Combinatorial Screening of Heterogeneous Catalysts in Selective

Aug 11, 2001 - Citation data is made available by participants in Crossref's Cited-by Linking service. For a more .... Combinatorial Methods in Cataly...
24 downloads 8 Views 316KB Size
Download PDF
Anal. Chem. 2001, 73, 4434-4440

Combinatorial Screening of Heterogeneous Catalysts in Selective Oxidation of Naphthalene by Laser-Induced Fluorescence Imaging Hui Su, Yongjin Hou, Robert S. Houk, Glenn L. Schrader, and Edward S. Yeung*

Ames LaboratorysUSDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011

Heterogeneous catalysis is one of the most important processes in the petroleum and the chemical industries. To be able to screen catalysts at high throughput will dramatically improve performance and reduce costs. Here we used laser-induced fluorescence imaging as a highthroughput screening technique in the combinatorial discovery of active catalysts for naphthalene oxidation. Binary catalysts of V-Mo-O, V-Sn-O, V-Ti-O, and V-W-O in various 15-member libraries were screened. Laser ablation ICPMS was employed to confirm the composition of the individual catalysts in the combinatorial library. The addition of MoO3, WO3, SnO2, and TiO2 to V2O5 did not improve the catalytic activity in the conversion of naphthalene to naphthoquinone, but the overall activity was found to increase for certain binary samples. The screening of ternary catalysts of V-SnMo-O revealed that the combination of V (45%)-Sn (45%)-Mo (10%) gave 70% higher catalytic activity than pure V2O5 in converting naphthalene to naphthoquinone. Reaction temperature and sample preparation effects on the activity and selectivity of catalysts are also studied in a combinatorial manner. Combinatorial chemistry and high-throughput screening (HTS) have fundamentally changed the concept and the process of discovery. In the development of heterogeneous catalysts, various promising techniques based on thin-film deposition and liquid dosing1 have been developed to synthesize a large number of catalytic materials in a high-throughput fashion. The HTS of heterogeneous catalysts has been more challenging compared to their synthesis. Recently IR thermography,2-4 laser-induced resonance-enhanced multiphoton ionization,5 microprobe sampling mass spectrometry,6,7 and fluorescence indicators 8,9 have been (1) Sun, X. D.; Wang, K. A.; Yoo, Y.; Wallace-Freedman, W. G.; Gao, C.; Schultz, P. G. Adv. Mater. 1997, 9, 1046-1049. (2) Moates, F. C.; Somani, M.; Annamalai, J.; Richardson, J. T.; Luss, D.; Wilson, R. C. Ind. Eng. Chem. Res. 1996, 35, 4801-4803. (3) Taylor, S. J.; Morken, J. P. Science 1998, 280, 267-270. (4) Wilson, R. C. PCT Int. Appl. 1997. CODEN: PIXXD2. WO9732208 A1 19970904. (5) Senkan, S. M. Nature 1998, 394, 350-353. (6) Cong, P.; Doolen, R. D.; Fan, Q.; Giaquinta, D. M.; Guan, S.; McFarland, E. W.; Poojary, D. M.; Self, K.; Turner, H. W.; Weinberg, W. H. Angew. Chem., Int. Ed. 1999, 38, 484-488. (7) Cong, P.; Dehestani, A.; Giaquinta, D.; Guan, S.; Markov, D.; Self, K.; Turner, H.; Weinberg H. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 11077-11080.

4434 Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

reported for screening catalysts. IR thermography is capable of monitoring the overall exothermicity of hundreds to thousands of catalysts either in gas-solid or in liquid-solid reactions by measuring the temperature change of the catalysts. Since IR thermography is not chemically specific, in certain circumstances, the information provided by it might be misinterpreted. Also, the selectivity of the reaction for competing pathways cannot be determined. Techniques based on multiphoton ionization and fluorescence indicators have limited applications because of their restricted operational requirements. So far, microprobe sampling MS is the most successful screening method for heterogeneous catalysts. However, it is basically a robust scanning technique and not a parallel screening methodology. Recently, we have demonstrated laser-induced fluorescence imaging (LIFI) as an alternative for HTS of heterogeneous catalysts.10 LIFI has good detection power and the high spatial and temporal resolution needed for in situ monitoring of heterogeneous catalysis. The method relies on the creation or destruction of chemical bonds, which is the case in most heterogeneous catalytic reactions, to alter the fluorescence properties of molecules. The combination of selective fluorescence detection and two-dimensional imaging allows screening of up to 250 × 250 subunits/cm2 (40-µm diameter/subunit) simultaneously.10 Furthermore, since the CCD camera is also sensitive to radiation in the near-infrared region, screening based on IR thermography can be performed by monitoring the blackbody radiation of the catalysts in the same system with the appropriate combination of optical filters. Unification of LIFI and IR thermography in a single instrument enables us to access unique information about the catalytic system in a simple and efficient way. Selective oxidation is a very important industrial process that is used to manufacture a variety of chemicals. Vanadium pentoxide catalysts are widely used as the selective catalysts for the oxidation of aromatic hydrocarbons. Various promoters are frequently added to vanadium pentoxide to improve the selectivity and activity to produce the desired products. The surface-active sites of V2O5 in the presence of WO3,11 MoO3,12 and SnO213 have been investigated. (8) Reddington, E.; Sapienza, A.; Gurau, B.; Viswanathan, R.; Sarangapani, S.; Smotkin, E. S.; Mallouk, T. E. Science 1998, 280, 1735-1737. (9) Miller, S. J.; Copeland, G. T. J. Am. Chem. Soc. 1999, 121, 4306-4307. (10) Su, H.; Yeung, E. S. J. Am. Chem. Soc. 2000, 122, 7422-7423. (11) Satsuma, A.; Hattori, A.; Mizutani, K.; Furuta, A.; Miyamoto, A.; Hattori, T.; Marukami, Y. J. Phys. Chem. 1988, 92, 6052-6058. (12) Satsuma, A.; Hattori, A.; Mizutani, K.; Furuta, A.; Miyamoto, A.; Hattori, T.; Marukami, Y. J. Phys. Chem. 1989, 93, 1484-1490. 10.1021/ac015513i CCC: $20.00

© 2001 American Chemical Society Published on Web 08/11/2001

Figure 2. Flow cell reactor with a 3 × 5 library sitting in the middle. The columns are labeled A-C starting from the left and the rows are labeled 1-5 starting from the top.

Figure 1. Experimental setup for screening heterogeneous catalysts by imaging. L1, L2, and L3 are cylindrical lenses.

The surface concentration of redox sites was found to increase significantly in the presence of these oxides. The activity and selectivity of the enhanced vanadium pentoxide catalysts in the oxidation of benzene to maleic anhydride (MA) have been evaluated by gas chromatography analysis.14 The addition of less electronegative elements, MoO3 and WO3, were found to increase the selectivity to MA by lowering the activity of redox sites in the selective oxidation of benzene. On the other hand, the addition of a more electronegative element, SnO2, resulted in decreased selectivity for MA. In this work, we use LIFI and IR thermography to screen the activity and selectivity of binary vanadium-based catalysts of V-Mo(VI)-O, V-W(VI)-O, V-Sn(IV)-O, and V-Ti(IV)-O for selective catalytic oxidation of naphthalene to naphthoquinone, which is the competing product of phthalic anhydride (PhA). We also prepared and screened ternary libraries of V-Sn-Mo and V-Sn-W. Composition of the components and sample preparation methods are two variables in constructing the libraries. Elemental analysis of the individual catalysts in the library was accomplished by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) with minimal damage to the library itself. The performance of the catalysts was also studied at different reaction temperatures. EXPERIMENTAL SECTION Laser-Induced Fluorescence Imaging System. A schematic diagram of the laser-induced fluorescence imaging system is shown in Figure 1. An Ar+ laser (Coherent Laser Group, Santa (13) Okada, F.; Satsuma, A.; Furuta, A.; Miyamoto, A.; Hattori, T.; Marukami, Y. J. Phys. Chem. 1990, 94, 5900-5908. (14) Satsuma, A.; Okada, F.; Hattori, A.; Miyamoto, A.; Hattori, T.; Marukami, Y. Appl. Catal. 1991, 72, 295-310.

Clara, CA) is used to irradiate the region right above the catalyst in a reactor. The fluorescence intensity of a selected product or reactant can be imaged by a frame transfer CCD camera (Roper Scientific, Trenton, NJ). The laser beam is focused into a horizontal sheet by a cylindrical lens combination. In the present work, the 488-nm wavelength is chosen to excite naphthoquinone in the oxidation reaction of naphthalene catalyzed by vanadium-based catalysts. A 50-mm lens (Nikon) collects the fluorescence emission of naphthoquinone from 515 to 545 nm. All images are stored and processed in a desktop computer by Winview software (Roper Scientific). This LIFI detection system can be switched to an infrared imaging detection mode simply by removing a hot mirror that blocks the infrared emission (>700 nm) in the LIFI mode. Laser excitation is not needed in IR thermal detection. Laser Sheet Excitation. To homogeneously excite the product or reactant immediately above the catalyst, a laser sheet instead of a laser beam is employed (Figure 1). The laser beam was converted into a laser sheet by a 2-m cylindrical lens, which has a 12-mm depth of focus (X) and 78-µm beam waist at the focal plane (Y). The Gaussian intensity distribution remains in the laser sheet, which contributes to a monotonic variation in laser power distribution in the Z direction. Data Calibration. Two calibrations are needed in LIFI detection. One is the laser scattering background, which can be corrected by subtraction of a stored image recorded before reaction from every fluorescence image taken during the reaction. Another is to account for the Gaussian distribution of energy in the laser sheet. The intensity of Rayleigh scattering is proportional to the laser energy, which provides a perfect calibration of the laser energy available in fluorescence excitation at each location. Calibration can be accomplished by dividing the backgroundsubtracted fluorescence image by the scattering background image. The intensity of the corrected image is then only proportional to the activity of the catalyst. The fluorescence and the thermal data are measured on separate intensity scales and do not immediately correlate with Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

4435

each other. Ideally, gas sampling at each spot will be used to relate the two scales. However, a gas sampling jet will interfere with the present optical geometry and will be difficult to operate at high throughput. For the data presented here, only the relative activity changes among a group of catalysts can be determined. In all plots shown here, each data point is the average of at least three independent runs. Different temperatures are tested in a random fashion from high to low or from low to high. The estimated uncertainty in the relative intensities within a given image is (2%. We have also shown previously10 that gas flow in the cell only broadens the images by 20 µm in the flow direction and that there is no depletion in the reactants that would lead to a change of intensity over the time scale of the experiment. Emissivity of the samples is assumed to be constant for the IR measurements. This is expected to be the case since the powder mesh as well as the colors of the samples in a series are nearly identical. Flow Reactor System. The reactor is a cylindrical stainless steel vessel with an inner diameter of 4 cm and a depth of 3 cm (Figure 2). The cylindrical sample holder with an outer diameter of 1.3 cm and a height of 2 cm is placed in the middle of the reactor. A cartridge heater (Omega, Stamford, CT) is attached to the sample holder cylinder. The small thermal mass of each sample means that they should all attain the temperature of the metal mounting plate. A thermocouple (Omega) is attached to the top of the sample holder, and a temperature control unit (Omega) is used to control the reaction temperature. Two optical windows are placed on the wall of the reactor to pass the laser beam immediately above the top of the sample holder. The mixture of gaseous reactants is introduced into the reactor from the entrance port and is pumped out from the exit port located on the other side of the reactor. A stainless steel oven with an additional heating device and temperature control unit (Omega) is used to provide reactant feed with desired concentration. In this work, 7% naphthalene is carried into the reactor by O2 by heating the naphthalene at 135 °C. Catalyst Preparation. The reactants are 99% naphthalene (Sigma, MO); 99.99% V2O5, 99.99% WO3, 99.99% MoO3, 99.9% SnO2, 99.9% (NH4)2MoO4, 99.9% (NH4)2WO4 (Aldrich, WI); and 99.9% cyclohexane (Fisher Scientific, NJ). Binary catalysts of V2O5MoO3, V2O5-WO3, V2O5-SnO2, and V2O5-TiO2 with atomic ratios of 9, 2.33, 1, 0.43, and 0.11 are prepared in two ways. In method A, the solid oxides of the two elements are simply mixed. In method B, the oxalic acid solutions of NH4VO3 with (NH4)2MoO4, (NH4)2WO4, Sn (OH)2, and TiO2, respectively, are mixed, followed by evaporation to dryness overnight and subsequent calcination in a flow of O2 at 773 K for 3 h. Ternary catalysts of V-Sn-W and V-Sn-Mo are prepared by method B. The compositions of these catalysts are a mixture of 5, 10, and 15% Mo or W with 9, 2.33, 1, 0.43, and 0.11 (atomic ratios) of V-Sn. Catalyst Library. A catalyst library is prepared by pipetting 1 µL of vanadium catalyst-cyclohexane slurry solution into small wells (1 mm wide and 0.5 mm deep with 1-mm spacing) on a stainless steel disk with a diameter of 1.3 cm. Eight libraries were tested in this work. Libraries 1-4 are made of binary samples and libraries 5-8 are ternary samples. A total of 15 sample wells in 3 columns (A-C) and 5 rows (1-5) are screened at one time. 4436

Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

Library 1: V (90%)/Sn (10%) (A1, B1), V (70%)/Sn (30%) (A2, B2), V (50%)/Sn (50%) (A3, B3), V (30%)/Sn (70%) (A4, B4), V (10%)/Sn (90%) (A5, B5), V2O5 (100%) (C1-C3), and SnO2 (100%) (C4-C5). Library 2 (V-Ti), library 3, (V-Mo) and library 4 (VW) have the same arrangement as that in library 1. Library 5: (A1) Mo (5%)/V (85.5%)/Sn (9.5%); (A2) Mo (5%)/V (66.5%)/Sn (28.5%); (A3) Mo (5%)/V (47.5%)/Sn (47.5%); (A4) Mo (5%)/V (28.5%)/Sn (66.5%); (A5) Mo (5%)/V (9.5%)/Sn (85.5%); (B1) Mo (10%)/V (81%)/Sn (9%); (B2) Mo (10%)/V (63%)/Sn (27%); (B3) Mo (10%)/V (45%)/Sn (45%); (B4) Mo (10%)/V (27%)/ Sn (63%); (B5) Mo (10%)/V (9%)/Sn (81%); (C1) Mo (15%)/V (76.5%)/Sn (8.5%); (C2) Mo (15%)/V (59.5%)/Sn (25.5%); (C3) Mo (15%)/V (42.5%)/Sn (42.5%); (C4) Mo (15%)/V (25.5%)/Sn (59.5%); (C5) Mo (15%)/V(8.5%)/Sn (76.5%). Library 6: (A1-A2) V2O5; (B1) V (90/Sn (10%); (B2) V (70%)/ Sn (30%); (B3) V (50%)/Sn (50%); (B4) V (30%)/Sn (70%); (B5) V (10%)/Sn (90%); (C1) Mo (10%)/V (81%)/Sn (9%); (C2) Mo (10%)/V (63%)/Sn (27%); (C3) Mo (10%)/V (45%)/Sn (45%); (C4) Mo (10%)/V (27%)/Sn (63%); (C5) Mo (10%)/V (9%)/Sn (81%). Library 7 and library 8 have the same arrangements as library 5 and library 6, but substituting Mo with W. Laser-Ablation ICPMS. The experiments were performed with a Finngann MAT (San Jose, CA) ICPMS and 266-nm laser ablation system (Cetac). The Nd:YAG laser was operated at 10 Hz with a typical pulse energy of 10 mJ. Single-spot sampling was used, and three samples were taken for each catalyst. The ICPMS was operated at medium resolution with rf power of 1200 W. RESULTS AND DISCUSSION Selective Oxidation of Naphthalene. The selective oxidation of naphthalene (NA) can be catalyzed by V2O5 at elevated temperatures. There are two major products in this reaction:

Here we screen the selective catalytic activity (relative) of catalysts by monitoring the fluorescence intensity of NQ produced in the reaction by LIFI and the overall catalytic activity (relative) by monitoring the temperature change of the surface of the catalysts by IR imaging, since both major reactions are exothermic. Combinatorial Screening of Binary Heterogeneous Catalysts. In search of active catalysts, usually one starts with binary samples and looks for possible trends in terms of composition and sample preparation methods. Then one goes to more complex systems: ternary, quaternary, etc. The high-throughput capability of LIFI provides for fast preselection of useful components from a diverse library. Vanadium pentoxide is an effective catalyst for the oxidation of naphthalene (NA) to naphthoquinone (NQ) and phathalic anhydride (PA). The addition of foreign oxides into V2O5 can change its catalytic activity and selectivity by modifying its surfaceactive sites. Sample preparation also has an important effect on the performance of catalysts. Catalytic reactions carried on libraries 1-4, which consist of binary catalysts prepared in two different ways, were monitored in both fluorescence and infrared detection modes.

Figure 3. Catalytic performance of V-Sn-O binary system (library 1): (a) fluorescence screening, (b) near-IR thermography screening, and (c) activity (NQ) and total activity of the catalysts.

In library 1, both fluorescence (Figure 3a) and IR image (Figure 3b) detection show that pure V2O5 is active in this catalytic reaction but pure SnO2 is inactive. The addition of SnO2 decreases the activity of V2O5 for NQ (Figure 3c). Larger depression of activity (NQ) is observed in catalysts prepared by method A than in catalysts prepared by method B. The total activity of A catalysts (Figure 3c) is enhanced by SnO2 (10%, 30%, 50%) but is depressed with further increase of SnO2. On the other hand, the total activity of B catalysts increases with the addition of SnO2 to a maximum of 67% for V2O5 (10%)-SnO2 (90%) (Figure 3c). The only explanation of the decrease in activity for NQ and increase in total activity for one catalyst is that the other product in the reaction, PA, is favored in the presence of certain amounts of SnO2. In ref 14, the reaction rate in the selective oxidation of benzene to produce MA reached a maximum in the presence of 40 or 80% of SnO2 in V2O5 prepared in two different ways, respectively. These compositions are similar to our screening results or 50 and 90%. Although different sample preparation methods are involved in the two experiments, we can conclude that the addition of SnO2 can increase the concentration of surface active sites while the amount of SnO2 affects the selectivity of these surface redox sites. Library 2 consists of 8 binary samples of V2O5-TiO2 with different ratios and prepared in different ways. From fluorescence detection (Figure 4a), infrared detection (Figure 4b), and their comparison (Figure 4c), TiO2 has a negative effect on the total activity and the NQ activity of V2O5 for both A and B catalysts. This can be caused by the dilution of active sites of V or by surface modification of the catalysts by TiO2. For As, similar decreases in activity for NQ and total activity are seen in Figure 4c. Dilution effects must therefore be dominant here because the modification of active sites should change the selectivity of the catalyst. For B

catalysts, the changes in total activity of samples B1-B3 are larger than changes in NQ activity. We can conclude that there is not only a dilution effect from inactive titanium oxide but also surface modification due to Ti. Furthermore, depression of the overall catalytic activity of vanadium pentoxide by the addition of TiO2 suggests that the concentration of surface-active sites cannot be increased by adding TiO2. The V-Mo binary catalyst is a particularly interesting system. Both V2O5 and MoO3 are catalytically active according to Figure 5a and b. As shown in Figure 5c, the addition of MoO3 up to 50% decreases the catalytic activity of vanadium in the selective oxidation of NA to NQ. Yet, further increase of Mo in the B samples results in increase in activity. This result agrees well with reported observations.14 However, the screening result from IR imaging does not support this conclusion (Figure 5b,c). The addition of MoO3 was reported to increase the overall activity as well as the activity of selective oxidation of benzene to MA.14 Here, the addition of Mo(VI) decreased the selective activity of V2O5 in the oxidation of NA to PhA. One explanation for this result is the difference in the type of surface sites involved in the production of MA and PhA. Samples A2 and B2 have the same composition and similar activity for NQ, but A2 shows higher overall activity. This means that the surface-active sites made via method B are less favorable to the formation of PhA. Catalyst B5 has a slightly higher catalytic activity than pure vanadium pentoxide in the selective oxidation of NA to NQ while IR detection shows that it is much less active than pure vanadium. This implies that IR thermography screening is not highly reliable in such systems. Still different catalytic behaviors for V-W catalysts prepared by different methods are seen (data not shown), but none of these combinations gives better performance than pure V2O5, either in Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

4437

Figure 4. Catalytic performance of V-Ti-O binary system (library 2): (a) fluorescence screening, (b) near-IR thermography screening, and (c) activity (NQ) and total activity of the catalysts.

Figure 5. Catalytic performance of V-Mo-O binary system (library 3): (a) fluorescence screening, (b) near-IR thermography screening, and (c) activity (NQ) and total activity of the catalysts.

the overall activity or the activity for selective oxidation of NQ to NA. These results are again in variance with those in ref 14. Reference 14 showed that the overall activity of promoted vanadium catalyst increased due to an increase in the surface concentration of redox sites (VdO). However, the addition of SnO2 was found in ref 14 to decrease the selectivity for MA, because (1) the addition of the ion less electronegative than V5+, Sn4+, enhanced electron delocalization around the VdO band and 4438

Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

therefore increased the activity of the redox site and (2) the activity of the redox site in the promoted catalyst (Sn) is too strong, making it more favorable for complete oxidation to COx rather than selective oxidation of benzene to MA. In our screening results for selective oxidation of naphthalene, we did see an increase in the overall activity of promoted catalysts and decrease in the catalytic activity in the oxidation of NA to NQ, but the catalytic activity toward PhA was enhanced by the

Figure 6. Screening of catalytic activity for V-Sn-Mo-O catalysts: (a) library 5, screening by fluorescence imaging (NQ), (b) library 5, screening by near-IR thermography, (c) library 6, screening by fluorescence imaging (NQ), and (d) library 6, screening by near-IR thermography.

addition of Sn instead of being depressed. This is probably because (1) based on Mar and van Krevelen’s assumption,15 there are two independent steps in this reaction. First, aromatics react with oxygen atoms at the surface of the catalyst. Then, gas-phase oxygen reoxidizes the partially reduced surface; (2) naphthalene is more active than benzene, but its size may hinder the reaction of NA with surface VdO groups. Therefore, NA is less active than benzene in the selective oxidation by vanadium catalysts. This explains why the major products of oxidation are NQ and PhA instead of COx; and (3) in ref 15, PhA is formed from naphthalene as well as naphthoquinone. So, the addition of Sn decreased the production of NQ but increased the production PhA. On the other hand, the addition of ions that are more electronegative than V5+, such as Mo6+, W6+, decreases the electron delocalization of VdO and therefore decreases the activity of VdO. Our screening results for V-Mo-O and V-W-O catalysts confirmed this assumption. Screening of Ternary Heterogeneous Catalysts. On the basis of above experimental results, Mo, Sn, and W seem to be able to modify the active sites of vanadium catalysts in different ways especially when prepared by method B. Thus, we prepared library 5, which consists of V-Sn-Mo-O catalysts in various ratios. The screening results of selective activity by LIFI as well as the overall activity by IR imaging for these catalysts are shown in Figure 6a and b. The combinations of V (63%)-Sn (27%)-Mo (10%) and V (45%)-Sn (45%)-Mo (10%) are the most active catalysts in the selective oxidation of NA to NQ. Comparing the selective activities of these two combinations with pure V2O5 in library 6 (Figure 6c,d), V (45%)-Sn (45%)-Mo (10%) provides a 70% increase compared to the conventional catalyst used in industry, V2O5. However, the overall activity of V (45%)-Sn (45%)(15) Mars, P.; Van Krevelen, D. W. Chem. Eng. Sci. 1954, 8, 41-59 (supplement).

Figure 7. Quantitative plot of temperature dependence of catalytic activity for NQ production.

Mo (10%) is depressed dramatically compared to pure V2O5 (Figure 6d). Libraries 7 and 8 consist of V-Sn-W ternary catalysts prepared by method B. Catalysts with 10% W in library 7 seem to be more active in the conversion of NA to NQ, but not as active as V2O5 by itself (data not shown). The addition of 10% W to V (30%)-Sn (70%) increases its activity for NQ and depresses its total activity. Therefore, we can predict that the conversion to PA is less favorable in the presence of W in V-Sn catalysts. Temperature Dependence of Catalytic Activity. It is important to locate a suitable temperature range for a given heterogeneous catalyst since its activity and selectivity are temperature dependent. When ternary library 5 was screened by LIFI at temperatures of 603, 613, 623, 633, and 643 K, the fluorescence images (data not shown) show that different materials demonstrate different sensitivities to temperature changes. In library 5, B samples with 10% Mo are among the most active catalysts. Therefore, library 6 was further screened at different reaction temperatures together with pure vanadium pentoxide and with catalysts that have the same V-Sn composition but no Mo. The screening results of ternary V-Sn-Mo are compared with vanadium oxide (Mo ) 0%, Sn ) 0%) in Figure 7. We can see that the addition of 10% Mo boosts the activity of V/Sn (7/3), V/Sn (5/5), and V/Sn (3/7), and the degree of enhancement varies as the temperature changes. At lower temperatures (up to 623 K), V (63%)-Sn (27%)-Mo (10%) is more active than V (45%)-Sn (45%)-Mo (10%), but at temperatures above 633 K, V (45%)-Sn (45%)-Mo (10%) is more active than V (63%)-Sn (27%)-Mo. All the catalysts in this library reach their highest activity around 633 K except for V (45%)-Sn (45%)-Mo (10%). Composition Analysis by LA-ICPMS. LIFI can assess the catalytic performance of individual catalyst in a large format library. This needs to be correlated with composition analysis for the catalysts in the library as prepared by thin-film or evaporation technologies. It is the stoichiometry and structure at the surface layer that is important to catalysis, not the bulk composition. A desirable characterization method should be able to sample solid Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

4439

Figure 8. Composition analysis of the surface of catalysts in library 3 by LA-ICPMS.

materials directly in the library but cause as little damage as possible to the library itself. Laser ablation ICPMS seems to be a good choice. In library 3, the catalysts were analyzed by LA-ICPMS after the imaging experiments. Single-spot sampling was used, and three spots were sampled for each catalyst (Figure 8). Since different elements have different absorption coefficients for 266nm light and also different sensitivities in MS, calibration is needed for compositional analysis. Ideally, intensities of the MS signals of individual compounds are linearly dependent on their concentrations. Therefore, the intensity ratios of any two elements should be proportional to their concentration ratios. According to the calibration curve, 74 and 33% Mo is found in samples B2 and B4. The actual percentages of Mo in these two catalysts are 70 and 30%, respectively, and the relative errors of the analysis are less than 6%. This demonstrates that ICPMS can be used to determine the local surface concentrations in these combinatorial libraries. Screening in Thin-Film Combinatorial Arrays. To test the sensitivity of this system, a thin-film combinatorial array was tested in the reaction chamber. The combinatorial arrays of catalytic materials for studying selective oxidation were produced using a reactive sputtering system. Up to three metal oxides can be deposited simultaneously, and additional components can be incorporated in a sequential processing strategy. Compositional variations in the arrays are induced by the geometrical arrangement within the sputtering chamber and by the power input to the rf and dc sputtering guns. Wide ranges of compositional gradients can be imposed by additional shutters which shield large portions of the metal targets. X-ray analysis of sputtered V and Mo samples indicates that samples having ratios of Mo/(Mo + V) ranging from 7.5 to 77% can be readily produced. Materials are isolated using masks that produce a 15 × 15 matrix of 2.0mm combinatorial compositions. Arrays can be deposited at temperatures greater than 700 °C, and annealing can be performed at even higher temperatures. The sputtering system operates with inert (Ar) or reactive (Ar plus O2) atmospheres. We have shown earlier10 that the catalytic activity of a given sample well depends on the thickness of the catalytic material up to a certain point. This is because the available surface area increases until the bottom layers are shielded from the reactants when a thick layer is present. For thin-film arrays made according to the above process, the thickness is limited. Nevertheless, useful signals can be obtained to assess relative activities within the array. Figure 9 shows the results of such an experiment. The signal-tonoise ratio is poorer than the other images presented in this article, but the integrated areas still produce meaningful quantitative results. Even in the gray scale image in Figure 9, differences in activity are evident. 4440 Analytical Chemistry, Vol. 73, No. 18, September 15, 2001

Figure 9. Screening of catalytic activity in a thin-film array. Left: optical image of 2-mm square array elements. Right: fluorescence image after background correction for 370-s exposure.

To obtain the image in Figure 9, a long exposure time is employed (370 s). The accumulated electrons in the CCD pixels were near saturation. The image threshold is set to a high level to represent the high background (scattering) level. Because of the good linearity and large dynamic range of this CCD camera, the small increase in intensity due to product fluorescence is readily extracted. Naturally, thicker films can be employed to increase the S/N ratio. CONCLUSIONS In this work, we demonstrated the application of laser-induced fluorescence imaging as a combinatorial screening technique for heterogeneous catalysts. Fluorescence imaging and infrared thermography can be unified in one instrument, and information about different reaction pathways can be obtained by comparing these side-by-side screening results. Several conclusions can be made based on our screening experiments: (1) vanadium catalysts are less active in the selective oxidation of naphthalene compared to benzene because of steric effects in the former case; (2) the addition of less electronegative ions increases the activity of redox sites on vanadium catalysts by increasing the electron delocalization of the VdO band; (3) the addition of more electronegative ions decreases the activity of vanadium catalysts by decreasing the electron delocalization of VdO; (4) different catalysts show different temperature dependencies in terms of catalytic activity; (5) LA-ICPMS is capable of providing surface composition analysis for catalysts in the library with little sample preparation and minimal damage to the library; and (6) by LIFI, we can screen solid heterogeneous catalysts at a speed of 15 s per library. This screening throughput is many times faster than conventional approaches (calculated on the basis of a 15-member library). Even thin catalytic films can be screened by sacrificing the temporal resolution. ACKNOWLEDGMENT We thank William Schroeder for preparing the thin-film samples. The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract W-7405Eng-82. This work was supported by the Director of Science, Office of Basic Energy Sciences, Division of Chemical Sciences. Received for review April 23, 2001. Accepted July 17, 2001. AC015513I