Time-Lag Focusing MALDI Time-of-Flight Mass Spectrometry for

We report a polymer characterization study by matrix-assisted laser desorption/ionization (MALDI) on a linear time-of-flight instrument equipped with ...
0 downloads 0 Views 252KB Size
Anal. Chem. 1997, 69, 2734-2741

Time-Lag Focusing MALDI Time-of-Flight Mass Spectrometry for Polymer Characterization: Oligomer Resolution, Mass Accuracy, and Average Weight Information Randy M. Whittal, David C. Schriemer, and Liang Li*

Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2

We report a polymer characterization study by matrixassisted laser desorption/ionization (MALDI) on a linear time-of-flight instrument equipped with pulsed ion extraction for time-lag focusing. It is demonstrated that timelag focusing MALDI provides improved mass resolution and mass accuracy over continuous extraction instruments. Oligomer resolution is extended to a much higher mass range than that observed even by continuous extraction reflectron systems. This allows new opportunities to study the chemical composition and determine the molecular weights of individual components in a mixture of higher molecular weight polymers. It is shown that oligomer resolution can be obtained for poly(ethylene glycol) (repeat unit mass of 44) of mass up to 25 000 u and poly(styrene) (repeat unit mass of 104) up to 55 000 u. Mass measurement accuracy of 80 ppm or better is demonstrated, and the relevance to end-group analysis is shown for two derivatives of poly(ethylene glycol) used as slow-release drugs. The analysis of the molecular weight distribution was investigated at several extraction pulse potentials to determine if there was an effect on the relative peak area. We found that the values of the number-average molecular weight (Mn) and the weightaverage molecular weight (Mw) do not change significantly for a poly(styrene) blend with oligomer masses between 2000 and 15 000 u and a polydispersity of 1.155. The values are within the 1.6% standard deviation observed for repeat analyses at the same extraction pulse. Polymer characterization often requires a host of analytical techniques. Among them, mass spectrometry has traditionally played an important role in providing structural and compositional information.1 Since the introduction of matrix-assisted laser desorption/ionization (MALDI), the role of mass spectrometry has been extended to include the determination of molecular weight, end-group mass, and molecular weight distribution of a wide range of polymers of narrow polydispersity.2-10 With the use of an appropriate matrix and sample preparation protocol, high (1) Lattimer, R. P.; Harris, R. E.; Schulten, H. R. In Determination of Molecular Weight; Cooper, A. R., Ed.; Wiley: New York, 1989; pp 391-412. (2) Bahr, U.; Deppe, A.; Karas, M.; Hillenkamp, F. Anal. Chem. 1992, 64, 28662869. (3) Danis, P. O.; Karr, D. E. Org. Mass Spectrom. 1993, 28, 923-925. (4) Whittal, R. M.; Li, L.; Lee, S.; Winnik, M. A. Macromol. Rapid Commun. 1996, 17, 59-64. (5) Lee, S.; Winnik, M. A.; Whittal, R. M.; Li, L. Macromolecules 1996, 29, 30603072.

2734 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

molecular weight polymers with masses up to 1.5 million Da can be analyzed.11 High-speed, direct weight measurement and the use of a minimum amount of solvent are the chief advantages of the MALDI method, compared with other traditional molecular weight characterization methods. For the analysis of a pure polymer, the weight-average and number-average molecular weights can be simultaneously determined using a MALDI instrument with low or moderate resolution, such as a linear timeof-flight mass spectrometer. MALDI can also be used for structural and compositional characterization of polymers, if oligomeric resolution is achieved to allow accurate mass assignment of individual peaks in the spectrum.4,5 Because of the limited mass resolution and mass measurement accuracy that can be attained with a conventional linear or reflectron time-of-flight instrument, this important area of application is confined to the characterization of low molecular weight polymers. It is often desirable to extend the upper mass limit where oligomeric resolution can be obtained. The recent development of time-lag focusing for MALDI time-of-flight mass spectrometers has provided a significant improvement in mass resolution and mass accuracy.12-15 The application of time-lag focusing MALDI to the characterization of several low molecular weight poly(ethylene glycol) derivatives has been recently demonstrated.4,5 However, the strength of the time-lag focusing MALDI instrument lies in its potential to extend oligomer resolution to higher molecular weight polymers. In this work, we describe a study of poly(ethylene glycol) and poly(styrene) by time-lag focusing MALDI with emphasis on illustrating the analytical capability of the technique. Oligomeric resolution is extended to ∼25 000 u for poly(ethylene glycol) and ∼55 000 u for poly(styrene). High mass measurement accuracy is demonstrated with application to the characterization of two slow-release drugs. The amplitude of the extraction pulse (6) Montaudo, G.; Montaudo, M. S.; Puglisi, C.; Samperi, F. Macromolecules 1995, 28, 4562-4569. (7) Lloyd, P. M.; Suddaby, K. G.; Varney, J. E.; Scrivener, E.; Derrick, P. J.; Haddleton, D. M. Eur. Mass Spectrom. 1995, 1, 293-300. (8) Belu, A. M.; DeSimone, J. M.; Linton, R. W.; Lange, G. W.; Friedman, R. M. J. Am. Soc. Mass Spectrom. 1996, 7, 11-24. (9) Jackson, C.; Larsen, B.; McEwen, C. Anal. Chem. 1996, 68, 1303-1308. (10) Danis, P. O.; Karr, D. E.; Xiong, Y.; Owens, K. G. Rapid Commun. Mass Spectrom. 1996, 10, 862-868. (11) Schriemer, D. C.; Li, L. Anal. Chem. 1996, 68, 2721-2725. (12) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995, 67, 1998-2003. (13) Whittal, R. M.; Li, L. Anal. Chem. 1995, 67, 1950-1954. (14) Colby, S. M.; King, T. B.; Reilly, J. P. Rapid. Commun. Mass Spectrom. 1994, 8, 865-868. (15) Vestal, M. L.; Juhasz, P.; Martin, S. A. Rapid Commun. Mass Spectrom. 1995, 9, 1044-1050. S0003-2700(97)00002-4 CCC: $14.00

© 1997 American Chemical Society

potential was adjusted to study the effect on the analysis of the molecular weight distribution. EXPERIMENTAL SECTION Materials. Poly(ethylene glycol)15000 and -20000 were purchased from Polysciences, Inc. (Warrington, PA). The bovine insulin B chain (oxidized form), bovine insulin, equine cytochrome c, bovine trypsinogen, bovine carbonic anhydrase II, bovine serum albumin, poly(oxyethylene) bis(acetaminophen), poly(oxyethylene) bis(ephedrine), and poly(ethylene glycol)3350 and -10 000, were purchased from Sigma Chemical Co. (St. Louis, MO). Poly(styrene)3250, -5050, -7000, -9240, and -11600 were purchased from Showa Denko KK (Tokyo, Japan). Poly(styrene)18700, -32660, and -45000 were purchased from Aldrich (Milwaukee, WI). All polymers were analyzed without further purification. The HP G2052A peptide standard containing oxytocin, arginine-8-vasopressin, angiotensin I, somatostatin, chicken atrial natriuretic peptide, human r insulin, and r hirudin was supplied by HewlettPackard Co. (Palo Alto, CA). The matrices 2-(4-hydroxyphenylazo)benzoic acid (HABA), 2,5-dihydroxybenzoic acid (DHB), trans-retinoic acid, sinapinic acid, and trans-3-indoleacrylic acid (IAA) were purchased from Aldrich (Milwaukee, WI) and used without purification. The cationizing reagents silver nitrate (Terochem Laboratories Ltd., Edmonton, Canada) and sodium chloride (BDH Laboratories, Toronto, Canada) were reagent grade. Analytical grade tetrahydrofuran was purchased from VWR Scientific (Toronto, Canada), treated with potassium hydroxide, and then distilled under nitrogen in the presence of sodium metal. HPLC grade methanol was purchased from EM Science (Gibbstown, NJ). Analytical grade 1,4-dioxane was purchased from BDH Laboratories (Toronto, Canada). Water used throughout was from a Milli-Q Plus Ultrapure water system (18.2 MΩ‚cm, Millipore, Bedford, MA). Sample Preparation. All polymer samples were prepared in glass vials. Poly(styrene) samples were dissolved in dry tetrahydrofuran to prepare stock solutions at concentrations between 0.1 and 7 mM.11 trans-Retinoic acid was used as the matrix, which was prepared as a 0.15 M solution in tetrahydrofuran. An aliquot of poly(styrene) solution was added to the matrix solution to give a final polymer concentration of 50-300 µM. To 100 µL of matrix/polymer solution was added 1 µL of silver nitrate saturated in ethanol. A 1 µL aliquot was added to the MALDI sample probe and allowed to dry. Poly(ethylene glycol)15000 and -20000 samples were dissolved in 1:1 methanol/water at a concentration of 1 mM. The poly(oxyethylene) bis(acetaminophen) and poly(oxyethylene) bis(ephedrine) were dissolved in methanol at a concentration of 1 mM. HABA was dissolved in 1,4-dioxane at a concentration of 50 mM. DHB was prepared as a 100 mM solution in 1:1 methanol/water, and IAA was prepared as a 200 mM solution in methanol. The polymer solution and matrix solution were mixed to give a final polymer concentration of 100 µM. To 100 µL of matrix/polymer solution was added 1 µL of 20 mM sodium chloride (aqueous). A 1 µL aliquot of the mixture was added to the MALDI sample probe and allowed to dry. Instrumentation. Mass spectra were collected on the previously described MALDI instrument, modified for operation up to 30 kV.13 The first extraction plate was redesigned, and a grid was placed on the repeller side. A single dc power supply was used to set the potential on the repeller and first extraction plate and the potential on the second extraction plate through a voltage

divider. Under normal operation, the power supply was set to 20 kV dc. In conjunction, a high-voltage pulser built in-house or a Hewlett-Packard pulser was used to generate the delayed extraction pulse. The extraction pulse and time lag were varied as required to focus the mass range of interest. A nitrogen laser (VSL-337ND, Laser Science, Inc., Newton, MA) with a 3 ns pulse width was used for desorption at 67.5° to the probe surface normal. A neutral density filter was used to attenuate the laser energy that reached the sample surface. A mass filter was added to the system. The mass filter can prevent saturation of the microchannel plate detector by deflecting matrix ions out of the usual ion trajectory, reducing the ion current from low-mass ions. Data Collection and Mass Calibration. The mass spectra were recorded using a LeCroy 9350M digital oscilloscope at a sampling rate from 5 × 108 samples s-1 to 1 × 109 samples s-1 in conjunction with Hewlett-Packard MALDI software. Peak centroids were determined with the software, and the mass calibration was done using a two-point linear calibration (i.e., m ) a(t + b)2, where t is the flight time, m is the mass of the ion, and a and b are the calibration coefficients). Calibrated spectra were transferred to a PowerMac and processed using Igor Pro (WaveMetrics, Inc., Lake Oswego, OR) data analysis software. Calibration of the instrument was done with proteins or peptides using sinapinic acid as the matrix. Peptide and protein samples were prepared in 0.5 mL Eppendorf microcentrifuge tubes. First, an aliquot of sinapinic acid in acetone (20 mg mL-1) was applied to the probe and allowed to air-dry, followed by an aliquot of protein or peptide solution mixed 1:1 (v/v) with a saturated sinapinic acid solution (33% acetonitrile/water (v/v)) that was allowed to air-dry. The low-mass poly(styrene) blend spectra were calibrated using the Hewlett-Packard G2052A peptide standard and cytochrome c. The poly(styrene)18700, -32660, and -45000 spectra were calibrated using cytochrome c, carbonic anhydrase, and bovine serum albumin. The poly(ethylene glycol)15000 spectra were calibrated using bovine insulin B chain (oxidized form) and cytochrome c. The poly(ethylene glycol)20000 spectrum was calibrated using cytochrome c and trypsinogen. The protein average mass was used for all calibrations. The poly(oxyethylene) bis(acetaminophen) and poly(oxyethylene) bis(ephedrine) spectra were calibrated using poly(ethylene glycol)3350 with the sample preparation method already described. The oligomer average mass was used for calibration. RESULTS AND DISCUSSION Mass Resolution. Figure 1 shows four spectra of poly(ethylene glycol)15000 using a number of matrices and collected under different experimental conditions for comparison. Figure 1 is separated into two mass ranges: the top of the figure shows the region between m/z 0 and 10 000, and the bottom between m/z 10 000 and 15 000. Figure 1A was obtained using HABA as matrix and continuous extraction; individual oligomers are not resolved above 9000 u. In contrast, the individual oligomers are well resolved in Figure 1B-D, where they were collected using time-lag focusing with DHB, IAA, and HABA as matrices, respectively. Figure 1 shows that the type of matrix used can significantly affect the apparent mass resolution in time-lag focusing MALDI. At 12 700 u, the resolution is 700, 1060, and 1400 fwhm for Figure 1B-D, respectively. Lower mass oligomers are not resolved with DHB, whereas oligomers are resolved across the whole spectrum using IAA and HABA as matrices. In addition, the overall detection sensitivity is higher with HABA or IAA than Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2735

Figure 1. MALDI mass spectra of poly(ethylene glycol)15000 obtained by operating the instrument in (A) the continuous extraction mode with HABA as the matrix and sodium cationization and (B), (C), and (D) the time-lag focusing extraction mode with (B) DHB, (C) IAA, and (D) HABA as the matrices and sodium cationization. For (D), Mn ) 9380 and D ) 1.20 by MALDI; the supplier certifies Mn ) 12 500 and D ) 1.20 by gel permeation chromatography.

with DHB. HABA provides a flat baseline, extending to the lowmass region, while the baselines elevate in the low-mass region for the spectra obtained with DHB and IAA. Intense matrix clusters are observed with IAA extending to m/z 3000, obscuring polymer peaks in this mass range. The extent of clustering observed with IAA could be reduced with a reduction in laser flux but would result in complete loss of the polymer signal. The reason for the difference in resolution for polymer spectra obtained with different matrices is not obvious using our linear time-lag focusing MALDI system. All the matrix preparations gave densely packed microcrystals on the probe surface. Well-packed microcrystals give improved mass accuracy, resolution, and shotto-shot reproducibility over preparations that produce large crystals.16,17 DHB required twice the laser flux used for HABA and IAA, and the shot-to-shot reproducibility was significantly poorer; lower sensitivity and resolution can be expected. However, the sensitivity and shot-to-shot reproducibility of HABA and IAA are similar. We speculate that the high abundance of matrix clusters observed with IAA indicates a higher charge density in the desorption plume, leading to increased Coulombic repulsion and an unanticipated spatial distribution that cannot be compensated with time-lag focusing, i.e., there is no longer an exact relationship between an ion’s position and its velocity at the time the extraction pulse is applied and slight peak broadening results. To illustrate the resolving power of time-lag focusing MALDI for higher molecular weight polymers, we examined two types of (16) Vorm, O.; Roepstorff, P.; Mann, M. Anal. Chem. 1994, 66, 3281-3287. (17) Dai, Y.; Whittal, R. M.; Li, L. Anal. Chem. 1996, 68, 2494-2500.

2736 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

polymers with different hydrophobicity and repeat unit mass. Figure 2 shows the mass spectrum of poly(ethylene glycol)20000 from Polysciences using HABA as the matrix. The extraction pulse potential was set to maximize resolution for ions at m/z 23 000, where the resolution is ∼1000 fwhm (see Figure 2B). Despite the high mass of the polymer and low mass of the repeat unit (44 u), oligomer peaks are still well resolved. Figure 2 illustrates that this sample has a bimodal mass distribution, with one distribution centered at m/z ∼23 000 and the other centered at m/z ∼18 500. The bimodal distribution was also observed with poly(ethylene glycol)23600 from Polymer Laboratories using continuous extraction in both linear and reflectron modes6 but not with Boehringer Mannheim’s poly(ethylene glycol)23000.2 However, under continuous extraction conditions with the same matrix and using a reflectron for energy compensation, the oligomer peaks are not resolved.6 Note that the bimodal distribution is not detected by gel permeation chromatography. While MALDI can distinguish bimodal and unimodal mass distributions, only time-lag focusing MALDI appears able to resolve poly(ethylene glycol) oligomers in this mass range. The analysis of less polar polymeric systems, such as poly(styrene), is of considerable interest to the polymer industry. Figure 3A-C gives mass spectra of narrowly disperse poly(styrene)18700, -32660, and -45000 obtained with time-lag focusing MALDI. The pulse voltage was set to optimize the resolution at the center of the distribution of each compound. In all cases, oligomer ions are well resolved. The mass resolution for poly(styrene)18700 is 725-905 fwhm, for poly(styrene)32660 the

Figure 2. (A) MALDI mass spectrum of poly(ethylene glycol)20000 using HABA as matrix with sodium cationization and time-lag focusing set to optimize the resolution of ions at m/z 23000. Mn ) 11 290 and D ) 1.48; the supplier certifies Mn ) 21 344 and D ) 1.06 by gel permeation chromatography. (B) Expansion of (A) in the high-mass region of the spectrum.

resolution is 870-1065 fwhm, and for poly(styrene)45000 the resolution is 550-670 fwhm. Figure 3D is an expansion of Figure 3C at the high-mass end of the distribution, illustrating that oligomer peaks are well resolved at masses beyond 50 000 u. The spectrum of poly(styrene)45000 represents, perhaps, the best mass resolution that has been reported by any mass spectrometric technique for polymers in this mass range. The data shown in Figure 3D also have a significant implication for biopolymer analysis. The results illustrate that time-lag focusing is still effective at least up to m/z 55 000. It is likely that the chemistry of high-mass proteins and oligonucleotides (adduction of alkali metals, matrix, etc.), combined with in-source fragmentation resulting in the loss of low-mass neutral species (e.g., loss of water or ammonia), limits the apparent resolution of these molecules. In polymer analysis by MALDI, the actual value of the upper mass limit where oligomers are resolved is dependent on several factors, including the number of components present in a polymer system, the mass(es) of the repeat unit(s), and the matrix/sample preparation method. For the latter, it is our experience that, whenever possible, the formation of two or more cation adducts should be avoided. The addition of a preferred cation suppresses other cationization and can improve the quality of the spectrum by reducing peak overlap. The above examples also indicate that the apparent resolution gradually decreases as the ion mass increases. The ultimate resolution is not limited by the isotopic distribution. For example, for poly(ethylene glycol) at m/z 23 000, we expect ∆m ) 10 u fwhm according to the calculated isotope distribution; thus, the expected resolution is 2300 fwhm. For poly(styrene) at m/z 50 000, the expected resolution is ∼3000 fwhm.

The resolution is also not limited by electronic jitter during signal averaging, since the electronic jitter is less than 1 ns in our system and the peak width is ∼36 ns for poly(ethylene glycol)20 000 at m/z 23 000 and ∼60 ns for poly(styrene)45000. As the mass of an ion increases, it is necessary to increase the extraction pulse potential to bring the ions into focus at the detector. As the extraction pulse potential increases, spatial aberrations (along the flight path axis) result in more peak broadening than at lower extraction potentials. The spatial aberrations can arise from several possible sources including the spatial distribution of the sample and the radial distribution of ions in the desorption plume. Mass Measurement Accuracy. Obtaining high mass accuracy is important in determining the end-group mass of polymers. We note that the mass of each oligomer in Figures 1 and 2 corresponds to sodium-cationized HO-(CH2CH2O)n-H, not to H-(CH2CH2O)n-H. For example, three peaks in Figure 2 are labeled with observed masses of 21 496.43, 23 037.57, and 24 314.32 u. If these peaks are sodium-cationized HO-(CH2CH2O)n-H, where n 487, 522, and 551, then the calculated masses are 21 495.04, 23 036.91, and 24 314.46 u, respectively. The mass errors are +65, +29, and -6 ppm, respectively. For Figures 1 and 2, sodium chloride was added during the MALDI sample preparation to ensure sodium cationization. If potassium chloride is added, the observed mass of each oligomer increases by 16 u (data not shown). Clearly, the oligomers are predominantly of the form HO-(CH2CH2O)n-H. The advantage of high-resolution and high-accuracy analysis of polymers is illustrated in the MALDI analysis of poly(ethylene glycol)20000. Inspection of the low-mass region of Figure 2 shows that there is a second type of polymer present in the sample. Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2737

Figure 3. MALDI mass spectra of (A) poly(styrene)18700, (B) poly(styrene)32660, (C) poly(styrene)45000, and (D) the expanded spectrum of poly(styrene)45000 using retinoic acid as the matrix with silver cationization. For (A), Mn ) 17 430 and D ) 1.008; the supplier certifies Mn ) 18 100 and D ) 1.066. For (B), Mn ) 29 350 and D ) 1.004; the supplier certifies Mn ) 30 000 and D ) 1.025. For (C), Mn ) 44 930 and D ) 1.008; the supplier certifies Mn ) 45 550 and D ) 1.053. The supplier used gel permeation chromatography for the analysis.

Figure 4. MALDI mass spectrum of poly(ethylene glycol)20000 using HABA as matrix with sodium cationization and time-lag focusing extraction set to optimize the resolution of ions at m/z 12 000.

Figure 4 is an expansion of a spectrum of poly(ethylene glycol)20000, obtained by adjusting the extraction pulse potential to achieve optimal resolution at m/z 12 000. The peaks at m/z 11 142.56 (calculated 11 142.48) and 11 935.54 (calculated 11 935.44) are from sodium-cationized HO-(CH2CH2O)n-H, where n ) 252 and 270, respectively. The peaks at m/z 11 126.33 and 11 919.28 likely correspond to H-(CH2CH2O)n-H, with calculated masses of 11 126.48 and 11 919.44 u, respectively. The sodiated form of this distribution has been verified, as the addition of potassium chloride also generated oligomer mass increases of 16 u. This second polymer distribution is observed in the low-mass region of the spectrum. The intensities of the oligomer peaks from the second distribution are about 80-90% of the intensities of the dihydroxy 2738 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

terminated oligomers up to m/z ∼15000. Beyond this point, the peak intensity gradually decreases until m/z ∼21500, where the oligomers from the second distribution are no longer detected. Extensive studies of poly(ethylene glycol) fragmentation patterns suggest that it is not possible to produce predominant fragment ions that arise from the poly(ethylene glycol) parent ion with a loss of 16 or 16 + 44n u (where n is an integer number of oligomers).18,19 The MALDI results shown here demonstrate that this sample contains two major poly(ethylene glycol) components with different end groups. Since limited information on the synthesis of this sample is provided by the supplier, the exact nature of the end group for the second distribution is unknown. The above example illustrates one major limitation of the current MALDI method for structural analysis. Some prior knowledge of the chemistry of the polymeric system under investigation is required. However, the MALDI method excels in its ability to provide structural confirmation and information on the distribution of individual components in mixtures. This is particularly true when the accuracy of mass measurement is sufficiently high. To demonstrate the enhanced capability of the time-lag focusing MALDI technique, we examined several poly(ethylene glycol) derivatives of pharmaceutical significance. Poly(ethylene glycol) derivatives where one or two of the end groups are functional molecules such as drugs, proteins, or liposomes have been developed for controlled release and selective delivery of drugs.20,21 The full characterization of these compounds is important in studying their function and metabolism and in quality (18) Lattimer, R. P. J. Am. Soc. Mass Spectrom. 1992, 3, 225-234. (19) Mattern, D. E.; Hercules, D. M. Anal. Chem. 1985, 57, 2041-2046. (20) Poly(Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications; Harris, J. M., Ed.; Plenum Press: New York, 1992.

Table 1. Mass Measurement Accuracy of Selected Oligomer Peaks of Poly(oxyethylene) Bis(ephedrine) in Figure 5Aa mass (M + Na)+ (u) n

two labels

obsd

mass error (ppm), two labels

60 70 80 90

2978.66 3419.19 3859.72 4300.26

2978.66 3419.03 3859.70 4299.99

0 -47 -5 -63

calcd nb 63 73 83 93

one

labelc

2963.59 3404.13 3844.66 4285.20

a Externally calibrated with poly(ethylene glycol)3350. b n is the number of oxyethylene repeat units in the polymer. c Label refers to the ephedrine moiety as the end group of the polymer.

Table 2. Mass Measurement Accuracy of Selected Oligomer Peaks of Poly(oxyethylene) Bis(acetaminophen) in Figure 5Ba mass (M + Na)+ (u) calcd nb

Figure 5. MALDI spectra of (A) poly(oxyethylene) bis(ephedrine) and (B) poly(oxyethylene) bis(acetaminophen) using HABA as matrix and sodium cationization. (A) Mn ) 3670 and D ) 1.007. (B) Mn ) 3520 and D ) 1.014.

40 50 60 70 80 90 100

one

labelc

1936.29 2376.83 2817.36 3257.90 3698.43 4138.97 4579.50

mass error (ppm)

n

two labels

obsd

one label

two labels

37 47 57 67 77 87 97

1937.28 2377.82 2818.35 3258.89 3699.42 4139.96 4580.49

1937.32 2377.89 2818.13 3258.65 3699.24 4140.07 4580.69

+532 +446 +273 +230 +219 +266 +260

+21 +29 -78 -74 -49 +27 +44

a Externally calibrated with poly(ethylene glycol)3350. b n is the number of oxyethylene repeat units in the polymer. c Label refers to the acetaminophen moiety as the end group of the polymer.

Chart 1

control and regulatory approval. Mass spectrometry can play a significant role in the chemical analysis of these molecules.22,23 Figure 5 shows the mass spectra of poly(oxyethylene) bis(ephedrine) and poly(oxyethylene) bis(acetaminophen), and their structures are shown in Chart 1. The desired product has ephedrine or acetaminophen linked to the polymer at both ends. Dosage determination is partially dependent on the success of derivatization of the polymer and on the number-average molecular weight, Mn. MALDI can determine if the polymer is derivatized on one or both ends and can determine Mn simulta(21) Roberts, M.; Scholes, D. F. In Chemical Aspects of Drug Delivery Systems; Karsa, D. R., Stephenson, R. A., Eds.; The Royal Society of Chemistry: Cambridge, 1996; pp 89-96. (22) Vestling, M. M.; Murphy, C. M.; Keller, D. A.; Fenselau, C.; Dedinas, J.; Ladd, D.; Olsen, M. Drug Metab. Dispos. 1993, 21, 911-917. (23) Chowdhury, S. K.; Doleman, M.; Johnston, D. J. Am. Soc. Mass Spectrom. 1995, 6, 478-487.

neously. This is trivial for poly(oxyethylene) bis(ephedrine). As shown in Table 1, the mass difference between a singly labeled polymer with n oxyethylene repeat units and a doubly labeled polymer with n - 3 oxyethylene units is 15.06 u. The data shown in Table 1 confirm that the sample has ephedrine linked to both ends of the polymer. In a similar fashion, it could be shown that unlabeled polymer with n + 3 oxyethylene units generates a mass difference with the singly labeled product of -15.06 u. Note that the inset to Figure 5A shows a peak at 15.9 u above the main distribution (3831.6 u, likely M + K+) and a peak of unknown origin at 24 u above the main distribution (3839.7 u). This peak is not from singly labeled polymer, which should have a mass of 3844.7 u. A MALDI system equipped with continuous extraction can distinguish nonlabeled, singly ephedrine-labeled, and doubly ephedrine-labeled poly(oxyethylene). However, when acetaminophen is the labeling agent, the demand on the mass accuracy of the instrument is much higher. Table 2 shows that the mass difference between the singly labeled poly(oxyethylene) with n repeat units and the doubly labeled poly(oxyethylene) with n 3 repeat units is only -0.99 u. In the mass range of interest, continuous extraction MALDI cannot provide sufficient mass accuracy to distinguish the two possibilities. With time-lag focusing MALDI, we can confirm that two acetaminophen groups are attached to the polymer. For example, the peak with an observed m/z of 3699.24 could represent a sodium-cationized oligomer with 80 oxyethylene repeat units and one acetaminophen end group. The expected mass would be 3698.43 u. However, if this peak represents a sodium-cationized oligomer with two acetaminophen end groups and 77 oxyethylene repeat units, then Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2739

the expected mass is 3699.42 u. If there are two acetaminophen end groups, the mass error is -49 ppm, whereas, if there is one acetaminophen end group, the mass error is +219 ppm. Since the average error in analysis by time-lag focusing MALDI for externally calibrated spectra is usually (70 ppm, two acetaminophen groups must be attached. This type of analysis is not possible in a conventional continuous extraction instrument, where external mass accuracy in this mass range is usually 500-1000 ppm. Note that unlabeled polymer with n + 3 repeat units generates a mass difference with the singly labeled product of 0.99 u (1.98 u difference with the nearest doubly labeled product). In a MALDI spectrum of the poly(oxyethylene) bis(acetaminophen), the singly labeled and unlabeled products, if present, would cause peak fronting and decreased resolution, in addition to a peak centroid shift. This is not observed in the spectrum, suggesting that these species are not present in significant abundance. Their presence in lower abundance, however, cannot be ruled out. The attachment of acetaminophen to poly(ethylene glycol) is expected to occur through the phenolic group of acetaminophen.24,25 The MALDI method does not provide information as to how the acetaminophen moiety is attached to poly(ethylene glycol). Other analytical methods are required in this regard. Tandem MS may provide additional structural information, but it is often limited to low-mass ions. Selective decomposition by laser desorption, followed by electron impact or multiphoton ionization, can provide useful structural information on end-group labeling.26 An alternative to the mass spectrometric approach is to use NMR. For poly(ethylene glycol) derivatives with pharmaceutical utility, the molecular weight of the poly(ethylene glycol) chain is often less than 25 000.21 NMR can determine the necessary structural information in this range21 and give the value of Mn but may not be able to distinguish singly labeled and unlabeled poly(ethylene glycol) present as a mixture in the doubly labeled sample. It should be cautioned that using the average molecular weight of the polymer distribution to deduce information on labeling chemistry is limited. Poly(ethylene glycol) can undergo hydrolysis or other decomposition processes during synthetic modification that can result in a less than expected shift in the average molecular weight distribution.5 Molecular Weight Distribution. Time-lag focusing is a massdependent initial energy compensation method. To optimize mass resolution for ions at different mass, the extraction pulse potential or time lag between ion desorption and ion extraction needs to be adjusted. Ions whose m/z is less than the optimally focused m/z receive more than the optimum energy from the extraction pulse, whereas ions with an m/z that is higher than the optimally focused m/z receive less than the optimum energy from the extraction pulse. In polymer analysis by MALDI, one of the major objectives is to measure the weight-average and number-average molecular weights, which are dependent on the relative peak area of each oligomer and the oligomer mass. Thus, it is important to determine if the extraction pulse potential affects the measurement of relative peak area. Figure 6 shows the MALDI spectra of a blend of narrowly disperse poly(styrene) compounds to give an overall polydispersity of ∼1.155. The sample used is a mixture of narrowly disperse (24) Weiner, B. Z.; Havron, A.; Zilkha, A. Isr. J. Chem. 1974, 12, 863-872. (25) Zalipsky, S.; Zilkha, A. Eur. Polym. J. 1983, 19, 1177-1183. (26) Schriemer, D. C.; Li, L. Anal. Chem. 1996, 68, 250-256.

2740 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

Figure 6. MALDI mass spectra of a poly(styrene) blend obtained by time-lag focusing with the extraction pulse potential set to optimize resolution at (A) m/z 4000 and (B) m/z 11 000. The matrix used was retinoic acid with silver cationization. Table 3. Mn and Mw Determined for a Blend of Poly(styrene)3250, -5050, -7000, -9240, and -11600 at Different Time-Lag Focusing Extraction Pulse Potentials optimization of extraction pulse (m/z)

obsd Mn

obsd Mw

4 000 11 000 20 000 25 000

7420 7390 7470 7260

8540 8420 8510 8330

poly(styrene)3250, -5050, -7000, -9240, and -11600. In Figure 6A, the extraction pulse potential was adjusted to optimize resolution for ions at m/z 4000. In Figure 6B, the extraction pulse potential was adjusted to optimize resolution for ions at m/z 11 000. The extraction pulse potential can significantly influence the shape of the distribution. However, the relative peak area of each oligomer remains approximately constant. In Figure 6A, the resolution decreases with increasing mass, requiring a drop in peak height to maintain the peak area. Table 3 summarizes the Mn and Mw values determined for the poly(styrene) blend at four extraction potentials. The standard deviation in determining Mn and Mw at a given extraction potential is 1.6%. The average standard deviation between extraction potentials is 1.2% and 1.1% for Mn and Mw, respectively. Thus, the extraction potential does not have a statistically significant effect on the relative peak area for the polydispersity studied here. The same conclusion holds for all narrowly disperse polymers studied so far with this instrument. The values of Mn and the polydispersity (D) determined by MALDI for the other polymers reported in this work are provided

in the figure captions. We note that, for the poly(styrene) samples, there is good agreement between MALDI and gel permeation chromatography results. However, for the poly(ethylene glycol) samples in Figures 1 and 2, the agreement between MALDI and gel permeation chromatography is poor. The problems associated with gel permeation chromatography for accurate measurement of molecular weight and molecular weight distribution of watersoluble polymers are well known.27 For poly(ethylene glycol)15000 and -20000, we observe significantly lower values of Mn and Mw. The sensitivity of detection is good for the poly(ethylene glycol)15000 in Figure 1D. We do not observe any oligomers above 15 000 u, and thus we find Mw is only 11 230. We note that poly(ethylene glycol)10000 from Sigma gives the same MALDI spectrum as poly(ethylene glycol)15000 from Polysciences. For the poly(ethylene glycol) 20000 sample it appears likely that differences in detection efficiency between low and high mass oligomers allows significant skewing of the relative peak area, overstating the relative peak area of low-mass oligomers. On the other hand, gel permeation chromatography would underestimate the relative contribution of low mass oligomers if they are present in low concentration. The study of more polydisperse polymers using time-lag focusing is currently underway in this laboratory and will be reported in the future.28 The advantage of MALDI for the determination of the molecular weight distribution of polymers is its ability to analyze the molecular weight of individual polymers in mixtures, providing that oligomer resolution is obtained. Such speciation information is more useful for quality control, reaction monitoring, and studying the relationship between the molecular weight and polymer properties. The ability to achieve oligomer resolution over an extended mass range by time-lag focusing will increase the utility of MALDI for molecular weight analysis of specific distributions in polymer mixtures.

advantage of the high-resolution capability of the instrument. With high resolution, individual components can be resolved in polymer mixtures, and detailed compositional information can be obtained for higher mass polymers. This is evident in the MALDI analysis of poly(ethylene glycol)20000 from Polysciences, where an unexpected, second polymer component is detected. Mass resolution between 550 and 670 fwhm was demonstrated for poly(styrene)45000 with molecular weights up to at least 55 000. This result indicates that time-lag focusing is effective up to m/z at least 55 000. Time-lag focusing MALDI provides high external mass measurement accuracy with a mass error that is usually 80 ppm or less for each oligomer. The mass accuracy can be used to provide structural confirmation, if one has prior knowledge of the polymer chemistry. This would be true for applications in reaction monitoring and product quality control. In addition, MALDI is poised to play a significant role in the development of polymers with pharmaceutical significance. These include poly(ethylene glycol) derivatives used for controlled drug delivery or slowrelease drugs. The analysis of poly(oxyethylene) bis(ephedrine) and poly(oxyethylene) bis(acetaminophen), shown in this work, demonstrates the unique ability of the MALDI technique to provide molecular speciation. Speciation is not readily accessible by conventional methods, such as NMR, since the separation of structurally similar polymer components in a mixture is difficult. Yet such information is often required for regulatory approval and for checking subtle batch-to-batch variations. Finally, we have shown that precise determination of the number-average and weight-average molecular weights of narrow polydisperse polymers can be readily achieved with time-lag focusing MALDI. Moreover, the precision is independent of the magnitude of the extraction pulse potential used for focusing ions for the narrow polydisperse polymers studied herein.

CONCLUSIONS We demonstrate that time-lag focusing time-of-flight mass spectrometry provides improved resolution and mass accuracy for polymer analysis compared to linear or reflectron MALDI systems equipped with continuous extraction. The upper mass limit of oligomer resolution is significantly extended. It is illustrated that, although the instrumental resolution of time-lag focusing time-of-flight mass spectrometry is improved, the apparent resolution of polymer spectra is still dependent on the matrix and sample preparation. Thus, one needs to pay particular attention in developing optimal matrix formulations to take full

ACKNOWLEDGMENT This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) through its Industrially Oriented Research Grant Program and by the Polymer Structure and Property Research Program of the Environmental Science and Technology Alliance of Canada. R.M.W. thanks NSERC for the postgraduate scholarship, and D.C.S. thanks the Killam Trust for a predoctoral fellowship.

(27) Styring, M. G.; Hamielec, A. E. In Determination of Molecular Weight; Cooper, A. R., Ed.; Wiley: New York, 1989; pp 263-300. (28) Schriemer, D. C.; Li, L. Submitted to Anal. Chem.

AC970002A

Received for review January 2, 1997. Accepted April 17, 1997.X

X

Abstract published in Advance ACS Abstracts, June 1, 1997.

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2741