Polymer MW and Branching.docPage 1 of 15

Polymer MW, Viscosity, and Branching Measurements by Mass Spectrometry

David J. Strumfels

DrexelUniversity Chemistry Dept., PhiladelphiaPA 19102

Summary:

Measurements of polymer molecular weights and MW distributions, intrinsic viscosities ([]), and branching parameters such as Mark-Houwink constants, have been historicallyperformed by a combination of Size Exclusion Chromatography (SEC), Laser Light Scattering (LALLS / MALLS), and/or viscosity techniques.1,2 Often these techniques are combined into a single system, e.g., a SEC with refractive index (RI) mass detection, in-line LALLS, and viscosity detection, the outputs from the three detectors used together to calculate all MW, viscosity, and branching results.3

This report will describe how branching properties might be determined by mass spectrometry (MS) alone, with or without SEC to fractionate polydisperse polymers. Although the measurement of molecular weight of polymers by MS is already an established science, thanks to developments in sample preparation (e.g., ESI and MALDI) and detection techniques such as TOF, there has been little in the way ofattempting branching measurements in this field. Basic principles of MS suggest however that such measurements should be possible. Given that MS instruments are already routinely employed for polymer analyses, the ability to extract this additional information would be valuable if it can be done practically; this is especially true in the case of chromatographic systems interfaced to MS detectors, where the use of additional detectors such as light scattering and viscosity would then be unnecessary. In addition, by making use of the theory used to calculate branching parameters from viscosity and MW, if MS can measure branching directly, it should also be able to calculate[]of a polymer distribution as well as other properties related to viscosity and MW, such as solution hydrodynamic volume.

Background:

A typical approach to explaining the behavior of polymeric molecules is to compare themto very long strands of (e.g.) spaghetti, strands that are often randomly spread in and around each other, inter-entangled, coiled (as in many elastomers), or sometimes rigidly/semi-rigidly aligned (in crystalline forms) or found in other arrangements. This is an apt analogy as far as it goes, which explains many of the physical and even chemical properties of polymers (though of course the chemical properties also require consideration of chemical properties too, such as van der Waals’ forces and hydrogen bonding). For example, the viscosity of a polymer solution is largely a function of the length of the strand, or chain; the longer the chain, the more entanglements between chains, and hence greater resistance to flow. Crystallinity and its macroscopic effects – melting and glass transition temperatures, chemical resistance and physical stiffness/ impermeability – are also explained in part as functions of chain length.5

The analogy is inapt or at least incomplete in several respects, however. Unlike spaghetti, polymer strands are literally chains, made of many repeating small chemical units, the MW of the polymer depending only on the MW of the unit and the number of units per chain. This allows for the phenomenon of branching in many polymers; side-chains which branch off the major chain. Branching is, in fact, a common feature of many synthetic polymers prepared by free-radical polymerization.5

The effects of branching on macroscopic properties, like polymer length, again can be understood well by a purely physical analysis. A linear polymer, being long and thin, has a relatively high surface area / MW, and therefore also possesses a large hydrodynamic volume (HV – the volume it effectively has in solution) / MW ratio because it can sweep out a large volume in its random tumblings and twistings in a solvent. A highly branched polymer, however, has a lower surface area / MW (as it is more spherical in shape), which leads in turn to a smaller solution HV / MW. Although this is not a universal feature of macromolecules (proteins – and some other water-soluble polymers – are linear, but have a small surface area / HV and thus HV / MW because intramolecular hydrogen bonding cause them to fold up into specific shapes), for many non-aqueous soluble synthetic polymers it accurately describes their solution behavior; so well in fact, that a plot of log (HV), or, in measurement terms, log (MW[]) versus retention time / volume can be used as a “universal” calibration in SEC for many polymers of this kind – calibrations with standards of the specified polymer are unneeded.6

The observed consequence of high HV / MW versus low HV / MW is a higher solution viscosity of the former compared to the latter. This is because the greater surface area, hence hydrodynamic volume, of linear polymers, leads to both more intermolecular entanglements and solute + solvent interactions, resulting in retarded solution fluidity – the essential definition of viscosity. The relatively non-spherical shape of linear polymers also yields a higher surface area to volume ratio, also increasing polymer interactions and so viscosity.

With this in mind, how to proceed is clear. Measure the MW of a polymer – either through light scattering, SEC, or other means – and the solution viscosity, and HV can be calculated directly: HV = MW[] (thus, hydrodynamic volume turns out to be the parameter used in the universal calibration technique described above). Having the HVs of both our 100% linear and branched polymers, we can calculate the so-called “g” factors, defined as the ratio of a polymer molecule’s unperturbed mean squared radius of gyration (roughly, the volume “swept out” in the solution by the freely moving polymer chain) to that of a linear polymer with the same composition and molecular weight:3

(Eq. 1)gMW = <S20,B / <S20,L

Experimentally, this ratio is calculated and used to determine either the number of branch points in the chain or their functionality:4

(Eq. 2)gMW = (HVB, MW / HVL, MW)1/ = ([]B, MW/[]L, MW)1/

where[]B, MW is the intrinsic viscosity of the branched polymer, and[]L, MW of the linear species at the same molecular weight, and  depends on the model of branching assumed plus various experimental factors, and where [] can be used in lieu of HV at the same MW values.

Other quantities related to branching can also be determined from MW and []. Mark-Houwink parameters, which are also useful in structural studies of polymers, can be calculated by plotting log([]) versus log(MW) for the linear polymer; in the resulting equation2

(Eq. 3)log ([]) = log (K) +  log (MW)

K and  are the Mark-Houwink constants for the given polymer, and comparisons of the plots of linear versus branched versions of the polymer will show the differences qualitatively. To construct a Mark-Houwink plot, it is of course necessary to fractionate the polymer first, generally by SEC. The relationship between Mark-Houwink constants and g factors is, combining equations (1) and (2)

(Eq. 4)gMW = [[]B, MW /(K  MW)]1/

Measurement of Polymer Molecular Weights by Mass Spectrometry:

MW measurements of high molar mass (~ 50,000 and up) synthetic polymers by mass spectrometry has been possible for well over a decade now, and a number of reviews7, 8, 9 and a book10 have been published on the subject during the last several years. Historically, the main obstacle in the way of MS analysis of polymers, prior to the 1980s, was the difficulty in enablingvery large molecules (e.g., 10,000 Daltons and up) to survive the volatilization and ionization steps necessary for MS: the main techniques available, such as electron ionization at high eV impact energies, and relatively soft, chemical ionization techniques, usually result in severe polymer fragmentation and degradation to the extent that measurement of parent ions is exceedingly difficult if not impossible using them. In addition, the MS designs then available did an inadequate job of achieving the mass range and resolution, and that only with difficulty. As for making measurements across polymer MW distributions, the techniques to directly couple MS with chromatographic instruments, e.g., fractionation by SEC, were not yet well developed; such techniques were required if a full analysis of polydisperse polymer systems was to be accomplished.

Over the last twenty years however, advances in sample preparation, new and/or improved MS techniques, and interfaces with other analytical instruments, have made it possible to measure polymer MW and MW distributions of increasingly larger polymers, both synthetic and natural, organic and aqueous soluble. On the sample preparation end, achieving the needed volatilization and ionization of even very high MW polymers has been made possible by the development of Matrix Assisted Laser Desorption Ionization (MALDI)7, while other soft ionization techniques, such as Fast Atom

Bombardment (FAB), and in particular Electro-Spray Ionization (ESI), have been improved to the point

of handling fairly large macromolecules when handled properly10*, although FAB is not used heavily anymore. On the technique and instrument interface end, particularly regarding HPLC and SEC, Mass Spectrometry Time Of Flight, or TOF, combined with ESI, is now a standard method of performing MS on the effluent of high pressure liquid chromatographic systems. This technique combination has proven essential in work with large biomolecules, such as proteins and nucleic acids.

For polymerswhich are the focus of this research proposal, which range in MW from approximately 104 – 105up to106Daltons, currently only MALDI (usually combined with TOF to give the acronym MALDI-TOF) generally possesses the molecular weight range and resolution needed for the MW measurements, particularly for non-aqueous soluble polymers (above this range TOF’s detection efficiency still suffers due to an excess of multiply charged species). For these molecules, other existing techniques still lead to unacceptable degrees of fragmentation and degradation. Although MALDI as a sample introduction method cannot be directlyconnected to anSEC (although TOF can be and often is the detection method of choice of LC-MS methods), a number of indirect techniques exist for interfacing, such as (semi) preparative fractionation and manual or robotic sample preparation, do exist, thus making it possible to obtain MW and other data on polydisperse systems. This of course still cannot not achieve the MW resolution a direct interface would be capable of.

As MALDI combined with TOF mass spectrometry is the only existing technique developed well enough for the polymer systems discussed here, it will be described first. The basic technique of MALDI is straightforward. The polymer of interest is first dissolved, dispersed, or otherwise intermixed13,14 in a suitable matrix material. The resulting mixture, usually a solution, is then deposited on the sample introduction surface and briefly irradiated by a pulsed laser of intense but short duration. The laser burst accomplishes the first two steps of any mass spectrometry technique: it vaporizes the analyte (along with the matrix), thereby allowing it to be swept into the MS acceleration chamber, and ionizes it, which is what allows the MS mechanism to separate the polymer molecules by their molecular weight differences.

*except where otherwise indicated, reviewed material comes from references 5., and 7. through 11., and personal communications and discussions.

Electrospray techniques of sample introduction turn a polymer solution (e.g., from an HPLC / SEC) into a spray of very fine droplets, which then obtain electric charge via an electron or chemical pathway, such as a chemical ionization technique or low energy electrons. The droplets are then injected into the MS chamber, where the solvent evaporates to leave behind ions of single and various multiple charges.

The multiply-charged species, ideally mostly parent ions, are separated by their mass to charge (m/z) ratio as in other forms of MS separation; here, however, the main separation factor is the varying number of charges on unfragmented parent ions, not the molecular weights of singly-charged ions, parent and fragments. The positions of the species in ESI are used to determine the parent molecule MW. This technique works well for narrow MW disperse polymers, whether prepared synthetically or by the SEC semi-preparative fractionation described above; the difficulty with more broadly polydisperse polymer systems is that the data are complicated when the molecules span a large range of weights, leading to peaks both of different m/z and MW. Also, again, for high MW polymers, especially synthetic systems, it is generally not as effective as MALDI in preventing fragmentation, adding still further to the spectrum’s complexity. ESI is still, as noted, for technical reasons the main technique for interfacing chromatographic systems with MS.

If TOF is the ion separation technique used, the volatilized and charged polymer ions, however prepared, are swept into the acceleration region of the MS, where an electric field gradient is used to boost the ions to their initial range of velocities. The velocities are MW dependent because this gradient boosts all the ions to the same kinetic energy ½mv2, thus necessitating differing velocities for molecules of different masses (more strictly, of different M /Z ratios). From the acceleration region, the ions move into the “drift” region, where their velocity spread allows them to separate by MW before striking the detector and being recorded. As all the polymer ions possess the same kinetic energy, higher MW species pass through the drift region slower than low MW species, and thus their molecular weights can be directly determined as a function of the time they spend in the region; ergo, the technique’s name.

Some history is needed here to explain how TOF became the instrument of choice for polymer analyses in chromatographic system. The problem with the basic TOF design as described above is its poor resolution, particularly when applied to increasingly higher MW materials. Thisproblem is mostly due to two interrelated flaws in the acceleration part of the basic design. Fundamental to high resolution mass spectrometry is the need for all the ions generated from a sample to be created at the same time in the same region of space, and to be accelerated at both the same times and with the same energies. In practice, of course, no operating MS instrument can perform all these tasks perfectly, but the unmodified, original TOF design was a notoriously poor performer precisely because, for example compared to magnetic sector designs, its separation method relies so directly on these parameters and their effects of ion time of flights in the drift region. Even the resolution losses from the small imperfections of all MS instruments are heavily exaggerated in TOF.

TOF is used so heavily now because these design flaws are now compensated for in a variety of ingenious ways, notably the use of reflection plates and acceleration delay times to correct for ion spreading in time and space (without, of course, reducing desired spread by differing m/z ratios). The net result is that modern TOF-MS has evolved into an technique with very high resolution, even at very high molecular weights. This is why it is commonly used as the instrument of choice in polymer analyses, particularly, as noted, in combination with the MALDI sample introduction method.

The combination of MALDI-TOF therefore allows quick, efficient, high-resolution analyses of polymer MWs up into the millions of Daltons (although MALDI still suffers from problems of detection efficiency above 106). Mw, Mn, and polydispersity indexes (Mw / Mn) can be readily determined for many synthetic polymers of even quite high MW, swiftly and with far less analyst time than many other techniques. Another advantage of MALDI-TOF is that is measures these quantities much more directly than other techniques; although calibration of the TOF instrument is usually necessary, only a single calibration is needed for all polymers,regardless of chemical composition. (Contrast this with straight SEC, where each polymer must be calibrated with known standards of the same chemical species, or where at best one can use the universal calibration technique already mentioned – a technique built around a number of variables, and not at all as universal as the name suggests.) Other MW measurement techniques, such as light scattering and viscosity techniques, although often called absolute, still require a variety of specific polymer inputs – such as solution refractive indexes and Mark-Houwink parameters – to give accurate results.

As mentioned, MALDI cannot be directly interfaced to a size exclusion chromatographic system, only indirectly by methods of collecting and prepping separate fractions (this can be highly automated). In one sense, this is irrelevant, as the MS technique itself, TOF or otherwise, separates the different MW species anyway. It is nevertheless desirable to perform an SEC separation of a polydisperse polymer sample prior to introduction into the MS, as even better resolution of the different MW species (leading to simpler spectra), can be obtained, simplifying analyses and in general providing more information on the system. Indeed, even the use of SEC alone may be inadequate here. The reason for this is a consequence of the SEC separation method, which is by hydrodynamic volume, not MW: when analyzing polymeric materials which vary by both MW and branching, even the narrowest fraction of the column eluent will represent not a narrowly disperse MW slice but in fact a mixture of lower MW linear and higher MW branched material. Thus, we are still looking at a fairly complex composition, complicating the MS spectra. If better separation is needed, one approach here would be to perform an additional chromatographic separation, this time via an absorption technique. Absorption-based chromatography can separate linear and branched polymers of similar hydrodynamic volume due to the fact that at a given HV branched molecules have lower surface area (because they are more spherical in solution) than linear.