (TGA) allows monitoring weight changes

    in the sample. It has been combined with IR to give new

    information on surface sites. Reliable quantitative information

    is the key to understanding the catalytic role of surface sites,

    and this combination of techniques offers just that. It was used

    to determine the molar absorption coefficients for IR bands of

    OH groups on silica and HY zeolites, as well as for adsorbed

    probe molecules.112 Important information was obtained

    about the quantity of the OH groups located in the different

    cages of HY zeolite and also about the quantity of inner andsurface silanol groups on silica. The number of silanol groups

    accessible to water molecules was shown to be constant

    whatever the sample of precipitated silica, as well as the ratio

    of water/silanol groups under room atmosphere. Combined

    thermogravimetry and IR was also applied to operando con-

    ditions, and denoted as AGIR (Analysis by Gravimetry and

    IR): a modified microbalance was used to follow mass changes

    (mg accuracy) inside a catalytic reactor equipped with infrared

    windows. The spectroscopic response of water and ammonium

    ions coadsorbed together on zeolites was shown to vary depending

    on the conditions. The molar absorption coefficients for d(H2O)

    andd(NH4+) at 1640 and 1540 cm1 for water and ammonia on

    a HY zeolite were studied under dry gas flow at variabletemperature. It showed the influence of coverage on the infrared

    response of adsorbed species in zeolites. Adsorption sites change

    with coverage, and bands are shifted and their shape and

    intensity are modified. Other interesting facts were observed:

    water modifies strongly the aspect of the d(NH4+) vibration

    band in ammoniated zeolites, without changing the absorption

    coefficient. Measuring the sample mass while at the same time

    recording its infrared spectrum showed the key importance of

    conditions under which the measurement is done. The presence

    of co-adsorbed species (water in particular) strongly modifies the

    spectrum of surface species. Under reaction condition, this new

    technique is especially important for a correct assignment of

    infrared features and catalytic behaviour, and above all it makesIR measurements really quantitative.114

    Other coupled techniques. As we have mentioned in the

    introduction, the purpose of this review is to critically discuss

    the characterization of the active site in catalytical processes

    by using IR analysis compared with catalytic tests. Sometimes,

    additional information can be provided by the coupling of

    complemental techniques able to yield enhanced insight into

    material local properties (structural, textural, physical, . . .)

    and reactional events at the nano-scale. For more information

    about this point we refer the reader to specific literature,

    notably within the present themed issue. We just want to

    briefly mention a few examples:

    Numerous techniques have been coupled with IR, thus

    broadening the experimental data set for subsequent mecha-

    nistic considerations. Of course, complementary information

    can be obtained by parallel experiments; nevertheless, to be

    sure that results are comparable, it is often preferred that the

    investigation is carried out in a single setup.115

    The most natural technique to associate with IR is certainly

    Raman, to cover the totality of the vibrational spectrum. The

    firstin situRaman and FTIR spectroscopic characterization of

    a catalytic system under reaction conditions using a single

    bench-top instrument with a dedicated cell was published in

    2003 by Payen et al.116 Following both the evolution of the

    molecular structure of the active phase by Raman spectro-

    scopy and the nature of the different surface adsorbed species

    by IR and Raman spectroscopy during deNOx reaction they

    were able to provide an exhaustive identification of the

    adsorbed species and the corresponding coordination sites.

    Similar information can also be obtained by coupling

    DRIFT with hard X-ray diffraction, which can highlight

    phenomena taking place in nanoparticulate catalysts, such as

    the formation of simultaneous surface and bulk species and

    their evolution.117

    Also combining, at high time resolution, a transmission

    based structural probes, dispersive EXAFS, with diffuse

    reflectance infrared spectroscopy, can highlight fundamental

    steps occurring during gassolid interactions, like that of

    oxidation and reduction of alumina-supported Rh at 573 K

    using NO and H2, and the structuralreactive role of linear

    Rh-nitrosyl species within these processes.118

    Chemometric tools.Chemometrics is of course a very efficient

    way of decomposing complex bands, when they are progressivelychanging during reaction or during progressive adsorption. CO

    adsorption on Pt sites leads to complex bands on zeolites where

    different metal particle sizes can exist.86 Small particles lead to a

    stronger influence of corners and edges compared to faces, and

    large particles produce bands with a stronger influence of large

    faces. Chemometrics allows separating contributions, and the

    spectra of CO on small and large particles can be distinguished

    quantitatively. This was revealed to be very important for

    studying the influence of ageing of the catalyst on the catalytic

    sites.86

    Active sites intermediates and spectators(discrimination of the aforementioned sites

    by using catalytic evidences)

    Intermediates, active sites and spectator species are well

    defined concepts in catalysis. We will summarize them again

    here for a clear discussion of their spectroscopic characteri-

    zation. Heterogeneous catalysis involves adsorption of

    reactants on the surface, their reaction on the active site, with

    the possible formation of intermediates species, and the

    formation of final products (with a possible change of adsorp-

    tion site), and the further desorption of the products which will

    go to the exit of the reactor.

  • 8/12/2019 b919543m

    16/24

    4942 Chem. Soc. Rev., 2010, 39, 49284950 This journal is c The Royal Society of Chemistry 2010

    During the adsorption step, several sites can participate.

    Some of them will indeed be the place for the reaction

    (and these are the active sites) while some others will only be

    adsorption sites on which no reaction will take place. These

    non-active sites can adsorb strongly reactants or products

    and block any movement of surface species, resulting in a

    deactivation of the catalyst. Several reactions can take place

    on the surface. Some species will have no role in the reaction

    under study, and will thus be spectator species. Some other

    species can hamper or slow down the reaction, either by

    staying adsorbed on the active site or by blocking the circula-

    tion of surface species, and will be poison species. Another

    type of species will be formed on the surface but will be later

    transformed into the final expected product, and these are

    intermediate species. Intermediates are stable species on the

    surface (at least for a short duration), and should not be

    confused with transition states. Transition states are postulated

    between two chemical species on the surface, but do not actually

    exist since they have no actual lifetime. They are only pathways

    between two states on the surface (on an active site).

    Correlation between probe studies and activity

    In situ analyses of a surface provide photographs of the

    material state, detailing the concentration of the superficial

    entities directly by their corresponding IR spectra or via the

    spectral response of a molecule adsorbed on the site and their

    relative strengths vs. the applied probe, as aforementioned.

    The most general way to link catalytic activity to a specific site

    is the correlation method. It consists of measuring activities of

    a series of materials (starting from the initial reaction rate, or

    the conversion at the steady state), and to compare this value

    to the number of sites estimated by the intensity of a specific

    IR band. In general, this site concentration corresponds to the

    entities able to adsorb probe molecules. If a linear correlation

    between these two sets of data is found, the activity can

    reasonably be attributed to the presence of such adsorption

    sites. Nevertheless, all the sites able to adsorb a molecule are

    not necessarily active: the presence of the correlation means

    that the ratio between the active and the adsorption sites

    remains constant when their global amount varies. Additionally,

    we should consider the limitations intrinsic to the probe

    molecule (probe sensitivity to the strength, nature and number

    of sites) and the difficulties to obtain for both spectroscopic

    and catalytic measurements in similar activation conditions.

    The use of appropriate probe molecules permits nevertheless

    to formulate reliable hypotheses on the nature of the corres-

    ponding sites. Moreover, when the quantitative analysis of the

    sites is possible, a calculation of the turn over frequency (TOF)

    can also been performed. A demonstrative example concerns

    active site study via CO adsorption in hydrotreating reactions

    such as hydrodesulfurization (HDS). They are usually related

    to anionic vacancies (coordinatively unsaturated sites, CUS)

    located on the edges of mixed sulfide nanosized particles

    (CoMoS or NiMoS) supported on high specific surface

    area alumina. Nitrogen monoxide has been the most employed

    but a partial oxidation of the sulfide phase may occur, even at

    very low temperatures. Infrared spectra of CO adsorbed on

    the sulfided promoted CoMo/Al2O3catalysts displays a strong

    nCO band at 2070 cm1 which is correlated to the HDS

    catalytic activity.119 It has been assigned to CO interacting

    either with a Co atom or with a Mo atom adjacent to a Co

    atom. Similar correlations between the intensity of specific nCO

    bands and the activity in HDS catalytic activity have been also

    found with Mo/Al2O3catalyst. Due to the drastic procedure of

    activation of sulfide catalyst, it is worth noting that some

    researchers of our laboratory have designed a new IR cell,

    called CellEx, in order to characterize in situ sulfided catalysts

    under a pertinent H2S/H2flow, with pressure varying from 0.1

    up to 4.0 MPa. This cell allows obtaining similar sulfidation

    procedure for both IR characterizations and catalytic tests.120

    Another significant example can be mentioned in the case of

    the NOx

    selective reduction by hydrocarbons on oxide-based

    catalysts. Working on Ag/Al2O3 samples for NOx reduction

    by ethanol, we observed the formation of cyanide and

    isocyanate species. The IR band assignment was not straight-

    forward: contrarily to what is usually reported in the litera-

    ture, n(NCO) of AgI(NCO) species is located at 2204 cm1

    (and not at 2230 cm1

    ), whereas that of Ag0

    (NCO) species is

    at 2243 cm1. Thus, during SCR reaction of NO on silver/

    alumina catalyst, the isocyanate species generally observed as

    intermediate compounds around 2230 and 2260 cm1 are not

    linked to silver sites but to the support, the main role of silver

    being to favour NCO formation and concentration on the

    support. The hypothesis that the two observed bands are due

    to different alumina coordination sites (Alocta and Altetra) for

    isocyanates was determined as the most probable by isotopic

    substitutions.121 This information was useful for the compre-

    hension of the NO SCR pathway. In fact comparing the

    selective catalytic reduction of NOx

    in excess of oxygen using

    ethanol as reducing agent on silver/alumina and on bare

    alumina showed the connection between the presence of silver

    and isocyanate species on a catalyst. Then, a detailed investi-

    gation concerning these groups was undertaken to understand

    their formation, their localization, and their reactivity in order

    to propose the pathway of the NOx

    SCR into N2. Three

    elemental sequences were suggested explaining, first, the

    formation of silver cyanide and its transformation into

    Al3+NCO, then isocyanate hydrolysis into ammonia, and

    finally the reaction of the latter species with NO in the

    presence of oxygen giving rise to nitrogen (Fig. 12).122

    However, a working surface can have a totally different

    behaviour, with sites changing their nature. The real valence of

    a site can be ascertained only catching it in full action, using

    operando techniques.

    Operando studies on Fe-FER indicated that these samples

    present interesting NOx

    SCR efficiency for temperature as low

    as 433 K. Our investigations also indicate that for such

    low reaction temperature ammonia and nitrogen monoxide

    compete for adsorption onto the Fe2+

    species (whose fine

    characterisation has been reported in the section Metallic and

    redox sites), which are active for the NO-to-NO2 oxidation.

    Our results are also consistent with this last reaction being

    the rate determining step of the global SCR process. We finally

    concentrated our study on the effect of SO2on this NO-to-NO2reaction. We must conclude that for the 2.5 wt% Fe-FER

    sample, the majority of iron sites are poisoned by sulfate

    formation, although some Fe2+ sites are thio-resistant.123

  • 8/12/2019 b919543m

    17/24

    This journal is c The Royal Society of Chemistry 2010 Chem. Soc. Rev., 2010, 39, 49284950 4943

    Operando IR spectroscopy shows out all its added values

    when very complicated and multifuncti