- Email: [email protected]
High throughput screening in monolith reactors for total oxidation reactions
Applied Catalysis A: General 254 (2003) 35–43
High throughput screening in monolith reactors for total oxidation reactions Martin Lucas, Peter Claus∗ Institute of Chemical Technology, Darmstadt University of Technology, Petersenstreet 20, D-64287 Darmstadt, Germany Received 31 October 2002; accepted 4 February 2003
Abstract We present here a powerful tool for high throughput screening of heterogeneous catalysts, successfully developed in our laboratory with the application focus on common monoliths as a special type of microreactors. It could be shown that it is possible to prepare in a reproducible manner catalytically active coatings on the wall of single channels of the monoliths by a channel-by-channel procedure enabling the application of these multichannel reactors, coupled with mass spectrometry or gas chromatography, for the fast screening of heterogeneous catalysts. Because time and spatially resolved sampling is needed and complex gaseous mixtures have to be analyzed, a special 3D positioning system, which allows measurements at high temperatures, is needed as the central element of the equipment. The high efficiency and reliability of the channel-by-channel preparation as well as the developed screening method is demonstrated for the total oxidation of hydrocarbons and carbon monoxide in the presence of further components like O2 , H2 O, CO2 , NO, SO2 and inert gas over precious metal catalysts. © 2003 Elsevier Science B.V. All rights reserved. Keywords: High throughput screening; Monolith reactor; Channel-by-channel preparation; Coating; Oxidation; Automobile-exhaust gas catalysts
1. Introduction With the development of high throughput methods for new catalysts a stormy progress could be observed in the last years. One of the main points of criticism at the systems developed for the heterogeneous catalysts is the unsatisfactory comparability of the results with those obtained in the technical reactor. The comparability is mainly limited by the necessary miniaturization of the screening devices. In the field of the combinatorial catalysis previous experiences with the ∗ Corresponding author. Tel.: +49-6151-16-5369; fax: + 49-6151-16-4788. E-mail address: [email protected] (P. Claus). URL: http://www.ct.chemie.tu-darmstadt.de/ak claus/index de.html.
analysis of heterogeneously catalyzed reactions can be split into two solutions. The first variant comprises the use of new analytical methods which are able to determine the selectivity to target products or, at least, the activity of several catalytic systems in a parallel way. For example, by emission corrected infrared (IR) thermography, Maier and co-workers [1] used IR cameras with high temperature resolution to spatially resolve reaction heats of the hydrogenation of 1-hexyne, the oxidation of hydrocarbons (isooctane, toluene) over catalysts of a library consisting of 37 combinations of transition metals on amorphous microporous mixed oxides. Disadvantage of this variant is that only limited information about the reaction examined can be achieved because, clearly, the identification of reaction products and, thus, the estimation of selectivities
0926-860X/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0926-860X(03)00261-8
36
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
is not possible. However, this technique is especially useful to detect the catalyst activity in a very sensitive and effective way during primary screening in combinatorial catalyst development. Among the optical screening methods, resonanceenhanced multiphoton ionization (REMPI) can, in principle, provide information about the reaction products, if the absorption and ionization features of the latter are known [2]. Senkan and Ozturk combined microreactors with REMPI as an optical screening tool to detect the catalytic properties of 72 Pt-Pd catalysts [2] and a ternary Pt-Pd-In/Al2 O3 library (66 catalysts [3]) for the simple dehydrogenation of cyclohexane to benzene. Up to date, the use of REMPI for screening reactions with more complicated product mixtures is not known. In the second variant, to overcome the drawbacks of the optical methods, high-throughput devices were developed which allows feeding of the gaseous educt mixture through the individual catalysts and the reaction products are analyzed by standard analyzing methods like gas chromatography and mass spectrometry [4–12] achieving quantitative informations about catalyst activity and even selectivity in an easy and economic way. Two basic concepts for the application of scanning mass spectrometers have been published: scanning a catalyst library wafer, which consists only of sites of thin film patches, with two concentrical capillaries in a transient mode [4] and scanning a continuously operated reactor array by spatially resolved analyses [5–12]. Catalyst screening techniques based on mass spectrometry were applied to several reactions like the partial oxidation of propene over a sol–gel derived catalyst library with 33 mixed oxides [7], the oxidation of CO over 120 ternary thin film Pt-Pd-Rh and Pd-Rh-Cu catalysts [4] and over supported gold catalysts [8,9], oxidative dehydrogenation of ethane over 66 ternary combinations of Mo, V and Nb oxides [10], and dehydrogenation of cyclohexane over 66 combinations of alumina supported Pt, Pd and In [11]. Due to the high number of catalysts which have to be examined it is fact for both variants that the structures into which these catalysts have to be deposited must be miniaturized in an appropriate manner. The results achieved by this way can only partially be compared with those achieved from technical reactors. We present here a powerful tool for high throughput screening of heterogeneous catalysts, successfully
developed in our laboratory with the application focus on common monoliths as a special type of microreactors where the channels are used as single plug-flow reactors containing different catalysts. Preconditions were to set a catalytically active layer on the channel walls of ceramic monoliths and to analyze the catalytic properties of all catalysts to be compared so that the comparability of all results, even for deactivating catalysts, could be guaranteed. That means that inside the monolith structure different catalytic assemblies shall be synthesized channel-by-channel and the catalytical activity shall be determined at the same values of time-on-stream. We wanted to achieve a good reproducibility and a high degree of automatism for all steps, beginning with the preparation, going through the measurement to the analysis, to prevent the user from being too much involved in extremely time-consuming serial measurements of the catalytic properties. The method was tested for the total oxidation of hydrocarbon and CO in the presence of further components like O2 , H2 O, CO2 , NO, SO2 and inert gas. 2. Experimental 2.1. Catalyst preparation in monolith structures Monolith structures have been used for a long time already as part of catalysts, e.g. as support of automobile-exhaust catalysts. The catalytic active layer is deposited onto the walls of the monolith by the wash-coat-procedure. The commonly used monolith materials are permeable for gases and fluids. For the use of each monolith channel as a single plug-flow reactor these characteristics of the common monolith material are rather restricting as diffusion, from channel-to-channel, through the channel walls will falsify the examinations or even will make them impossible. Therefore we chose a monolith material (Corderit 410, Inocermic GmbH Hermsdorf) which is impermeable for gases and fluids. The disadvantage is the bad wettability of the material with water as this can lead to problems during the coating with the support material or catalyst precursors. The monolith structure used consists of 10 × 20 channels with a diameter of 2.6 mm (72 cpsi). The channel length is 75 mm. A number of 128 of the existing 200 channels
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
can be used for the catalytical tests which are arranged in an array of 8 rows and 16 columns. By a wash-coat procedure [13] the monolith is coated uniformly with the catalyst support material (Al2 O3 , SiO2 , TiO2 etc.). The thickness of oxide layer can be adjusted variably. Afterwards the coated monolith is saturated channel-by-channel individually with aqueous solutions of the metal precursors (most of all we used the nitrates, in case of Ru and Au the chlorides) after it is sealed up at one side. For example in case of the Al2 O3 support material, with an absorption capacity of water of 500 ml/kg, 260 l of a precursor solution with a content of 2 wt.% metal will be filled in a channel to prepare a catalyst with 1 wt.% metal. After a short time (2–10 s) which is necessary for saturating the oxide layer completely, the dilutions are removed. The supply of these 128 individual solutions, the filling of the channels and the removing of the solutions after a defined time is effected by a robot (Tecan Miniprep 60). If the pore volume respectively the capacity of water absorption of the oxide layer is already known the desired content of a metal can be adjusted with high accuracy (Table 1); this is analog to the dry impregnation or incipient-wetness method [14]. Thus, the prepared monolith is now ready for the common pretreatment methods like calcinations and reduction in flowing air and hydrogen (both 3 h at 400 ◦ C, flow = 50 l/h), respectively. Beside this impregnation procedure, the immobilization of metals
Table 1 Comparison between the desired contents of Pt and Zr (theoretical metal content) and the experimental data obtained by ICP-OES in every eight channels Experimental (wt.%) (ICP-OES)
Pt column 11
Pt column 12
Zr column 12
Channel 1 Channel 2 Channel 3 Channel 4 Channel 5 Channel 6 Channel 7 Channel 8 Average value (wt.%) S.D. (wt.%) Theoretical (wt.%)
1.08 1.06 0.96 1.22 1.19 1.23 1.08 1.04 1.108 0.096 1.0
0.53 0.45 0.74 0.65 0.65 0.64 0.47 0.63 0.595 0.101 0.5
0.64 0.57 0.56 0.65 0.61 0.65 0.64 0.61 0.616 0.035 0.5
Support material Al2 O3 with an absorption capacity of water of 500 ml/kg.
37
could be also achieved by deposition–precipitation using urea. In this case, the monolith was again sealed up at one side, filled channel-by-channel with aqueous solutions of the metal precursors, followed by mixing with urea and finally heating to 80 ◦ C. At this point, decomposition of urea starts yielding hydroxid ions which give the corresponding hydroxides. Subsequently, again calcination and reduction steps produce a library of supported oxide or metal catalysts. 2.2. Development of screening device and analysis The experimental setup for the HT-screening is shown in Fig. 1. The inlet capillary of a quadrupole mass spectrometer (QMS, Quadstar Pfeiffer Vacuum) is moved in x- and y-direction of the monolith and then into the particular channel (z-direction) by a three-dimensional (3D) positioning system (Amtec). The latter contains the monolith to be tested and is the central element of the equipment. The position of the capillary above a monolith channel can be automatically changed by using the three stepping motors of the positioning system supervised by two CCD cameras and an inspection window and is exactly controlled by the software. The inspection window restricts the available temperature range to 250 ◦ C and the pressure range to normal pressure during the test runs. The gaseous educts are taken from gas mixtures stored in gas bottles by means of mass-flow-controller (FiC01 and 02, Hitec) and conducted to a evaporator. Together with a GC-sampling valve this evaporator is situated in a thermostat housing. A defined amount of water is dosed into the evaporator by a liquid-flow-controller (Hitec), then vaporized there and conducted to the 3D positioning system after being mixed with the reaction gases. The interior of the 3D positioning system is rinsed with argon. Inside the 3D positioning system the monolith is situated in a stainless steel housing which is heated to the reaction temperature. With respect to the approaches described so far in the literature [14] the difference of this equipment is the type of feeding of the gaseous educt mixture via a 1/16 in. tube positionable in the x–y–z direction (Fig. 2). This tube is brought to a single channel of the monolith while a teflon disk, situated at the end of the tube, is sealing this channel towards its surroundings. Consequently the gaseous educt mixture only flows through this single channel
38
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
Fig. 1. Principle of the developed apparatus for high throughput screening of heterogeneous catalysts in monolith reactors.
of the monolith. Inside this dosing tube there is 1/32 in.-sample-drawing capillary which juts 48 mm out over the end. The sample drawing is made by sucking a defined gas amount which is controlled by a mass-flow-controller (FIC04) out of the single channel feeded with gaseous educt mixture using a vacuum pump. According to the length of the sample-drawing capillary, the sample drawing is made at a distance of 48 mm from the channel entry. On its way to the vacuum pump the gas probe passes the GC-sampling valve mentioned above which injects the contents of the sample loop (PS in Fig. 1) computer-controlled into the GC. Behind the GC-sampling valve and following the flow direction some part of the gas probe
taken from the monolith channel is supplied to a mass spectrometer. Alternatively a three-way switching valve (H3) can be switched in a way that the gaseous educt mixture can be analyzed. Prompt and exact analyses are achieved by using the mass spectrometer unless there are overlaying spectra (MZ 28) as it happened with CO and CO2 which cannot be separated numerically if there is a big CO2 excess. In that case we used a gas chromatograph (HP 5890A) with 30 m capillary column (Poraplot Q). Before the flame ionization detector used for educt/product detection a selfmade methanizer (5 wt.% ruthenium on Al2 O3 ) was switched. At 450 ◦ C this methanizer converts, via hydrogen excess, carbonaceous substances eluting
Fig. 2. Principle sketch of gas dosage and sampling in the monolith. The gaseous educt mixture (reaction gas) is proportioned channel-by-channel into the monolith. At the same time, a gas probe is taken in 48 mm of depth.
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
from the column completely into methan. Thus it is possible to determine the contents of CO, CO2 , propene and propane by one GC-column within the analysis cycle of 4.5 min. By using a micro GC (Agilent “G2890A” Micro GC with 10 m Molsieve 5A and 8 m Poraplot U column) the time for an analysis cycle can be reduced to 1.5 min, if necessary. The screening of the monolith takes place after the following algorithm: the sampling and dosing (SAD) equipment goes into the first channel to be examined and stays there for 4.5 min. During this time mass spectra are constantly recorded to determine the temporary course of the components’ contents (without CO). After these 4.5 min there is a sample drawing by the GC and thereafter the change to another channel where the SAD equipment stays for another 4.5 min respectively until the GC-analysis of the previous channel has been finished and the GC is able to analyze another gas probe. During this time again mass spectra are constantly recorded. This algorithm takes place for all channels to be examined. Altogether the determination of the catalyst activity of all 128 monolith channels takes ∼10 h, whereas the time-on-stream is the same for each channel and takes only 4.5 min at the present example. For the rest of the time, each channel is flooded by argon. If the quantity of the taken gas probe is adapted to the quantity of gas given to each channel, the gas will
39
flow exclusively through the channel chosen. Additionally it can be checked by this method whether the admitted gas is completely dosed into the channel or whether it partly passes the teflon gasket and ends in the 3D positioning system. If that should be the case a considerable amount of argon which is flowing the neighbor channels is to be found in the gas probe. Fig. 3 shows the signal of the mass spectrometer for mass numbers 4 and 40. Every 4.5 min the channel for the gas probe was changed. In the figure this phenomena is to be seen as a vertical line at the argon signal as the sample-taking capillary goes, when changing the channels, outside of the monolith through the 3D positioning system filled with argon. It is clearly to be seen that argon is only marginally detected in the gas probes taken from the channels and thus the teflon gasket is sealing the channels reliably. When using monolith structures for HT-screening the temperature gradient within the monolith is a big problem for the comparison of results which were won with possibly immensely varying temperatures respectively a bad temperature distribution within the channels. This temperature gradient results from the minimal thermal conductivity of the ceramic material when heating the monolith from outside and from the high space velocity necessary for many applications the monolith structure was made for. Should a test run of a monolith take place in a conventional method
Fig. 3. Exemplary results of channel-by-channel feeding of 200 ml/min helium into the monolith followed by taking a gas probe (180 ml/min) and analysis by MS (Tmonolith = 150 ◦ C). Dotted vertical lines mark the move to the next monolith channel.
40
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
where the complete monolith is feeded and the sum of the conversion of all channels is determined as in the automobile-exhaust gas catalyst, a comparability regarding the average temperature distribution from monolith to monolith is usually given. In order to get the desired steady temperature distribution in all channels of the monolith which is necessary for the application to be done, preheated argon is led through all channels in reverse flow (except the channel feeded by gaseous educt mixture). The heating device which is coated around the monolith is not used for heating the monolith to reaction temperature; it is only necessary to balance the heat loss via the exterior wall. The temperature distribution gained by this method within the monolith was measured by inserting a thermal element into different channels of the monolith depending on the depth and is shown in Fig. 4. The monolith is divided into a channel pattern of 16 columns in height and 8 rows in width. Fig. 4 shows the temperature distribution across the monolith in different depths. It is obvious the outer channels show higher temperatures than the inner ones, however, the deviation is with ±1 K in radial direction quite tolerant. Larger deviations of ±4 K were measured in axial direction, but we have to state that all channels examined show a very similar course of the temperature profile in axial direction and thus the effects to the comparability of the catalytic activities should be slight. Moreover, the reaction heat which has a big influence on the medium temperature of the monolith during total oxidation of hydrocarbons examined can be transported off easily by the reaction stream in the
Fig. 4. Temperature profiles in the monolith measured in different depths and channels along the middle row of the monolith.
Table 2 Influence of the capillary material, used for feeding the gaseous educt mixture, on the activity of a monolith coated only with Al2 O3 (blind reaction): conversion of CO and propene when using stainless steel capillary or a capillary passivated with SiO2 T (◦ C)
150 200 225 250
Stainless steel capillary ¯ CO (%) X
¯ propene (%) X
12.05 34.69 97.11
8.73 11.69 24.92
Capillary passivated by SiO2 ¯ CO (%) X ¯ propene (%) X 0.79 1.47 3.21
0.20 0.23 0.40
single channel which is surrounded directly by eight channels flooded with argon. Table 2 shows conversion data of the blind reaction measured in a monolith coated only with Al2 O3 which were relatively high when using the stainless steel capillary for feeding the gaseous educt mixture and taking gas probes during the reaction. By installing capillary tubes passivated by SiO2 this amount of the blind reaction could nearly be eliminated completely.
3. Results and discussion Nevertheless, the evaluation of the HTS procedure cannot be made via the measurement of the single gradients respectively the irregularities of the parameters temperature, gas dosing, analysis incl sample taking and preparation. The very fact is the comparison of the catalytic properties of the channels among one another. For this purpose we produced catalytically active wall coatings of identical composition for every eight channels of the monolith. Table 1 shows two examples for such preparations. In column 11 a layer with 1 wt.% Pt/Al2 O3 should be produced, in column 12 it should be a catalyst with 0.5 wt.% Pt and Zr. As to be seen from the quantitative analysis via ICP-OES the metal contents can be adjusted relatively exactly and reproducible. Statistics show a content which is 0.1 wt.% higher than the content desired; but this can be corrected during the preparation by a metal precursor solution being accordingly lower. The Pt-Zr catalyst was not produced by simultaneous saturating with a mixture of Pt and Zr precursor solution but by impregnation of the oxide layer with Pt first, drying
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
41
Fig. 5. Conversion of CO measured in two different columns of the monolith. The eight channels of a column should have the same catalytic composition (white chart: 0.5 wt.% Pt and 0.5 wt.% Zr, supported on Al2 O3 ; grey chart: 1 wt.% Pt/Al2 O3 ).
at 300 ◦ C for 3 h and followed by the deposition of Zr. After the second preparation step, the monolith is calcinated at 400 ◦ C in air (flow = 50 l/h). A variable mixture of CO, propene, O2 , H2 O, CO2 , NO, SO2 , and inert gas with a space velocity of 50,000 h−1 was given into the channel for the measurements (200 ml/min total volumetric rate). The total oxidation of CO and propen was examined at normal pressure in a temperature range of 100–250 ◦ C. Fig. 5 shows the conversion of CO at 150 ◦ C for each of those channels in which the above mentioned catalytically active layers were prepared. For the monometallic Pt catalyst the average CO conversion was determined with 38.2% at a S.D. of 3.3%. For the bimetallic Pt-Zr catalyst we have a much larger S.D. of 12% at an average CO conversion of 23.9%. Clearly to be seen in Fig. 5 is that the conversion of the Pt-Zr catalyst is extremely decreasing in the middle of the monolith whereas the column with the monometallic Pt catalyst, which is directly alongside, does not show such a geometric gradient of the conversion. Parameters of the screening like temperature gradients can thus be eliminated as reason for this phenomenon which up to now has only been seen at catalysts containing Zr and Mn or with converse effect at some catalyst combinations containing Rh. As a possible explanation we assume a temperature gradient during the catalyst pretreatment which should be done in an oven free of temperature gradients as recently shown by Moulijn
and co-workers [15]. It is very difficult to confirm this assumption by determining the catalyst composition in dependence of their axial position in these small channels and has not yet been executed (each channel contains approximately 20–30 mg of catalyst). The next step was to prepare 32 different catalysts within the monolith in such a way that always four channels contain identical catalyst compositions. In this case, we prepared 1 wt.% Pt, Rh, Ru, Pd, Au, Zr, Mn and Ag onto Al2 O3 layers as well as selected bimetallic combinations of the mentioned metals with a content of 0.5 wt.% of each component. The CO conversion obtained at 150 ◦ C are shown in Fig. 6. For each catalyst we analyzed two successive GC-probes. Thus an estimate of the deactivating respectively activating of the catalyst of 4.5 min time-on-stream (first run) compared to 9 min time-on-stream (second run) could be gained. Note, that the mass spectrometer produces concentration versus time curves with high resolution with respect to the content of the individually dosed components (except CO). Part of the catalyst screening of the above mentioned monolith with 32 different catalysts is to be seen in Fig. 7. The vertical lines in the curves mark again the change of each channel after nine minutes. Every four channels had, as mentioned above, an identical catalyst composition and were synthesized directly one after the other. These identical compositions agree quite well with the time processes especially of the components propene, SO2
42
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
Fig. 6. Results of quantitative MS analyses of gas probes from different channels of the monolith with Al2 O3 supported catalysts at T = 150 ◦ C. Dotted vertical lines mark the move to the next monolith channel. Note, that every four channels following one after another in sequence include the same catalyst composition, i.e. the first four channels include catalysts with 0.5% Pt-0.5% Au, the next four channels 0.5% Pt-0.5% Mn, etc. Every catalyst composition gave characteristic pattern in product mass spectroscopy.
Fig. 7. CO conversion at 150 ◦ C for a catalyst library containing 32 different materials within the monolith (see text). Values are averages for the four channels with desired identical catalyst composition. The S.D. is related to the conversion rate of the second run.
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
and H2 in Fig. 6 as nearly identical patterns of these components were analyzed in the mass spectrometer at every four successive channels. 4. Conclusions It could be shown that it is possible to prepare in a reproducible manner catalytically active coatings in the single channels of monoliths by a channelby-channel procedure enabling the application of these structures for high throughput screening of heterogeneous catalysts. Moreover an equipment was developed which is able to reproducibly feed any gaseous educt mixture into single channels of a monolith and to take representative probes of the reaction products. It is of great advantage that the method presented here enables a HT-screening with the common analytical equipment (GC or MS) of a research lab in catalysis. In addition to the application of the presented tool for the oxidation of CO and hydrocarbons shown above this equipment was successfully used for the selective oxidation from propene to propene oxide. During these tests one-hundredth of the usual gas volume of the total oxidation was dosed into the single channels. In spite of the low volumetric feed of 2 ml/min reproducible results could be achieved [16]. References [1] A. Holzwarth, H.-W. Schmidt, W.F. Maier, Angew. Chem. Int. Ed. 37 (1998) 2644–2647.
43
[2] S.M. Senkan, Nature 394 (1998) 350–353. [3] S.M. Senkan, S. Ozturk, Angew. Chem. Int. Ed. 38 (1999) 791–795. [4] P. Cong, R.D. Doolen, Q. Fan, D.M. Giaquinta, S. Guan, E.W. McFarland, D.M. Poojary, K. Self, H.W. Turner, W.H. Weinberg, Angew. Chem. Int. Ed. 38 (1999) 483–488. [5] T. Zech, A. Lohf, K. Golbig, T. Richter, D. Hönicke, in: Proceedings of the Third International Conference on Microreaction Technology, Frankfurt, Main, Germany, April 1999. [6] T. Zech, J. Klein, S. A. Schunk, D. Demuth, D. Hönicke, in: Proceedings of the XXXV, vol. 20, Jahrestreffen Deutscher Katalytiker, Weimar, Germany, (2002) 49–51. [7] M. Orschel, J. Klein, H.-W. Schmidt, W.F. Maier, Angew. Chem. Int. Ed. 38 (1999) 2791–2794. [8] C. Hoffmann, A. Wolf, F. Schüth, Angew. Chem. Int. Ed. 38 (1999) 2800–2803. [9] U. Rodemerck, P. Ignaszewski, M. Lucas, P. Claus, Chem. Ing. Tech. 71 (8) (1999) 873–877. [10] P. Cong, A. Dehestani, R. Doolen, D.M. Giaquinta, S. Guan, V. Markov, D. Poojary, K. Self, H. Turner, W.H. Weinberg, Proc. Natl. Acad. Sci. 96 (1999) 11077–11080. [11] S. Senkan, K. Krantz, S. Ozturk, V. Zengin, I. Onal, Angew. Chem. Int. Ed. 38 (1999) 2794–2799. [12] P. Claus, D. Hönicke, T. Zech, Cat. Today 67 (2001) 319– 339. [13] Xu Xiaoding, H. Vonk, A. Cybulski, J.A. Moulijn, in: G. Pocelet, J. Martens, B. Delmon, P.A. Jacobs, P. Grange (Eds.), Stud. Surf. Sci. Catal., Preparation of Catalysts VI 1995, pp. 1069–1078. [14] M. Che, O. Clause, Ch. Marcilly, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis, vol. 1, VCH, 1997, pp. 191–207. [15] Th. Vergunst, F. Kapteijn, J.A. Moulijn, Appl. Catal. A. 213 (2001) 179–187. [16] M. Lucas, P. Claus, in: Proceedings of the 18th NACS, Cancun, Mexico, June 1–6, 2003, submitted for publication.
High throughput screening in monolith reactors for total oxidation reactions Martin Lucas, Peter Claus∗ Institute of Chemical Technology, Darmstadt University of Technology, Petersenstreet 20, D-64287 Darmstadt, Germany Received 31 October 2002; accepted 4 February 2003
Abstract We present here a powerful tool for high throughput screening of heterogeneous catalysts, successfully developed in our laboratory with the application focus on common monoliths as a special type of microreactors. It could be shown that it is possible to prepare in a reproducible manner catalytically active coatings on the wall of single channels of the monoliths by a channel-by-channel procedure enabling the application of these multichannel reactors, coupled with mass spectrometry or gas chromatography, for the fast screening of heterogeneous catalysts. Because time and spatially resolved sampling is needed and complex gaseous mixtures have to be analyzed, a special 3D positioning system, which allows measurements at high temperatures, is needed as the central element of the equipment. The high efficiency and reliability of the channel-by-channel preparation as well as the developed screening method is demonstrated for the total oxidation of hydrocarbons and carbon monoxide in the presence of further components like O2 , H2 O, CO2 , NO, SO2 and inert gas over precious metal catalysts. © 2003 Elsevier Science B.V. All rights reserved. Keywords: High throughput screening; Monolith reactor; Channel-by-channel preparation; Coating; Oxidation; Automobile-exhaust gas catalysts
1. Introduction With the development of high throughput methods for new catalysts a stormy progress could be observed in the last years. One of the main points of criticism at the systems developed for the heterogeneous catalysts is the unsatisfactory comparability of the results with those obtained in the technical reactor. The comparability is mainly limited by the necessary miniaturization of the screening devices. In the field of the combinatorial catalysis previous experiences with the ∗ Corresponding author. Tel.: +49-6151-16-5369; fax: + 49-6151-16-4788. E-mail address: [email protected] (P. Claus). URL: http://www.ct.chemie.tu-darmstadt.de/ak claus/index de.html.
analysis of heterogeneously catalyzed reactions can be split into two solutions. The first variant comprises the use of new analytical methods which are able to determine the selectivity to target products or, at least, the activity of several catalytic systems in a parallel way. For example, by emission corrected infrared (IR) thermography, Maier and co-workers [1] used IR cameras with high temperature resolution to spatially resolve reaction heats of the hydrogenation of 1-hexyne, the oxidation of hydrocarbons (isooctane, toluene) over catalysts of a library consisting of 37 combinations of transition metals on amorphous microporous mixed oxides. Disadvantage of this variant is that only limited information about the reaction examined can be achieved because, clearly, the identification of reaction products and, thus, the estimation of selectivities
0926-860X/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0926-860X(03)00261-8
36
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
is not possible. However, this technique is especially useful to detect the catalyst activity in a very sensitive and effective way during primary screening in combinatorial catalyst development. Among the optical screening methods, resonanceenhanced multiphoton ionization (REMPI) can, in principle, provide information about the reaction products, if the absorption and ionization features of the latter are known [2]. Senkan and Ozturk combined microreactors with REMPI as an optical screening tool to detect the catalytic properties of 72 Pt-Pd catalysts [2] and a ternary Pt-Pd-In/Al2 O3 library (66 catalysts [3]) for the simple dehydrogenation of cyclohexane to benzene. Up to date, the use of REMPI for screening reactions with more complicated product mixtures is not known. In the second variant, to overcome the drawbacks of the optical methods, high-throughput devices were developed which allows feeding of the gaseous educt mixture through the individual catalysts and the reaction products are analyzed by standard analyzing methods like gas chromatography and mass spectrometry [4–12] achieving quantitative informations about catalyst activity and even selectivity in an easy and economic way. Two basic concepts for the application of scanning mass spectrometers have been published: scanning a catalyst library wafer, which consists only of sites of thin film patches, with two concentrical capillaries in a transient mode [4] and scanning a continuously operated reactor array by spatially resolved analyses [5–12]. Catalyst screening techniques based on mass spectrometry were applied to several reactions like the partial oxidation of propene over a sol–gel derived catalyst library with 33 mixed oxides [7], the oxidation of CO over 120 ternary thin film Pt-Pd-Rh and Pd-Rh-Cu catalysts [4] and over supported gold catalysts [8,9], oxidative dehydrogenation of ethane over 66 ternary combinations of Mo, V and Nb oxides [10], and dehydrogenation of cyclohexane over 66 combinations of alumina supported Pt, Pd and In [11]. Due to the high number of catalysts which have to be examined it is fact for both variants that the structures into which these catalysts have to be deposited must be miniaturized in an appropriate manner. The results achieved by this way can only partially be compared with those achieved from technical reactors. We present here a powerful tool for high throughput screening of heterogeneous catalysts, successfully
developed in our laboratory with the application focus on common monoliths as a special type of microreactors where the channels are used as single plug-flow reactors containing different catalysts. Preconditions were to set a catalytically active layer on the channel walls of ceramic monoliths and to analyze the catalytic properties of all catalysts to be compared so that the comparability of all results, even for deactivating catalysts, could be guaranteed. That means that inside the monolith structure different catalytic assemblies shall be synthesized channel-by-channel and the catalytical activity shall be determined at the same values of time-on-stream. We wanted to achieve a good reproducibility and a high degree of automatism for all steps, beginning with the preparation, going through the measurement to the analysis, to prevent the user from being too much involved in extremely time-consuming serial measurements of the catalytic properties. The method was tested for the total oxidation of hydrocarbon and CO in the presence of further components like O2 , H2 O, CO2 , NO, SO2 and inert gas. 2. Experimental 2.1. Catalyst preparation in monolith structures Monolith structures have been used for a long time already as part of catalysts, e.g. as support of automobile-exhaust catalysts. The catalytic active layer is deposited onto the walls of the monolith by the wash-coat-procedure. The commonly used monolith materials are permeable for gases and fluids. For the use of each monolith channel as a single plug-flow reactor these characteristics of the common monolith material are rather restricting as diffusion, from channel-to-channel, through the channel walls will falsify the examinations or even will make them impossible. Therefore we chose a monolith material (Corderit 410, Inocermic GmbH Hermsdorf) which is impermeable for gases and fluids. The disadvantage is the bad wettability of the material with water as this can lead to problems during the coating with the support material or catalyst precursors. The monolith structure used consists of 10 × 20 channels with a diameter of 2.6 mm (72 cpsi). The channel length is 75 mm. A number of 128 of the existing 200 channels
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
can be used for the catalytical tests which are arranged in an array of 8 rows and 16 columns. By a wash-coat procedure [13] the monolith is coated uniformly with the catalyst support material (Al2 O3 , SiO2 , TiO2 etc.). The thickness of oxide layer can be adjusted variably. Afterwards the coated monolith is saturated channel-by-channel individually with aqueous solutions of the metal precursors (most of all we used the nitrates, in case of Ru and Au the chlorides) after it is sealed up at one side. For example in case of the Al2 O3 support material, with an absorption capacity of water of 500 ml/kg, 260 l of a precursor solution with a content of 2 wt.% metal will be filled in a channel to prepare a catalyst with 1 wt.% metal. After a short time (2–10 s) which is necessary for saturating the oxide layer completely, the dilutions are removed. The supply of these 128 individual solutions, the filling of the channels and the removing of the solutions after a defined time is effected by a robot (Tecan Miniprep 60). If the pore volume respectively the capacity of water absorption of the oxide layer is already known the desired content of a metal can be adjusted with high accuracy (Table 1); this is analog to the dry impregnation or incipient-wetness method [14]. Thus, the prepared monolith is now ready for the common pretreatment methods like calcinations and reduction in flowing air and hydrogen (both 3 h at 400 ◦ C, flow = 50 l/h), respectively. Beside this impregnation procedure, the immobilization of metals
Table 1 Comparison between the desired contents of Pt and Zr (theoretical metal content) and the experimental data obtained by ICP-OES in every eight channels Experimental (wt.%) (ICP-OES)
Pt column 11
Pt column 12
Zr column 12
Channel 1 Channel 2 Channel 3 Channel 4 Channel 5 Channel 6 Channel 7 Channel 8 Average value (wt.%) S.D. (wt.%) Theoretical (wt.%)
1.08 1.06 0.96 1.22 1.19 1.23 1.08 1.04 1.108 0.096 1.0
0.53 0.45 0.74 0.65 0.65 0.64 0.47 0.63 0.595 0.101 0.5
0.64 0.57 0.56 0.65 0.61 0.65 0.64 0.61 0.616 0.035 0.5
Support material Al2 O3 with an absorption capacity of water of 500 ml/kg.
37
could be also achieved by deposition–precipitation using urea. In this case, the monolith was again sealed up at one side, filled channel-by-channel with aqueous solutions of the metal precursors, followed by mixing with urea and finally heating to 80 ◦ C. At this point, decomposition of urea starts yielding hydroxid ions which give the corresponding hydroxides. Subsequently, again calcination and reduction steps produce a library of supported oxide or metal catalysts. 2.2. Development of screening device and analysis The experimental setup for the HT-screening is shown in Fig. 1. The inlet capillary of a quadrupole mass spectrometer (QMS, Quadstar Pfeiffer Vacuum) is moved in x- and y-direction of the monolith and then into the particular channel (z-direction) by a three-dimensional (3D) positioning system (Amtec). The latter contains the monolith to be tested and is the central element of the equipment. The position of the capillary above a monolith channel can be automatically changed by using the three stepping motors of the positioning system supervised by two CCD cameras and an inspection window and is exactly controlled by the software. The inspection window restricts the available temperature range to 250 ◦ C and the pressure range to normal pressure during the test runs. The gaseous educts are taken from gas mixtures stored in gas bottles by means of mass-flow-controller (FiC01 and 02, Hitec) and conducted to a evaporator. Together with a GC-sampling valve this evaporator is situated in a thermostat housing. A defined amount of water is dosed into the evaporator by a liquid-flow-controller (Hitec), then vaporized there and conducted to the 3D positioning system after being mixed with the reaction gases. The interior of the 3D positioning system is rinsed with argon. Inside the 3D positioning system the monolith is situated in a stainless steel housing which is heated to the reaction temperature. With respect to the approaches described so far in the literature [14] the difference of this equipment is the type of feeding of the gaseous educt mixture via a 1/16 in. tube positionable in the x–y–z direction (Fig. 2). This tube is brought to a single channel of the monolith while a teflon disk, situated at the end of the tube, is sealing this channel towards its surroundings. Consequently the gaseous educt mixture only flows through this single channel
38
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
Fig. 1. Principle of the developed apparatus for high throughput screening of heterogeneous catalysts in monolith reactors.
of the monolith. Inside this dosing tube there is 1/32 in.-sample-drawing capillary which juts 48 mm out over the end. The sample drawing is made by sucking a defined gas amount which is controlled by a mass-flow-controller (FIC04) out of the single channel feeded with gaseous educt mixture using a vacuum pump. According to the length of the sample-drawing capillary, the sample drawing is made at a distance of 48 mm from the channel entry. On its way to the vacuum pump the gas probe passes the GC-sampling valve mentioned above which injects the contents of the sample loop (PS in Fig. 1) computer-controlled into the GC. Behind the GC-sampling valve and following the flow direction some part of the gas probe
taken from the monolith channel is supplied to a mass spectrometer. Alternatively a three-way switching valve (H3) can be switched in a way that the gaseous educt mixture can be analyzed. Prompt and exact analyses are achieved by using the mass spectrometer unless there are overlaying spectra (MZ 28) as it happened with CO and CO2 which cannot be separated numerically if there is a big CO2 excess. In that case we used a gas chromatograph (HP 5890A) with 30 m capillary column (Poraplot Q). Before the flame ionization detector used for educt/product detection a selfmade methanizer (5 wt.% ruthenium on Al2 O3 ) was switched. At 450 ◦ C this methanizer converts, via hydrogen excess, carbonaceous substances eluting
Fig. 2. Principle sketch of gas dosage and sampling in the monolith. The gaseous educt mixture (reaction gas) is proportioned channel-by-channel into the monolith. At the same time, a gas probe is taken in 48 mm of depth.
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
from the column completely into methan. Thus it is possible to determine the contents of CO, CO2 , propene and propane by one GC-column within the analysis cycle of 4.5 min. By using a micro GC (Agilent “G2890A” Micro GC with 10 m Molsieve 5A and 8 m Poraplot U column) the time for an analysis cycle can be reduced to 1.5 min, if necessary. The screening of the monolith takes place after the following algorithm: the sampling and dosing (SAD) equipment goes into the first channel to be examined and stays there for 4.5 min. During this time mass spectra are constantly recorded to determine the temporary course of the components’ contents (without CO). After these 4.5 min there is a sample drawing by the GC and thereafter the change to another channel where the SAD equipment stays for another 4.5 min respectively until the GC-analysis of the previous channel has been finished and the GC is able to analyze another gas probe. During this time again mass spectra are constantly recorded. This algorithm takes place for all channels to be examined. Altogether the determination of the catalyst activity of all 128 monolith channels takes ∼10 h, whereas the time-on-stream is the same for each channel and takes only 4.5 min at the present example. For the rest of the time, each channel is flooded by argon. If the quantity of the taken gas probe is adapted to the quantity of gas given to each channel, the gas will
39
flow exclusively through the channel chosen. Additionally it can be checked by this method whether the admitted gas is completely dosed into the channel or whether it partly passes the teflon gasket and ends in the 3D positioning system. If that should be the case a considerable amount of argon which is flowing the neighbor channels is to be found in the gas probe. Fig. 3 shows the signal of the mass spectrometer for mass numbers 4 and 40. Every 4.5 min the channel for the gas probe was changed. In the figure this phenomena is to be seen as a vertical line at the argon signal as the sample-taking capillary goes, when changing the channels, outside of the monolith through the 3D positioning system filled with argon. It is clearly to be seen that argon is only marginally detected in the gas probes taken from the channels and thus the teflon gasket is sealing the channels reliably. When using monolith structures for HT-screening the temperature gradient within the monolith is a big problem for the comparison of results which were won with possibly immensely varying temperatures respectively a bad temperature distribution within the channels. This temperature gradient results from the minimal thermal conductivity of the ceramic material when heating the monolith from outside and from the high space velocity necessary for many applications the monolith structure was made for. Should a test run of a monolith take place in a conventional method
Fig. 3. Exemplary results of channel-by-channel feeding of 200 ml/min helium into the monolith followed by taking a gas probe (180 ml/min) and analysis by MS (Tmonolith = 150 ◦ C). Dotted vertical lines mark the move to the next monolith channel.
40
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
where the complete monolith is feeded and the sum of the conversion of all channels is determined as in the automobile-exhaust gas catalyst, a comparability regarding the average temperature distribution from monolith to monolith is usually given. In order to get the desired steady temperature distribution in all channels of the monolith which is necessary for the application to be done, preheated argon is led through all channels in reverse flow (except the channel feeded by gaseous educt mixture). The heating device which is coated around the monolith is not used for heating the monolith to reaction temperature; it is only necessary to balance the heat loss via the exterior wall. The temperature distribution gained by this method within the monolith was measured by inserting a thermal element into different channels of the monolith depending on the depth and is shown in Fig. 4. The monolith is divided into a channel pattern of 16 columns in height and 8 rows in width. Fig. 4 shows the temperature distribution across the monolith in different depths. It is obvious the outer channels show higher temperatures than the inner ones, however, the deviation is with ±1 K in radial direction quite tolerant. Larger deviations of ±4 K were measured in axial direction, but we have to state that all channels examined show a very similar course of the temperature profile in axial direction and thus the effects to the comparability of the catalytic activities should be slight. Moreover, the reaction heat which has a big influence on the medium temperature of the monolith during total oxidation of hydrocarbons examined can be transported off easily by the reaction stream in the
Fig. 4. Temperature profiles in the monolith measured in different depths and channels along the middle row of the monolith.
Table 2 Influence of the capillary material, used for feeding the gaseous educt mixture, on the activity of a monolith coated only with Al2 O3 (blind reaction): conversion of CO and propene when using stainless steel capillary or a capillary passivated with SiO2 T (◦ C)
150 200 225 250
Stainless steel capillary ¯ CO (%) X
¯ propene (%) X
12.05 34.69 97.11
8.73 11.69 24.92
Capillary passivated by SiO2 ¯ CO (%) X ¯ propene (%) X 0.79 1.47 3.21
0.20 0.23 0.40
single channel which is surrounded directly by eight channels flooded with argon. Table 2 shows conversion data of the blind reaction measured in a monolith coated only with Al2 O3 which were relatively high when using the stainless steel capillary for feeding the gaseous educt mixture and taking gas probes during the reaction. By installing capillary tubes passivated by SiO2 this amount of the blind reaction could nearly be eliminated completely.
3. Results and discussion Nevertheless, the evaluation of the HTS procedure cannot be made via the measurement of the single gradients respectively the irregularities of the parameters temperature, gas dosing, analysis incl sample taking and preparation. The very fact is the comparison of the catalytic properties of the channels among one another. For this purpose we produced catalytically active wall coatings of identical composition for every eight channels of the monolith. Table 1 shows two examples for such preparations. In column 11 a layer with 1 wt.% Pt/Al2 O3 should be produced, in column 12 it should be a catalyst with 0.5 wt.% Pt and Zr. As to be seen from the quantitative analysis via ICP-OES the metal contents can be adjusted relatively exactly and reproducible. Statistics show a content which is 0.1 wt.% higher than the content desired; but this can be corrected during the preparation by a metal precursor solution being accordingly lower. The Pt-Zr catalyst was not produced by simultaneous saturating with a mixture of Pt and Zr precursor solution but by impregnation of the oxide layer with Pt first, drying
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
41
Fig. 5. Conversion of CO measured in two different columns of the monolith. The eight channels of a column should have the same catalytic composition (white chart: 0.5 wt.% Pt and 0.5 wt.% Zr, supported on Al2 O3 ; grey chart: 1 wt.% Pt/Al2 O3 ).
at 300 ◦ C for 3 h and followed by the deposition of Zr. After the second preparation step, the monolith is calcinated at 400 ◦ C in air (flow = 50 l/h). A variable mixture of CO, propene, O2 , H2 O, CO2 , NO, SO2 , and inert gas with a space velocity of 50,000 h−1 was given into the channel for the measurements (200 ml/min total volumetric rate). The total oxidation of CO and propen was examined at normal pressure in a temperature range of 100–250 ◦ C. Fig. 5 shows the conversion of CO at 150 ◦ C for each of those channels in which the above mentioned catalytically active layers were prepared. For the monometallic Pt catalyst the average CO conversion was determined with 38.2% at a S.D. of 3.3%. For the bimetallic Pt-Zr catalyst we have a much larger S.D. of 12% at an average CO conversion of 23.9%. Clearly to be seen in Fig. 5 is that the conversion of the Pt-Zr catalyst is extremely decreasing in the middle of the monolith whereas the column with the monometallic Pt catalyst, which is directly alongside, does not show such a geometric gradient of the conversion. Parameters of the screening like temperature gradients can thus be eliminated as reason for this phenomenon which up to now has only been seen at catalysts containing Zr and Mn or with converse effect at some catalyst combinations containing Rh. As a possible explanation we assume a temperature gradient during the catalyst pretreatment which should be done in an oven free of temperature gradients as recently shown by Moulijn
and co-workers [15]. It is very difficult to confirm this assumption by determining the catalyst composition in dependence of their axial position in these small channels and has not yet been executed (each channel contains approximately 20–30 mg of catalyst). The next step was to prepare 32 different catalysts within the monolith in such a way that always four channels contain identical catalyst compositions. In this case, we prepared 1 wt.% Pt, Rh, Ru, Pd, Au, Zr, Mn and Ag onto Al2 O3 layers as well as selected bimetallic combinations of the mentioned metals with a content of 0.5 wt.% of each component. The CO conversion obtained at 150 ◦ C are shown in Fig. 6. For each catalyst we analyzed two successive GC-probes. Thus an estimate of the deactivating respectively activating of the catalyst of 4.5 min time-on-stream (first run) compared to 9 min time-on-stream (second run) could be gained. Note, that the mass spectrometer produces concentration versus time curves with high resolution with respect to the content of the individually dosed components (except CO). Part of the catalyst screening of the above mentioned monolith with 32 different catalysts is to be seen in Fig. 7. The vertical lines in the curves mark again the change of each channel after nine minutes. Every four channels had, as mentioned above, an identical catalyst composition and were synthesized directly one after the other. These identical compositions agree quite well with the time processes especially of the components propene, SO2
42
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
Fig. 6. Results of quantitative MS analyses of gas probes from different channels of the monolith with Al2 O3 supported catalysts at T = 150 ◦ C. Dotted vertical lines mark the move to the next monolith channel. Note, that every four channels following one after another in sequence include the same catalyst composition, i.e. the first four channels include catalysts with 0.5% Pt-0.5% Au, the next four channels 0.5% Pt-0.5% Mn, etc. Every catalyst composition gave characteristic pattern in product mass spectroscopy.
Fig. 7. CO conversion at 150 ◦ C for a catalyst library containing 32 different materials within the monolith (see text). Values are averages for the four channels with desired identical catalyst composition. The S.D. is related to the conversion rate of the second run.
M. Lucas, P. Claus / Applied Catalysis A: General 254 (2003) 35–43
and H2 in Fig. 6 as nearly identical patterns of these components were analyzed in the mass spectrometer at every four successive channels. 4. Conclusions It could be shown that it is possible to prepare in a reproducible manner catalytically active coatings in the single channels of monoliths by a channelby-channel procedure enabling the application of these structures for high throughput screening of heterogeneous catalysts. Moreover an equipment was developed which is able to reproducibly feed any gaseous educt mixture into single channels of a monolith and to take representative probes of the reaction products. It is of great advantage that the method presented here enables a HT-screening with the common analytical equipment (GC or MS) of a research lab in catalysis. In addition to the application of the presented tool for the oxidation of CO and hydrocarbons shown above this equipment was successfully used for the selective oxidation from propene to propene oxide. During these tests one-hundredth of the usual gas volume of the total oxidation was dosed into the single channels. In spite of the low volumetric feed of 2 ml/min reproducible results could be achieved [16]. References [1] A. Holzwarth, H.-W. Schmidt, W.F. Maier, Angew. Chem. Int. Ed. 37 (1998) 2644–2647.
43
[2] S.M. Senkan, Nature 394 (1998) 350–353. [3] S.M. Senkan, S. Ozturk, Angew. Chem. Int. Ed. 38 (1999) 791–795. [4] P. Cong, R.D. Doolen, Q. Fan, D.M. Giaquinta, S. Guan, E.W. McFarland, D.M. Poojary, K. Self, H.W. Turner, W.H. Weinberg, Angew. Chem. Int. Ed. 38 (1999) 483–488. [5] T. Zech, A. Lohf, K. Golbig, T. Richter, D. Hönicke, in: Proceedings of the Third International Conference on Microreaction Technology, Frankfurt, Main, Germany, April 1999. [6] T. Zech, J. Klein, S. A. Schunk, D. Demuth, D. Hönicke, in: Proceedings of the XXXV, vol. 20, Jahrestreffen Deutscher Katalytiker, Weimar, Germany, (2002) 49–51. [7] M. Orschel, J. Klein, H.-W. Schmidt, W.F. Maier, Angew. Chem. Int. Ed. 38 (1999) 2791–2794. [8] C. Hoffmann, A. Wolf, F. Schüth, Angew. Chem. Int. Ed. 38 (1999) 2800–2803. [9] U. Rodemerck, P. Ignaszewski, M. Lucas, P. Claus, Chem. Ing. Tech. 71 (8) (1999) 873–877. [10] P. Cong, A. Dehestani, R. Doolen, D.M. Giaquinta, S. Guan, V. Markov, D. Poojary, K. Self, H. Turner, W.H. Weinberg, Proc. Natl. Acad. Sci. 96 (1999) 11077–11080. [11] S. Senkan, K. Krantz, S. Ozturk, V. Zengin, I. Onal, Angew. Chem. Int. Ed. 38 (1999) 2794–2799. [12] P. Claus, D. Hönicke, T. Zech, Cat. Today 67 (2001) 319– 339. [13] Xu Xiaoding, H. Vonk, A. Cybulski, J.A. Moulijn, in: G. Pocelet, J. Martens, B. Delmon, P.A. Jacobs, P. Grange (Eds.), Stud. Surf. Sci. Catal., Preparation of Catalysts VI 1995, pp. 1069–1078. [14] M. Che, O. Clause, Ch. Marcilly, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis, vol. 1, VCH, 1997, pp. 191–207. [15] Th. Vergunst, F. Kapteijn, J.A. Moulijn, Appl. Catal. A. 213 (2001) 179–187. [16] M. Lucas, P. Claus, in: Proceedings of the 18th NACS, Cancun, Mexico, June 1–6, 2003, submitted for publication.