High-Throughput Screening of Selectivity of Melt Polymerization

Anal. Chem. 2003, 75, 4676-4681

High-Throughput Screening of Selectivity of Melt Polymerization Catalysts Using Fluorescence Spectroscopy and Two-Wavelength Fluorescence Imaging Radislav A. Potyrailo,* John P. Lemmon, and Terry K. Leib

Global Research Center, Combinatorial Chemistry and Characterization Technologies, General Electric Company, Schenectady, New York 12301

A new general approach for rapid assessment of polymerization catalysts is introduced. Native fluorescence emission of solid polymers is measured directly in combinatorial 96-microreactor arrays and polymers produced in a laboratory-scale validation reactor. Fluorescence features collected with a CCD-based spectrofluorometer are correlated with chemical properties of interest such as polymer molecular weight, amount of branching, and catalyst selectivity. The approach is illustrated by screening of selectivity of melt polymerization catalysts used in synthesis of an aromatic bisphenol A polycarbonate. Selectivity of catalysts correlated with the ratio of fluorescence intensities at 400 and 500 nm at 340-nm excitation. The relative standard deviation (RSD) in spectroscopic serial measurements was 1-12.5%. This spread included instrument variability (e1% RSD) and sample inhomogeneity. Parallel quantitative screening of catalyst selectivity in combinatorial 96-microreactor arrays was performed as a two-wavelength ratiometric fluorescence imaging through 400- and 500-nm interference filters and showed a good correlation (R2 ) 0.994) with serial screening. Our approach is an attractive alternative to traditional separation-based techniques if speed and nondestructive nature of analysis are critical and when the high cross-linking or solvent resistance of polymers complicates traditional analysis. At present, combinatorial and high-throughput (HT) screening methods have found applications beyond the pharmaceutical industry for discovery of new materials.1-4 One of the most convincing demonstrations of the value of these screening methodologies is in the discovery of new catalysts.3,4 Performance of catalysts is evaluated using a variety of impressive innovative * Corresponding author. E-mail: [email protected]. (1) Malhotra, R., Ed. Combinatorial Approaches to Materials Development; American Chemical Society: Washington, DC, 2002. (2) Takeuchi, I., Newsam, J. M., Wille, L. T., Koinuma, H., Amis, E. J., Eds. Combinatorial and Arficial Intelligence Methods in Materials Science; Materials Research Society: Warrendale, PA, 2002. (3) Schlo ¨gl, R. Angew. Chem., Int. Ed. 1998, 37, 2333-2336. (4) Crabtree, R. H. Chem. Commun. 1999, 1611-1616.

4676 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

screening techniques.5-13 In the discovery of new polymerization catalysts, it is essential to assess their effectiveness by the built polymer molecular weight (MW), composition, and amount of branching. Analysis of polymer MW is typically done using conventional and HT gel permeation chromatography (GPC).14-19 Branching content and composition are determined using NMR, IR, and Raman spectroscopies.15,18,20-25 Examples of other evaluations of combinatorial polymers include melting temperatures by DSC, crystallinity by powder diffractometry, surface properties by SIMS-TOF, contact angle by sessile drop, and others.14,21-23 Unfortunately, dissolution of polymers for GPC and other types of analysis could be a rate-limiting step or could even be impossible if the polymers are highly cross-linked, consist of a (5) Holzwarth, A.; Schmidt, H.-W.; Maier, W. Angew. Chem., Int. Ed. 1998, 37, 2644-2647. (6) Connolly, A. R.; Sutherland, J. D. Angew. Chem., Int. Ed. 2000, 39, 42684271. (7) Senkan, S. M. Nature 1998, 394, 350-353. (8) Su, H.; Yeung, E. S. J. Am. Chem. Soc. 2000, 122, 7422-7423. (9) Snively, C. M.; Oskarsdottir, G.; Lauterbach, J. Angew. Chem., Int. Ed. 2001, 40, 3028-3030. (10) Taylor, S. J.; Morken, J. P. Science 1998, 280, 267-270. (11) Reddington, E.; Sapienza, A.; Gurau, B.; Viswanathan, R.; Sarangapani, S.; Smotkin, E. S.; Mallouk, T. E. Science 1998, 280, 1735-1737. (12) Busch, O. M.; Hoffmann, C.; Johann, T. R. F.; Schmidt, H.-W.; Strehlau, W.; Schu ¨ th, F. J. Am. Chem. Soc. 2002, 124, 13527-13532. (13) Morris, N. D.; Mallouk, T. E. J. Am. Chem. Soc. 2002, 124, 11114-11121. (14) Brocchini, S.; James, K.; Tangpasuthadol, V.; Kohn, J. J. Am. Chem. Soc. 1997, 119, 4553-4554. (15) Boussie, T. R.; Coutard, C.; Turner, H.; Murphy, V.; Powers, T. S. Angew. Chem., Int. Ed. 1998, 37, 3272-3275. (16) Klaerner, G.; Safir, A. L.; Chang, H.-T.; Petro, M.; Nielsen, R. B. Polym. Prepr. 1999, 40, 469. (17) Bosman, A. W.; Heumann, A.; Klaerner, G.; Benoit, D.; Frechet, J. M. J.; Hawker, C. J. J. Am. Chem. Soc. 2001, 123, 6461-6462. (18) Komon, Z. J. A.; Diamond, G. M.; Leclerc, M. K.; Murphy, V.; Okazaki, M.; Bazan, G. C. J. Am. Chem. Soc. 2002, 124, 15280-15285. (19) Pasch, H.; Kilz, P. Macromol. Rapid Commun. 2003, 24, 104-108. (20) Tian, J.; Coates, G. W. Angew. Chem., Int. Ed. 2000, 39, 3626-3629. (21) Stork, M.; Herrmann, A.; Nemnich, T.; Klapper, M.; Mu ¨ llen, K. Angew. Chem., Int. Ed. 2000, 39, 4367-4369. (22) Belu, A. M.; Brocchini, S.; Kohn, J.; Ratner, B. D. Rapid Commun. Mass Spectrom. 2000, 14, 564-571. (23) Pijpers, T. F. J.; Mathot, V. B. F.; Goderis, B.; van der Vegte, E. Proc. NATAS Annu. Conf. Therm. Anal. Appl. 2000, 28. (24) Tuchbreiter, A.; Marquardt, J.; Zimmermann, J.; Walter, P.; Mu ¨ lhaupt, R.; Kappler, B.; Faller, D.; Roths, T.; Honerkamp, J. J. Comb. Chem. 2001, 3, 598-603. (25) Gabriel, C.; Lilge, D.; Kristen, M. O. Macromol. Rapid Commun. 2002, 24, 109-112. 10.1021/ac034296d CCC: $25.00

© 2003 American Chemical Society Published on Web 07/24/2003

Scheme 1. Melt Polymerization Reaction of an Aromatic Bisphenol-A Polycarbonate

large number of repeating units, or are solvent-resistant. Also, by dissolving polymers for analysis, the information about the spatial heterogeneity of polymers in microreactors is lost. Such information is a powerful indicator of the reproducibility of combinatorial polymerizations.26 Thus, it is highly desirable to measure polymer MW and other chemical properties in a nondestructive fashion. This also permits us to screen the same library of polymerized materials for additional properties of interest such as mechanical, physical, and dielectric.27 In this report, we introduce a new method for rapid evaluation of combinatorially developed polymers. We measure the native fluorescence emission of solid polymers directly in microreactors using spectroscopic and imaging techniques and correlate the fluorescence features with the chemical properties of interest such as polymer MW, amount of branching, and catalyst selectivity. This fluorescence screening method can be applicable to diverse types of polymers where fluorescence originates from chromophores situated as in-chain and end-chain groups or from chromophores present in the repeat units in the backbone polymer structure.28-31 We demonstrate our method for screening of selectivity of melt polymerization catalysts used in synthesis of an aromatic bisphenol A polycarbonate (BPA-PC). This engineering thermoplastic has the desirable combination of toughness and transparency, which makes it attractive for demanding applications where resistance to hydrolytic and photochemical degradation is important.32 (26) Potyrailo, R. A.; Wroczynski, R. J.; Lemmon, J. P.; Flanagan, W. P.; Siclovan, O. P. J. Comb. Chem. 2003, 5, 8-17. (27) Potyrailo, R. A. In Encyclopedia of Materials: Science and Technology; Buschow, K. H. J., Cahn, R. W., Flemings, M. C., Ilschner, B., Kramer, E. J., Mahajan, S., Eds.; Elsevier: Amsterdam, The Netherlands, 2001; Vol. 2, pp 1329-1343. (28) Gachkovskii, V. F. Vysokomol. Soedin. 1965, 7, 2199-2205 (English). (29) Morawetz, H. Science 1979, 203, 405-410. (30) Chipalkatti, M. H.; Laski, J. J. In Structure-Property Relations in Polymers. Spectroscopy and Performance; Urban, M. W., Craver, C. D., Eds.; Advances in Chemistry 236; American Chemical Society: Washington, DC, 1993; pp 623-642. (31) Barashkov, N. N.; Gunder, O. A. Fluorescent Polymers; Ellis Horwood: New York, 1994. (32) LeGrand, D. G., Bendler, J. T., Eds. Handbook of Polycarbonate Science and Technology; Marcel Dekker: New York, 2000.

EXPERIMENTAL SECTION Melt Polymerization Reaction System. Synthesis of BPAPC was initially performed in a laboratory-scale reactor by reaction of an aromatic diphenol such as 2,2′-bis(4-hydroxyphenyl)propane (1), also known as bisphenol A, with a derivative of carbonic acid, such as diphenyl carbonate (2), in the presence of a catalyst candidate (see Scheme 1). During the condensation polymerization of 1 and 2, phenol 3 is evolved and is removed from the reaction volume with a flow of an inert gas in a combinatorial 96microreactor array26 or with vacuum in the laboratory-scale validation reactor.33 As a result of the melt polymerization reaction, linear BPA-PC 4 is formed followed by formation of salicylatecontaining BPA-PC 5 and branched side product 6 (known as Fries product34). Selectivity (S) of formation of linear BPA-PC 4 versus branched side product 6 is controlled by the catalyst.35 Selectivity is defined as the ratio of polymer number-average molecular weight (Mn) to the concentration of Fries product 6, S ) Mn/[Fries]. The properties of BPA-PC are affected by the amount of product 6 present, and it is required to minimize its content for consistent melt flow and ductility of BPA-PC.33 Parallel Melt Polymerization in 96-Microreactor Arrays. The starting reaction components 1 and 2 were obtained from GE Plastics. Catalyst candidates were obtained from an in-house collection. Melt polymerization reactions were performed in glass 96-well microtiter plates purchased from Spike International (Wilmington, NC) that served as 96-microreactor arrays36 in a sequence of steps of increasing temperature33 with a maximum temperature of 280 °C. Additional details of combinatorial polymerizations in 96-microreactor arrays are available from our recent (33) King, J. A., Jr. In Solvent-Free Polymerizations and Processes. Minimization of Conventional Organic Solvents; Long, T. E., Hunt, M. O., Eds.; ACS Symposium Series 713; American Chemical Society: Washington, DC, 1998; pp 49-77. (34) Factor, A. In Polymer Durability. Degradation, Stabilization, and Lifetime Prediction; Clough, R. L., Billingham, N. C., Gillen, K. T., Eds.; Advances in Chemistry Series 249; American Chemical Society: Washington, DC, 1996; pp 59-76. (35) Brunelle, D. J.; Boden, E. P.; Shannon, T. G. J. Am. Chem. Soc. 1990, 112, 2399-2402. (36) Carnahan, J. C.; Lemmon, J. P.; Potyrailo, R. A.; Leib, T. K.; Warner, G. L. Method for parallel melt-polymerization. U.S. Patent 6,307,004 B1, 2001.

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4677

Figure 1. Photograph of a 96-microreactor array under a broadband UV excitation. Scale bar is 10 mm.

report.26 A photograph of a 96-microreactor array with solid melt polymerized BPA-PC under a UV excitation is depicted in Figure 1. High-Throughput Fluorescence Scanning System. Measurements of fluorescence spectra of solid polymers in each microreactor of 96-microreactor arrays were performed using a modular automatic scanning system that consisted of a white light source, a monochromator, a portable spectrofluorometer, and a translation stage. A schematic of the automatic scanning system is shown in Figure 2A. The white light source (450-W Xe arc lamp, SLM Instruments, Inc., Urbana, IL, model FP-024) was coupled to the monochromator (SLM Instruments, Inc., model FP-092) for selection of the excitation wavelength. The light was further focused into one of the arms of a “six-around-one” bifurcated fiberoptic reflection probe (Ocean Optics, Inc., model R400-7-UV/vis). Emission light from the polymeric material in each microreactor was collected when the common end of the fiber-optic probe was positioned in proximity to a microreactor. The second arm of the probe was coupled to the portable spectrofluorometer (Ocean Optics, Inc., model ST2000) through an in-line optical filter holder (Ocean Optics). The holder contained a long-pass optical filter to block excitation light from entering the spectrofluorometer. The spectrofluorometer was equipped with a 200-µm slit, 600 grooves/ mm grating blazed at 400 nm and covering the spectral range from 250 to 800 nm with efficiency greater than 30%, and a linear CCD array detector. Fluorescence spectra reported here were not corrected by the spectral response of the used optical system. Data analysis was performed using KaleidaGraph software (Synergy Software, Reading, PA) and PLS_Toolbox software (Eigenvector Research, Inc., Manson, WA) operated with Matlab software (The Mathworks Inc., Natick, MA).

Fluorescence Imaging System. A schematic of an imaging system assembled for screening of the 96-microreactor arrays is shown in Figure 2B. The imaging system included the white light source (450-W Xe arc lamp, SLM Instruments, Inc., model FP024), the monochromator (SLM Instruments, Inc., model FP-092) for selection of the excitation wavelength used for the spectroscopic measurements, a beam expander for the efficient illumination of the combinatorial 96-microreactor arrays, and a cooled CCD camera (Roper Scientific, Trenton, NJ, model TE/CCD 1100 PF/ UV). Fluorescence images were collected under a 340-nm excitation through the 400- and 500-nm interference filters (Melles Griot, Irvine, CA). Image analysis was performed using a software provided with the CCD camera and Matlab software. RESULTS AND DISCUSSION Spectral Conditions for Polymer Screening. In exploring new polymerization catalysts for a wide range of polymers, it is highly desirable to employ a catalyst-screening technique that is nondestructive, capable of determinations of branching from low part-per-million to percent levels, and unaffected by possible gelation effects.27 We addressed these needs by developing spectroscopic and imaging systems that measure native fluorescence of solid polymerized materials directly in each microreactor of the thin-film combinatorial 96-microreactor array and produced in a laboratory-scale validation reactor. In the initial experiments, excitation-emission fluorescence maps of solid BPA-PC produced in the laboratory-scale validation reactor were examined to reveal different species. A typical excitation-emission fluorescence map of a solid BPA-PC is presented in Figure 3. Excitation at 260-280 nm preferentially produced 300-320-nm emission from π-π* transition of the carbonyl group in the BPA-PC chain. Excitation at 300-470 nm produced 400-500-nm emission from linear and branched polycarbonate chains, oligomers, and cyclics.37 In the past, correlation of fluorescence intensity with the polymer molecular weight has been reported for vinyl polymers of the polystyrene and polyacrylonitrile series, poly(methacrylic acid) derivates, and highly conjugated polymers of the polyphenylacetylene type28,31 Our analysis of the excitation-emission fluorescence maps of solid BPA-PC permitted determination of the fluorescence conditions for quantitation of polymer molecular weight and catalyst selectivity. Observed emission from π-π* transition of the carbonyl group in BPA-PC chain at 305 nm under a 280-nm excitation correlated well with the number-average

Figure 2. Automatic systems for the determination of selectivity of melt polymerization catalysts in solid BPA-PC in combinatorial 96-microreactor arrays by fluorescence. (A) Scanning modular system. (b) Imaging system. 4678

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

Figure 6. Correlation between catalyst selectivity S measured using reference techniques and new HT fluorescence technique for different NaOH catalyst loadings. Fluorescence excitation 340 nm. Figure 3. Typical excitation-emission fluorescence map of solid BPA-PC.

Figure 4. Correlation between fluorescence of solid BPA-PC samples produced using a variety of catalysts and their numberaverage molecular weight Mn measured by traditional GPC analysis (reference, polycarbonate). Fluorescence excitation 280 nm and emission 305 nm. Error bars, 1 SD (n ) 3).

Figure 5. Normalized emission spectra of solid BPA-PC samples with variable S. Fluorescence excitation 340 nm.

molecular weight of BPA-PC. This correlation is shown in Figure 4. Importantly, this correlation was established with solid BPAPC samples produced using a variety of catalysts. Validation of Quantitation of Catalyst Selectivity by Fluorescence. Selectivity S of catalysts was found to correlate with the ratio of fluorescence intensities at 400 and 500 nm (I400/I500) at 340-nm excitation. Fluorescence spectra normalized to the intensity at 500 nm and baseline-corrected demonstrate this correlation in Figure 5. These spectra show that variations in the independently determined ratio of number-average molecular weight to Fries content (Mn/[Fries]) are correlated with the observed ratio of fluorescence intensities at 400 and 500 nm.

Determination of catalyst selectivity by the ratiometric fluorescence measurements is an important advancement over the measurements of fluorescence intensity. By using the ratiometric measurements of fluorescence emission at different wavelengths but from the same region of polymer, it was possible to reduce the requirements for control of the amount of the polymer in each microreactor. However, the need still remained to have an adequate signal-to-noise ratio in the spectral measurements. The signal-to-noise ratio was kept high (cf. Figure 5) by using a highintensity lamp (450-W Xe arc) and long integration time (100200 ms). While in this study we used the white source filtered to produce a 340-nm emission, if needed, it is straightforward to use a nitrogen laser instead. Such laser emits at 337 nm and, depending on its type, produces at least 100 times more intense light. The new high-throughput screening fluorescence technique showed a good correlation with the determinations of catalyst selectivity S performed using reference techniques. Reference values for polymer Mn were measured by GPC. Branched product 6 was quantified by alkaline hydrolysis and HPLC.34 Typical results of these correlations for a single type of a catalyst are shown in Figure 6. These data was obtained when a control catalyst, such as NaOH, was used at different catalyst equivalents (loadings) to produce several polymer samples with different catalysts selectivity. These data also illustrate the reproducibility of the polymerizations as indicated by multiple data points at a catalyst loading of 5 equiv. Even these small variations in the catalyst selectivity were picked very well with the new technique. We further evaluated precision (as percent relative standard deviation, RSD) of measurements of the ratio I400/I500. These evaluations were performed with replicate measurements on a number of solid samples with different selectivity S using the fiberoptic probe. Measurements were performed at different distances from the samples and repositioning samples between measurements. We have found that the instrument reproducibility was at least 1% RSD or better (n ) 11) when measurements were performed at 0-5-mm distances between the fiber-optic probe and the sample. The slight improvement in precision of measurements from 1 to 0.6% RSD was observed upon increasing the probesample distance from 0 to 5 mm. Such improvement was likely (37) Potyrailo, R. A.; Lemmon, J. P. Method for direct measurement of polycarbonate compositions by fluorescence. U.S. Patent 6,193,850, 2001.

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4679

Figure 7. Ratio of fluorescence intensities I400/I500 of solid BPAPC samples (117 catalyst/loading combinations, validation reactor) as a function of catalyst selectivity S. Error bars 1 SD (n ) 3). Empirical fitted line is a quadratic fit.

due to the sample averaging effect consistent with the methods for sampling of solid materials.38,39 This high measurement precision makes possible nondestructive determinations of spatial inhomogeneities of combinatorially developed solid polymers as has been recently reported.26 Often, the reproducibility of replicate combinatorial synthesis reactions may be the main source of variability between the combinatorial samples.40 Such variability can be reduced using, for example, multiparameter optimization algorithms26 and statistical methods for evaluating the variation sources.40 It is worth noting that it may also be possible to apply “ratiometric” fluorescence measurements for the determinations of MW of BPA-PC. Such measurements could involve collection of fluorescence spectra at several excitation wavelengths followed by multivariate analysis where certain multiple excitationemission conditions can be found for further simple ratio calculations of emissions at different wavelengths and different excitation conditions. Detailed results of such studies will be reported in the future. However, we have already demonstrated a multivariate correlation between fluorescence spectra, the branched side product 6, and polymer MW.41,42 Quantitative Fluorescence Spectroscopic Screening of Catalyst Selectivity. We applied the fluorescence ratiometric technique under our determined excitation conditions for screening of a variety of catalysts. We evaluated solid BPA-PC samples produced with 117 catalyst/loading conditions. Examples of groups of screened catalysts include alkali and alkaline earth metal (e.g., Ba, Ca, Na, Li, and Cs) hydroxides and salts of nonvolatile acids (e.g., H3PO3, H3PO4, H2SO4, and EDTA) and some others. Screening results are presented in Figure 7 where the ratio of fluorescence intensities I400/I500 was measured in three randomly selected different regions of each of the samples to reduce the effects of inhomogeneity of solid samples. All solid samples with catalysts of different nature and loadings showed a good correlation between the catalyst selectivity (38) Carr-Brion, K. G.; Clarke, J. R. P. Sampling Systems for Process Analyzers; Butterworth-Heinemann: Oxford, England, 1996. (39) ASTM E 1655-97, Standard Practices for Infrared, Multivariate, Quantitative Analysis; ASTM: West Conshohocken, PA, 1997. (40) Spivack, J. L.; Cawse, J. N.; Whisenhunt, D. W., Jr.; Johnson, B. F.; Shalyaev, K. V.; Male, J.; Pressman, E. J.; Ofori, J.; Soloveichik, G.; Patel, B. P.; Chuck, T. L.; Smith, D. J.; Jordan, T. M.; Brennan, M. R.; Kilmer, R. J.; Williams, E. D. Appl. Catal., A 2003, ASAP Article. (41) Potyrailo, R. A.; Lemmon, J. P.; Leib, T. K. Aromatic polycarbonate characterization. U.S. Patent 6,541,264 B1, 2003. (42) Potyrailo, R. A.; May, R. J.; Shaffer, R. E.; Lemmon, J. P.; Wroczynski, R. J. Method for high-throughput fluorescence screening of polymerization reactions. U.S. Patent 6,589,788 B1, 2003.

4680 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

determined using traditional and our nondestructive fluorescence ratiometric techniques. The sensitivity of measurements was increasing with the increase in catalyst selectivity S. The range of selectivity measured by our fluorescence method was 1.13305.8 and was limited by the type of polymerized materials rather than by the measurement method. The RSD in these measurements on individual samples was 1-12.5%. This spread included instrument variability (e1% RSD, measured on reference materials) and sample inhomogeneity. Quantitative Two-Wavelength Fluorescence Imaging of Catalyst Selectivity. Parallel quantitative screening of catalyst selectivity is highly desired for HT applications.5-13,43 Because we were able to determine catalyst selectivity from fluorescence ratiometric measurements, it was attractive to image the whole array of 96 microreactors at once. The availability of cooled CCD cameras, multiple-cavity interference filters with steep band slope, improved near-band rejection, square pass-band peak, and a 100nm difference in the emission wavelengths needed for twowavelength imaging dramatically supported these experiments. Under the used excitation conditions (340 nm), the background fluorescence of the empty 96-microreactor arrays was negligible compared to the polymer fluorescence. Fluorescence images were collected through the 400- and 500nm interference filters at room temperature after the polymerization in the 96-microreactor arrays. Results of these experiments are presented in Figure 8A and B. Performance of imaging (parallel) and spectroscopic (serial) systems was compared by collecting fluorescence spectra from several microreactors (see Figure 8C) that contained solid BPA-PC polymerized with different levels of catalysts selectivity. Figure 8D illustrates the good correlation (R2 ) 0.994) between the serial and parallel HT screening tools. A small deviation from a unity slope in this plot can be easily corrected for because it originates simply from instrumental differences between the imaging and the spectroscopic setups. The differences include transmission of interference filters and fiber-optics and spectral response of photodetectors. Once the correlation between two systems is established, the quantitation of catalyst selectivity can be performed using fluorescence imaging with all advantages of parallel screening tools. Such measurements can be also performed in a straightforward manner during the dwell time periods of the melt polymerization reaction33 because our polymerization reactor has a quartz-viewing window that permits the observation of the whole 96-microreactor array.44 Of course, in molten samples, the fluorescence ratio can be different, which will simply require a separate calibration procedure. CONCLUSIONS In summary, fluorescence spectroscopy and imaging can be an attractive alternative to traditional and HT techniques for rapid evaluation of polymerization catalysts, measurements of MW, and other chemical parameters of polymers if speed, nondestructive analysis, and its sensitivity are critical. Compared to NMR and IR (43) Lettmann, C.; Hinrichs, H.; Maier, W. F. Angew. Chem., Int. Ed. 2001, 40, 3160-3164. (44) Potyrailo, R. A.; Lemmon, J. P.; Flanagan, W. P.; Johnson, R. N. Method and apparatus for high-throughput chemical screening; U.S. Patent 6,572,828 B1, 2003.

Figure 8. Determination of selectivity of melt polymerization catalysts in combinatorial 96-microreactor arrays by fluorescence. Fluorescence images of microreactors through (A) 400- and (B) 500-nm interference filters. (C) Emission spectra from polymers in highlighted column of microreactors. (D) Correlation of I400/I500 values from serial spectroscopic analysis and parallel two-wavelength fluorescence imaging.

analysis of polymer branching, fluorescence measurements are at least 1000 times more sensitive and are done in a noncontact mode without any sample preparation. The noncontact and spatially resolved nature of these measurements could provide previously unavailable insights on the reaction kinetics from in situ observations of combinatorial polymerizations. Although fluorescence does not yield the detailed determinations of chemical structure as being done by NMR, IR, and Raman spectroscopies, this fluorescence screening method can be applicable to diverse types of polymers where fluorescence originates from chromophores situated as in-chain and end-chain groups or from chromophores present in the repeat units in the backbone polymer (45) Hwang, I.-W.; Song, N. W.; Kim, D.; Park, Y. T.; Kim, Y.-R. J. Polym. Sci., B 1999, 37, 2901-2908. (46) Nango, M.; Kimura, Y.; Ihara, Y.; Kuroki, N. Macromolecules 1988, 21, 2330-2335. (47) Lee, H. J.; Nakayama, Y.; Matsuda, T. Macromolecules 1999, 32, 69896995. (48) Evans, C. A.; Miller, S. J. Curr. Opin. Chem. Biol. 2002, 6, 333-338.

structure.28-31 If needed, picosecond time-resolved fluorescence spectroscopy can be also applied to improve further the measurements of polymer branching.45 The use of reactive fluorescent dyes46-48 for the evaluation of polymer branching may also be possible, however, with an addition of an extra step in the postpolymerization screening and possible difficulties in the in situ polymer screening during polymerization. ACKNOWLEDGMENT The authors thank Patrick McCloskey for his help in validating the HT capabilities, William Flanagan and Roger Johnson for design and fabrication of the melt polymerization reactor system, and Oltea Siclovan for technical assistance.

Received for review March 24, 2003. Accepted June 18, 2003. AC034296D

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4681