Online Screening of Homogeneous Catalyst Performance using

Aug 13, 2008 - Citation data is made available by participants in Crossref's Cited-by Linking service. For a more comprehensive list of citations to t...
0 downloads 0 Views 669KB Size
Download PDF
Anal. Chem. 2008, 80, 7121–7127

Online Screening of Homogeneous Catalyst Performance using Reaction Detection Mass Spectrometry Cornelius T. Martha,† Niels Elders,‡ Johannes G. Krabbe,† Jeroen Kool,† Wilfried M. A. Niessen,† Romano V. A. Orru,‡ and Hubertus Irth*,† Department of Analytical Chemistry and Applied Spectroscopy and Department of Organic and Inorganic Chemistry, Faculty of Sciences, Vrije Universiteit Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands An integrated online screening system was developed to rapidly screen homogeneous catalysts for activity toward a selected synthesis. The continuous-flow system comprises standard HPLC pumps for the delivery of substrates, an HPLC autosampler for the injection of homogeneous catalysts, a thermostated reactor to mediate synthesis, and a single-stage quadrupole mass spectrometer (MS) equipped with atmospheric pressure chemical ionization for the determination of product formation. MS detection offers sensitivity, specificity, and speed when applied to the analysis of dynamic processes in the condensed phase. By applying the present methodology for the study of substrate conversion mediated by homogeneous catalysts, the concentration of substrates and reaction product could be monitored while information about the catalysts could also be obtained. In an initial screening application, the performance of a selected number of Lewis acids in the multicomponent synthesis of a highly substituted 2-imidazoline was determined. Limit of detection and limit of quantitation were determined by injecting different concentrations of 2-imidazoline standards and proved to be 1.6 and 5.2 nM, respectively. The results obtained with the new screening method were in good agreement with a traditional bench-scale experiment. Moreover, the system was capable of determining catalyst performance with very low catalyst and solvent consumption while the ruggedness of the system was exhibited with a 24-h continuous analysis of 280 successive catalyst injections with a peak area variation within 7% relative standard deviation. The high-throughput experimental syntheses and screening methods that were first introduced in the pharmaceutical industry to accelerate the discovery process have also been applied for catalyst development.1-4 The high information density libraries * To whom correspondence should be addressed. E-mail: [email protected]. Fax: +31 20 5987543. † Department of Analytical Chemistry and Applied Spectroscopy. ‡ Department of Organic and Inorganic Chemistry. (1) Hagemeyer, A.; Jandeleit, B.; Liu, Y.; Poorjary, D. M.; Turner, H. W.; Volpe, A. F.; Weinberg, W. H. Appl. Catal., A 2001, 221, 23–43. (2) Dahmen, S.; Bra¨se, S. Synthesis 2001, 10, 1431–1449. (3) Jandeleit, B.; Schaefer, D. J.; Powers, T. S.; Turner, H. W.; Weinberg, W. H. Angew. Chem., Int. Ed. 1999, 38, 2494–2532. 10.1021/ac801003h CCC: $40.75  2008 American Chemical Society Published on Web 08/13/2008

that are created with these techniques are composed of a large number of homogeneous catalysts. In order to cope with these large pools of potential catalysts, sophisticated high-throughput screening methods are essential. Spectroscopy-based detection techniques are often employed for catalyst screening. However, these methods rely on the presence of chromophores and fluorophores in either substrate or product for detection of catalytic activity.5-7 Consequently, in order to expand the application range, alternative detection methodologies are being developed for catalyst screening.8-12 In this perspective, mass spectrometry (MS) offers exceptional specificity and selectivity for the sensitive detection of (multiple) selected target molecules in complex matrixes.13 The applicability of mass spectrometric detection and its impact on highthroughput screening was reviewed by Niessen, who concluded that the widespread use and increasing number of publications in the field of high-throughput screening is an indicator for the progress that has been made in terms of both applications and instrumentation.14 In the past decade, a similar advance in the utilization of MS for the characterization of organometallic compounds is described.15,16 Although there are still classes of compounds that can not be identified, commonly reported conversions of neutral molecules to ionic species (i.e., protonation, metalation, chemical reaction, quaternization, and oxidation/ reduction) increase the potential to apply MS for the identification of neutral catalyst complexes. An elegant MS-based approach in the development of catalysts is the so-called “screening before synthesis” methodology pro(4) Pescarmona, P. P.; van der Waal, J. C.; Maxwell, I. E.; Maschmeyer, T. Catal. Lett. 1999, 63, 1–11. (5) Potyrailo, R. A. Trends Anal. Chem. 2003, 22, 374–384. (6) Onaran, M. B.; Seto, C. T. J. Org. Chem. 2003, 68, 8136–8141. (7) Reetz, M. T.; Zonta, A.; Schimossek, K.; Liebeton, K.; Jaeger, K.-E. Angew. Chem., Int. Ed. 1997, 36, 2830–2832. (8) Tielmann, P.; Boese, M.; Luft, M.; Reetz, M. T. Chem. Eur. J. 2003, 9, 3882–3887. (9) Brown, J. M. J. Organomet. Chem. 2004, 689, 4006–4015. (10) Reetz, M. T.; Eipper, A.; Tielmann, P.; Mynott, R. Adv. Synth. Catal. 2002, 344, 1008–1016. (11) Klein, J.; Stichert, W.; Strehlau, W.; Brenner, A.; Demuth, D.; Schunk, S. A.; Hibst, H.; Storck, S. Catal. Today 2003, 81, 329–335. (12) Wolf, C.; Hawes, P. A. J. Org. Chem. 2002, 67, 2727–2729. (13) Fabris, D. Mass Spectrom. Rev. 2005, 24, 30–54. (14) Niessen, W. M. A. J. Chromatogr., A 2003, 1000, 413–436. (15) Traeger, J. C. Int. J. Mass Spectrom. 2000, 200, 387–401. (16) Henderson, W.; Nickleson, B. K.; McCaffrey, L. J. Polyhedron 1998, 17, 4291–4313.

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7121

posed by Chen.17 Instead of screening randomly assembled libraries of spatially resolved catalyst candidates, this methodology offers the possibility to obtain mechanistic information from pooled catalyst libraries. The approach is based on catalyst reaction monitoring in the gas phase by a modified tandem mass spectrometer to provide mechanistic insights of the selected synthesis. With this approach, valuable information for the creation of corresponding catalyst libraries is generated. Several application examples reported good agreement between gas- and solutionphase reactivities. An illustrative example of the possibilities that the enhanced resolution and selectivity of MS detection offer is presented by Liesener et al.18 They described a system designed for the determination of biocatalytic activities of a selected number of proteases for a mixture of substrates. In this so-called multiplexed bioassay, seven different substrate conversions were monitored in parallel. By the repetitive injections of reactor solution and by monitoring both substrate and product concentrations, they were able to obtain selective activities and kinetics for a broad range of proteases. In this paper, a novel integrated approach to achieve high sample throughput for the screening of homogeneous catalysts is described. In recent years, several online integrated methodologies were developed and applied.19,20 In previous work, similar principles were used for the monitoring of biospecific interactions,21-24 for the determination of protein kinase A phosphorylation,25 for trace analysis of proteins using solution-phase digestion,26 and for the characterization of metal complexes.27,28 In traditional catalyst performance methods, the catalysts are rated based on an elevated slope of the initial rate period. In order to obtain kinetic data, the product formation has to be determined at several time intervals. In the present concept, the ranking of different homogeneous catalysts is performed by determining the peak area of formed product shortly after the online synthesis reaches the initial rate period. The increased throughput is then accomplished by reducing the number of data points that are required to characterize catalyst performance. This concept is only feasible when the methodology allows for the quantification of small amounts of synthesis product with high accuracy. In the present online system, all system parameters (i.e., sampling time, reaction time, reaction pressure, and reaction temperature) are (17) Chen, P. Angew. Chem., Int. Ed. 2003, 42, 2832–2847. (18) Liesener, A.; Perchuc, A-M.; Schoni, R.; Wilmer, M.; Karst, U. Rapid Commun. Mass Spectrom. 2005, 19, 2923–2928. (19) Valca´rcel, M.; Ca´rdenas, S.; Gallego, M. Trends Anal. Chem. 2002, 21, 251–258. (20) Lenz, E.; Taylor, S.; Collins, C.; Wilson, I. D.; Louden, D.; Handley, A. J. Pharm. Biomed. Anal. 2002, 27, 191–200. (21) Hogenboom, A. C.; de Boer, A. R.; Derks, R. J. E.; Irth, H. Anal. Chem. 2001, 73, 3816–3823. (22) van Elswijk, D. A.; Tjaden, U. R.; van der Greef, J.; Irth, H. Int. J. Mass Spectrom. 2001, 210/211, 625–636. (23) Derks, R. J. E.; Hogenboom, A. C.; van der Zwan, G.; Irth, H. Anal. Chem. 2003, 75, 3376–3384. (24) de Boer, A. R.; Bruyneel, B.; Krabbe, J. G.; Lingeman, H.; Niessen, W. M. A.; Irth, H. Lab Chip 2005, 5, 1286–1292. (25) de Boer, A. R.; Letzel, T.; Lingeman, H.; Irth, H. Anal. Bioanal. Chem. 2005, 381, 647–655. (26) Bruyneel, B.; Hoos, J. S.; Smoluch, M. T.; Lingeman, H.; Niessen, W. M. A.; Irth, H. Anal. Chem. 2007, 79, 1591–1598. (27) Krabbe, J. G.; Lingeman, H.; Niessen, W. M. A.; Irth, H. J. Chromatogr., A 2005, 1093, 36–46. (28) Krabbe, J. G.; de Boer, A. R.; van der Zwan, G.; Lingeman, H.; Niessen, W. M. A.; Irth, H. J. Am. Soc. Mass. Spectrom. 2007, 18, 707–713.

7122

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

either fixed or accurately controlled, and therefore, a very high system repeatability and reproducibility can be achieved. The experiments described in this paper are conducted in both tetrahydrofuran (THF) and dichloromethane. Although these solvents are commonly used solvents in synthetic organic chemistry, the combination with mass spectrometric detection is less common. In our proof of principle study, a limited number of Lewis acid catalysts were screened for activity toward the synthesis of a highly substituted 2-imidazoline. The synthesis starts with the condensation reaction of benzylamine and acetone to form an intermediate imine, which subsequently reacts with an R-acidic p-nitrobenzylisocyanide to form the 2-imidazoline derivative.29-31 2-Imidazolines are interesting synthetic targets due to their broad application range derived from their affinity toward imidazoline binding sites.32 This affinity is related to a wide variety of biological functions like hypertension, blood pressure regulation, insulin secretion control, and numerous human brain disorders.33-40 The 2-imidazoline synthesis used in our study incorporates most of the atoms of the starting materials in the final product, and only water is produced as waste. Due to this superior atom economy, this multicomponent reaction approach does not require the isolation and purification of product intermediates.41-47 Moreover, by variation of the substrate substituents, different analogous products can be easily obtained and can be used to generate focused libraries of potential therapeutics. Important to this study is that ketones (like acetone) only yield the corresponding 2-imidazolines under standard reaction conditions in the presence of a catalyst. EXPERIMENTAL SECTION Standards and Reagents. GC-grade (>99%) THF and benzylamine were purchased from Fluka (Buchs, Switzerland). Purified water (MQ-quality) was produced in-house using a Millipore (Molsheim, France) Academic water purifier and degassed using a Branson ultrasonic bath (Danbury, CT). HPLC-grade methanol and dichloromethane were purchased from Biosolve (Valken(29) Bon, R. S.; Hong, C.; Bouma, M. J.; Schmitz, R. F.; de Kanter, F. J. J.; Lutz, M.; Spek, A. L.; Orru, R. V. A. Org. Lett. 2003, 5, 3759–3762. (30) Bon, R. S.; van Vliet, B.; Sprenkels, N. E.; Schmitz, R. F.; de Kanter, F. J. J.; Stevens, C. V.; Swart, M.; Bickelhaupt, F. M.; Groen, M. B.; Orru, R. V. A. J. Org. Chem. 2005, 70, 3542–3553. (31) Elders, N.; Schmitz, R. F.; de Kanter, F. J. J.; Ruijter, E.; Groen, M. B.; Orru, R. V. A. J. Org. Chem. 2007, 72, 6135–6142. (32) Bousquet, P.; Feldman, J.; Schwartz, J. J. Pharmacol. Exp. Ther. 1984, 230, 232–236. (33) Parini, A.; Gargalidis Moudanos, C.; Pizzinat, N.; Lanier, S. M. Trends Pharmacol. Sci. 1996, 17, 13–16. (34) Dardonville, C.; Rozas, I. Med. Res. Rev. 2004, 24, 639–661. (35) Bousquet, P.; Feldman, J. Drugs 1999, 58, 799–812. (36) Ueno, M.; Imaizumi, K.; Sugita, T.; Takata, I.; Takeshita, M. Int. J. Immunopharmacol. 1995, 7, 597–603. (37) Doyle, M. E.; Egan, J. M. Pharmacol. Rev. 2003, 55, 105–131. (38) von Rauch, M.; Schlenk, M.; Gust, R. J. Med. Chem. 2004, 47, 915–927. (39) Holt, A. J. Psychiatry Neurosci. 2003, 28, 409–414. (40) Do ¨mling, A. Chem. Rev. 2006, 106, 17–89. (41) Armstrong, R. W.; Combs, A. P.; Tempest, P. A.; Brown, S. D.; Keating, T. A. Acc. Chem. Res. 1996, 29, 123–131. (42) Tietze, L. F.; Modi, A. Med. Res. Rev. 2000, 20, 304–322. (43) Do ¨mling, A.; Ugi, I. Angew. Chem., Int. Ed. 2000, 39, 3168–3210. (44) Do ¨mling, A. Comb. Chem. High Throughput Screening 1998, 1, 1–22. (45) Orru, R. V. A.; de Greef, M. Synthesis 2003, 10, 1471–1499. (46) Zhu, J. Eur. J. Org. Chem. 2003, 7, 1133–1144. (47) Bienayme´, H.; Hulme, C.; Oddon, G.; Schmitt, P. Chem. Eur. J. 2000, 6, 3321–3329.

Figure 1. Schematic overview of the online analytical system. The system comprises two HPLC pumps (P1 and P2) that pump the imine (substrate A) formed after the condensation of acetone and benzylamine and the p-nitrobenzylisocyanide (substrate B) into the coiled reactor. An injector is used to introduce the homogeneous catalysts into the stream of substrates. A third HPLC pump (P3) is used to add a water makeup flow to quench the reaction and increase APCI-MS detector compatibility.

swaard, The Netherlands). Formic acid (98%) and acetone (>99%) were purchased from J.T. Baker (Deventer, The Netherlands). Silver trifluoromethanesulfonate (AgOTf, >98%), nickel nitrate (>98.5%), scandium trifluoromethanesulfonate (Sc(OTf)3, >97%), copper trifluoromethanesulfonate (Cu(OTf)2, >97%) were purchased from Fluka (Buchs, Switzerland). Magnesium iodide (98%), magnesium bromide (98%), magnesium perchlorate (Mg(ClO4)2, 99%), and magnesium bromide ethyl etherate (99%) were purchased from Aldrich (Steinheim, Germany). The p-nitrobenzylisocyanide and 2-imidazoline product standards were produced in-house according to the protocol described by Elders et al.30 Continuous-Flow System. The online experiments were performed using an online continuous-flow system that consisted of a substrate delivery unit, a temperature-controlled reactor, and a mass spectrometer for the reaction detection (scheme, see Figure 1). The substrate delivery configuration was based on the solvent used. In the case of dichloromethane, due to equipment compatibilities, the substrate delivery was performed using two Gilson model 302 (Villiers-le-Bel, France) HPLC pumps. When THF-based substrate solutions were applied, the substrate delivery was performed with two Shimadzu LC-20AD HPLC pumps (‘sHertogenbosch, The Netherlands). The HPLC pumps delivered the preformed imine and isocyanide substrate solutions at a flow rate of 200 µL/min into the mixer (Valco International, Schenkon, Switzerland). Subsequently, a selected homogeneous catalyst was injected using a Gilson 234 autoinjector. The substrates were converted into the imidazoline derivative product in an in-houseproduced reactor immersed in a temperature-controlled water bath from Grant (Cambridge, England). The product formation was hereafter quenched by the addition of 200 µL/min water by a Shimadzu LC-20AD HPLC pump. In the experiments conducted with dichloromethane, the water makeup flow was omitted. Finally, the detection of substrate conversion, product formation, and identification of catalyst were performed with a LCMS-2010A single-stage quadrupole mass spectrometer from Shimadzu. Measurements were performed by alternating between a selected

ion monitoring (SIM) of the imidazoline derivative (m/z ) 310.2) and a full-spectrum total ion current (TIC). From the TIC data, the conversion of substrates can be monitored and catalysts identified. Bench-Scale Sample Preparation. Labware was rinsed with GC-grade THF prior to use. In order to determine bench-scale benchmark kinetics, a selected Lewis acid catalyst (2 m/m %) was placed in a glass tube equipped with a stir bar and placed on a heated (40 °C) magnetic stirrer. Subsequently, 5 mL of 1 mM acetone in THF, 5 mL of 1 mM benzylamine in THF, and 5 mL of 1 mM isocyanide in THF were added. At certain time intervals (0, 10, 30, 60, 120, 300, and 1440 min), 100-µL aliquots were taken from the reactor and the reaction was quenched by the addition of 900 µL of water. Finally, the samples were stored at 4 °C prior to analysis. Bench-Scale Sample Analysis. In order to obtain bench-scale kinetic data, the homogeneous catalyst-mediated 2-imidazoline derivative formation (Figure 2) was determined with a 20-min HPLC-MS analysis. Optimal conditions for separation and quantification by a LCMS-2010A single-stage quadrupole HPLC-MS from Shimadzu involved injection of 10 µL of sample obtained from the bench-scale experiments onto a 50 × 2 mm Phenomenex 125-Å Aqua 3-µM C-18 column. The flow rate was 200 µL/min. Gradient elution conditions were as follows: starting conditions 80% A (water with 0.1% of formic acid)-80% B (100% methanol). The data were obtained by the SIM of the imidazoline (m/z ) 310.2), and product formation was quantified with Labsolutions/LCMS solution software from Shimadzu. Finally, kinetic data of the different homogeneous Lewis acid catalysts was obtained by determining the slope of the initial rate period. RESULTS AND DISCUSSION Setup of the Continuous-Flow Screening System. The present screening methodology is based on the implementation of mass spectrometric detection for the detailed study of homogeneous catalyst-mediated product formation as well as for the determiAnalytical Chemistry, Vol. 80, No. 18, September 15, 2008

7123

Figure 2. Multicomponent reaction where benzylamine (a) and acetone (b) condensate into an imine (d) that subsequently reacts in a Lewis acid catalyzed reaction with p-nitrobenzylisocyanide (c) into a highly substituted 2-imidazoline derivative (e).

nation of substrate concentrations and the detection of solvated catalysts. In this system, standard HPLC pumps are utilized for the delivery of individual substrates, an HPLC autosampler is used for the introduction of homogeneous catalysts in the solvent stream, and a thermostated reactor is applied for the in-solution product formation and mass spectrometry for the determination of the synthesis product. A main requirement to successfully monitor solution-phase reactions with mass spectrometry is the MS compatibility of the applied solvents. Organic syntheses usually involve solvents that are not commonly associated with mass spectrometric detection. While a water makeup flow can increase the mass spectrometry compatibility of a reaction in a hydrophilic solvent, this approach is not feasible when water-immiscible solvents are used. In order to investigate the usability of this continuous-flow system for syntheses in different solvents, we calibrated the performance under synthetic conditions by determining the limit of detection (LOD), the limit of quantification (LOQ), and the linearity of the online system for the 2-imidazoline derivative in both THF and dichloromethane. In order to determine the detection limits in THF, varying concentrations of in-house synthesized and purified 2-imidazoline product standards (500 pM-10 µM) were injected into the online system where the continuous flow was composed of the imine and isocyanide substrates but in the absence of catalysts. The data obtained for the standards are mean values of triplicate determinations. The repeatability is very good with the highest variation coefficient determined to be 4.1%. The LOD and LOQ, defined as 3 and 10 times the standard deviation of the noise, were determined to be 1.6 and 5.2 nM, respectively. Moreover, from 5 to 10 000 nM, the method proved to give excellent linear responses with a correlation coefficient of 0.9997. For the dichloromethane experiments, the same approach was used although the water makeup flow had to be omitted. Triplicates of nine product standards (1 nM-1 µM) were injected with a signal-tonoise ratio of the lowest product standard of 135 with a RSD of 9%. From 1 nM to 1 µM, the method proved to be linear with a correlation coefficient of 0.9955. These results confirm the applicability of this online approach for the sensitive and repeatable determination of 2-imidazoline formation. Additionally, the performance in THF and dichloromethane demonstrates that the monitoring of catalyst performance is not restricted to syntheses in hydrophilic solvents, and because of the low detection limits and variation coefficients, a short reaction time can be used in further experiments. An online continuous-flow reaction detection system using MS as detector was previously published by Hogenboom et al.20 and 7124

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

used for the screening of receptor-ligand interactions. By applying a similar configuration for the screening of homogeneous catalysts, most system parameters are compromises between an optimum analyte response, system ruggedness, and sample throughput. In the current reaction detection system, the formation of product can be enhanced by increasing substrate concentrations, by increasing catalyst concentrations and catalyst injection volumes, by increasing reaction times, and by elevating reaction temperatures. However, elevated analyte and homogeneous catalyst concentrations lead to restricted continuous mass spectrometric analysis times due to mass spectrometric source pollution. Moreover, in this continuous-flow system, an extended reaction time causes increased analyte diffusion, decreased analyte peak heights, and limited catalyst injection repetition rates. One example of a compromise in the analytical system optimization is presented in Figure 3 where the optimization of reactor temperature of the synthesis of 2-imidazoline product in THF is shown. The baseline in this figure demonstrates an increased noncatalyzed formation of product at elevated reactor temperatures, and although the highest reaction temperature yields the highest amount of formed product, considering the high flammability of THF, the system was further operated at a reactor temperature of 40 °C. System Ruggedness. One of the key issues when applicability of a screening system is concerned is the ruggedness of the applied methodology. In mass spectrometry, the ruggedness is mainly limited by the introduction of nonvolatiles into the mass spectrometer. The ruggedness of the current screening system was investigated by the repetitive injecting of silver triflate catalyst every 4 min for 24 h. The analysis of successive catalyst injections is presented in Figure 4. In 1440 min, the system was capable of performing 305 silver triflate analyses with a peak area variation within 7% residual standard deviation (RSD). The high intraday repeatability and the ruggedness demonstrate the potential of this new method. Catalyst Screening. In order to benchmark the new system, a selected number of metal-based Lewis acid catalysts were screened for activity. The results of the continuous-flow system were compared with the results obtained with a more traditional bench-scale catalyst assessment. For the screening, eight representative Lewis acid catalysts (Cu(OTf)2, MgBr2, MgBr2 · O(C2H5)2, MgI2, Mg(ClO4)2, NiNO3, Sc(OTf)3, AgOTf) and a blank were tested for activity toward a multicomponent synthesis of a highly substituted 2-imidazoline derivative. The traditional bench-scale catalyst screening method consisted of mixing of benzylamine, acetone, and p-nitrobenzyliso-

Figure 3. Online reactor temperature optimization. The silver triflate-induced formation of 2-imidazoline product in THF is determined with APCI-MS at increasing reactor temperatures. a-i: 10-80 °C with 10 °C increment.

Figure 4. Ruggedness of the online methodology investigated by a 24-h consecutive analysis of a substrate conversion mediated with AgOTf Lewis acid catalyst. The chromatogram presents 280 successive analyses with a RSD within 7%. The inset presents a zoom-in of the chromatogram and demonstrates the excellent peak area repeatability.

cyanide and the subsequent addition of a selected homogeneous Lewis acid catalyst. After mixing, an aliquot was collected, the reaction quenched and diluted by the addition of water, and the sample stored at 4 °C prior to analysis. Furthermore, at certain time intervals (0, 10, 30, 60, 120, 300, and 1440 min) this procedure

was repeated until all the samples were acquired. Subsequently, the amount of 2-imidazoline was determined by using a gradient HPLC separation applying the SIM mode of a single-stage quadrupole mass spectrometer. From the chromatogram, the peak area of catalyst-mediated 2-imidazoline synthesis product was Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7125

Figure 5. Comparison of the two different catalysts performance determination approaches. The dark gray bars present the results (slope) obtained with the traditional off-line benchtop approach while the light gray bars present the results (peak area) obtained with the online system.

determined. Finally, from the peak areas at certain intervals, conversion kinetics were calculated and performance of different homogeneous catalysts could be compared. In the continuous-flow screening experiments, the different homogeneous catalysts were injected into the substrate flow with a repetition rate of 4 min. The catalyst-mediated conversion of substrates into product occurred in the thermostated reactor, and the amount of product was instantly determined by peak integration of the chromatogram obtained with mass spectrometric detection utilizing the SIM option of the single-stage quadrupole mass spectrometer. The results of the traditional off-line screening method and continuous-flow screening approach are summarized in Figure 5. For the traditional bench-scale screening method, the performances of the different catalysts are presented as system blankcorrected slopes of the initial rate period. For the continuousflow screening system, the performances of the different catalysts are presented as peak areas of the 2-imidazoline product. The results present excellent agreement between the two screening procedures. In both methodologies, it is found that silver triflate is the best performing homogeneous catalyst in the multicomponent synthesis of the substituted 2-imidazoline. In both screening methods, the two best performing catalysts (AgOTf and CuOTf) as well as the two catalysts yielding the least product formation enhancement (ScOTf and NiNO3) are identical. Although the rating of equally performing catalysts diverges to some extent, both screening methodologies rate magnesium-based homogeneous catalyst as medium performing catalyst for this multicomponent reaction. CONCLUSIONS We have presented a continuous-flow screening system for the determination of homogeneous catalyst activity and demonstrated the performance of the methodology with a screening application for homogeneous catalysts in multicomponent reactions. Because mass spectrometric detection is incorporated into the online method, the determination of product formation and substrate conversion as well as catalyst identification can be performed simultaneously. 7126

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

When the continuous-flow screening method is utilized, the assessment of homogeneous catalysts for a selected synthesis requires 4 min of analysis time per sample. Compared to the traditional bench-scale catalyst assessment, a considerable time advantage is achieved. Although the analysis time of the HPLC separation used in the bench-scale screening can be reduced significantly, even the fastest HPLC separation methods are not extensively faster than the proposed continuous-flow screening method. The sample preparation that the continuous screening technique requires consists of the preparation of substrate and sample solutions. This is comparable to the sample preparation required in the off-line screening approach with the remark that a multireactor setup used in the bench-scale method requires significantly more sample and substrate transfers. The design of the current bench-scale experiment was such that reaction kinetics were constructed from 2-imidazoline product determination at seven different reaction time intervals. This approach was selected because homogeneous catalyst assessments using the product formation at one fixed reaction time proved to not be successful due to high variation in formed 2-imidazoline product caused by poor reaction time and sample handling repeatabilities. The repeatability of the off-line procedure can possibly be improved by applying high-precision robotics for the accurate transfer of fluids. For the continuous-flow system, the high system repeatabilities are the result of fixed or accurately controlled system parameters, and therefore, a single reaction time catalyst assessment was feasible. Other virtues of our automated continuous-flow system are the low sample consumption, high sensitivity, and broad solvent, substrate, and product applicability. The experiments described in this paper are all conducted in both THF and dichloromethane. Although these are commonly used solvents in synthetic organic chemistry, the combination with mass spectrometric detection is less common and a demonstration of the versatility of the novel concept. The results obtained with the screening method are in good agreement with a traditionally applied bench-scale experiment and unambiguously demonstrates the power

of the methodology for the screening of homogeneous catalysts. The current system can easily be adapted to other synthetic conversions by adjusting the continuous flows of the reaction substrates with the standard HPLC equipment that this system is based upon.

edged for the loan of a LCMS-2010A single-stage quadrupole mass spectrometer.

ACKNOWLEDGMENT This work was financially supported with an ECHO grand of the Dutch Scientific Society NWO. Shimadzu Benelux is acknowl-

Received for review May 16, 2008. Accepted July 16, 2008. AC801003H

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7127