Polymer 42 32001) 9949±9953 www.elsevier.com/locate/polymer ªCovalent cationization methodº for the analysis of polyethylene by mass spectrometry Barry J. Bauer*, William E. Wallace, Bruno M. Fanconi, Charles M. Guttman Polymers Division, National Institute of Standards and Technology, 100 Bureau Dr., Stop 8541, Gaithersburg, MD 20899-8541, USA Received 31 May 2001; received in revised form 23 July 2001; accepted 25 July 2001 Abstract Polyethylene and other polyole®ns have not been amenable to mass spectrometric characterization of molecular mass distribution due to the ineffectiveness of conventional methods of cationization. The lack of polar groups, unsaturation, and aromaticity excludes cationization methods that are used commonly for polymers. A new method is introduced in which an organic cation is covalently bonded to the polymer to produce the necessary ionization for successful matrix-assisted laser desorption/ionization 3MALDI) time-of-¯ight mass spectrometry. A strong MALDI signal results from modi®ed polymers that give no response in their unmodi®ed form. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: Matrix-assisted laser desorption/ionization; Polyethylene; Polyole®n 1. Introduction such as by the extraction of sodium ions from glassware for poly3methyl methacrylate) [3], or by addition of metal salts Matrix-assisted laser desorption/ionization 3MALDI) such as silver or copper to polystyrene [4]. Such procedures time-of-¯ight 3TOF) mass spectrometry 3MS) of synthetic can provide MALDI signals, but the total amount and polymers can yield qualitative information about end groups stability of the cationization is poorly understood. and repeat units, and also quantitative information, such as Polyethylene has been cationized by this approach with molecular masses and molecular mass distributions 3MMD) silver, but only for technical waxes, M , 2000 g mol21 [5], [1,2]. However, such measurements can only be made on a saturated crude oil fractions [6], long chain alkanes [7], and select group of polymers. Polyole®ns, which include poly- for hydrogenated polybutadiene, M , 2000 g mol21 [8]. ethylene 3PE), polypropylene, polyisobutylene, poly3methyl Higher molecular mass characterization was not possible. pentene), etc. have not been analyzed by MALDI for Polyethylene with `living' Ziegler±Natta catalyst has been molecular masses in excess of 2000 g mol21. analyzed for M , 2000 g mol21 [9], but was speci®c to the Polyole®ns cannot be analyzed by MALDI using existing system and not possible for higher molecular masses. techniques due to the inef®ciency of the cationization Polyole®ns dominate the market share of synthetic process. Cationization by metals such as sodium, potassium, polymer production. While it is one of the polymer classes silver, copper, etc. involves association of the cations with with the longest production history, new developments in functional groups on the polymers. Examples of functional metallocene catalysts have caused a great resurgence of groups include the carbonyl groups of methacrylate activity and have resulted in a wide variety of new applica- polymers, the double bonds of diene polymers, and the tions and increased raw material production [10]. The new phenyl groups of styrenics. Polymers that lack polar, catalysts can provide unprecedented control of polymer unsaturated, or aromatic groups have not yet been analyzed molecular mass and MMD, co-polymerization control, by MALDI. stereochemistry, etc. MALDI has provided valuable infor- Cationization of MALDI samples of synthetic polymers mation on the nature of the polymerization process and the is provided through addition of metal cations to the resultant structure for many other polymer types, but not for polymers. The addition can be inadvertent and uncontrolled polyole®ns. If polyole®ns could be routinely analyzed by MALDI, development time for new catalysts and polymers * Corresponding author. Tel.: 11-301-975-6849; fax: 11-301-975-3928. could be greatly reduced. E-mail address: [email protected] 3B.J. Bauer). In this report, we describe a novel method of placing 0032-3861/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S0032-3861301)00539-0 9950 B.J. Bauer et al. / Polymer 42 32001) 9949±9953 Fig. 1. Difference FTIR spectra of brominated PE 2 PE 3dashed curve) and TPP PE 2 PE 3solid curve). Numbers on plot are values of spectral features in cm21. charges that are covalently bonded to the PE. The modi®ed Dithranol or all-trans retinoic acid were each shown to PE produces strong MALDI signals capable of being perform satisfactorily as MALDI matrices. The data processed by conventional MALDI data analysis presented were taken using all-trans retinoic acid. Toluene techniques. or xylene were used as the solvents and heated to .1008Cto put the analyte into solution. Solution concentrations were typically 40 mg/ml for the matrices and 1 mg/ml for the 2. Experimental analyte. The two solutions were mixed in a 1:1 ratio and hand spotted from a glass micropipette onto the steel target. The chemical modi®cation involves two synthetic steps The pipettes were heated to limit cooling of the solution on a PE, NIST standard reference material, SRM 2885. A during spotting. The dried droplets showed a ®nely divided solution of PE in toluene is prepared by heating the mixture crystalline structure. to 1108C. Excess bromine 3.10 £ molar excess) is added MS was performed on a Bruker REFLEX II instrument in and kept at 1108C for 4 h and then the reaction product is re¯ectron mode using delayed extraction.1 Ions were precipitated into methanol. The polymer is redissolved in generated using a 337 nm wavelength nitrogen laser with xylene and excess triphenyl phosphine 3TPP) 3.10 £ molar a pulse duration of the order of 3 ns and an average energy excess) is added. It is then precipitated into methanol and of approximately 5 mJ spread over a spot size of dried in vacuum. 200 £ 50 mm2. The laser power used was slightly higher Products of the reaction of PE with bromine and TPP than typically used for other polymers 3e.g. polystyrene were analyzed by Fourier transform infrared spectroscopy. with silver cationization). All data shown are for positive The samples were puri®ed by precipitation into a large ions: negative ion spectra typically produced only matrix excess of methanol to remove any unreacted TPP. The infra- ions and their clusters, e.g. M2,2M2, etc. The instrument red samples were ®lms, approximately 0.015 cm thick, was periodically calibrated with bovine insulin using the prepared by hot pressing at approximately 1308C. Spectra [M 1 H]1 and [M 1 2H]12 peaks which provide uncertain- 21 were recorded at 2 cm resolution on a Nicolet Magna ties of less than a few mass units. System 550 FTIR equipped with a DTGS detector.1 The co-addition of 100 scans gave adequate signal-to-noise to perform spectral subtraction to enhance spectral features 3. Results and discussion resulting from chemical modi®cation of the samples. 3.1. FTIR 1 Certain commercial materials and equipments are identi®ed in this paper in order to specify adequately the experimental procedure. In no Fig. 1 shows the difference spectra obtained by subtract- case does such identi®cation imply recommendation by the National Insti- tute of Standards and Technology nor does it imply that the material or ing the spectrum of SRM 2885 from the brominated PE, and equipment identi®ed is necessarily the best available for this purpose. from the product of addition of TPP to the brominated PE. B.J. Bauer et al. / Polymer 42 32001) 9949±9953 9951 Fig. 2. MALDI±TOF±MS of chemically modi®ed PE, in this case NIST Standard Reference Material 2885. The normalized absorbance curves are subtracted from each The vinyl content of SRM 2885 is determined from the other to emphasize the differences due to the chemical 909 cm21 band based on an integrated absorbance obtained modi®cation. The bottom curve shows the subtraction of from analysis of the infrared spectrum of 1-octene. The the as-polymerized PE from the brominated sample in the determined value, 0:125 ^ 0:01 CyC per 100 C, corre- frequency range of n 1700±500 cm21 that contains bands sponds to approximately one vinyl group for every two characteristic of vinyl groups 3909 and 991 cm21), carbon± molecules as estimated from the certi®ed mass average carbon double bonds, 1642 cm21, and bromine±carbon molecular mass of SRM 2885, 6280 g mol21, and poly- bonds 3574 and 657 cm21). The negative peaks 3909, 991 dispersity of 1.13 as characterized by light scattering and and 1642 cm21) indicate that vinyl groups have been gel permeation chromatography [13]. The vinyl content is consumed in the bromination step. Bromination of PE has reduced by bromination to a fraction 0:32 ^ 0:03 of its been used in the analysis of unsaturation [11], including value in SRM 2885. A qualitative assessment of the extent resolution of vinylidene and methyl components of the of reaction with TPP is made from the infrared data in the 888 cm21 band [12]. The appearance of vinyl bands at absence of spectral data on a model compound of TPP. The 909 and 991 cm21, and the carbon±carbon double bond intensity of the carbon±bromine stretch band at 574 cm21 in stretch at 1642 cm21 in the spectrum of the brominated TPP treated PE decreases to approximately 0:3 ^ 0:1 of its PE indicates that the bromination reaction was incomplete. value in brominated PE. Since two bromines should be Although the positive peak at 963 cm21 in the difference attached to the PE on adjacent carbons, but only one spectrum, Fig. 1 3bottom curve), may be due to bromination would be removed by the formation of the phosphonium it was not observed by Rueda et al.