вход по аккаунту


Heterogeneous CatalysisЧStill Magic or Already Science.

код для вставкиСкачать
Heterogeneous Catalysis-Still
Magic or Already Science?**
By Robert Schlogl*
The optimization of industrial heterogeneous catalytic
processes with respect to the minimization of energy consumption and pollution is of increasing importance. Although the results of these efforts may be put into operation
only slowly because of the high capital costs, the scientific
efforts required for these optimizations, though rewarding,
are very substantial. The underlying problem is understanding the operation of a catalytic system at an atomic level; it
is not adequate to have only a model. In the historic approach to catalysis research the complexity of a practical
reaction system is reduced to a level that can be adequately
described. This approach was very successful in the analysis
of a large number of details of heterogeneous catalysis,
giving rise to the hope that someday entire processes may be
understood. This was thought to be possible by extrapolations of experimental details from model investigations to
the real world of an operating system.
Studies on ammonia synthesis, the traditional testing
ground for catalysis science, are the best known examples of
this strategy.['] The culminating success of decades of research was a kinetic theory that allowed the precise prediction of the yield of ammonia under technical conditions using as parameters experimentally determined properties and
a set of experimentally verified elementary reaction steps.[21
Unfortunately, this strategy was not successful in any other practical application as yet. Actual heterogeneous catalytic processes could not be designed up to now. All of the
catalytic processes vital to industry were developed by purely
empirical methods and countless screening experiments.
This heuristic approach has led to the opinion that catalysis
research is not a science but rather "black magic". Industrial
chemists tend to be rather skeptical towards inductive-scientific strategies in heterogeneous catalysis, since it is evident
that our understanding of the relationships between structure and reactivity is inadequate.
One significant obstacle to understanding structure-reactivity relationships is the complexity of the solid catalyst.
Active materials are multiphase compounds with dynamic
behavior with respect to both surface and bulk structure.
These dynamics were recognized only recently and relate to
geometric structure, composition, and the local electronic
state"' of the active center. The analyst of a catalyst should
thus not be preoccupied with a priori distinctions between
important and unimportant components of the system but
should consider all constituents (active components, support, impurities, poisons, promotors) of a practical catalyst.
The long history of catalyst characterization should have
taught the lesson that all analysis should be done in situ; all
monographs on the subject repeat this statement.[31Nevertheless only a fraction of the analyses are conducted under in
situ conditions, since only few techniques can be used for
[*] Prof Di-. R. Schlogl
lnstitut fur Anorganische Chemie der Universitat
Niedcrurseler Hang. D-W-6000 Frankfurt 50 ( F R G )
This work was supported by the Deutsche Forschungsgemeinschaft, the
Max-Planck Society, and the Fonds der Chemischeu Industrie The author
thanks the manyco-workers involved with the projects mentioned. We also
thank BASF (Ludwigshafen) for a longstanding cooperation.
such investigations and even then the technical reaction conditions must be modified.
Thus these experiments place certain technical constraints
o n the reaction conditions. A realistic description of an in
situ experiment may be that a property of a catalyst is investigated as a function of reaction conditions characteristic
of the practical application under simultaneous observation
of the reaction kinetics (conversion, selectivity). Only then is
it ensured that the active state of the catalyst is characterized
and that the data are relevant to the reaction mechanism.
Knowing the mechanism is crucial to understanding the
catalyst operation, which in turn is central to the scientific
approach towards catalyst optimization. Extrapolations
from experimental to real operation conditions must be
limited, as the mechanism is dependent on interconnected
sets of parameters whose quantitative relations are unknown
(Fig. 1 ) . Quantitative knowledge of these interconnections
allows the full understanding of the mode of catalyst operation.
Activity and selectivity data d o not allow conclusions concerning the reaction mechanism. The kinetics are determined
by chernisorption processes (only a part of the mechanism)
and by the micro- and macrotexture of the catalyst. The
limitations of mass- and energy-exchange processes and noncatalytic consecutive reactions tend to obscure the relationship between kinetics and mechanism. The same holds for
the relationship between the elementary steps and the
microkinetics composed of them.
Analytical in situ experiments and all non in situ studies
require deviations from the empirically optimized reaction
conditions which affect the four groups of parameters in
Figure 1 simultaneously. These deviations cause parameter
gaps. If the gaps are closed by inadequate extrapolations the
interconnection of all factors may lead to possibly dramatic
Fig 1. Relationships between the catalytic performance of a solid (middle) and
the main determining factors. The gaps arise from simplifications in the
parameter sets that are required for catalysis experiments. Extrapolations from
the data obtained from the simplified model to the actual catalytic reaction may
endanger the relevance of a whole mechanistic reconstruction.
mechanistic errors. This is the reason for the failure of catalysis research as a science which has to prove its validity in the
industrial application. As an example, much information has
been accumulated on the elementary steps of chemical reactions on single crystals under high-vacuum conditions. Yet
the “pressure gap” and “material gap” have been bridged
inadequately, and a n interpretation of the formal kinetics is
not possible. As a further illustration for this point, the famous prediction arising from surface science of a theoretical
limit for the selectivity of ethylene epoxidation was proven
false in industrial practice.
The gaps shown in Figure 1 are all relevant in the design
of in situ experiments. The transport gap describes the influences of mass- and energy-transport limitations on the kinetics. These phenomena are different for a catalyst in a fixedbed reactor, for transparent wafers of the same material, and
for catalysis experiments with single crystals. The pressure
gap describes differences in the total pressure in experiments.
The reaction mechanism frequently changes profoundly in a
pressure range between millibars and ultrahigh-vacuum conditions because of threshold limits in the chemical potential
of the gas-phase
The material gapLs1describes the
difference in local microstructure and defect properties between single crystals (well-defined, easy to analyse, low
dynamics) and polycrystalline materials (ill-defined, highly
reactive, difficult to analyze). The model gap finally addresses the problem of finding the physical relevance of parameters obtained in a formal mathematical analysis of catalyst
performance. Typically, gradients of specific activity, chemical transformations, and structural modifications are
assumed to exist along the reactor tube or within a single
grain of catalyst. Their experimental verification is in most
cases, however, not possible.
In a fashionable approach to understanding a process a
model system whose complexity is intermediate to that of a
single crystal and a real system is defined and examined.
Although this approach is partly successful, experimental
verification of the relation between model and reality is very
This list of interconnected problems was addressed in a
recent study of the methanol synthesis catalyst,r71in which
the authors describe their attempts at determining the nature
of the active material of the catalyst experimentally. It is not
known if in the hydrogenation of C O the active copper surface is metallic o r oxidic. This reaction is the present testing
ground in catalysis research, in analogy to the ammonia
synthesis.[*] Certainly if the chemical nature of the active
species is not known,’’] the mechanism of this reaction cannot be worked out scientifically.
For the difficult task of determining the valence state of
the active copper species, which was attempted frequently
before, a new spectroscopic technique has been developed
which can provide information concerning the catalyst surface in situ. The results of the method are compared to and
calibrated against the results obtained by more traditional
techniques. The crucial experiment is a variation of the wellknown extended X-ray absorption fine structure spectroscopy (EXAFS) experiment. This method allows the determination of the local geometric structure of a solid for a
specific element without requiring long-range order as in
conventional diffraction experiments. For metal atoms the
VCH VerluggeseNscha/t mbH, W-6940 Weinheim.1993
specific X-ray radiation is hard enough to penetrate a gas
atmosphere and the windows of a reaction cell, allowing in
situ experiments on supported metal catalysts. In these
studies of the Cu catalyst for the hydrogenation of CO the
bond lengths determined allowed the differentiation between
copper metal and copper oxides.
The experiment was conducted with a technical, polycrystalline, supported catalyst at ambient pressure in a gas atmosphere typical for industrial conditions (ca. 60 bar pressure).
The bulk-sensitive EXAFS data clearly show that the catalyst is present in the metallic form under the conditions
chosen. This was confirmed in a similar in situ experiment
with conventional X-ray diffraction. The two sets of complementary structural information indicate that neither crystalline nor amorphous oxide are present in the catalyst.
Nothing has been said so far about the state of the copper
surface, the essential interface for catalytic operation. The
new experiment helped here: X-ray absorption was detected
by measuring the yieid of secondary electrons caused by the
photoabsorption process, which converted the method from
bulk-sensitive to surface-sensitive (information from a few
monolayers instead of several micrometers). This technique
also showed very convincingly that no copper oxide was
present on the surface.
The nature of the active catalyst would have been clarified
at this point if the experiments had been truely in situ. Unfortunately, as the authors did not determine the activity and
selectivity of their catalyst, they describe an arbitrary state of
the catalyst and give no evidence that the catalyst was indeed
active. This point is not trivial (in the light of the gas phase
applied), as the complexity of a working catalyst system
outlined in Figure 1 does not permit this far-reaching extrapolation. It is frequently observed that the top layer of a
catalyst bed is deactivated by poisons from the feed. In the
experiments discussed only the top layer of the catalyst bed
was investigated. N o information on the gas purity, the
shape of the catalyst bed, and the influence of the catalyst
geometry (differential flow reactor) on the kinetic performance was given. Because of this lack of experimental information the investigation is not an in situ experiment, and the
nature of the active copper species is open for further discussion.
At this stage the question arises of which other methods
may be used to obtain information relevant to catalysis.
Instationary conversion experiments are often suitable. In
such truely in situ experiments the catalytic performance is
analyzed as function of a process variable that changes with
a defined time profile. The determination of dissolved
atomic nitrogen in the catalysts used in the technical ammonia synthesis serves as an example.“0’ Its interpretation required, however, a detailed knowledge of the surface chemistry of elemental iron.[”]
Such detailed chemical information is a prerequisite for a
mechanistic analysis of a catalytic process and can only be
obtained with certain reliability if several complementary
methods are used for their determination. Only then can
misinterpretations occurring as internal inconsistencies
within one dataset be recognized and possibly corrected. An
example of the erroneous overinterpretation of a single analytical method is the theory of paracrystallinity in catalysis.
Carefully measured X-ray powder diffraction data of ammo-
0570-0833~9310303-03X2$/0.00+ .25/0
Angew. Chem. lnr. Ed. Engl. 1993, 32, No. 3
nia synthesis catalysts revealed characteristic anomalies in
the lineprofiles.1121These were correlated with molecular dispersions of promoter oxides within the bulk iron, and special
catalytic effects were ascribed to this unusual state of matter.
Today we
that the lineprofile anomalies were
caused by special crystallographic artifacts which were
known to exist then but were not considered. No correlation
exists between the bulk X-ray diffraction data and the catalytic performance.
A further example of incorrect structural data from an in
situ experiment on the active phase of the ammonia synthesis
catalyst was caused by the inadvertant activation of the catalyst within the X-ray diffraction system. The resulting
amorphous state was concluded['41 to be the active principle
in ammonia synthesis, again without testing the catalytic
activity during the diffraction experiment. Today we know
that the activated catalyst contains highly oriented, crystalline iron metal particIes.['O1
As outlined in Figure 1 the detailed knowledge of microstructure alone is inadequate for a complete understanding of the mode of operation of a catalyst; information about
its surface constitution is essential. Unfortunately, all surface science techniques require ultrahigh-vacuum conditions
because of the nature of the information-bearing particles
(photoelectrons, ions, atoms). Real on-line in situ experiments with these techniques are thus not possible. The most
adequate solution to this dilemma is the combination of a
high-pressure microreactor with a surface analytical instrument. The suitable design of the transfer system between the
high-pressure and ultrahigh-vacuum sections ensures that
the working catalyst surface is frozen.[15] Any reactive
chemisorbed molecular species will, however, be lost in the
evacuation process. This technique was used for the investigation of the nature of the active phaserL6]of the iron oxide
catalyst used for the dehydrogenation of ethylbenzene to
styrene. A published model of the catalyst operation1"] predicted that the conversion should increase with the surface
abundance of potassium (an essential promoter) and should
decrease as the surface becomes covered with carbon (from
the products). A summary of X-ray photoelectron spectroscopic data collected from catalysts operated at various levels of conversion in a steady state prior to analysis['*] is given
in Figure 2. It can be seen that both predictions from the
catalyst model are incorrect. The parameter controlling the
conversion is the abundance of hydroxyl ions indicating the
presence of potassium in the deactivated form of KOH,
which results from disintegration of the active ternary iron
oxide KFe0,.[l6' A number of other promoters that are
added to technical catalyst systems to improve the selectivity
and stability of the catalyst are either not abundant at the
surface or their bulk concentration does not appear to correlate with the surface catalytic process. This example demonstrates the value of an in situ experiment including simultaneous characterization of the catalyst performance. Its
results were corroborated by a variety of complementary
experiments probing both the bulk and the surface of the
In consideration of this discussion if we try to answer the
question asked in the title the realistic reply must be-still
magic. The obstacles to a scientific understanding of hetero-
A n w ! ' . Chem. Int. Ed. Engl. 1993,32.N o . 3
Fig. 2. Results of quantitative XPS analyses on an iron oxide catalyst active in
the dehydrogenation of ethylbenzene to styrene. The data were measured at
given steady-state conversion levels. The conversions were carried out under
technical conditions (873 to 893 K, 6.1 mixture of water ethylbenzene, 1 bar
total pressure, liquid hourly space velocity (LHSV) 0.5 Lh-'). XPS data were
obtained after rapid transfer of a fraction of the catalyst sample into ultrahigh
vacuum. All numbers are normalized to 100. x Axis: Conversion to ethylbenmne in %. y Axis: Spectroscopic parameter in arbitrary units. Light gray
columns: carbon abundance, dark gray columns: potassium to iron ratio, black
columns: hydroxyl abundance.
geneous catalysis are considerable. The perspectives of penetrating the magic of catalysis by scientific means is, however,
promising owing to the increasingly frequent attempts at
circumventing these obstacles. Catalysis research today is on
its way to becoming an interdisciplinary science incorporating in a rational way the knowledge from surface science,
solid-state physics, and chemical engineering. Its progress
will be measured by its contribution to the efficient solution
of the strategies problems of future catalyst applications.
German version: Angew. Chem. 1993,105, 403
[ l ] G. Ertl, Angew. Chem. 1990,102,1258; Angew. Chem. I n t . Ed. Engl. 1990,
[2] P. Stoltze, Phys. Scr. 1987,36,824.
[3] Characterization ofcatalysts (Eds.: J. M. Thomas, R. M. Lambert), Wiley, New York, 1980;J. R. Anderson, K. C. Pratt, Inrroduction to Characterization and Testing ofCalalysts. Academic Press, 1985;Characterization
of Heterogeneous Catalysts (Ed.: F. Delannay), Dekker, New York, 1984.
[4] C. Rehren, M. Muhler, X. Bao, R. Schlogl, G. Ertl, 2. Phys. Chem. 1991,
174,1 1 .
[S] R. Schlogl, R. C. Schoonmaker, M. Muhler, G. Ertl, Catal. Lerf. 1988,I,
[6] See for example: D. R. Strongin, G. A. Somorjai in Catalytic Ammonia
Synfhesis (Ed.: J. R. Jennings), Plenum, New York, 1991, p. 133; G. A.
Somorjai. Chemistry in Two Dimensions, Cornell University Press, Ithaca.
[?] G. T. Moggridge, T. Rayment, R. M. Ormerod, M. A. Morris, R. M. Lambert, Nature 1992,358,658.
[S] G. Ertl, J. Vac. Sci. Technol. A 1983,1, 1247; A. Ozaki, K. Aika in Catalysis. Science and Engineering, Vol. I (Eds.: J. R. Anderson, M. Boudart),
Springer, Berlin, 1981,p. 88.
[9] B. S. Clausen, J. Catal. 1991, 132, 524; G.C. Chincen, M. S. Spencer, K.
Waugh, D. A. Whan, J. Chem. Soc. Faraday Trans. I 1987.83,2193.
[lo] W. Mahdi, J. Schiitze, G. Weinberg, R. Schoonmaker, R. Schlogl. G. Ertl,
Catal. Lett. 1991, if, 19.
Ill] F. Bozso, G. Ertl, M. Weiss, J. Catal. 1977,50,519.
1121 R. Hosemann, A. Preisinger, W. Vogel, Ber. Bunsen Ges. Phys. Chem. 1966,
70, 797.
[13] W. S. Borghard, M. Boudart, J. Cafal. 1983.80,194.
1141 T. Rayment, R. Schlogl, J. M. Thomas, G. Ertl, Nature 1985,315,311.
1151 M. Muhler, R. Schlogl, S. Eder, G. Ertl, Surf. Sci. 1987,189,69.
[16] M. Muhler, R. Schlogl, A. Reller, G. Ertl, Catal. Lerf. 1989,2, 201; M.
Muhler, R. Schlogl, G. Ertl, SIA Surf. Interface Anal. 1988,i2, 233; M.
Muhler, J. Schiitze, M. Wesemann, T. Rayment, A. Dent, R. Schlogl, G.
Ertl, J. Catal. 1990. 128. 339.
[17] D. Mross, Catal. Rev. Sci. Eng. 1983,25, 591; W. Mross in DECHEMA
Statusseminar Fortschritte der Katalyseforschung, 1987,p. 173.
1181 C Abundance: Integral of the C-1s line; K:Fe ratio: ratio of the corrected
integrals of the K-2p and Fe-2p line; OH abundance: relative area of the
OH emission in the 0 - 1 s spectrum at 531.7 eV relative to the emission of
the metal oxides at 529.9 eV.
Verlagsgesellschafi mhH. W-6940 Weinheim, i993
S iO.00+ .25/0
Без категории
Размер файла
508 Кб
catalysisчstill, already, heterogeneous, magii, science
Пожаловаться на содержимое документа