usy,coking and deactivation ,paweewan,1999

Upload: 271207

Post on 04-Jun-2018

218 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    1/10

    Applied Catalysis A: General 185 (1999) 259268

    Coking and deactivation duringn-hexane cracking in

    ultrastable zeolite Y

    Boontham Paweewan, Patrick J. Barrie , Lynn F. GladdenDepartment of Chemical Engineering, University of Cambridge, Pembroke Street, Cambridge CB2 3RA, UK

    Received 25 February 1999; received in revised form 5 May 1999; accepted 6 May 1999

    Abstract

    A multi-technique approach has been used to investigate coke formation and the effects of coke formation duringn-hexane

    cracking on ultrastable zeolite Y. The product analysis reveals that propane and propene are the major products, and this

    suggests that the reaction initiation step may be direct protonation at very strong Brnsted acid sites. The coke formed does

    not change in composition significantly over the course of the reaction. Infrared spectroscopy reveals that the vast majority

    of Brnsted and Lewis acid sites are still present in the catalyst, even when the catalyst activity has become low. Diffusion

    measurements using pulsed field gradient (PFG) NMR show that the mobility of adsorbed n-butane or n-hexane is not affected

    by the presence of the coke, and so shows that pore blockage is not a significant factor. Based on the experimental results,

    it is argued that selectivesite poisoning of a few very strong acid sites is the main deactivation mechanism for this reaction

    under the conditions employed. 1999 Elsevier Science B.V. All rights reserved.

    Keywords:Coking; Deactivation; Cracking; Zeolite Y; Poisoning, selective site

    1. Introduction

    Catalytic cracking over zeolite-based catalysts is an

    important reaction in the petrochemical industry. It

    is generally accepted that the initial cracking activity

    is principally due to the Brnsted acidity of the zeo-

    lite, and the reaction mechanism has been investigated

    extensively [14]. However, the reaction is a com-

    plex one and results from one zeolite-reactant system

    are not necessarily directly applicable to another. The

    catalyst may have a range of different Brnsted acid

    sites present with varying degrees of acidity, and the

    Lewis acid sites present may also have an influence.

    Characterisation of acid sites in zeolites has thus been

    Corresponding author. Fax: +44-1223-334796.

    studied using a wide variety of different techniques[510]. The combined influence of the range of differ-

    ent strength acid sites and pore diffusion has also been

    considered [11]. The situation is complicated by the

    fact that catalyst deactivation takes place during the

    cracking reaction. The major cause of deactivation un-

    der laboratory conditions is the formation of carbona-

    ceous residues known as coke within the catalyst. A

    wide variety of analytical techniques have been used

    to study both the formation of coke and its influence

    on deactivation [1217]. Despite previous work, there

    is still no generally accepted model for coke forma-

    tion and, more importantly, for predicting the effects

    of the coke on catalyst activity and selectivity [18].Thus, an important research topic is the identification

    of the mechanism as to how coke causes deactiva-

    0926-860X/99/$ see front matter 1999 Elsevier Science B.V. All rights reserved.

    PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 1 4 3 - X

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    2/10

    260 B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268

    tion during a particular cracking reaction on a specific

    zeolite.

    While n-hexane and n-heptane cracking on zeo-

    lite Y have previously been studied by several groups

    of workers [13,17,1927], there remain differences

    in interpretation of the coke formation and deacti-

    vation mechanism for this reaction. Coke can affect

    the catalytic activity of zeolites through both site poi-

    soning and pore blocking mechanisms, and it is im-

    portant to be able to distinguish between the two.

    Further, the coke formed may not necessarily be dis-

    tributed homogeneously throughout the zeolite parti-

    cles. For instance, a recent paper found that a large

    decrease in activity occurred even though only few

    acid sites were poisoned, and it was proposed that the

    reaction was diffusion limited [25]. For a diffusion-

    limited reaction, n-hexane cracking and coke deposi-

    tion will occur primarily near the external surface of

    the zeolite crystallites, with little reaction at the par-

    ticle interior. Other workers have suggested that theinitial deactivation for this reaction is based on site

    poisoning [20,21], while others have found that the

    coke limits access of the reactant to the active sites

    [13,17,23].

    In a previous paper, we adopted a multi-technique

    approach to investigate coke formation and the deac-

    tivation mechanism during the conversion of ethene

    on ultrastable zeolite Y at 773 K [28]. It was found

    that both site poisoning and pore blockage played

    significant roles. In this paper, we report results

    on n-hexane cracking in ultrastable zeolite Y using

    the same range of techniques. The products were

    analysed using gas chromatography, and the chem-ical nature of the coke formed, and its influence

    on the number of acid sites and molecular trans-

    port, were studied using infrared and NMR spec-

    troscopic methods. The results shed new light on

    the deactivation mechanism for this particular reac-

    tion, and resolve some of the contradictions in the

    literature.

    2. Experimental

    The sample of steamed zeolite Y, denoted HUSY,

    used in this work came from the same batch as thatused in our study on coke formation during ethene

    conversion on ultrastable zeolite Y [28]. The sample

    has been well characterised previously [29] and has

    a framework Si/Al ratio of 4.15 (based on unit cell

    parameter) and a bulk Si/Al ratio of 2.65 from X-

    ray fluorescence measurements. Coke was generated

    by passing nitrogen at 50 ml/min through a satura-

    tor containing n-hexane at 303 K, and then on to a

    fixed bed reactor containing 500 mg of calcined cat-

    alyst at 773 K. The products from the reaction were

    analysed using a Phillips Pye 4500 gas chromato-

    graph (GC) equipped with a flame ionisation detec-

    tor and a Porapak Q column. After a specified time

    on stream, the saturator was bypassed, and the re-

    actor cooled down to room temperature under the

    nitrogen atmosphere. Coked samples were obtained

    in this way after 5, 10, 15, 25, 65, 125, 185 and

    245 min of time on stream. The integrity of the coked

    samples was checked by powder X-ray diffraction,

    and the samples were then examined by a variety of

    techniques.

    One portion of the coked sample after 125 minon stream was checked for catalytic activity for n-

    dodecane cracking. In this experiment, n-dodecane

    was fed at a rate of 0.1 ml/min using a peristaltic

    pump in a flow of nitrogen at 50 ml/min to the re-

    actor at 773 K, and the light products were analysed

    using a gas chromatograph as above. Negligible ther-

    mal cracking was observed when the catalyst was

    absent.

    The BET surface area was obtained by analysing

    nitrogen adsorption at 77 K in a conventional volu-

    metric apparatus (Micromeritics ASAP 2000). Each

    coked sample was pre-treated by heating to 673K

    at 105

    Torr for 12h in order to remove all wa-ter prior to starting the BET adsorption isotherm

    experiment.

    The weight percent of coke present in the samples

    was estimated by thermogravimetric analysis (TGA).

    The sample was heated to 973 K at a rate of 10 K/min

    in an inert gas to remove water, and the weight loss at

    that temperature when the gas was switched to air was

    measured.

    Infrared spectra were obtained on a Nicolet Magna

    750 FT-IR spectrometer equipped with a Spectra Tech

    0030-102 diffuse reflectance assembly. Each coked

    sample was heated in a helium gas flow to 673 K to

    remove water present in the pore space before record-ing the spectrum at elevated temperature. Infrared

    spectra were also obtained on pyridine adsorbed within

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    3/10

    B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268 261

    the catalyst. For these measurements, the sample were

    exposed to pyridine vapour at 573 K for 10 min, and

    then left in a flow of helium gas at that temperature

    for 5 min, before recording the spectrum at room tem-

    perature. The spectra are plotted assuming the model

    of Kubelka and Munk for diffuse reflectance [30],

    with the KubelkaMunk function being analogous to

    absorbance [31].13C solid state NMR spectra were obtained at

    50.32 MHz on the coked samples using cross-

    polarisation (CP), magic-angle spinning (MAS) and

    high-power proton decoupling on a Bruker MSL-

    200 spectrometer. Spectra were acquired overnight

    using a contact time of 1 ms and a recycle delay

    between scans of 1 s. The sample spinning rate was

    4 kHz, and spinning sidebands were suppressed us-

    ing the five -pulse TOSS method [32]. Chemical

    shifts are quoted relative to external tetramethylsilane

    (TMS).

    Diffusion measurements on n-butane and n-hexaneadsorbed within the fresh zeolite and the coked sam-

    ples were made using pulsed field gradient (PFG)

    NMR spectroscopy [3335]. For the diffusion mea-

    surements, eithern-butane orn-hexane were adsorbed

    into each sample at a loading of approximately 10

    molecules per unit cell. The 1H NMR spectra were

    acquired at 200.13 MHz. After measuring the T1 and

    T2 nuclear relaxation times to aid choice of the ac-

    quisition parameters, diffusivities were measured us-

    ing the PFGLED pulse sequence. This uses a stim-

    ulated echo and minimises any errors due to eddy

    current effects [36]. For single-component diffusion,

    the signal intensity is expected to follow the formln(I/I0) =2g22D( /3)whereIis the observedintensity, I0 the intensity in the absence of gradient

    pulses, the gyromagnetic ratio of1H,gis the applied

    field gradient amplitude, is the length of the gradi-

    ent pulse, the interval between the gradient pulses,

    andD the diffusion coefficient [3436]. For measure-

    ments on adsorbedn-butane, was varied while main-

    tainingg at 1 T/m and at 10 ms. For measurements

    on adsorbedn-hexane, was again varied, with g and

    fixed at values of 2 T/m and 160 ms, respectively.

    Due to the presence of both intercrystalline and in-

    tracrystalline diffusion, it was necessary to analyse the

    experimental data assuming the presence of two dis-tinct diffusion components using a least-squares fitting

    routine.

    Fig. 1. Plot of n-hexane conversion (circles, left axis) and coke

    content (squares, right axis) as a function of time on stream.

    3. Results and discussion

    3.1. The reaction

    A typical plot ofn-hexane conversion against time

    on stream, together with the coke loading, is shown

    in Fig. 1. It is known that significant deactivation oc-

    curs during the first minute of time on stream [22]. It

    can be seen that the activity continues to drop rapidly

    during the first 60 min or so of the reaction, and the

    conversion becomes close to zero after 180 min. The

    line shown in Fig. 1 for the conversion uses the ex-

    ponential decay equation X= X0exp(kdt), whereX is the conversion at time t, X0 conversion at zero

    time,kdthe deactivation rate constant andtthe time on

    stream.The product selectivities during the reaction are

    shown in Fig. 2. The major products are the C3 hy-

    drocarbons propene and propane (not resolved by the

    gas chromatograph column used), with the other prod-

    ucts being in the C1C4 range. During the first 60

    minutes, an increasing proportion of the products are

    C1 and C3 species, while the relative amount of C4

    species decreases. The proportion of C2 species re-

    mains roughly constant.

    The generally accepted theory of the cracking of

    alkanes on a solid with acid sites is that cracking pro-

    ceeds via formation of a tricoordinated carbenium ion

    as an intermediate, which then undergoes -scissionto form an alkene fragment and a carbenium ion frag-

    ment [37,38]. However, it has been suggested that for-

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    4/10

    262 B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268

    Fig. 2. Plot showing product selectivity as a function of time on

    stream.

    mation of primary carbenium ions is unlikely to oc-

    cur, and that the intermediate may in fact be a pro-

    tonated cyclopropane species [39,40]. Direct protona-

    tion of hydrocarbons to form pentacoordinated carbo-

    nium ions provides an alternative cracking mechanism

    and this has been suggested to occur when very strongBrnsted acid sites are present at high temperatures;

    this route is expected to be significant when the reac-

    tant is difficult to crack [39,41]. A combination of the

    mechanisms has been proposed: initial formation of

    a pentacoordinated ion will be followed by protolytic

    fission to give an alkene and an adsorbed carbenium

    ion that will then react further [3,42]. Bimolecular re-

    actions, involving hydrogen transfer between the reac-

    tant and adsorbed carbenium ions, may also then occur

    in a chain mechanism to produce a range of products

    [3]. Coke may be formed within the catalyst from bi-

    molecular interactions between adsorbed species [3].

    The main area of uncertainty in the proposed reactionmechanisms is the initiation step in which the first re-

    active ion is formed. Reaction at Lewis acid sites has

    been proposed as an alternative initiation step for the

    cracking mechanisms [4345], but other groups have

    found that Lewis acid sites do not play a significant

    role during the cracking of light hydrocarbons [46].

    Thermal cracking has been suggested as a possible

    initiation step during the cracking of heavy hydrocar-

    bons [47].

    The product distribution is shown in Fig. 2 and is

    similar to that obtained recently in the literature under

    conditions favouring monomolecular cracking [27]. It

    is worth observing that -scission of the most stable

    C6 carbenium ions that can be formed (in principle

    at Lewis acid sites or by bimolecular hydrogen trans-

    fer reactions) would give C2 and C4 species, and that

    these are only observed in relatively small quantities

    in the reaction under investigation. The observation of

    C3 species as dominant products in Fig. 2 thus sug-

    gests that the major mechanism occurring in this cat-

    alyst is in fact direct protonation of one of the mid-

    dle carbons of the n-hexane molecule at very strongBrnsted acid sites to form a non-classical pentacoor-

    dinate carbonium ion. The protonation would then be

    followed by fission into propane and a C3 carbenium

    ion. The carbenium ion may react further, or lose a

    proton to form propene. Direct protonation followed

    by fissionis also a sourceof the C2 and C4 products, as

    is the carbenium ion mechanism. It is, of course, pos-

    sible that C4 species can react further in the catalyst

    (e.g. to give C1 and C3 species). Thus, the increasing

    selectivity towards methane as the reaction proceeds

    is due to transformation of products rather than direct

    protolytic cracking. These results are consistent with

    some recent studies on n-hexane cracking in dealu-minated zeolite Y [24] and for n-hexane cracking in

    ZSM-5 zeolite [4850].

    The possibility that initiation of the reaction may

    occur only at a few very strong active sites, which

    is suggested by the observed product distribution, is

    discussed further below when the deactivation mech-

    anism is considered.

    3.2. The chemical nature of the coke formed

    The total amount of coke formed was determined

    by thermogravimetric analysis and is shown in Fig. 1.It can be seen that it only takes a small amount of coke

    to have a large effect on catalytic activity. It can also be

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    5/10

    B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268 263

    Fig. 3. 13C MAS NMR spectra of coked samples after a time on stream of: (top) 25min; (middle) 125 min; (bottom) 185 min.

    seen that coke continues to form even after 120 min re-

    action time when the catalyst activity is low. However,

    the amount of coke formed during n-hexane crack-

    ing is significantly less than that formed during ethene

    oligomerisation on the same catalyst under identical

    conditions [28]. In order to probe the chemical iden-

    tity of the coke formed, both 13C solid state NMR

    spectroscopy and infrared spectroscopy were used.13C solid state NMR spectroscopy has been used

    previously to characterise coke formed in zeolite cata-

    lysts [28,5155]. In this work, we achieved high qual-

    ity13 C spectra using the cross-polarisation (CP) tech-

    nique at natural abundance without any chemical treat-

    ment. Spectra are shown in Fig. 3. Two major spec-

    tral regions can be identified: the asymmetric peak at

    130 ppm is associated with aromatic carbon environ-

    ments (though polyalkenes also resonate in this re-

    gion), while the peaks at 1050 ppm are associatedwith saturated aliphatic carbon species. One frequently

    used parameter in coal chemistry is the aromaticity,

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    6/10

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    7/10

    B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268 265

    Fig. 6. FTIR measurements of adsorbed pyridine on HUSY after

    a time on stream of: (a) 0 min (fresh catalyst); (b) 65 min; (c)

    245min.

    acid sites in the smaller sodalite cages, respectively

    [62,63]. There may also be a contribution from extra-framework AlOH groups in this region. It can be

    seen that there is no reduction in the number of acid

    groups detected with increasing time on stream, even

    after four hours reaction time. This indicates that the

    majority of active centres are not removed during de-

    activation by coking under the reaction conditions em-

    ployed.

    Some workers have reported the presence of an

    OH stretching band at 3600 cm1 in zeolite Y af-ter chemical dealumination, or after steam dealumi-

    nation followed by acid leaching, and this band has

    been assigned to particularly strong Brnsted acid sites

    [6469]. The intensity of this band has been foundto decrease during n-hexane cracking, and so it may

    well be these sites that are removed during the cok-

    ing reaction [68,69]. However, this band is only rarely

    observed for steam dealuminated zeolite Y samples

    (unless acid leaching is also performed), and it is not

    resolved in the spectra shown in Fig. 5. No firm con-

    clusion can therefore be drawn from our FTIR spectra

    alone about whether any particularly strong acid sites

    are affected by coking.

    One method to quantify Brnsted and Lewis acid

    sites is by using pyridine adsorption in conjunction

    with infrared spectroscopy. FTIR spectra on pyridine

    within the fresh and coked catalysts are shown in Fig.6. The band at 1546 cm1 is characteristic of pyridinebonded to a Brnsted acid site, while that at 1453 cm1

    is characteristic of pyridine bonded to a Lewis acid

    site [5]. Other bands in this spectral region are due

    to other vibrational modes for pyridine which overlap

    and so do not easily distinguish between the different

    acid sites present. For the coked sample after 65 min

    reaction time, there is no significant reduction in signal

    intensity of either of the two characteristic bands, and

    adsorbed pyridine can still easily be observed in the

    FTIR spectra of the sample that had 245 min time on

    stream.

    The BET surface area of the coked samples was

    found to remain high, being 610 m2/g for the fresh

    catalyst, and 518 m2/g after 125 min on stream. How-

    ever, the small size of the nitrogen molecule means

    that the BET surface area is not necessarily a good

    indicator of pore blockage or restricted diffusion for

    larger molecules.

    In order to shed more light on the deactivation

    mechanism, self-diffusion measurements on adsor-

    bates within the coked catalyst were performed usingPFG NMR to see whether the coke is influencing

    molecular transport within the zeolite pore space.

    n-Butane was used as a probe molecule, as was also

    the case in our previous work [28], due to its diffu-

    sivity and favourable NMR relaxation characteristics

    making the PFG NMR experiment straightforward.

    Results are shown in Fig. 7. For single-component

    diffusion, the plot of ln(I/I0) against2g22( /3)

    is expected to be linear with a gradient of D, where

    D is the diffusion coefficient [3335]. However, the

    log attenuation plots in Fig. 7 show a curve. This

    Fig. 7. PFG NMR intensity results for n-butane adsorbed in zeolite

    HUSY after a time on stream of: (a) 0 min (fresh catalyst); (b)

    65min; (c) 185 min.

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    8/10

    266 B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268

    is because the length-scale being probed by the

    PFG NMR measurement, which is of the order of6D, is greater than the particle size (0.5 1m

    for the sample used) [34,35]. Hence both interpar-

    ticle and intraparticle effects are being observed.

    The initial slope depends on the interparticle dif-

    fusivity and the relative weighting of how long the

    adsorbed molecule spends in each region, while the

    final slope gives the intraparticle diffusivity only [35].

    Thus, the curves may be fitted using a least-squares

    fitting routine in order to obtain the intraparticle

    diffusivity.

    The slower diffusion components, corresponding to

    intraparticle diffusivity, for the fresh sample and those

    after 65 min and 185 min of reaction time were found

    to be 2.6 109, 2.2 109 and 1.6 109 m2/s(all 0.2 109 m2/s), respectively. This is aboutthe same value, 2 109 m2/s, found by PFG NMRfor n-butane in zeolite NaX [70]. Thus, there is little

    change in the diffusivity of n-butane, even after 4 hon stream. This may be contrasted with our previous

    work on ethene oligomerisation which showed a sig-

    nificant drop in n-butane intracrystalline diffusivity

    with increasing time on stream [28].

    It was also possible to measure the diffusion con-

    stant of the reactant itself by performing PFG NMR

    experiments on n-hexane adsorbed within the coked

    zeolite. However, the NMR relaxation times of ad-

    sorbed n-hexane are smaller than those for n-butane,

    and this results in slightly more experimental scatter

    in the data points. As was the case with the measure-

    ments on n-butane, a two-component fit is necessary

    to fit the experimental data as shown in Fig. 8. It can beseen that the signal attenuation does not depend on the

    time on stream and fitting a single line to the data gives

    the average intraparticle diffusivity ofn-hexane within

    ultrastable zeolite Y to be 4.5 (0.5) 1012 m2/s.

    3.4. Discussion of the deactivation mechanism

    The infrared spectroscopic results clearly show that

    removal of Brnsted and Lewis acid sites does not

    occur to a significant extent, and thus that uniform site

    poisoning is not responsible for the deactivation of ul-

    trastable zeolite Y duringn-hexane cracking. Further,

    it is clear that the deactivation is not due to the poreblockage mechanism, as the diffusion rate of the re-

    actant through the bulk particle is unaffected by the

    Fig. 8. PFG NMR intensity results for n-hexane adsorbed in zeolite

    HUSY after a time on stream of 0 min (circles), 65 min (squares),

    and 125 min (diamonds).

    presence of coke. It should be noted that these obser-

    vations are consistent with those of some other work-

    ers [25]. It is also clear that the deactivation mech-anism during n-hexane cracking on ultrastable zeo-

    lite Y is very different to the one that takes places

    during ethene conversion, in which both uniform site

    poisoning and pore blockage effects were observed

    [28].

    The analysis of the product distribution discussed

    above indicates that the initiation step in the n-hexane

    cracking reaction is direct protonation of the reac-

    tant by a very strong Brnsted acid site. It is possi-

    ble that only a relatively small number of such active

    sites are present in ultrastable zeolite Y. For instance,

    they may only be present at defects, or at Brnsted

    sites where an adsorbed molecule can simultaneouslyinteract with a Lewis acid site. This leads to the pos-

    sibility that selectivesite poisoning is the main deac-

    tivation mechanism: the few highly active sites would

    be removed, resulting in considerable loss of activ-

    ity, but the majority of the acid sites would be unaf-

    fected and diffusion within the pore space would also

    be unaffected. Another possible explanation, as pro-

    posed by Hopkins et al. [25], is that the reaction is

    diffusion limited with pore mouth poisoning as the

    main deactivation mechanism. In this case, virtually

    all the reaction would take place near the external

    surface of the crystallites, and only this region would

    become coked. Again transport and the number ofactive sites present in the bulk solid would be barely

    affected.

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    9/10

    B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268 267

    In order to distinguish between the selective site poi-

    soning and the pore mouth poisoning models, a sample

    of coked zeolite Y (deactivated by 125 min on stream

    for then-hexane cracking reaction) was tested for cat-

    alytic activity in the cracking of n-dodecane. If pore

    mouth poisoning had been responsible for the deacti-

    vation, it would be expected that n-dodecane conver-

    sion would be very low (other than a small amount of

    thermal cracking) as the reaction would be more dif-

    fusion limited than the n-hexane cracking case. On the

    other hand, if selective site poisoning was responsi-

    ble, it would be expected that the coked catalyst would

    be highly active, as n-dodecane would be cracked

    on relatively low strength acid sites compared to n-

    hexane. It was found that the coked sample was highly

    active, giving a conversion to light products com-

    parable to a fresh catalyst sample. This observation

    confirms that the selective site poisoning mechanism

    must be the one responsible for deactivation during

    n-hexane cracking.

    4. Conclusions

    Coking and deactivation during n-hexane cracking

    on ultrastable zeolite Y has been studied using a va-

    riety of techniques. The product distribution suggests

    that the initiation step is direct protonation at very

    strong Brnsted acid sites, and it may be the removal of

    these that is the principal cause of deactivation for this

    reaction. The coke formed has an aromaticity of about

    7880%, and does not become highly polyaromaticeven after four hours on stream under the conditions

    employed. It is found that the bulk of the acid sites

    present in the catalyst are not removed, even though

    the catalyst activity drops significantly, while diffusion

    measurements show that pore blockage is not a fac-

    tor. A selective site poisoning model for deactivation

    under these conditions is thus proposed. It should be

    noted that the deactivation mechanism is very different

    to that observed during ethene conversion on the same

    catalyst. This work highlights the need for a multi-

    technique approach to the study of coking and deac-

    tivation in zeolite catalysts. Caution is needed when

    drawing conclusions on the deactivation mechanismfrom one reaction to another, and from one zeolite to

    another, as the initiation step may be different.

    Acknowledgements

    We thank the Cambridge Overseas Trust and the

    Cambridge Thai Foundation for funding Boontham

    Paweewans studentship. We are grateful to Dr. Mick

    Mantle and Dr. Sunil Ashtekar for their experimental

    help with the NMR and the infrared measurements.

    References

    [1] B.C. Gates, Catalytic Chemistry, Wiley, New York, 1992.

    [2] M.L. Occelli, ACS Symp. Ser. 375 (1988) 1.

    [3] K.A. Cumming, B.W. Wojciechowski, Catal. Rev. Sci. Eng.

    38 (1996) 101.

    [4] A. Corma, Curr. Opin. Solid State Mater. Sci. 2 (1997) 63.

    [5] A. Corma, Chem. Rev. 95 (1995) 559.

    [6] A. Zecchina, C.O. Areau, Chem. Soc. Rev. 25 (1996) 187.

    [7] M. Stocker, Microporous Mater. 6 (1996) 235.

    [8] A. Boreave, A. Auroux, C. Guiman, Microporous Mater. 11

    (1997) 275.

    [9] M. Hunger, Catal. Rev. Sci. Eng. 39 (1997) 345.[10] S.P. Bates, R.A. Van Santen, Adv. Catal. 42 (1998) 1.

    [11] B.A. Williams, S.M. Babitz, J.T. Miller, R.Q. Snurr, H.H.

    Kung, Appl. Catal. A: General 177 (1999) 161.

    [12] E.G. Derouane, Stud. Surf. Sci. Catal. 20 (1985) 221.

    [13] M. Guisnet, P. Magnoux, Appl. Catal. 54 (1989) 1.

    [14] D.M. Bibby, R.F. Howe, G.D. McLellan, Appl. Catal. 93

    (1992) 1.

    [15] H.G. Karge, Stud. Surf. Sci. Catal. 58 (1991) 531.

    [16] M. Guisnet, P. Magnoux, K. Moljord, in: P. OConnor, T.

    Takatsuka, G.L. Woolery (Eds.), Deactivation and Testing

    of Hydrocarbon-Processing Catalysts, ACS Symp. Ser. 634

    (1996) 77.

    [17] M. Guisnet, P. Magnoux, D. Martin, Stud. Surf. Sci. Catal.

    111 (1997) 1.

    [18] J.B. Butt, E.E. Petersen, Activation, Deactivation andPoisoning of Catalysts, Academic Press, London, 1988.

    [19] P. Magnoux, P. Cartraud, S. Mignard, M. Guisnet, J. Catal.

    106 (1987) 235.

    [20] N. Mori, S. Nishiyama, S. Tsuruya, M. Masai, Appl. Catal.

    74 (1991) 37.

    [21] C. Li, Y.-W Chen, S.-J. Yang, R.-B. Yen, Appl. Surf. Sci. 81

    (1994) 465.

    [22] A. Corma, P.J. Miguel, A.V. Orchills, Appl. Catal. A: General

    117 (1994) 29.

    [23] K.P. Mller, M. Kojima, C.T. OConnor, Chem. Eng. J. 54

    (1994) 115.

    [24] N.P. Rhodes, R. Rudham, N.H.J. Stanbridge, J. Chem. Soc.,

    Faraday Trans. 92 (1996) 2817.

    [25] P.D. Hopkins, J.T. Miller, B.L. Meyers, G.J. Ray, R.T.

    Roginski, M.A. Kuehne, H.H. Kung., Appl. Catal. A: General

    136 (1996) 29.

    [26] K.Y. Cheah, P. Alexander, L.F. Gladden, Appl. Catal. A:

    General 148 (1997) 387.

  • 8/13/2019 USY,Coking and Deactivation ,Paweewan,1999

    10/10

    268 B. Paweewan et al. / Applied Catalysis A: General 185 (1999) 259268

    [27] S.M. Babitz, B.A. Williams, J.T. Miller, R.Q. Snurr, W.O.

    Haag, H.H. Kung, Appl. Catal. A: General 179 (1999) 71.

    [28] B. Paweewan, P.J. Barrie, L.F. Gladden, Appl. Catal. A:

    General 167 (1998) 353.

    [29] A.P. Matharu, Ph.D. thesis, University of Cambridge, 1995.

    [30] P. Kubelka, K. Munk, Z. Tech. Phys. 12 (1931) 593.

    [31] T. Burger, J. Kuhn, R. Caps, J. Fricke, Appl. Spectrosc. 51

    (1997) 309.

    [32] Z. Song, O.N. Antzutkin, X. Feng, M.H. Levitt, Solid StateNucl. Magn. Reson. 2 (1993) 143.

    [33] J. Krger, D.M. Ruthven, Diffusion in Zeolites and other

    Microporous Solids, Wiley, New York, 1992.

    [34] J. Caro, H. Jobic, M. Blow, J. Krger, B. Zibrowius, Adv.

    Catal. 39 (1993) 351.

    [35] J. Krger, H. Pfeifer, W. Heink, Adv. Magn. Reson. 12 (1988)

    1.

    [36] S.J. Gibbs, C.S. Johnson, J. Magn. Reson. 93 (1991) 395.

    [37] B.S. Greensfelder, H.H. Voge, G.M. Good, Ind. Eng. Chem.

    41 (1949) 2573.

    [38] C.L. Thomas, Ind. Eng. Chem. 41 (1949) 2564.

    [39] S.T. Sie, Ind. Eng. Chem. Res. 31 (1992) 1881.

    [40] S.T. Sie, Ind. Eng. Chem. Res. 32 (1993) 397.

    [41] W.O. Haag, R.M. Dessau, Proc. 8th Congress on Catalysis,

    Verlag-Chemie, Weinheim, 1984, p. 305.[42] J. Abbot, B.W. Wojciechowski, J. Catal. 115 (1989) 1.

    [43] A. Corma, J. Planelles, J. Sanchez-Marin, F. Tomas, J. Catal.

    93 (1985) 30.

    [44] V.L. Zholobenko, L.M. Kustov, V.B. Kazansky, E. Loeffler,

    U. Lohser, G. Oehlmann, Zeolites 11 (1991) 132.

    [45] T.F. Narbeshuber, A. Brait, K. Seshan, J.A. Lercher, J. Catal.

    172 (1997) 127.

    [46] S.M. Babitz, M.A. Kuehne, H.H. Kung, J.T. Miller, Ind. Eng.

    Chem. Res. 26 (1997) 3027.

    [47] J. Scherzer, R.E. Ritter, Ind. Eng. Chem. Prod. Res. Dev. 17

    (1978) 219.

    [48] J. Abbot, Appl. Catal. 57 (1990) 105.

    [49] D.B. Lukyanov, V.I. Shtral, S.N. Khadzhiev, J. Catal. 146

    (1994) 87.

    [50] S. Jolly, J. Saussey, M.M. Bettahar, J.C. Lavalley, E. Benazzi,

    Appl. Catal. A: General 156 (1997) 71.

    [51] E.G. Derouane, J.P. Gilson, J.B. Nagy, Zeolites 2 (1982) 42.

    [52] J. Weitkamp, S. Maixner, Zeolites 7 (1987) 6.

    [53] M.C. Barrage, F. Bauer, H. Ernst, J. Fraissard, D. Freude, H.

    Pfeifer, Catal. Lett. 6 (1990) 201.

    [54] W.A. Groten, B.W. Wojciechowski, B.K. Hunter, J. Catal.

    138 (1992) 343.

    [55] C.E. Snape, B.J. McGhee, S.C. Martin, J.M. Andresen, Catal.

    Today 37 (1997) 285.

    [56] D.E. Axelson, Solid State Nuclear Magnetic Resonanceof Fossil Fuels, Multiscience Publications Ltd., Canada,

    1985.

    [57] P.E. Eberly, J. Phys. Chem. 71 (1967) 1717.

    [58] M.C. Barrage, J.L. Bonardet, J. Fraissard, Catal. Lett. 5 (1990)

    143.

    [59] S.-B. Liu, S.S. Prasad, J.-F. Wu, L.-J. Ma, T.-R. Yang, J.-

    T Chiou, J.-Y. Chang, T.-C. Tsai, J. Catal. 142 (1993)

    664.

    [60] A.R. Pradhan, J.F. Wu, S.J. Jong, W.H. Chen, T.C. Tsai, S.B.

    Liu, Appl. Catal. A: General 159 (1997) 187.

    [61] P.J. Barrie, Annu. Rep. NMR Spectrosc. 30 (1995) 37.

    [62] J.W. Ward, J. Catal. 13 (1969) 364.

    [63] D.W. Breck, Zeolite Molecular Sieves, Wiley, New York,

    1974, p. 475.

    [64] G. Garraln, A. Corma, V. Forns, Zeolites 9 (1989) 84.[65] A. Corma, V. Forns, A. Martinez, A.V. Orchills, in: W.H.

    Flank, T.E. Whyte Jr. (Eds.), Perspectives in Molecular Sieve

    Science, ACS Symp. Ser. 368 (1988) 542.

    [66] A. Janin, J.C. Lavalley, A. Macedo, F. Raatz, in: W.H.

    Flank, T.E. Whyte Jr. (Eds.), Perspectives in Molecular Sieve

    Science, ACS Symp. Ser. 368 (1988) 117.

    [67] A. Chambellan, T. Chevreau, S. Khabtou, M. Marzin, J.C.

    Lavalley, Zeolites 12 (1992) 306.

    [68] S. Jolly, J. Saussey, J.C. Lavalley, N. Zanier, E. Benazzi, J.F.

    Joly, Ber. Bunsenges. Phys. Chem. 97 (1993) 313.

    [69] S. Jolly, J. Saussey, J.C. Lavalley, J. Mol. Catal. 86 (1994)

    401.

    [70] J. Krger, H. Pfeifer, M. Rauscher, A. Walter, J. Chem. Soc.,

    Faraday Trans. I 76 (1980) 717.