Coupled Liquid Chromatographic Techniques in Molecular Characterization. Peter Kilz and Harald Pasch

Transcription

1 Coupled Liquid Chromatographic Techniques in Molecular Characterization Peter Kilz and Harald Pasch in Encyclopedia of Analytical Chemistry R.A. Meyers (Ed.) pp John Wiley & Sons Ltd, Chichester, 2000

2

3 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 1 Coupled Liquid Chromatographic Techniques in Molecular Characterization Peter Kilz PSS Polymer Standards Service, Mainz, Germany Harald Pasch Deutsches Kunststoffinstitut, Darmstadt, Germany 1 Introduction 1 2 Coupled Techniques in Polymer Analysis 3 3 Coupling of Liquid Chromatography with Information-rich Detectors Introduction Coupling with Molar Mass-sensitive Detectors Coupling with Mass Spectroscopy Coupling with Fourier Transform Infrared Spectroscopy Coupling with Nuclear Magnetic Resonance Spectroscopy 22 4 Multidimensional Liquid Chromatography Introduction Experimental Aspects of Multidimensional Separations Separation Techniques for the First Dimension Separation Techniques for the Second Dimension State-of-the-art of On-line Coupled Two-dimensional Chromatography Conclusions and Future Developments 43 List of Symbols 44 Abbreviations and Acronyms 44 Related Articles 45 References 45 Complex polymers are distributed in more than one direction of molecular heterogeneity. In addition to the molar mass distribution (MMD), they are frequently distributed with respect to chemical composition, functionality, and molecular architecture (see Size-exclusion Chromatography of Polymers). For the characterization of the different types of molecular heterogeneity it is necessary to use a wide range of analytical techniques. Preferably, these techniques should be selective towards a specific type of heterogeneity. The combination of two or more selective analytical techniques is assumed to yield multidimensional information on the molecular heterogeneity. The present review presents the fundamental ideas of combining liquid chromatography (LC) with other analytical techniques in multidimensional analysis schemes (see Size-exclusion Chromatography of Polymers; Gas Chromatography in Analysis of Polymers and Rubbers; Field Flow Fractionation in Analysis of Polymers and Rubbers). The capabilities and limitations of different coupling techniques are discussed and a number of relevant applications are given. It is shown that multidimensional structural information can be obtained when different chromatographic techniques are combined. Another approach is the hyphenation of LC with information-rich detectors. These detectors include molar mass-sensitive detection systems, such as on-line viscometry (VISC) and light scattering (LS). Information on the chemical composition of complex polymers can be obtained when spectroscopic techniques, like Fourier transform infrared (FTIR) (see Infrared Spectroscopy in Analysis of Polymer Structure Property Relationships), nuclear magnetic resonance (NMR) or mass spectrometry (MS) are coupled to LC. The basics and applications of multidimensional LC are addressed rather extensively. A brief introduction to different separation mechanisms is given and the particular requirements for the first and second dimensions are discussed. In conclusion, state-of-the-art examples for online coupled two-dimensional (2-D) chromatography are demonstrated, and future developments are reviewed. 1 INTRODUCTION Today s polymeric materials are designed to meet very specific requirements defined by the application. Therefore, most synthetic polymers are highly complex multicomponent materials. They are composed of macromolecules varying in chain length, chemical composition, and architecture. By definition, complex polymers are heterogeneous in more than one distributed property (for example, linear copolymers are distributed in molar mass and chemical composition). In general, the molecular structure of a macromolecule is described by its size, its chemical structure, and its architecture. The chemical structure characterizes the constitution of the macromolecule, its configuration and its conformation. For a complete description of the constitution the chemical composition of the polymer chain and the chain ends must be known. In addition to the type and quantity of the repeat units their sequence Encyclopedia of Analytical Chemistry R.A. Meyers (Ed.) Copyright John Wiley & Sons Ltd

4 2 POLYMERS AND RUBBERS of incorporation must be described (alternating, random, or block in the case of copolymers). Macromolecules of the same chemical composition can still have different constitutions due to constitutional isomerism (1,2- versus 1,4-coupling of butadiene, head-to-tail versus headto-head coupling, linear versus branched molecules). Configurational isomers have the same constitution but different steric patterns (cis- versus trans configuration; isotactic, syndiotactic and atactic sequences in a polymer chain). Conformational heterogeneity is the result of the ability of fragments of the polymer chain to rotate around single bonds. Depending on the size of these fragments, interactions between different fragments, and a certain energy barrier, more or less stable conformations may be obtained for the same macromolecule (rod-like versus coil conformation). Depending on the composition of the monomer feed and the polymerization procedure, different types of heterogeneities may become important. For example, in the synthesis of tailor-made polymers frequently telechelics or macromonomers are used. These oligomers or polymers usually contain functional groups at the polymer chain end. Depending on the preparation procedure, they can have a different number of functional endgroups, i.e. be mono-, or bifunctional and so forth. In addition, polymers can have different architectures, i.e. they can be branched (star- or comb-like), and they can be cyclic. The structural complexity of synthetic polymers can be described using the concept of molecular heterogeneity, see Figure 1, meaning the different aspects of MMD, chemical composition distribution (CCD), functionality type distribution (FTD) and molecular architecture Molar mass distribution Molecular heterogeneity Distribution of chemical composition Molecular architecture Functionality type distribution Figure 1 Schematic representation of the molecular heterogeneity of complex polymers. distribution (MAD). They can be superimposed one on another, i.e. bifunctional molecules can be linear or branched, linear molecules can be mono- or bifunctional, copolymers can be block or graft copolymers etc. In order to characterize complex polymers it is necessary to know the MMD within each other type of heterogeneity. Using the traditional methods of polymer analysis, such as NMR, one can determine the type and concentration of monomers or functional groups present in the sample. However, the determination of functional endgroups is complicated for long-chain molecules because of low concentration. On the other hand, these methods do not yield information on how different monomer units or functional groups are distributed in the polymer molecule. Finally, these methods in general do not provide molar mass information. With respect to methods sensitive to the size of the macromolecule, one can face other difficulties. Size exclusion chromatography (SEC), which is most frequently used to separate polymer molecules from each other according to their molecular size in solution, must be used very carefully when analyzing complex polymers. The molecular size distribution of macromolecules can in general be unambiguously correlated with MMD only within one heterogeneity type. For samples consisting of a mixture of molecules of different functionality, the distribution obtained represents a sum of distributions of molecules having a different functionality and, therefore, cannot be attributed to a specific functionality type without additional assumptions. For the analysis of copolymers by SEC either the chemical composition along the molar mass axis must be known or detectors must be used which, instead of a concentration information, can provide molar mass information. To this end, SEC has to be coupled to composition-sensitive or molar mass-sensitive detectors. Another option for the analysis of complex polymers is the separation with respect to chemical composition or functionality by means of interaction chromatography. In this case, functionally or chemically homogeneous fractions are obtained which then can be subjected to molar mass determination. To summarize, for the complete analysis of complex polymers a minimum of two different characterization methods must be used. It is most desirable that each method is sensitive towards a specific type of heterogeneity. Maximum efficiency can be expected when, similar to the 2-D distribution in properties, 2-D analytical techniques are used. A possible approach in this respect is the coupling of different chromatographic modes in 2-D chromatography or the coupling of a separation technique with selective detectors, such as molar mass-sensitive or spectroscopic detectors.

5 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 3 2 COUPLED TECHNIQUES IN POLYMER ANALYSIS Coupled techniques (also termed hyphenated techniques) are very frequently used in low molar mass organic chemistry. Using high-resolution chromatographic techniques, such as capillary gas chromatography (GC), gradient high-performance liquid chromatography (HPLC) and capillary electrophoresis (CE), complex mixtures are separated into single components which are then identified by MS. By hyphenated GC/MS, HPLC/MS, and CE/MS up to several hundreds of different components can be separated and identified in one run with very high sensitivity. This is particularly important for environmental and biological samples, where frequently only very limited sample amounts are available. Polymers are typically complex mixtures in which the composition depends on polymerization kinetics and mechanism and process conditions. To obtain polymeric materials of desired characteristics, polymer processing must be carefully controlled and monitored. Furthermore, one needs to understand the influence of molecular parameters on polymer properties and end-use performance. MMD and average chemical composition may no longer provide sufficient information for process and quality control nor define structure property relationships. Modern characterization methods now require multidimensional analytical approaches rather than average properties of the whole sample. 1 Different from low molar mass organic samples, where single molecules are to be determined, for complex synthetic polymers, the analytical task is the determination of a distributed property. The molecular heterogeneity of a certain complex polymer can be presented in either a three-dimensional (3-D) diagram or a so-called contour plot. For a telechelic polymer these presentations are given in Figure 2. Using appropriate analytical methods, the type and concentration of the different functionality fractions must be determined and, within each functionality, the MMD has to be obtained. To do this, two different methods must be combined, each of which preferably is selective towards one type of heterogeneity. For example, a chromatographic method separating solely with respect to functionality could be combined with a molar mass selective method. Another approach would be the separation of the sample into different molar mass fractions which are then analyzed with respect to functionality. For copolymers, in particular random copolymers, instead of discrete functionality fractions a continuous drift in composition is present, see Figure 3. To determine this chemical composition drift in interrelation with the MMD, a number of classical methods have been used, including precipitation, partition, and (a) Functionality (b) W R R Molar mass A R A A Molar mass A A A R R R Functionality Figure 2 Representation of the molecular heterogeneity of a telechelic polymer in a 3-D diagram (a) and a contour plot (b). (a) Composition M i Molar mass (b) W A i Molar mass Composition (A in copolymer AB) Figure 3 Representation of the molecular heterogeneity of a random copolymer in a 3-D diagram (a) and a contour plot (b). cross-fractionation. 2 The aim of these very laborious techniques was to obtain fractions of narrow composition and/or MMD which are then analyzed by spectroscopy and SEC. During the last 20 years a number of techniques have been introduced in organic chemistry and applied to

6 4 POLYMERS AND RUBBERS polymer analysis, combining chromatographic separation with spectroscopic detection. 3 GC/MS has been used in polymer analysis, 4 11 but, due to the low volatility of high molar mass compounds it is limited to the oligomer region. The combination of pyrolysis and gas chromatography/mass spectrometry (GC/MS), however, is of great value for polymer characterization. 12,13 It provides for the analysis of complex polymers with respect to chemical composition. For a number of polymer systems characteristic low molar mass pyrolysis products are obtained, which yield information of the average composition and the blockiness of the polymer chain. Molar mass information, however, is not available from pyrolysis-gc/ms. Much more important for polymer analysis than GC are the different techniques of LC. Using SEC, liquid adsorption chromatography (LAC), or liquid chromatography at the critical point of adsorption (LC/CC) polymers can be fractionated with respect to different aspects of molecular heterogeneity, including molar mass, functionality, and chemical composition. The advantage of these techniques over GC is that intact macromolecules are separated and analysed. As will be shown in the next sections, LC can be efficiently coupled to infrared (IR) spectroscopy, to MS, and to NMR spectroscopy. 20,21 Another most efficient approach is the chromatographic separation of complex polymers by combining different separation mechanisms. This can be done by coupling two chromatographs in an off-line or on-line mode. Each of these chromatographs must operate in a mode which is selective towards one type of molecular heterogeneity. This 2-D chromatography has been termed orthogonal chromatography assuming the selectivity of each separation method with respect to one distribution function, e.g. MMD, FTD, or CCD. 22 The first truly automated 2-D chromatography set-up for polymer analysis was proposed by Kilz et al., 23 who coupled gradient HPLC and SEC. The need for such analysis protocols results from the fact that in complex polymers in addition to chemical heterogeneity of the first kind, another type of chemical heterogeneity may exist: chemical heterogeneity of the second kind, in which polymers of different composition and chain length have similar hydrodynamic volumes and, hence, co-elute in SEC, see Figure 4. A possible separation protocol for a complex polymer mixture is presented in Figure 5. The sample under investigation comprises molecules of different chemical compositions (different colors) and different sizes. In a first separation step this mixture is separated according to composition, yielding fractions which are chemically homogeneous. These fractions are transfered to a sizeselective separation method and analyzed with respect to Polymer concentration (a) (b) AAAAA BBAB ABABA BBB Log M Figure 4 SEC fractionation showing composition or architecture at a given retention volume. (Reproduced by permission from Barth. 1 ) molar mass. As a result of this 2-D separation, information on both types of molecular heterogeneity is obtained. 3 COUPLING OF LIQUID CHROMATOGRAPHY WITH INFORMATION-RICH DETECTORS 3.1 Introduction LC of polymers is often understood to be synonymous with SEC. SEC separates polymers according to the size of the macromolecules by entropic interactions and enables the MMD of a sample to be evaluated. However, in addition to size exclusion phenomena, other types of interaction of the macromolecules and the stationary and mobile phases of a chromatographic system can be used for separation. LAC uses enthalpic interactions to separate substances such as copolymers according to chemical composition. Finally, LC/CC can be used for functionality type separation by balancing entropic and enthalpic interactions in the chromatographic system.

7 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 5 1 V r Figure 5 Schematic separation protocol for the analysis of a complex polymer mixture. SEC is the premier polymer characterization method for determining MMD. In SEC, the separation mechanism is based on molecular hydrodynamic volume. For homopolymers, condensation polymers and strictly alternating copolymers, there is a correspondence between elution volume and molar mass. Thus, chemically similar polymer standards of known molar mass can be used for calibration. However, for SEC of random and block copolymers and branched polymers, no simple correspondence exists between elution volume and molar mass because of possible compositional heterogeneity of these materials. The dimensional distribution of macromolecules can, in general, be unambiguously correlated with MMD only within one heterogeneity type. For samples consisting of molecules of different chemical composition, the distribution obtained represents an average of dimensional distributions of molecules having a different composition and, therefore, cannot be attributed to a certain type of macromolecule. The inadequacy of using SEC without further precaution for the determination of MMD of polymer blends or copolymers results from the following consideration: for a linear homopolymer distributed only in molar mass, fractionation by SEC results in one molar mass being present in each retention volume. The polymer at each retention volume is monodisperse. If a blend of two linear homopolymers is fractionated then two different molar masses can be present in one retention volume. If now a copolymer is analyzed then a multitude of different combinations of molar mass, composition, and sequence length can be combined to give the same hydrodynamic volume. In this case, fractionation with respect to molecular size is completely ineffective in assisting the analysis of composition or MMD. Three on-line methods are used to try to characterize copolymers by SEC with respect to MMD and composition: ž ž ž conventional SEC utilizing multiple concentration detection on-line analysis of SEC fractions with a LS detector. VISC. The experimentally simplest approach is the combination of SEC with multiple concentration detectors. If the response factors of the detectors for the components of the polymer are sufficiently different, the chemical composition of each slice of the elution curve can be determined from the detector signals. Typically, a combination of ultraviolet (UV) and refractive index (RI) detection is used; another possibility is the use of a diode-array detector. In the case of non-uv absorbing polymers, a combination of RI and density detection yields information on chemical composition

8 6 POLYMERS AND RUBBERS The principle of dual detection is rather simple: when a mass m i of a copolymer, which contains the weight fractions w A and w B (D1 w A ) of the monomers A and B, is eluted in the slice i (with the volume 1V) of the peak, the area x i,j of slice i obtained from detector j depends on the mass m i (or the concentration c i D m i /1V) of polymer in the slice, its composition (w A ), and the corresponding response factors f j,a and f j,b,whereinj denotes the individual detector, as in Equation (1) below: x i,j D m i w A f j,a C w B f j,b The weight fractions w A and w B of the monomers can be calculated using Equation (2) below: { } 1 x1 /x 2 f 2,A f 1,A D 1 2 w A x 1 /x 2 f 2,B f 1,B Once the weight fractions of the monomers are known, the correct mass of polymer in the slice can be calculated using Equation (3) as follows: x i m i D 3 w A f 1,A f 1,B C f 1,B and the molar mass M C of the copolymer is obtained by interpolation between the calibration lines of the homopolymers 27 which is given in Equation (4): 1 ln M C D ln M B C w A ln M A ln M B 4 wherein M A and M B are the molar masses of the homopolymers, which would elute in this slice of the peak (at the corresponding elution volume V e ). It is clear that the interpolation between the calibration lines cannot be applied to mixtures of polymers (polymer blends): if the calibration lines are different, different molar masses of the homopolymers will elute at the same volume. The universal calibration is not capable of eliminating these errors, either, which originate from the simultaneous elution of two polymer fractions with the same hydrodynamic volume, but different composition and molar mass. The architecture of a copolymer (random, block, graft) has also to be taken into account, as Revillon 28 has shown by SEC with RI, UV, and viscosity detection. Intrinsic viscosity varies largely with molar mass according to the type of polymer, its composition, and the nature of its components. Tung 29 found that for block copolymers in good SEC solvents the simpler first approach (Equation 4) is more precise. Further information on quantitative aspects of SEC with dual detection can be obtained from Trathnigg et al. 30 Different applications of dual detection SEC in the analysis of segmented copolymers, 31 block copolymers, 32,33 star polymers, 34 and polymer blends 35,36 are also available. The limitation of SEC with dual detection is that only binary combinations of monomers can be investigated successfully. In the case of ternary combinations, more than two detectors must be used or one of the detectors must be able to detect two components simultaneously. To overcome the problems related to classical SEC of complex polymers, molar mass-sensitive detectors are coupled to the SEC instrument. Since the response of such detectors depends on both concentration and molar mass, they have to be combined with a concentration-sensitive detector. The following types of molar mass-sensitive detectors are used frequently: ž ž ž differential viscometer low angle laser light scattering (LALLS) detector multiangle laser light scattering (MALLS) detector. 3.2 Coupling with Molar Mass-sensitive Detectors As has been pointed out, for SEC of complex polymers no simple correspondence exists between elution volume and molar mass. It is, therefore, useful to determine the molar mass not via a calibration curve but directly from the SEC effluent. This can be done by using molar mass-sensitive detectors based on Rayleigh LS or intrinsic viscosity measurements. 41 In a LS detector, the scattered light of a laser beam passing through the cell is measured at angles different from zero. The (excess) intensity of the scattered light at the angle 2 (R 2)) is related to the weight-average of molar mass M w as expressed by Equation (5): K Ł c R 2 D 1 M w P 2 C 2A 2c 5 wherein c is the concentration of the polymer, A 2 is the second virial coefficient, and P 2 describes the scattered light angular dependence. K Ł is an optical constant containing Avogadro s constant N A, the wavelength l 0, the RI n 0 of the solvent, and the RI increment dn/dc of the sample. Their relationship is described by Equation (6): K Ł D 4p2 n 2 0 dn/dc 2 l 4 0 N 6 A In a plot of K Ł c/r 2 versus sin 2 2/2, M w can be obtained from the intercept and the radius of gyration (R g ) from the slope. A multiangle measurement provides additional information. In most cases the injected concentration is small and A 2 can be neglected. Thus, if the optical properties (n 0 and dn/dc) of the polymer solution are known, the molar mass at each elution volume increment can be determined as expressed by Equation (7): M w,i D R 2 i 7 K Ł P 2 i c i

9 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 7 If a low-angle LS instrument is used, P 2 is close to unity and M w,i can be calculated directly. For a multiangle LS instrument, the mean-square radius of gyration hr 2 g i at each elution volume can also be obtained from P 2 as shown in Equation (8): 1 D 1 C q2 hr 2 g i i P 2 i 3 ( ) ( ) 4p P 2 q D sin 2 l 0 In practice, however, the radius of gyration can only be determined for molecules larger than 20 nm in diameter. By measuring radius of gyration as a function of M w, insight into the molecular conformation of the polymer can be obtained. 1 Molar mass determination requires the knowledge of the specific RI increment dn/dc which in the case of complex polymers depends on chemical composition. Copolymer RI increments dn/dc copo can be calculated accurately for chemically monodisperse fractions, if comonomer weight fractions w i and homopolymer values are known, as described in Equation (9): ( dn dc ) copo D w i ( dn dc However, in some cases additional effects on dn/dc copo must be considered. Due to cooperative interactions between the monomer units in the polymer chain, copolymer RI increments may deviate from the summation scheme. As a result of different sequence length distributions, different dn/dc copo can be obtained for the same gross composition. Copolymer dn/dc copo values can be obtained by multiple detection SEC providing the chemical composition at each slice of the elution curve. Unfortunately, LS investigations of copolymers are complicated even further by the fact that SEC does not separate into chemically monodisperse fractions. Accordingly, due to compositional heterogeneity the RI increment of a particular scattering center may be different from the total dn/dc of the corresponding SEC slice. Therefore, in general, only apparent molar masses for copolymers can be measured. 34 Another influencing factor is the RI of the solvent. As has been shown by Kratochvil, 42 the solvent RI should be significantly different from the values of the copolymer fractions and the corresponding homopolymers. The evaluation of LS detectors for SEC was conducted by Jeng et al. with respect to precision and accuracy 43 and the proper selection of the LS equation. 44 The results obtained for polystyrene (PS) and polyethylene were compared for a low-angle and a multiangle LS instrument. The application of SEC/LS has been discussed in a multitude of papers. In addition to determining ) i 8 9 M w values, the formation of microgels has been studied by Pille and Solomon. 45 Mourey and Coll investigated high molar mass PS and branched polyesters, and discussed the problems encountered in molar mass and radius of gyration determination. 46,47 Grubisic- Gallot et al. proved that SEC/LS is useful for analysing micellar systems with regard to determining molar masses, qualitative evaluation of the dynamics of unimer-micelles re-equilibration, and revealing the mode of micelle formation Another very useful approach to molar mass information of complex polymers is the coupling of SEC to a viscosity detector The viscosity of a polymer solution is closely related to the molar mass (and architecture) of the polymer molecules. The product of polymer intrinsic viscosity [h] times molar mass is proportional to the size of the polymer molecule (the hydrodynamic volume). Viscosity measurements in SEC can be performed by measuring the pressure drop 1P across a capillary, which is proportional to the viscosity h of the flowing liquid (the viscosity of the pure mobile phase is denoted as h 0 ). The relevant parameter [h] is defined as the limiting value of the ratio of specific viscosity (h sp D h h 0 /h 0 ) and concentration c for c! 0, as shown by Equation (10): [h] D lim h h 0 h 0 c D lim h sp c for c! 0 10 The viscosity of a polymer solution as compared to the viscosity of the pure solvent is measured by the pressure drop 1P across an analytical capillary-transducer system. The specific viscosity is obtained from 1P/P, wherepis the inlet pressure of the system. As the concentrations in SEC are usually very low, [h] can be approximated by h sp /c. A simple approach using one capillary and one differential pressure transducer will not work very well, because the viscosity changes 1h D h h 0 will typically be very small compared to h 0, which means that one has to measure a very small change of a large signal. Moreover, flow-rate fluctuations due to pulsations of a reciprocating pump will lead to much greater pressure differences than the change in viscosity due to the eluted polymer. Instruments of this type should be used with a positive displacement pump. A better approach is the use of two capillaries (C1 and C2) in series, each of which is connected to a differential pressure transducer (DP1 and DP2), and a sufficiently large holdup reservoir (HR) in between. With this approach, one measures the sample viscosity h from the pressure drop across the first capillary, and the solvent viscosity h 0 from the pressure drop across the second capillary. Pulsations are eliminated in this set-up, because they appear in both transducers simultaneously. Another design is that of the differential viscometer, in which four

10 8 POLYMERS AND RUBBERS From column C1 HR DP1 C1 From column C2 P same frictional properties or enhanced viscosity to the same degree as the actual polymer in solution. Assuming the validity of this approach and in agreement with the SEC mechanism, similar elution volumes correspond to similar hydrodynamic volumes, as shown in Equation (11): V e,1 D V e,2! M 1 [h] 1 D M 2 [h] 2 11 C2 (a) DP2 (b) Figure 6 Schematic representation of differential viscometers. P is inlet pressure transducer: C1 C4 are flow restriction capillaries. capillaries are arranged in a manner similar to that of a Wheatstone bridge. In Figure 6, both designs are shown schematically. In the bridge design, a holdup reservoir in front of the reference capillary (C4) makes sure that only pure mobile phase flows through the reference capillary, when the peak passes the sample capillary (C3). This design offers considerable advantages: the detector actually measures the pressure difference 1P at the differential pressure transducer (DP) between the inlets of the sample capillary and the reference capillary, which have a common outlet, and the overall pressure P at the inlet of the bridge. The specific viscosity h sp D 1h/h 0 is thus obtained from 1P/P. One concern with this type of detector is that the flow must be divided in the ratio of 1 : 1 between both arms of the bridge. This is achieved by capillaries C1 and C2, which must have a sufficiently high back pressure. Nevertheless, when a peak passes through the sample capillary, a slight deviation of the 1 : 1 ratio will be observed. A problem of flow-rate variations exists also in a single capillary viscometer: when the polymer peak passes through the measuring capillary, the increased back pressure leads to a peak shift. 57 Being able to determine [h] as a function of elution volume, one can now compare the hydrodynamic volumes (V h ) for different polymers. The hydrodynamic volume is, through Einstein s viscosity law, related to intrinsic viscosity and molar mass by V h D [h]m/2.5. Einstein s law is, strictly speaking, valid only for impenetrable spheres at infinitely low volume fraction of the solute (equivalent to concentration at very low values). However, it can be extended to particles of other shapes, defining the particle radius then as the radius of a hydrodynamically equivalent sphere. In this case V h is defined as the molar volume of impenetrable spheres which would have the C3 DP C4 H Inaplotoflog(M[h] versus V e identical calibration lines should be found for the two polymers 1 and 2, irrespective of their chemical composition. This universal calibration approach has been predicted and experimentally proved by Benoit et al. 58 As a consequence, using the universal calibration curve established with known calibration standards (for example PS), one can obtain the SEC-molar mass calibration for an unknown polymer sample. The intrinsic viscosity is a function of molar mass given by the Mark Houwink relationship (Equation 12), wherein K and a are coefficients for a given polymer in a given solvent at a given temperature. This leads to Equation (13): [h] D KM a 12 K 1 M a 1 C1 1 D K 2 M a 2 C If a column has been calibrated with polymer 1 (e.g. PS), the calibration line for polymer 2 can be calculated, provided that the coefficients K and a are known for both polymers with sufficient accuracy. This is shown by Equation (14): ( 1 ln M 2 D 1 C a 2 ) ln ( K1 K 2 ) C ( 1 C a1 1 C a 2 ) ln M 1 14 Thus, the concept of universal calibration provides an appropriate calibration also for polymers for which no calibration standards exist. The limiting factor of this approach is the accuracy of determining K and a. There are very high variations in the values reported in the literature. 59,60 Even for such common polymers as PS and polymethyl methacrylate (PMMA) the values may differ considerably. If the Mark Houwink coefficients are not available, a universal calibration curve is established using PS calibration standards and the SEC viscometer combination. The basic steps involved in the MMD analysis are summarized in Figure 7. First, the universal calibration curve of the SEC separation system has to be established by using narrow molar mass standards as indicated by the top arrow pointing to the right. Once the universal calibration curve is established, one can then reverse the procedure, by going from right to left following the bottom arrow, to

11 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 9 To obtain universal calibration (PS standard ) Molar mass-calibration [η]-calibration (from viscosity detector) Universal Log M PS Log [η] PS Log [η] M V R V R V R To obtain absolute molar mass calibration (unknown polymer) Figure 7 Determination of absolute molar masses via universal SEC calibration. obtain the molar mass calibration curve of any unknown polymer. The calibration curve is obtained literally by substracting the [h] calibration curve of the unknown sample from the universal calibration curve. The [h] calibration curve for the unknown sample is obtained from the on-line viscometer. 61 The application of RI and differential viscometer detection in SEC has been discussed by a number of authors Lew et al. presented the quantitative analysis of polyolefines by high-temperature SEC and dual RI viscosity detection. 65 They applied a systematic approach for multidetector operation, assessed the effect of branching on the SEC calibration curve, and used a signal averaging procedure to define better intrinsic viscosity as a function of retention volume. The combination of SEC with RI and viscosity detectors was used to determine molar mass and functionality of polytetrahydrofurane simultaneously. 66 Long chain branching in ethylene propylene diene rubber (EPDM) copolymers by SEC viscometry was analysed by Chiantore et al. 67 One of the difficult problems in characterizing copolymers and polymer blends by SEC viscometry is the accurate determination of the polymer concentration across the SEC elution curve. The concentration detector signal is a function of the chemical drift of the sample under investigation. To overcome this problem, Goldwasser proposed a method where no concentration detector is required for obtaining number-average molar mass (M n ) data. 68 In the usual SEC viscometry experiment, the determination of the intrinsic viscosity at each slice of the elution curve requires a viscosity and a concentration signal as shown by Equation (15): ( ) ln hrel [h] i D 15 c i where ln h rel is the direct detector response of the viscometer. One calculates the molar mass averages by the expressions given in Equation (16) and in Equation (17): ci M n D [ci / V h,x /[h] i ] M w D ci V h,x /[h] i ci where V h,x D [h] x M x is the data retrievable from the universal calibration curve. By rearranging Equation (17) using Equation (16) the following expression (Equation 18) is obtained: ci M n D ln hrel /V h,x i M n D sample amount ln hrel /V h,x i or 18 The sample amount can be determined easily from the injection volume and the sample concentration and no information from a concentration detector is required. With this approach, the M n value of any polymer sample can be determined by SEC using only the viscosity detector. Other molar mass averages, however, cannot be determined. The advantage of the Goldwasser M n method is that it can access much wider molar mass ranges than other existing methods like osmometry or endgroup methods. Due to the problems encountered with SEC/LALLS and SEC viscometry, a triple-detector SEC technology has been developed, where three on-line detectors are

12 10 POLYMERS AND RUBBERS used together in a single SEC system. In addition to the concentration detector, an on-line viscometer and a LALLS instrument are coupled to the SEC: this arrangement is known as TriSEC. With TriSEC, absolute molar mass determination is possible for polymers that are very different in chemical composition and molecular conformation. The usefulness of the TriSEC approach has been demonstrated in a number of applications. It was shown by Pang and Rudin that only by using both viscometer and LS detection are accurate MMDs obtained. 69 Wintermantel et al. have developed a custom-made multidetector instrument and demonstrated that it has great potential not only for absolute molar mass determinations but also for structure characterization of linear flexible, semiflexible, and branched polymers. 70 Degoulet et al. characterized polydisperse solutions of branched PMMA, 71 while Jackson et al. investigated linear chains of varying flexibility in order to prove universal calibration. 72 Yau and Arora discussed the advantages of TriSEC for the determination of Mark Houwink coefficients, long-chain branching, and polymer architecture. 73 Finally, several attempts have been made to develop an absolute molar mass detector based on osmotic pressure measurements. Commercially available membrane osmometers are designed for static measurements, and the cell design with a flat membrane is not suited for continuous flow operation. Yau 61,74 developed a detector which is different from the conventional design; it measures the flow resistance of a column caused by osmotic swelling and deswelling of soft gel particles used for the packing, see Figure 8. With a microbore gel column an M n sensitive detector with a fast response was obtained which could be coupled to the SEC equipment. However, since the change in flow resistance could not easily be related to the osmotic pressure of the solution, absolute calibration was lost. Recently, an osmometer based on a concentric design with a capillary-shaped membrane has been developed by Köhler et al. 75 and Lehmann et al. 76 The flow cell volume is 12.2 µl, the response time approximately 15 s, and the molar mass cut-off is below g mol 1.The design of the cell is given in Figure 9. The cylinder symmetry and stiffness of the osmometer and the favorable properties of the membrane were combined to meet the requirements for on-line detection. Testing the instrument in both batch and continuous flow operation with PS standards yielded reproducible results and good agreement with the nominal molar masses. However, the osmometer still caused a certain peak broadening, and the pressure noise level still strongly exceeded the noise of the concentration detector. As has been discussed, the combination of SEC and molar mass-sensitive detectors is a powerful tool for the High P gel Solvent,, Solvent Polymer solution,,,,, Solution Osmotic effects of polymer solution: Shrinkage of soft gel particles More open flow channels Lower flow resistance Lower pressure drop P gel Low P gel Figure 8 Differential pressure measurement of osmotic effect on a soft gel column. Flush,,,,,,,, Flush Gasket Membrane,,,, Pressure transducer Solvent Solution,,,,,,, Seal Glass tube Holder Out Figure 9 Design of a concentric osmometric flow cell. (Reproduced by permission from Lehmann et al. 76 ) analysis of complex polymers. However, it is important to distinguish between claimed versus actual capabilities and between potential expectations and demonstrated performance. Tables 1 and 2 below, taken from a critical review of different techniques summarize the information content and additional details of SEC/LS and SEC viscometry coupling. 61 The information content is classified into two categories. Primary information is of high precision and accuracy, insensitive to SEC operation variables, and does not require molar mass or universal calibration. Secondary information is less precise and requires calibration.

13 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 11 Table 1 SEC analysis using molar mass-sensitive detectors Method Primary Information content Secondary Conventional SEC MMD SEC/LALLS MMD SEC/MALLS MMD R g distribution SEC/VISC [h] distribution MMD R g distribution Copolymer M n SEC/VISC/LS [h] distribution Copolymer M n MMD R g distribution Table 2 Generalization of molar mass-sensitive detectors Intended LALLS/MALLS Viscometer measurements MMD Requires precise n and dn/dc values Not affected by nonexclusion effects Requires universal calibration and K, a-parameters [h] distribution Directly from experiment Not affected by nonexclusion effects R g distribution MALLS only Calculable from Chain conformation and branching Chemically heterogeneous polymer analysis Noise, particulates, bubbles R g vs M plot, MALLS only Limited Strongly affected [h]m [h]vsmplot, R g vs M plot Better Less affected In addition, the complex procedures related to SEC/LS and SEC viscometry coupling are a potential source of error. According to Jackson and Barth 77 these include: 1. Accuracy of the universal calibration curve. 2. Detector configuration: arrangement of multiple detectors in series or in parallel can cause additional peak broadening, flow rate variations, back pressure variations. 3. Interdetector volume: detectors are placed at different physical positions and their signals must be aligned very precisely. 4. Detector sensitivity: LS and viscosity detectors are very sensitive towards higher molar masses, while the RI detector is most sensitive at lower molar masses. 5. Low molar mass fractions: polymer molecules may not adopt random coil conformation, the Mark Houwink coefficients become functions of molar mass. To summarize, although the principal limitation of SEC separating according to hydrodynamic volume and not molar mass cannot be overcome, the advantages of multidetector SEC in the accurate characterization of complex polymers are significant. However, in order to generate reproducible and accurate results on a routine basis, special care must be taken regarding the added complexity of the instrumentation. In addition to improving the design of multidetector SEC setups, important advances are expected from methods for determining the chemical composition across the MMD by interfacing SEC with FTIR spectroscopy, MS, and NMR. 3.3 Coupling with Mass Spectroscopy From the very early stages of development of modern MS, the value of its combination with chromatography was quickly recognized. The coupling of GC with MS was a natural evolution since they are both vapor phase techniques, and very quickly GC/MS has been accepted as a standard component of the organic analytical laboratory. It has taken considerably longer to achieve a satisfactory and all-purpose mode of HPLC/MS coupling. The difficulties with HPLC/MS were associated with the fact that vaporization of typically 1mLmin 1 from the HPLC translates into a vapor flowrate of approx ml min 1. Other difficulties related to the eluent composition as a result of the frequent use of nonvolatile modifiers, and the ionization of nonvolatile and thermally labile analytes. However, during the past several years commercial interfaces have been developed which have led to a broad applicability of HPLC/MS The techniques necessary for the successful introduction of a liquid stream into a mass spectrometer are based on the following principles: electrospray ionization (ESI), 81 atmospheric pressure chemical ionization, 82 thermospray ionization, 83 and particle beam ionization. 84 From the point of view of polymer analysis, a mass spectrometric detector would be a most interesting alternative to the conventional detectors because this detector could provide absolute molar masses of polymer components. 85,86 Provided that fragmentation does not occur, intact molecular ions could be measured. The measured mass of a particular component could then be correlated with chemical composition or chain length. However, the major drawback of most conventional HPLC/MS techniques is the limited mass range, preventing higher

14 12 POLYMERS AND RUBBERS oligomers (molar mass above g mol 1 )tobe ionized without fragmentation The use of MS for detailed polymer analysis has become increasingly established due to the introduction of soft ionization techniques that afford intact oligomer or polymer ions with less fragmentation One of these techniques, ESI/MS, has been widely applied in biopolymer analysis. Proteins and biopolymers are typically ionized through acid base equilibria. When a protein solution (the effluent from an HPLC separation) is exposed to an electrical potential it ionizes and disperses into charged droplets. Solvent evaporation upon heat transfer leads to the shrinking of the droplets and the formation of analyte ions. Larger molecules acquire more than one single charge, and, typically, a mixture of differently charged ions is obtained. Unfortunately, up to now ESI/MS has had limited application in polymer analysis. 94,95 Unlike biopolymers, most synthetic polymers have no acidic or basic functional groups that can be used for ion formation. Moreover, each molecule gives rise to a charge distribution envelope, thus complicating the spectrum further. Therefore, synthetic polymers that can typically contain a distribution of chain lengths and have a variety in chemical composition or functionality furnish complicated mass spectra, making interpretation nearly impossible. To overcome the difficulties of ESI/MS, Prokai and Simonsick added sodium cations to the mobile phase to facilitate ionization. 96,97 To simplify the resulting ESI spectra, the number of components entering the ion source was reduced. Prokai et al. implemented microcolumn SEC for the separation of polydisperse mixtures prior to ESI detection. 98 They used a 250 ð 0.5 mm internal diameter SEC column for the molar mass separation of octylphenoxy polyethylene oxide (PEO). Applying a flow of 4 µl min, they were able to supply the effluent from the column directly into the ESI source. To promote ionization, a sheath liquid of sodium iodide in methanol was delivered to the ESI interface. Figure 10 shows a representative chromatogram and mass spectrum from the SEC/ESI analysis. The mass spectrum was obtained by averaging between 6.9 and 9.2 min. It shows singly and doubly charged molecules in the molar mass range of g mol 1. The analysis of PEOs by SEC/ESI/MS with respect to chemical composition and oligomer distribution was discussed by Simonsick. 99 In a similar approach, aliphatic polyesters, 100 phenolic resins, 101 methyl methacrylate macromonomers 101 and polysulfides 102 have been analysed. The detectable mass range for different species, however, was well below 5000 g mol 1, indicating that the technique is not really suited for polymer analysis. The quantitative analysis of PMMA-butyl acrylate copolymers by coupled LC and particle beam MS has been described by Murphy et al. 103 For separation with respect to chemical composition gradient HPLC was used. The copolymer composition was determined by monitoring several low-mass fragments formed by thermal decomposition and electron impact ionization in the particle beam interface. Matrix-assisted laser desorption/ionization (MALDI)- time-of-flight (TOF) MS is one of the newest soft ionization techniques that allows desorption and ionization of very large molecules even in complex mixtures. In 100 Total ion chromatogram 50 Relative abundance 0 0 (a) Time (min) (b) Figure 10 Micro-SEC/ESI analysis of octyloxy PEO, (a) TIC chromatogram, (b) averaged mass spectrum. (Reproduced by permission from Prokai et al. 98 ) m/z

15 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 13 polymer analysis, the great promise of MALDI/TOF/MS is to perform the direct identification of mass-resolved polymer chains, including intact oligomers within an MMD, and the simultaneous determination of structure and endgroups in polymer samples. This most promising method for the ionization of large molecules and analysis according to their molar mass and functionality has been introduced by Karas and Hillenkamp 104,106 and by Beavis and Chait. 105 Compared to other MS techniques, the accessible mass range has been extended considerably, and the technique is fast and instrumentally very simple. Moreover, relatively inexpensive commercial instrumentation has become accessible. In principle, the sample to be investigated and a matrix solution are mixed in such a ratio that matrix separation of the sample molecules is achieved. After drying, a laser pulse is directed onto the solid matrix to photo-excite the matrix material. This excitation causes the matrix to explode, resulting in the expulsion and soft ionization of the sample molecules. Once the analyte is ionized, it is accelerated and analysed in a TOF mass spectrometer. As a result, the analyte is separated according to the molar mass of its components, and in the case of heterogeneous polymers additional information on chemical composition may be obtained. In a number of papers it was shown that polymers may be analysed up to relative molar masses of about Da It was shown in a number of applications that functionally heterogeneous polymers can be analysed with respect to the degree of polymerization and the type of functional groups. The on-line combination of LC and MALDI/TOF/MS would be of great value for polymer analysis. In particular, for chemically or functionally heterogeneous polymers LC could provide separation with respect to chemical composition while MALDI/TOF would analyse the fractions with respect to oligomer distribution or molar mass. Unfortunately, MALDI/TOF is based on the desorption of molecules from a solid surface layer and, therefore, a priori not compatible with LC. In an attempt to take advantage of the MALDI/TOF capabilities, a number of research groups carried out off-line LC separations and subjected the resulting fractions to MALDI/TOF measurements. Although this is laborious, it has the advantage that virtually any type of chromatographic separation can be combined with MALDI/TOF. The different options for using MALDI/TOF as an off-line detector in LC have been discussed by Pasch and Rode. 116 In SEC of low molar mass samples the separation into individual oligomers and the quantitative determination of the MMD via an oligomer calibration could be achieved, see Figure 11 for oligo(caprolactone). The lower oligomers appeared as well separated peaks at the high retention time end of the chromatogram. For the analysis of the peaks, i.e. the assignment of a certain degree of polymerization (n) to each peak, MALDI/TOF/MS was used. The SEC separation was conducted at the usual analytical scale and the oligomer fractions were collected, resulting in amounts of 5 20 ng substance per fraction in tetrahydrofuran (THF) solution. The solutions were directly mixed with the matrix solution, placed on the sample slide and subjected to the MALDI experiments. As a large number of fractions may be introduced into the mass spectrometer at one time, sample preparation and MALDI/MS measurements take a very short period of time. In total, nine fractions were collected from SEC and measured by MALDI/MS. For the lower oligomers the spectra consisted of a number of peaks of high intensity, having a peakto-peak mass increment of 114 Da, which equals the mass of the caprolactone repeating unit. These peaks represented the M C Na C molecular ions, whereas the peaks of lower intensity in their vicinity were due to the formation of M C K C molecular ions. M C Na C and M C K C molecular ions were formed due to the presence of small amounts of Na C and K C ions in the samples and/or the matrix. Further peaks of low intensity indicated a functional heterogeneity in the samples. From the masses of the M C Na C peaks the degree of polymerization of the corresponding oligomer was calculated. By this procedure, the first peak in the chromatogram was assigned to n D 1, the second peak to n D 2, and so on. From the elution time and the degree of polymerization of each oligomer peak an oligomer calibration curve of log molar mass vs elution time was constructed. The conventional calibration curve based on PS standards differed remarkably from this oligomer calibration curve. A much more demanding task is the analysis of fractions from LC not only with respect to molar mass but also with respect to chemical structure. The separation of a technical fatty alcohol ethoxylate (FAE) by LC, under conditions where the chain length as well as the endgroups direct the separation, is presented in Figure 12. Using this chromatographic technique, the FAE was separated into three main fractions, the first fraction appearing as one peak at a retention time of about 60 s and the second and third fractions showing oligomer separations. Fraction 1 was collected in total, whereas for fractions 2 and 3 the individual oligomer peaks were collected. The MALDI/MS spectra of all three fractions gave a peak-to-peak mass increment of 44 Da, thus indicating that all fractions consisted of species with an ethylene oxide-based polymer chain. From the masses assigned to the peaks and the peak-to-peak mass increment of the ethylene oxide repeating unit the mass of the endgroup for the different fractions was calculated. Provided the sample was a pure FAE, the endgroups of fractions 1 3 could be identified as being polyethylene glycol

16 14 POLYMERS AND RUBBERS n = n = 4 30 Time (min) 40 n = 3 (a) 50 n = 2 (b) m/z (Da) Figure 11 SEC of oligo(caprolactone) and MALDI/TOF analysis of fractions (a) and SEC calibration graphs (b). (Reproduced by permission from Pasch and Rode. 116 ) (PEG) (a,w-dihydroxy endgroups), C 13 -terminated PEO (a-tridecyl-w-hydroxy endgroups) and C 15 -terminated PEO (a-pentadecyl-w-hydroxy endgroups), respectively. Using MALDI/TOF the oligomer distribution of the PEG fraction was measured directly. For fractions 2 and 3 by determining the degree of polymerization of the oligomer peaks oligomer calibration curves were obtained, which were used for the molar mass calculation of the fractions. Thus, by combining LC and MALDI/MS detection, complex samples can be analysed with respect to chemical structure and molar mass. Other examples of successful off-line combinations of LC and MALDI/TOF were given by Krüger et al., separating linear and cyclic fractions of polylactides by LC/CC. 117 Just and Krüger were able to separate cyclic siloxanes from linear silanols and to characterize their chemical composition. 118 The calibration of an SEC system by MALDI/TOF was discussed by Montaudo et al. 119 Polydimethyl siloxane (PDMS) was fractionated by SEC into different molar mass fractions. These fractions were subjected to MALDI/TOF for molar mass determination. The resulting peak maximum molar masses were combined with the elution volumes of the fractions from SEC to give a PDMS calibration curve log M vs. V e. The calibration of SEC by MALDI/TOF/MS for PMMA, polyvinyl acetate and

17 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 15 n = 9 n = 7 0 PEG n = C 13 H 27 O(CH 2 OCH 2 O) n H t (s) 9 C 15 H 31 O(CH 2 CH 2 O) n H n = n = 5 n = n = m/z (Da) Figure 12 Separation of a technical PEO by LC and analysis of fractions by MALDI/TOF, peak assignment indicates degree of polymerization n. (Reproduced by permission from Pasch and Rode. 116 ) vinyl acetate copolymers has been discussed by Danis et al. In addition to obtaining proper calibration curves, band broadening of the SEC system was detected. 120 The analysis of random copolyesters has been described recently by Montaudo et al. 121 To overcome the difficulties of the off-line analysis of SEC fractions, recently interfaces were introduced where the SEC effluent was sprayed onto a moving matrix-coated substrate. Kassis et al. used a modified LC-Transform SEC/FTIR interface, 122 while Nielen applied a robotic interface of Bioanalytical Instruments where the effluent was spotted on the MALDI target. 123 A novel interface for coupling SEC and MALDI/TOF has been developed recently by Lab Connections Inc. 124 In this interface, the effluent from the SEC is sprayed through a heated capillary nozzle continuously on a slowly moving MALDI target precoated with the appropriate matrix, resulting in a uniform surface layer of sample fraction and matrix. The matrix can be deposited manually or automatically on the MALDI target from an appropriate solution. When necessary, a salt is added to the matrix solution. The characterization of PMMA by SEC/MALDI/TOF is shown in Figure Prior to fraction deposition the target was precoated with the matrix dithranol and a small amount of LiCl to enhance the formation of M C Li C molecular ions. Since the fraction deposition was carried out through a heated capillary nozzle, a solid fraction/matrix film was obtained on the MALDI/TOF target. The MALDI/TOF target had a length of 70 mm

18 16 POLYMERS AND RUBBERS Pulses Pulses Pulses Pulses Pulses Pulses Mass/Charge Figure 13 MALDI/TOF spectra of PMMA fractions obtained from SEC/MALDI/TOF analysis. (Reproduced by permission from Pasch. 125 ) and was scanned continuously with 3500 laser pulses. Every 50 pulses were summarized to give a complete MALDI/TOF spectrum. With SEC as the preseparation technique, low positions on the target correspond to high molar masses, while high positions are equivalent to low molar masses. Selected spectra from different positions of the polymer/matrix track of the PMMA sample are given in Figure 13. In the present experiment, a sample amount of 10 µg (100 µl of a 0.1mgmL 1 solution) was injected into the SEC. An amount of 10% of the total effluent was sprayed onto the MALDI target, resulting in a total amount of deposited sample of 1 µg. As can be seen, for all fractions high quality spectra were obtained giving the oligomer distributions of the different fractions. Depending on the complexity of a specific sample, MALDI/TOF is more or less capable of resolving different chemical structures. While this technique is very powerful in determining different endgroups in macromonomers and telechelics, it has its limitations when it comes to analysing copolymers. Due to the fact that the number of possible oligomers increases exponentially with the degree of polymerization, even for low molar masses very complex product mixtures are obtained which cannot be analysed solely by MALDI/TOF. In these cases it is unavoidable to combine a chromatographic prefractionation with a MALDI/TOF analysis. The usefulness of such a combination shall be demonstrated for a diblock copolymer of n-butyl methacrylate and methyl methacrylate, i.e. poly-n-butyl methacrylate (PnBMA)-b-PMMA. The sample under investigation was prepared by group transfer polymerization (GTP) resulting in structure (1) which follows and is typical. Typical spectra for fractions of different molar masses obtained from the SEC/MALDI/TOF experiment are

19 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION % Intensity (a) Mass/Charge Pulses % Intensity (b) Mass/Charge % Intensity Mass/Charge % Intensity Pulses % Intensity Pulses (c) Mass/Charge (d) Mass/Charge Figure 14 MALDI/TOF spectra of fractions obtained from SEC/MALDI/TOF analysis of a PMMA/PnBMA block copolymer. (Reproduced by permission from Pasch. 125 ) CH 3 CH 3 H CH 2 C CH 2 C COOC 4 H 9 COOCH 3 X (1) given in Figure 14. The higher molar mass fractions in Figure 14(a), (b) and (c) are characteristic for copolymer structures exhibiting typical mass increments of 100 Da for the MMA repeat unit and 142 Da for the nbma repeat unit. Even these narrow disperse fractions exhibit a multitude of different mass peaks (usually more than 100) indicating the high complexity of the fractions. The lower molar mass fraction in Figure 14(d) is very uniform with Y H respect to composition and thus differs from the molar mass fraction in Figure 14(a), (b) and (c). For the fraction in Figure 14(d), only peak-to-peak mass increments of 142 Da were observed which are typically for PnBMA. The chemical composition of the block copolymer was studied in detail by analysing the different mass peaks (see zoomed part of the spectrum in the insert of Figure 14b). Each peak in the spectrum could be assigned to one individual oligomer composition (nbma) X (MMA) Y. Considering the potential of MALDI/TOF in terms of versatility and sensitivity, the direct interfacing of LC and MALDI/TOF would be a highly attractive possibility. Given the experiences with the direct introduction of small matrix-containing liquid streams into high-vacuum

20 18 POLYMERS AND RUBBERS Injection valve Solvent syringe pump (a) (b) GPC column TEE Matrix syringe pump Reflectron Nebulizer Detector Nd: yttrium aluminum laser 43 min 45 min 47 min 49 min 51 min m /z (u) Computer Oscilloscope Figure 15 Aerosol MALDI apparatus configured for on-line SEC/MS (a) and five spectra obtained during the separation of PEG 1000 (b). GPC, gel permeation chromatography; TEE, T-shaped flow-splitter. (Reproduced by permission from Fei and Murray. 128 ) instruments, it took surprisingly long before a device for liquid introduction to MALDI was described. In some recent papers Murray and Russell 126,127 and Fei and Murray 128 discussed the on-line coupling of SEC and MALDI/TOF/MS. In an aerosol MALDI/SEC experiment, the effluent from the SEC column was combined with a matrix solution and sprayed directly into a TOF/MS. Ions were formed by irradiation of the aerosol particles with pulsed 355 nm radiation from a frequencytripled Nd : yttrium aluminum garnet laser. The ions were mass separated in a two-stage reflectron TOF instrument, and averaged mass spectra were stored every 11 sec throughout the SEC/MS experiment. Well-resolved MALDI/TOF spectra were obtained from commercial PEG 1000 and poly(propylene glycol) (PPG) 1000, see Figure Coupling with Fourier Transform Infrared Spectroscopy When analysing a complex polymer, very frequently the first step must be the determination of the gross composition. Only when the chemical structures of the polymer components are known can sophisticated separation techniques such as gradient HPLC or LC/CC be adapted to a specific analysis. The most frequently used techniques for a flash analysis are IR spectroscopy and SEC. IR spectroscopy provides information on the chemical substructures present in the sample, while SEC gives a first indication of the molar mass range. Information on both molar mass and composition is obtained when SEC or a comparable chromatographic method is combined with an IR detector. In the past, numerous workers have tried to use IR detection of the SEC column effluent in liquid flow cells. The problems encountered relate to obtaining sufficient signal-to-noise (S/N) ratio even with FTIR instruments, flow-through cells with minimum path lengths and mobile phases with sufficient spectral windows. Attempts to use FTIR detection with liquid flow-through cells and high performance columns have not been very successful due to the requirement of considerably less sample concentration for efficient separation. A rather broad applicability of FTIR as a detector in LC can be achieved when the mobile phase is removed from the sample prior to detection. In this case the sample fractions are measured in pure state without interference from solvents. Experimental interfaces to eliminate volatile mobile phases from HPLC effluents have been tried with some success but the breakthrough towards a powerful FTIR detector was achieved only by Gagel and Biemann, who formed an aerosol from the effluent and sprayed it on a rotating aluminum mirror. The mirror was then deposited in an FTIR spectrometer and spectra were recorded at each position in the reflexion mode. Recently, Lab Connections Inc. introduced the LC- Transform, a direct HPLC/FTIR interface based on the invention of Gagel and Biemann and discussed its first applications in polymer analysis The design concept of the interface is shown in Figure 16. The system is composed of two independent modules, the sample collection module and the optics module. The effluent of the LC column is split with a fraction (frequently 10% of the total effluent) going into the heated nebulizer nozzle located above a rotating sample collection disc. The nozzle rapidly evaporates the mobile phase while depositing a tightly focused track of the solute. When a chromatogram has been collected on the sample collector disc, the disc is transferred to the optics module in the

21 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 19 Sample separation Sample identification RI-detector Spectra GPC/ HPLC FTIR-spectrophotometer Collection module Optics module Figure 16 Schematic representation of the principle of coupled LC and FTIR spectroscopy. FTIR detector for analysis of the deposited sample track. A control module defines the sample collection disc position and rotation rate in order to be compatible with the run time and peak resolution of the chromatographic separation. Data collection is readily accomplished with software packages presently used for GC/FTIR. The sample collection disc is made from germanium which is optically transparent in the range cm 1.The lower surface of the disc is covered with a reflecting aluminum layer. As a result of the investigation a complete FTIR spectrum for each position on the disc and, hence, for each sample fraction is obtained. This spectrum bears information on the chemical composition of each sample fraction. The set of all spectra can be arranged along the elution time axis and yields a 3-D plot in the coordinates elution time FTIR frequency absorbance. One of the benefits of coupled SEC/FTIR is the ability to identify directly the individual components separated by chromatography. A typical SEC separation of a polymer blend is shown in Figure Two separate elution peaks 1 and 2 were obtained, indicating that the blend contained at least two components of significantly different molar masses. A quantification of the components with respect to concentration and molar mass, however, could not be carried out as long as the chemical structure of the components is unknown. The analysis of the chemical composition of the sample was conducted by coupled SEC/FTIR using the LC- Transform. After separating the sample with respect to molecular size, the fractions were deposited on the germanium disc and FTIR spectra were recorded continuously along the sample track. In total, a set of about 80 spectra was obtained which was presented in a 3-D plot, see Figure 18. The projection of the RI-response V e (ml) Figure 17 SEC separation of a binary blend, stationary phase: Ultrastyragel 2 ð linear C 10 5 Å, eluent: THF. Absorbance RT Wavenumber (cm 1 ) Figure 18 SEC/FTIR analysis of a binary blend, Waterfall representation.

22 20 POLYMERS AND RUBBERS Time (min) Wavenumber (cm 1 ) ABS Figure 19 SEC/FTIR analysis of a binary blend, Contour plot representation. 3-D plot on the retention time IR frequency coordinate system yielded a 2-D representation, where the intensities of the absorption peaks were given by a color code. Such a contour plot readily provides information on the chemical composition of each chromatographic fraction, see Figure 19. It was obvious that the chromatographic peaks 1 and 2 had different chemical structures. By comparison with reference spectra which are accessible from corresponding data bases, component 1 could be identified as PS, while component 2 was polyphenylene oxide. With this knowledge, appropriate calibration curves could be used for quantifying the composition and the component molar masses of the blend. Coupled SEC/FTIR becomes an inevitable tool when blends comprising copolymers have to be analysed. Very frequently components of similar molar masses are used in polymer blends. In these cases the resolution of SEC is not sufficient to resolve all component peaks: see Figure 20 for a model binary blend containing an additive. The elution peaks of the polymer components 1 and 2 overlapped and, thus, the molar masses could not be determined directly. Only the additive peak 3 at the low molar mass end of the chromatogram was well separated and could be quantified. A first indication of the composition of the present sample could be obtained from the contour plot in Figure 21. Component 3 showed typical absorption peaks of a phenyl benzotriazole and could be identified as a UV stabilizer of the Tinuvin type. Component 2 exhibited absorption peaks which were characteristic for nitrile groups and styrene units, while component 1 showed a strong ester carbonyl peak and peaks of styrene units. In agreement with the peak pattern of literature spectra, component 2 was identified as styrene acrylonitrile (SAN) copolymer. Component 1 could have been a mixture of PS and PMMA or a styrene methyl methacrylate copolymer. Since the FTIR

23 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 21 Rl-response V e (ml) Figure 20 SEC separation of a blend of two copolymers and an additive, chromatographic conditions (see Figure 17). spectra over the entire elution peak were uniform, it is more likely that component 1 was a copolymer. 3 One important feature of the SEC/FTIR software is that from the contour plot specific elugrams at one absorption frequency can be obtained. Taking the elugram at 2230 cm 1, which is specific for the nitrile group, the elution peak of the SAN copolymer could be presented individually. For the presentation of component 1 the elugram at the carbonyl absorption frequency was drawn. Thus, via the chemigram presentation the elution peak of each component is obtained, see Figure 22. In a relatively short period of time the LC-Transform system found its way into a large number of laboratories. Applications of the technique have been discussed in various fields. Willis and Wheeler demonstrated the determination of the vinyl acetate distribution in ethylene vinyl acetate copolymers, the analysis of branching in high-density polyethylene, and the analysis of the chemical composition of a jet oil lubricant. 147 Provder et al. 148 showed that in powder coatings all additives were positively identified by SEC/FTIR through comparison of the known spectra. Even biocides could be analysed in commercial house paints. The comparison Time (min) Wavenumber (cm 1 ) ABS Figure 21 Contour plot of the SEC/FTIR analysis of a blend of two copolymers and an additive.

24 22 POLYMERS AND RUBBERS Intensity Chemigram: cm 1 (Total) Chemigram: cm 1 (PS-b-PMMA) Chemigram: cm 1 (SAN) Chemigram: cm 1 (Tinuvin 326) Retention time (min) Figure 22 Chemigrams taken from the contour plot in Figure 21. of a PS/PMMA blend with a corresponding copolymer gave information on the chemical drift. In the analysis of a competitive modified vinyl polymer sample by SEC/FTIR some of the components of the binder could be identified readily (vinyl chloride, ethyl methacrylate and acrylonitrile), and an epoxidized drying oil additive was detected. 148 The analysis of styrene butadiene copolymers by combining interaction chromatography and FTIR has been demonstrated. 149,150 By using LC/CC it was possible to separate block copolymers and technical rubber mixtures with respect to chemical composition. The determination of the styrene : butadiene ratio and the fine structure of the butadiene units (cis/trans-, 1,2/1,4- units) was achieved by FTIR spectroscopy. The quality of the results from SEC/FTIR strongly depend on the surface quality of the deposited sample fractions. Cheung et al. demonstrated that the surfacewetting properties of the substrate dominate the deposit morphology 151 and the spectra fidelity, film quality, resolution and polymer recovery were considered. 152 For different interface designs it was found that the morphology of the deposited polymer film was a key parameter for quantitative measurements. 3.5 Coupling with Nuclear Magnetic Resonance Spectroscopy NMR spectroscopy is by far the most powerful spectroscopic technique for obtaining structural information about organic compounds in solution. Its particular strength lies in its ability to differentiate between most structural, conformational and optical isomers. NMR spectroscopy can usually provide all necessary information to identify unambiguously a completely unknown compound. The NMR detection technique is quantitative with individual areas in spectra being proportional to the number of contributing nuclei. The major drawback of NMR is the relatively low sensitivity in comparison to MS, another is the fact that structure elucidation of mixtures of unknown compounds with overlapping NMR signals is difficult and may be nearly impossible in cases with overcrowding signals in a small chemical shift region of the NMR spectrum. Therefore, in many cases it would be useful that a separation is performed prior to the use of NMR. For more efficient procedures, a direct coupling of separation with NMR detection would be the method of choice. 153 The direct coupling of LC with proton NMR has been attempted numerous times. Early experiments of coupled HPLC- 1 H-NMR were conducted in a stop-flow mode or with very low flow rates. This was necessary to accumulate a sufficient number of spectra per sample volume in order to improve the S/N ratio. Other problems associated with the implementation of online HPLC/NMR have included the need for deuterated solvents. However, with the exception of deuterium oxide the use of deuterated eluents is too expensive for routine analysis. Therefore, proton-containing solvents such as acetonitrile (ACN) or methanol must be used. To get rid of the solvent signals in the spectra, the proton NMR signals of the solvents have to be suppressed. Recent rapid advances in HPLC/NMR provide evidence that many of the major technical obstacles have been overcome. 157,158 With the development of more powerful NMR spectrometers combined with new NMR techniques for solvent suppression it became much easier to obtain well-resolved spectra in an on-flow mode. In particular, very efficient solvent-suppression techniques significantly improved the spectra during the HPLC/NMR run. 159,160 These techniques combine shaped radio frequency pulses, pulsed-field gradients, and selective 13 C decoupling to acquire high-quality spectra at on-flow conditions even with high HPLC gradients. Recently, even the direct coupling of supercritical fluid chromatography (SFC) with 1 H-NMR together with the monitoring of supercritical fluid extraction 164 as well as the coupling of CE and H-NMR have been reported. An overview of the applications of on-line HPLC- 1 H-NMR in organic chemistry was given by Albert. 153 The first steps of polymer analysis into coupled LC- 1 H-NMR were performed by Hatada et al. 168 They linked a size exclusion chromatograph to a 500 MHz proton NMR spectrometer and investigated isotactic PMMA. Using deuterated chloroform as the eluent and running the chromatography at a rather low flowrate of 0.2 ml min 1 they were able to accumulate well resolved proton spectra. From the intensities of the proton signals of the endgroups and the monomer units they

25 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 23 determined the number-average molar mass across the elution curve. In further investigations they developed an absolute calibration method for direct determination of molar masses and MMDs by on-line SEC- 1 H-NMR. 169 Ute reported on the chemical composition analysis of EPDM copolymers as a function of molar mass, and the monitoring of stereocomplex formation for isotactic and syndiotactic PMMA. 170 The analysis of a technical PEO with respect to chemical composition and degree of polymerization has been performed by Pasch and Hiller. 171 This investigation has been conducted under conditions which are common for HPLC separations, i.e. sufficiently high flow-rate, moderate sample concentration, and on-flow detection. Using an octadecyl-modified silica gel as the stationary phase and an eluent of ACN deuterium oxide 50 : 50 (v/v) the sample was separated into different functionality fractions, see Figure 23. The major fraction of the sample eluting between 14 and 25 min exhibited a partial oligomer separation. For structural identification of the fractions, the 1 H- NMR spectrometer was directly coupled via capillary tubing to the HPLC system. The injection of the sample into the HPLC system was automatically initiated by the NMR console via a trigger pulse when starting to acquire NMR data. Using an appropriate pulse sequence, both solvent resonances (ACN at 2.4 ppm and water at 4.4 ppm) could be suppressed simultaneously. As a result of the on-line HPLC/NMR experiment a contour plot of 1 H chemical shift vs retention time could be generated, see Figure 24. Owing to the efficient solvent suppression, the obtainable structural information relates to the entire chemical shift region. From the contour plot, four different elution peaks could clearly be identified and analysed with respect to chemical composition. The remarkable feature of this investigation was that even the low concentration components in peaks 1 3 could clearly be identified in the contour plot. Detailed structural information could be obtained from the individual NMR spectra of the fractions at the peak maximum, see Figure 25. This also gave the relevant structures (2 and 3). The first peak was identified as being PEG while the other fractions were alkylphenoxy PEOs. From the intensities of the endgroups and the ethylene oxide repeat units the average degree of polymerization for each fraction was calculated. Based on the total intensity distribution, a calculated chromatogram (or chemigram) was generated from the NMR contour plot. Comparing the real chromatogram (Figure 23) with the chemigram (Figure 24) an excellent agreement was obtained even recalling the oligomer separation pattern of the major fraction UV-response Retention time (min) Figure 23 HPLC chromatogram of a technical PEO. Stationary phase: RP-18; eluent: ACN deuterium oxide 50 : 50 (v/v). (Reproduced by permission from Schlotterbeck et al. 171 )

26 24 POLYMERS AND RUBBERS ACN Retention time (s) δ (ppm) Figure 24 Contour plot of chemical shift vs. retention time and chemigram of the on-line HPLC/NMR analysis of a technical PEO. (Reproduced by permission from Pasch and Hiller. 171 ) The analysis of FAE based surfactants by online HPLC- 1 H-NMR has been described by Schlotterbeck et al. 172 Using a reversed stationary phase and ACN deuterium oxide as the eluent, surfactant mixtures were separated with respect to the fatty alcohol endgroups. 1 H-NMR detection revealed the number of components, the chemical structure of the components, endgroups, and the chain length. Finally, the investigation of the tacticity of oligostyrenes by on-line HPLC- 1 H-NMR has been reported by Pasch et al. 173 The oligomer separation was carried out by hydrophobic interaction chromatography using isocratic elution with ACN on a reversed phase (RP) column RP-18. The chromatogram of an oligostyrene is shown in Figure 26. The first oligomer peak could be identified as being the dimer (n D 2), the next peak was identified as the trimer (n D 3) and, accordingly, the following peaks could be assigned to the tetramer, pentamer etc. The dimer peak appeared uniform, whereas for the following oligomers a splitting of the peaks was obtained. For n D 3 and n D 4 a splitting into two peaks was observed. For n D 5 and further, a splitting into three or more peaks occurred, which could be attributed to the formation of different tactic isomers. The analysis of the isomerism of the oligomers by HPLC/NMR is given in Figure In this experiment conventional HPLC grade ACN was used as the eluent and no deuterium lock was applied. These conditions required high stability of the NMR instrument and a very efficient solvent-suppression technique since 100% ACN must be suppressed. The obtainable structural information related to the entire chemical shift region: however, residual signals of the eluent were obtained at ppm and 1.3 ppm due to ACN and its impurities. The contour plot clearly revealed two signal regions, which could be used for analysis. These were the region of the methyl protons of the sec-butyl endgroup at ppm and the aromatic proton region of the styrene units at ppm. For the generation of the contour plot every 8 seconds a complete spectrum was produced by co-adding 8 scans. Accordingly, for the structural analysis 128 spectra were available over the entire retention time range. For the analysis of a separated oligomer, a minimum of four spectra could be used. These spectra bear selective information on the tacticity, even without completely separating the tactic isomers chromatographically. As has been shown recently by Krämer et al., online coupled SEC- 1 H-NMR can be used to monitor the chemical composition of random copolymers across the molar mass axis. 174 They investigated high-conversion poly(styrene-co-ethyl acrylate)s using dichloromethane as the solvent for SEC. The contour plot for a typical sample with a styrene ethyl acrylate (EA) ratio of 40 : 60 indicates that all characteristic spectral regions are accessible for analysis: see Figure 28. Residual solvent signals at ppm do not overlap with resonances of the polymer molecules. The chemical composition across the elution peak of the copolymers is shown in Figure 29. The data were calculated from the peak

27 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 25 HO(CH 2 CH 2 O) n H CH 2 O ACN CH 2 OH H 2 O (a) δ (ppm) (c) δ (ppm) CH 3 CH 3 CH O(CH 2 CH 2 O) n H 3 C 3 CH 3 CH 3 CH 3 C C CH 2 OCH 2 CH 2 O (CH 2 CH 2 O) m CH 2 CH 2 OH 7 CH 3 CH CH 3 Ar H Ar H (b) δ (ppm) (d) δ (ppm) Figure 25 Individual fraction spectra taken from Figure 24. (a) NMR spectrum obtained from peak 1 in Figure 24; (b) NMR spectrum from peak 2 in Figure 24; (c) NMR spectrum from peak 3 in Figure 24; (d) NMR spectrum from peak 4 in Figure 24. areas of the aromatic protons (d D ppm) and the oxymethylene protons (d D ppm) which were recorded during the SEC/NMR experiment. Owing to the low S/N ratio at the start and the end of the SEC elution curves, the chemical composition determination is less accurate than in the peak maximum. As can be seen, most of the copolymers exhibit constant chemical composition across the elution curves. This corresponds to the average chemical composition of the bulk sample. An exception is the EA-richest copolymer (styrene-ea 10.95), where at the high molar mass end of the elution curve an increased EA content is detected. At the same time, at the low molar mass end of the elution curve a strong tailing and an increased EA content is obtained. This strongly indicates that at high concentrations of EA in the monomer feed, copolymers with increased chemical and molar mass heterogeneity are obtained. 4 MULTIDIMENSIONAL LIQUID CHROMATOGRAPHY 4.1 Introduction Despite the fact that substantial progress has been achieved in recent years in size-exclusion and interaction modes of polymer chromatography, the need and use for multidimensional separation systems has increased. The main reason for that is the fact that nowadays most classes of macromolecules posses property distributions in more than one parameter (e.g. molar mass and chemical composition at the same time). It is obvious that n independent properties require n-dimensional methods for accurate (independent) characterization of all those parameters. Moreover, the separation efficiency of any single separation method is limited by the efficiency and selectivity of

28 26 POLYMERS AND RUBBERS UV (260 nm) Retention time (min) Figure 26 HPLC chromatogram of an oligostyrene PS 530. Stationary phase: RP-18; eluent: ACN. (Reproduced by permission from Pasch et al. 173 ) Retention time (s) Heptamer Hexamer Pentamer Tetramer Trimer Dimer F2 (ppm) Figure 27 Contour plot of chemical shift vs retention time of the on-line HPLC/NMR analysis of PS 530. (Reproduced by permission from Pasch et al. 173 ) the separation mode, i.e. the plate count of the column and the phase system selected. Adding more columns will not overcome the need to identify more components in a complex sample, due to the limitation of peak capacities. The peak capacity in an isocratic separation can be described, following Grushka, 175 as in Equation (19): p N n D 1 C 4 ln V p 19 V 0 The corresponding peak capacity in an n-dimensional separation is considerably higher due to the fact that ppm min Figure 28 Contour plot of chemical shift vs retention time of the on-line SEC/NMR analysis of a random copolymer of styrene and EA. (Reproduced by permission from Krämer et al. 174 ) each dimension contributes to the total peak capacity as a factor and not as an additive term for single dimension methods as described in Equation (20): n total D n i sin i 1 # i 20 where n total represents the total peak capacity, n i the peak capacity in dimension i and # i is the angle between two dimensions. The angle between dimensions is determined by the independence of the methods; a 90 angle is obtained by two methods, which are completely independent of each other and will, for example, separate two properties solely on a single parameter without influencing themselves. In the case of a 2-D system the peak capacity is given by Equation (21): n 2D D n 1 n 2 sin # ( p )( N1 D 1 C ln V p ) p,1 N2 1C ln V p,2 sin # 4 V 0,1 4 V 0,2 21 This effect is schematically illustrated in Figure 30. Multidimensional chromatography separations can be done in planar systems or coupled-column systems. Examples of planar systems include 2-D thinlayer chromatography (TLC), 176,177 where successive 1-D TLC experiments are performed at 90 angles with different solvents, and 2-D electrophoresis, where gel electrophoresis is run in the first dimension followed by isoelectric focusing in the second dimension. Hybrids of these systems where chromatography and electrophoresis are used in each spatial dimension were reported nearly 40 years ago. 181 The main problem using planar methods is the difficulty in detection and collection of fractions among other

29 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 27 Intensity PDA (a) (b) (c) SEA SEA SEA Retention time (min) EA in copolymer (mol-%) Figure 29 Chemical composition as a function of molar mass for random styrene EA copolymers of different average composition. (Reproduced by permission from Krämer et al. 174 ) less critical problems, such as homogeneous preparation of chromatographic media. However, the detection problem exists also for the coupled-column methods, mainly because of fraction dilution by each stage in a multidimensional separation system. Another aspect is the adjustment of chromatographic time bases between the different dimensions so that first dimension peaks may be sampled an adequate number of times by the next dimension separation system. This aspect has been recently studied in detail. 182 In 2-D column chromatography systems an aliquot from a column or channel is transferred into the next separation method in a sequential and repetitive manner. Storage of the accumulating eluent is typically provided by sampling loops connected to an automated valve. Many variations on this theme exist which use various chromatographic and electrophoretic methods for one of the dimensions. In addition, the simpler heart cutting mode of operation takes the eluent from a first dimension peak or a few peaks and manually injects this into another column during the first dimension elution process. A partial compilation of these techniques has been given in several places. The use of different modes of LC facilitates the separation of complex samples selectively with respect to different properties like hydrodynamic volume, molar mass, chemical composition or functionality. Using these techniques in combination, multidimensional information on different aspects of molecular heterogeneity can be obtained. If, for example, two different chromatographic techniques are combined in a crossfractionation mode, information on CCD and MMD can be obtained. Literally, the term chromatographic cross-fractionation refers to any combination of chromatographic methods capable of evaluating the distribution in size and composition of copolymers. An excellent overview on different techniques and applications involving the combination of SEC and gradient HPLC was published by Glöckner. 60 In SEC mode the separation occurs according to the molecular size of a macromolecule in solution, which is dependent on its chain length, chemical composition, solvent and temperature. Thus, molecules of the same chain length but different composition may have different hydrodynamic volumes. Since SEC separates according to hydrodynamic volume, SEC in different eluents can separate a copolymer in two diverging directions. This principle of orthogonal chromatography was suggested by Balke and Patel The authors coupled two SEC instruments together so that the eluent from the first one flowed through the injection valve of the second one. At any desired retention time the flow through SEC 1 could be stopped and an injection made into SEC 2. The first instrument was operated with THF as the eluent and PS gel as the packing, whereas for SEC 2 polyether bondedphase columns and THF heptane were used. Both instruments utilized SEC columns. However, whereas the first SEC was operating so as to achieve conventional molecular size separation, the second SEC was used to fractionate by composition, utilizing a mixed solvent to encourage adsorption and partition effects in addition to size exclusion. Consequently, independent information on both MMD and CCD could not be obtained from such an experiment.

30 28 POLYMERS AND RUBBERS Resolution enhancement by 2-D separation 1. Dimension peak capacity: 4 ϑ 2-D peak capacity: Dimension peak capacity: 3 Figure 30 Schematic contour map representation of increased resolution and peak capacity in 2-D separations (peaks in each dimension are indicated by bars at the axes). Since orthogonality requires that each separation technique is totally selective towards an investigated property, it seems to be more advantageous to use a sequence of methods, in which the first dimension separates according to chemical composition. In this way quantitative information on CCD can be obtained and the resulting fractions eluting from the first dimension are chemically homogeneous. These homogeneous fractions can then be analyzed independently in SEC mode in the second dimension to get the required MMD information. In such cases, SEC separation is strictly separating according to molar mass, and quantitative MMD information can be obtained. Examples illustrating the potential of multidimensional separations will be given in section Experimental Aspects of Multidimensional Separations Setting up a 2-D chromatographic separation system is actually not as difficult as one might think at first. As long as well-known separation methods exist for each dimension the experimental aspects can be handled quite easily in most cases. Off-line systems just require a fraction-collection device and something or someone who reinjects the fractions into the next chromatographic dimension. In on-line 2-D systems the transfer of fractions is preferentially done by automatic injection valves as was proposed by Kilz et al. 23,188,194 Figure 31 shows a general set-up for an automated 2-D chromatography system. The focal point in 2-D chromatography separations is the transfer of fractions eluting from the first dimension into the second dimension. This can be done in various ways. The most simplistic approach is by collecting Solvent delivery system 1 First dimension (horizontal) Inj1 Column 1 Detector 1 TV To waste TV: transfer valve Solvent delivery system 2 Column 2 Detector 2 To waste Figure 31 General experimental set-up for an 2-D chromatography system. fractions from one separation and manually transferring them into the second separation system. Obviously, this approach is prone to many errors, labor intensive and quite time-consuming. A more efficient way of fraction transfer can be achieved by using electrically (or pneumatically) actuated valves equipped with two injection loops. Such a set-up allows one fraction to be injected and analyzed from one loop while the next fraction is collected at the same time in the second loop (see Figure 32). Total mass transfer from the first to the second dimension can be guaranteed by proper selection of flow rates in both dimensions. 195 This is a very beneficial situation as compared to heart-cut transfers, since by-products and trace-impurities can be separated even if they are not visible (VIS) in the first dimension separation. Table 3 shows a summary of potential fraction transfer options. Second dimension (vertictal)

31 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 29 HPLC HPLC THF Waste THF Waste SEC SEC Position INJECT Position LOAD Figure 32 Fraction transfer between chromatography dimensions using a dual-loop 8-port valve. There are some other important aspects which have to be considered for optimum 2-D experiment design Selection of Separation Techniques Despite the fact that, historically, planar chromatography played an important role in multidimensional separations, this article will not discuss these aspects because they represent the past. The foreseeable future is with columnbased techniques, which allow a well-controlled transfer of samples between different methods. Obviously, destructive methods like GC and SFC, which destroy the chromatographic phase system, play a more limited role in multidimensional separations as they can only be used in the last separation step. Schure recently published a theoretical paper 196 which discussed different chromatographic method combinations on the basis of efficiency, sample dilution and detectability. He investigated CE, GC, LC, SEC and field-flow fractionation (FFF) in detail, while omitting other methods, which are potential candidates for method hyphenation, such as SFC and temperature-rising elution fractionation (TREF). Schure highlights several universal experimental factors (including plate count, injection volume, injected mass and injection band dilution), which should be taken into account when designing multidimensional separations. Table 4 summarizes Schure s results for the applicability of a given method in a multidimensional experiment. It is obvious that a low resolution, low injected mass method with high dilution of the Table 3 Summary of 2-D transfer injection options Transfer Mode Advantages Disadvantages Example Manual Off-line Very simple Fast set-up Automatic Off-line Simple Easy Fast set-up Single-loop On-line Correct concentrations Correct transfer times Automation Dual-loop On-line Correct concentrations Correct transfer times Quantitative transfer automation Time-consuming Not for routine work Not precise No correlation of fraction elution to transfer time Not quantitative Less precise No correlation of fraction elution to transfer time Not quantitative Transfer not quantitative Set-up time Set-up time Special valve Test tube Fraction collector storage valve Injection valve (with actuation) 8-port actuated valve Combination of 2 conventional 6-port injection valves Table 4 Synopsis of typical conditions and dilution factors in 1-D separations (Schure 196 ) Separation Experimental parameters Result mode dimensions (mm) d p (µm) N (1 col 1 Dilution f ) V inj (µl) GC 2500 ð 0.25 n/a LC 250 ð SEC 300 ð FFF 600 ð 2 ð n/a CE 400 ð 0.05 n/a e

32 30 POLYMERS AND RUBBERS Table 5 Calculated dilution factors for 2-D separations (Schure 196 ) assuming splitless transfer injections (experimental conditions similar to those of Table 4) 1-D 2-D GC LC SEC FFF CE a GC 226 n/a n/a n/a n/a LC SEC FFF CE b a Splitless transfer injections very difficult. b Transfer concentrations very small. injection band is a poor candidate for a multidimensional experiment. Schure also calculated parameters to estimate the potential of 2-D method combinations. Results for splitless transfer injections are given in Table 5. CE, LC and SEC were rated best. This theoretical result agrees very well with experimental results and the actual number of published papers on 2-D separations. The most widely used method combination currently is that of LC with SEC Sequence of Separation Methods This is an important aspect in order to get the best resolution and most accurate determination of property distributions. It is advisable to use the method with highest selectivity for the separation of one property as the first dimension. This ensures the highest purity of eluting fractions being transfered into the subsequent separation. In the case of gradient HPLC and GPC as separation methods, authors of early publications ,197,198 used GPC as the first separation, because it took much longer than a subsequent HPLC analysis. This is not the best set-up, however, because the GPC fractions are only monodisperse in hydrodynamic volume, not in molar mass, chemical composition, etc. On the other hand, HPLC separations can be fine-tuned using gradients to fractionate only according to a single property, which can then be characterized for molar mass without any bias. In many cases, interaction chromatography as the first dimension separation method is the best and most adjustable choice. From an experimental point of view, high flexibility is required for the first chromatographic dimension. In general, this is also more easily achieved when running the interaction chromatography mode in the first dimension, because: (i) More parameters (mobile phase, mobile phase composition, mobile phase modifiers, stationary phase, temperature etc.) can be used to adjust the (ii) (iii) separation according to the chemical nature of the sample. Better fine-tuning in interaction chromatography allows for more homogeneous fractions. Sample load on such columns can be much higher as compared to SEC columns, for instance Detectability and Sensitivity in the Second Dimension Because of the consecutive dilution of fractions, detectability and sensitivity become important criteria in 2-D experiment design. If byproducts and trace impurities have to be detected, only the most sensitive and/or selective detection methods can be employed. Evaporative light scattering detection (ELSD), despite several drawbacks, has been used mostly due to its high sensitivity for compounds which will not evaporate or sublime under detection conditions. Fluorescence and diode array UV/VIS are also sensitive detection methods, which can pick up samples at nanogram level. Mass spectrometers have a high potential in this respect too: however, they are currently not developed to a state where they would be generably usable. Only in rare cases has RI detection, otherwise very popular in SEC, been used in multidimensional separations, because of its low sensitivity and strong dependence on mobile phase composition. As a general rule, the higher the inject band dilution of a given separation method the more sensitive a subsequent detection method has to be. Such type of model calculations can be done easily; refer to Table 4 in section and the paper by M. Schure 196 for further details Other Experimental Factors Affecting Multidimensional Separations Depending on the specific type of the multidimensional experimental set-up, there are a number of other parameters to control and care about. Some are listed here, but because they are specific to the method combination, this list reflects only those techniques in most common use Influence of Eluent Transfer from First to Second Dimension A very important aspect in multidimensional chromatography design is the compatibility of mobile phases which are transferred between the different dimensions. It is a necessity that the mobile phases in two consecutive stages in multidimensional separations are completely miscible. Otherwise the separation in the second method is dramatically influenced and the fraction transfer is restricted or completely hindered. In gradient

33 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 31 systems, this requirement has to be verified for the total composition range. In SEC separations the transfer of mixed mobile phases can affect molar mass calibration. In order to get proper molar mass results, the calibration curves have to be measured using the extremes of mobile phase composition and tested for changes in elution behavior and pore-size influence in the SEC column packing. The better the thermodynamic property of the SEC eluent, the less influence is expected on the SEC calibration when the transfer of mobile phase from the previous dimension occurs. It has been shown to be advantageous to use the SEC eluent as one component of the mobile phase in the previous dimension to avoid potential interference and mobile phase incompatibility Time Consumption Time is an important issue when designing multidimensional experiments. Setup time itself plays only a lesser role, but the time needed for the multidimensional separations themselves can be considerable. This is especially true for 2-D separations using quantitative mass transfer via tandem-loop transfer valves. Heart-cut experiments require much less time and are often sufficient to check out the applicability of the approach. Cutting down on time consumption for multidimensional experiments is currently a heavily investigated topic. Several approaches are investigated and allow investigators to be optimistic and reduce experiment times by a factor of about 10 for complete mass transfer experiments using optimized column sets and flow conditions. Another time requirement in multidimensional separations is that needed to reduce the amount of data and present them in an instructive way. With several dozen transfers between dimensions, data reduction and presentation can be very time-consuming and has been a real burden for those who performed the first crossfractionation experiments. There is a clear ,198 need for specialized multidimensional software which does all the data acquisition, fraction transfer, valve switching, data reduction, data consolidation and presentation of results. Currently, there is only one 2-D chromatography system commercially available 23,199 which is widely used. A few laboratories use in-house solutions, which are specific to their own chromatography and data capture hardware and specific also to result calculation and report creation. 4.3 Separation Techniques for the First Dimension For an in-depth description of individual separation techniques used for multidimensional separations, please refer to the respective sections in this encyclopedia. This chapter deals with the specific aspects of the separation methods, which will help the reader to understand how to select one of them for a given multiple separation experiment Liquid Chromatography This is the most often used technique for multidimensional separations. LC can be performed in normal phase or RP systems using isocratic or gradient elution. There is an abundance of stationary phases with different types of surface modifications of different polarities. This flexibility in experimental parameters is a very important consideration when using LC as a first dimension method, since it can be fine-tuned to separate according to a given property more easily than most other chromatographic techniques. Gradient HPLC has been useful for the characterization of copolymers In such experiments careful choice of separation conditions is imperative. Otherwise, low resolution for the polymeric sample will obstruct the separation. On the other hand, the separation in HPLC, dominated by enthalpic interactions, perfectly complements the entropic nature of the SEC retention mechanism in the characterization of complex polymer formulations. LC separation is based on an enthalpic interaction between the solute and the surface of the stationary phase. In pure interactive LC separations entropic contributions to the retention are absent. The distribution coefficient K d can be derived from basic thermodynamics and can be measured from the activity of the analyte in the mobile (a m ) and stationary (a s ) phase as shown in Equation (22): K d LC D a s D e 1H RT 22 a m The enthalpy change of the analyte corresponds to dispersion, polarization and charge-transfer interactions as well as H-bonding and ion exchange. Obviously, the distribution coefficient is larger than unity in LC separations. The retention in pure LC separations can be calculated according to Equation (23): V e D V 0 C K d LC V pore 23 The absence of entropic contributions to the separation is only possible if the stationary phase consists of nonporous beads or if the analyte molecules cannot penetrate into any pore in the stationary phase because of their size or interaction energy (e.g. ionic repulsion). In general, it will not be possible to avoid entropy changes in LC experiments with samples of different molar masses or sizes. In such cases it is best to select either a column which has very small or very large pores, which will force

34 32 POLYMERS AND RUBBERS the molecules to be excluded totally from the stationary phase or to permeate totally the pores of the packing. In both situations entropic contributions to the separation can be minimized. In the general case, the distribution coefficient can be written as in Equation (24): K d D a s D e 1G RT a m D e 1H RT e 1S R D K d LC K d SEC 24 The elution in mixed-mode LC separations can be calculated according to Equation (25): V e D V 0 C [K d LC K d SEC ]V pore 25 This equation describes the general behavior of solutes in porous stationary phases. However, the dominant separation mechanism in LC with entropic contributions is the enthalpy term. It is interesting to study the retention dependence on molar mass in LC separations. This is especially important when applying this technique to macromolecules, where chain statistics, chemical composition and molecular size play an important role in chromatographic behavior. Figure 33 shows the adsorption characteristics of molecules of different chain length or molar mass. The adsorption process is determined by the interaction enthalpy which itself is governed by active sites on the analyte molecule. If molecules are small only very few interactive sites are present, which may differ in nature and possess different interaction energies. In such cases, retention is relatively small (K d less than about 10). Larger molecules, especially macromolecules, are composed of repeating units (usually called monomers), which can totally adsorb on the column packing. This is due to the fact that each and every repeating unit can potentially Log M Complete adsorption interact with the chromatographic surface. The longer the chains get, the higher the total interaction enthalpy and the higher the retention time to elude from the column (see Figure 33). The statistics of chain adsorption are determined by the magnitude of the entropy loss of the adsorbed macromolecule. If the entropy loss of the chain is small as compared to the adsorption enthalpy gain, the molecules will completely attach to the surface and the Gaussian chain will collapse into a 2-D layer. This scenario will dominate in cases where the solutes contain functional groups with strong interactions and where the eluent strength is relatively poor. In cases where the adsorption energy is relatively small, the entropy loss due to adsorption can be larger than the enthalpy gain. The chain only attaches selectively at the surface on the column packing forming loops. In such cases the macromolecules can desorb again. The adsorption/desorption process can be controlled by the nature of the stationary phase or even more easily by the thermodynamic properties of the eluents and gradient composition. Recently, temperature was also used to moderate adsorption behavior in chromatography. 205 Details of chain statistics in adsorption chromatography can be found in the book by Glöckner Liquid Chromatography at the Critical Point of Adsorption In LC and SEC chromatography modes either the enthalpy or the entropy dominate the separation. Several years ago Russian scientists found that homopolymers of different molar mass show exactly identical retention behavior on TLC plates 207,208 and on silica columns, 209 if a special eluent mixture was used for that macromolecule. They found that under critical conditions the sorbent did not see the polymeric nature of the chain. The separation was dependent only on the enthalpic interaction of the sample-sorbent pair (so-called critical chromatography or liquid chromatography at critical conditions. The LC/CC mode relates to a chromatographic situation where the entropic and enthalpic interactions of the macromolecule and the column packing compensate each other, as shown in Equation (26): 1G D 1H T1S D 0 26 Therefore, we can derive Equation (27): Retention Figure 33 Adsorption of macromolecules of different molar mass on interactive LC columns. K d D e 1G RT D 1 27 Rewriting this result for retention in elution volume terms, we directly get the experimentally observed result that the chromatographic peak position is independent

35 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 33 of the molar mass of the analyte and equal to the accessible volume of the stationary phase, which is stated in Equation (28): V e D V 0 C K d LC/CC V pore D V 0 C V pore D V t 28 The Gibbs free energy of the macromolecule remains constant when it penetrates the pores of the stationary phase (1G D 0). The distribution coefficient K d is unity, regardless of the size of the macromolecules, and all macromolecules of equal chemical structure elute from the chromatographic column in one peak. The term chromatographic invisibility is used to refer to this phenomenon. This means that the chromatographic behavior is not directed by the size, but by the heterogeneities (chemical structure, branching point, endgroup, etc.) in the macromolecular chains. In general, as the Gibbs free energy is influenced by the length of the polymer chain and its chemical structure, contributions G i for the polymer chain and G j for the heterogeneity may be introduced, as stated in Equation (29): 1G D n i 1G i C n j 1G j 29 For a perfectly uniform homopolymer chain the free energy change is determined by the contribution of the repeating units of the polymer chain (Equation 30): 1G D n i 1G i 30 At the critical point of adsorption of the polymer chain of a complex polymer, however, the contribution G i becomes zero and chromatographic behavior is exclusively directed by imperfections in the macromolecular chain (Equation 31): 1G D n j 1G j 31 This chromatographic effect can be employed to determine imperfections in the polymer chain selectively and without any contribution by the repeating units themselves. LC/CC has been successfully used for the determination of the FTD of telechelics and macromonomers, for the analysis of block copolymers, macrocyclic polymers, 221 and polymer blends Thus, LC/CC represents a chromatographic separation technique yielding fractions which are homogeneous in one property (e.g. chemical composition) but polydisperse in a different property (e.g. molar mass). These fractions can readily be analysed by SEC, which for chemically homogeneous fractions provides true MMDs without interference of CCD or FTD. Therefore, the combination of LC/CC and SEC in a 2-D chromatography experiment can be regarded as orthogonal chromatography in the strict sense provided that LC/CC is used as the first dimension separation mode. Consequently, for functional homopolymers being distributed in functionality and molar mass, coupling LC/CC with SEC can yield combined information on FTD and MMD. Such property information is important, e.g. for the quality control of amphiphilic polyalkylene oxides. There is another area where the 2-D combination of LC/CC and SEC is extremely useful and can give results no other technique can provide in a single experiment. The 2-D separation of segmented copolymers (such as block- or graft- or comb-shaped copolymers) allows the complete molecular characterization of the copolymer with regard to individual segment molar masses and composition. It has been demonstrated that with such a set-up the polydispersities of copolymer segments can be determined independently. 225 The thermodynamics of segmented copolymers is based on the idea that free energy change 1G AB of a segmented copolymer molecule, A n B m, is the sum of the contributions of segments A and segments B, 1G A and 1G B, respectively, which can be expressed by Equation (32): 1G D n A 1G A C n B 1G B C c AB 32 where c AB is the Flory Huggins parameter describing the interactions between segments A and B. It has been demonstrated for a number of block copolymers 27,34 that no specific interactions between the heterosegments AandB(c AB D 0) can be measured by chromatography. Using this assumption the change in the Gibbs free energy is solely dependent on the energy contributions of segments A and B, as is shown in Equation (33): 1G D n A 1G A C n B 1G B 33 Applying experimental conditions, which correspond to the critical point of homopolymer A, the A segment in the segmented copolymer will become chromatographically invisible, as expressed in Equation (34): 1G A D 0 34 Consequently, the retention of the segmented copolymer will be determined solely by the chromatographic properties of segment B. This is shown in Equation (35): 1G AB D n B 1G B 35 This also means that the distributions coefficient for this system, Kd AB can be reduced to the very simple term which is used for homogeneous molecules (Equation 36): K AB d D K B d 36

36 34 POLYMERS AND RUBBERS Repetition of such an experiment using critical conditions for segment B allows the determination of molecular parameters for the other segment A in the copolymer. The same equations derived above apply, just the parameters for segments A and B are exchanged. Critical conditions for segment B mean that Equation (37) holds: 1G B D 0 37 and the copolymer shows only the chromatographic behavior of segment A (Equation 38): Solvent Degasser Pump TREF column Concentration Detector Elution temperature 1G AB D n A 1G A The retention is then given by Equation (39): 38 Pulse dampener Temperature controller K AB d D K A d 39 The characterization of ABA triblock copolymers can be done in a similar way to the analysis of diblock copolymers. The two possible cases for this type of investigation are: (a) analysis with respect to the center block B using the critical conditions of the outer blocks A (b) the analysis of the terminal A segments using the critical conditions of the center block B. It is particularly useful to carry out experiments at the critical point of A. The separation occurs then with respect to the chain length of B, yielding fractions which are monodisperse with respect to B and polydisperse withrespecttoaanda 0. These fractions can be analyzed selectively with respect to the outer blocks A and A 0 in separate experiments Temperature-rising Elution Fractionation This is not, strictly speaking, a chromatographic technique but it uses the same equipment and leads to detector traces which resemble chromatograms. This method is widely used for the short-chain branching characterization of polyolefins in the petrochemical industry. Many polyolefins are in fact copolymers; the second comonomer introduces short-chain branches into the macromolecule. The copolymer properties depend strongly on the average composition and its sequence length distribution. In the case of polyethylene copolymers, changes in composition change the ease and temperature at which the polymer chains can crystallize. This property allows for the determination of composition distribution by measuring TREF. The TREF techniques rely on the fact that the redissolution temperature of a precipitated sample depends strongly on the number of short-chain branches. A chromatographic pump will transport the redissolved species through the column Figure 34 Schematic layout of a TREF instrument with thermostated TREF column. loaded with precipitated polymer into a concentration detector, which in turn will measure the concentration of the fraction redissolved at a given temperature, T, as shown in Figure Separation Techniques for the Second Dimension The LC methods described in section 4.3 can also be used in the subsequent separation stage in a multidimensional chromatography set-up. However, as pointed out earlier (cf. section 4.2), it is advantageous to use gradient LC or LC/CC in the first separation dimension. On the other hand, SEC is preferentially used as the second method to retrieve molar mass information. In theory, TREF can also be used in a later separation step; however, for practical reasons, this is not advisable. There are a number of chromatographic separation methods which can only be used in the last stage of a multidimensional experiment Size-exclusion Chromatography SEC is described in the section of techniques for the second dimension, because of its primary benefits there (cf. section 4.2.2). However, it can also be used with lower efficiency and more biased results in the first separation dimension. This technique was developed for the separation and characterization of large molecules. It is also called Gel Filtration Chromatography for natural and biopolymers and known as GPC for synthetic polymers. The principles are the same in both cases and rely on the fact that the macromolecule can only partially penetrate the porous packing, depending on its molecular size (not molar mass). The molecular size of a macromolecule in solution, or more accurately its hydrodynamic volume,

37 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 35 will depend on its chain length, the nature of the repeating units, chemical composition, molecular topology and the thermodynamic quality of the solvent. The same dependencies exist for the distribution coefficient in SEC, which can lead to the co-elution in SEC of species having identical hydrodynamic radius, but different composition, molecular architecture, and so forth. Large molecules can only access pores larger than the hydrodynamic radius of the molecule and will elute from the column early. If macromolecules are larger than the biggest pores, they will be totally excluded from the pore volume and will not be separated into fractions of different molar mass. On the other hand, the smallest molecules might be able to penetrate into all pores in the packing. They also will not be separated and will all co-elute at the total permeation volume. This chromatographic behavior is illustrated in Figure 35. Similarly to the thermodynamic treatment of LC, retention in SEC can be described by basic thermodynamic entities and can be determined by measuring the concentration of the molecule in the stationary and mobile phases, as expressed in Equation (40): K d SEC D a s D e 1S R 40 a m In ideal SEC separations the retention in the column is only governed by the entropy loss when the macromolecule enters the pore of the packing. No enthalpic interaction should be present in order to allow for accurate molar mass determinations. The retention in ideal SEC experiments is given by Equation (41): Log M Complete exclusion V 0 V e D V 0 C K d SEC V pore Retention Total penetration V t 41 Figure 35 Size-exclusion behavior of macromolecules of different size on a porous column packing. This equation looks very similar to Equation (23). The only difference is the magnitude of K d, which is always less than unity for SEC separations and reaches a maximum value of unity for molecules which enter all pores. Ideal SEC conditions are difficult to obtain for real macromolecules, however. As in the case of LC, there are contributions from both entropic and enthalpic terms to the distribution coefficient (cf. Equation 25). In such cases, the retention for molecules with enthalpic interactions is higher and they will elute later from the column. This behavior can be described by Equation (42): V e D V 0 C [K d SEC K d LC ]V pore 42 In this equation the term K d LC V pore describes the delayed elution from the column as compared to the ideal case. This equation is mathematically identical to the respective equation for LC. However, the dominant parameter is K d (SEC) and the entropy change governs the penetration of the analyte into the pores Capillary Electrophoresis CE is a very efficient microseparation method (typically N > ), which uses a strong electric field to create an electro-osmotic flow in which the species will migrate. The reason for that is that the surface of the silicate glass capillary contains negatively-charged functional groups that attract positively-charged counterions. The positively-charged ions migrate towards the negative electrode and carry solvent molecules in the same direction. This overall solvent movement is called electroosmotic flow. During a separation, uncharged molecules move at the same velocity as the electro-osmotic flow (with very little separation). Positively-charged ions move faster and negatively-charged ions move more slowly. CE can also be used as a first set in multidimensional separation, but its practical use here is very limited due to the minute sample amounts injected. For further information and details on the CE technique, please consult the articles in this encyclopedia Supercritical Fluid Chromatography SFC is a relatively recent chromatographic technique which was commercialized in the early 1980s. In SFC, the sample is carried through a capillary or packed column by a supercritical fluid (typically carbon dioxide). The properties of the mobile phase can be modified easily by polar additives and/or pressure programming, just as in gradient HPLC, to optimize selectivity. All three basic modes of chromatography (interaction, size-exclusion and critical conditions) have been verified in SFC separations. 226 SFC is a very

38 36 POLYMERS AND RUBBERS efficient separation technique, which has most of its applications in low molecular weight separations. SFC has several advantages over conventional chromatographic techniques. SFC separations can be done considerably faster than HPLC separations, because the diffusion of solutes in supercritical fluids is about 10 times greater than that in liquids (and about 3 times less than in gases). This results in a decrease in resistance to mass transfer in the column and allows for fast high resolution separations. Compared with GC, capillary SFC can provide high resolution chromatography at much lower temperatures. This allows fast analysis of thermolabile and nonvolatile compounds. These advantages make SFC a good choice for multidimensional chromatography set-ups. Since SFC is a mobile phase-destroying technique, it can only be used in the last separation step in multidimensional separations Gas Chromatography As with SFC, GC is a mobile phase-destroying technique and can only be used in the last stage of multidimensional separations. It also shares with SFC the high efficiency and speed of separations. However, it is limited to relatively low molecular weight compounds which are volatile without degradation and thermostable. Most multidimensional applications reported in the literature use the first dimension for precleaning and removal of high molar mass species and not for complete characterization of the samples. Examples of such applications are the removal of humic acids from pesticides in soil extracts by SEC/GC and of high molecular byproducts from mono, di, and triglycerides. For an indepth description of the technique, please refer to the Gas Chromatography in this encyclopedia. 4.5 State-of-the-art of On-line Coupled Two-dimensional Chromatography This section will illustrate the current state and future potential of 2-D chromatography by reviewing separations published in the literature. Examples will be given for different separation techniques and combinations of these Two-dimensional Separations by Liquid Chromatography and by Size Exclusion Chromatography Much work on chromatographic cross-fractionation was carried out with respect to the combination of SEC and gradient HPLC. In early experiments SEC was used as the first separation step, followed by HPLC. In a number of early papers the cross-fractionation of model mixtures was discussed. Investigations of this kind demonstrated the efficiency of gradient HPLC for separation by chemical composition. Mixtures of random copolymers of styrene and acrylonitrile were separated by Glöckner et al. 198 In the first dimension an SEC separation was carried out using THF as the eluent and PS gel as the packing. In total, about 10 fractions were collected and subjected to the second dimension, which was gradient HPLC on a CN bondedphase using iso-octane/thf as the mobile phase. Model mixtures of random copolymers of styrene and 2- methoxyethyl methacrylate were separated in a similar way, the mobile phase of the HPLC mode being isooctane/methanol in this case. 227 This procedure was also applied to real-world copolymers. 198 Graft copolymers of methyl methacrylate onto EPDM were analyzed by Augenstein and Stickler; 228 whereas Mori reported on the fractionation of block copolymers of styrene and vinyl acetate. 229 For all these experiments the same limitation with respect to the SEC part holds true: when SEC is used as the first dimension, true MMDs are not obtained. From the theoretical point of view, a better copolymer separation set-up is prefractionation through HPLC in the first dimension and subsequent analysis of the fractions by SEC. HPLC was found to be rather insensitive towards molar mass effects and yielded very uniform fractions with respect to chemical composition. 230,231 The major disadvantage of all early investigations on chromatographic cross-fractionation was related to the fact that both separation modes were combined with each other either off-line or in a stop-flow mode. Regardless of the separation order SEC vs HPLC or HPLC vs SEC, in the first separation step fractions were collected, isolated, and then subjected to the second separation step. This procedure, of course, is very time-consuming and the reliability of the results, at least to a certain extent, depends on the skills of the operator. A fully automated 2-D chromatographic system was developed by Kilz et al. 23,188,194 The operation of the column-switching device is automatically driven by the software, which at the same time organizes the data collection from the detector. One of the very few applications of 2-D gradient HPLC/SEC was published by Kilz et al. and described the analysis of styrene butadiene star polymers. 188 Fourarm star polymers based on poly(styrene-b-butadiene) were prepared by anionic polymerization to give samples with well-known structure and molar mass control. In a first reaction step, a poly(styrene-b-butadiene) with a reactive chain end at the butadiene was prepared. This precursor reacted with a tetrafunctional terminating agent to give a mixture of linear (of molar mass M), 2-arm (2M), 3-arm (3M) and 4-arm (4M) species. Four samples with varying butadiene content (about 20%, 40%, 60%, and

39 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 37 80%) were prepared in this way. A mixture of these samples was used for the 2-D experiment. Accordingly, a complex mixture of 16 components, resulting from the combination of four different butadiene contents and four different molar masses (M, 2M, 3M, 4M) had to be separated with respect to chemical composition and molar mass. Initially, the 16-component star block copolymer was investigated by SEC. As can be seen in Figure 36, four peaks were obtained. They correspond to the four molar masses of the sample consisting of species with one to four arms. The molar masses are defined by the number of arms and were in the ratio M 2M 3M 4M. Despite the high resolution, the chromatogram did not give any indication of the very complex chemical structure of the sample. Even when pure fractions with different chemical composition were investigated, the retention behavior did not show significant changes as compared to the sample mixture. In each case a tetramodal MMD was visible, indicating the different topological species. The SEC separation alone did not show any difference in chemical composition of the samples, which varied from 20% to 80% butadiene content. Running the sample mixture in gradient HPLC mode gave poorly resolved peaks, which might suggest different composition, but gave no clear indication of different molar mass and topology, see Figure 37. UV-response Retention time (min) Figure 37 Gradient HPLC chromatogram of the 16-component star block copolymer mixture; separation is dominated by chemical composition and does not reveal the complexity of the sample. A RI-response V e (ml) Injections B C M (g mol 1 ) Figure 36 SEC chromatograms of the 16-component star block copolymer mixture (A) overlayed with styrene butadiene 60 : 40 (B) star copolymer and 20 : 80 (C); peaks correspond to molar mass separation of different arm numbers. Figure 38 3-D view of the 16-component star block copolymer mixture; each trace represents an SEC chromatogram from a transfer injection. The combination of the two methods in the 2-D set-up dramatically increased the resolution of the separation system and gave a clear picture of the complex nature of the 16-component sample. A 3-D representation of the gradient HPLC/SEC separation is given in Figure 38. Each trace represents a fraction transferred from HPLC to SEC and reflects the result

40 38 POLYMERS AND RUBBERS of the SEC analysis in the second dimension. Based on the composition of the sample, a contour map with the coordinates chemical composition and molar mass is expected to show 16 spots, equivalent to the 16 components. Each spot would represent a component which is defined by a single composition and molar mass. The experimental evidence of the improved resolution in the 2-D analysis is given in Figure 39. This contour plot was calculated from experimental data based on 28 transfer injections. The contour plot clearly revealed the broad chemical heterogeneity (y-axis, chemical composition) and the wide MMD (x-axis) of the mixture. The relative concentrations of the components were represented by colors. Sixteen major peaks were resolved with high selectivity. These correspond directly to the components. For example, peak 1 corresponds to the component with the highest butadiene content (80%) and the lowest molar mass (molar mass 1M) whereas peak 13 relates also to a molecule with 80% butadiene content but a molar mass of 4M. Accordingly, peak 16 is due to the component with the lowest butadiene content and a molar mass of 4M, representing a 4-arm star block copolymer with a styrene butadiene content of 80 : 20. A certain molar mass dependence of the HPLC separation is indicated by a drift of the peaks for components of similar chemical composition: see peaks , for example. This kind of behavior is normal for polymers, because pores in the HPLC stationary phase lead to size-exclusion effects which overlap with the enthalpic interactions at the surface of the stationary phase. Consequently, 2-D separations of this type will in general be not orthogonal but skewed, depending on the pore size distribution of the stationary phase and the nature of the sample. The quantitative amount of butadiene in each peak could be determined via an appropriate calibration with samples of known composition. The molar masses could be calculated based on a conventional molar mass calibration of the second dimension. The mapping of ethoxylated fatty alcohols and ethylene oxide propylene oxide block copolymers by 2-D chromatography was discussed by Trathnigg et al. 232 They Styrene content (%) Relative concentration SEC molar mass (D) Figure 39 Contour plot of the 16-component styrene butadiene star block copolymer mixture characterized by the 2-D HPLC/SEC separation.

41 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION 39 mv Alk Alk Cycles Alk OH HOOC COOH HO COOH HO OH Ether Ether Retention time (min) Figure 40 LC chromatogram of an aliphatic polyester sample at critical conditions; peak labels identify endgroups or functionality. HD combined LAC and SEC and were able to determine the CCD and the MMD of the polyethers Two-dimensional Separations by Liquid Chromatography at Critical Conditions and Size Exclusion Chromatography The analysis of aliphatic polyesters with respect to FTD and MMD has been demonstrated. 188,233 Polyesters from adipic acid and 1,6-hexanediol are manufactured for a wide field of applications with an output of thousands of tonnes per year. They are intermediates for the manufacture of polyurethanes, and their FTD is a major parameter affecting the quality of the final products. In particular, nonreactive cyclic species are responsible for the fogging effect in polyurethane foams. For the separation of the polyesters with respect to functionality, LC/CC was used, the critical point of adsorption of the polymer chain corresponding to an eluent composition of acetone hexane 51 : 49 (v/v) on silica gel. The critical chromatogram of a polyester sample together with the functionality fraction assignment is given in Figure 40. The ether peaks could be attributed to the formation of ether structures in the polyester samples by a condensation reaction. The MMDs of the functionality fractions could be determined by preparatively separating the fractions and subjecting them to SEC. The SEC chromatograms of fractions 1 9 are summarized in Figure 41. For a number of fractions oligomer separations were obtained, which could be used to calibrate the SEC system. The very complex nature of the sample could be verified in a 2-D experiment using LC/CC as the first dimension to separate for functionality and SEC in the second dimension to determine molar masses. The contour plot in Figure 42 reveals the structural complexity of the sample including the functionality fractions from LC/CC and the oligomer separations from SEC which were wellrecognizable. The sample was prepared from an adipic acid-rich reaction mixture resulting in an acid number of about 5. The high content of dicarboxylic acid endgroups is clearly reflected in the contour map. Quantification of Fr.2 Fr.1 Fr.3 Fr.4 Fr.6 Fr.5 Fr.9 Fr.8 Fr.7 V e (ml) Figure 41 SEC separation of different fractions taken from the LC/CC separation of the aliphatic polyester sample (cf. Figure 40). the contour plot yielded quantitative information on both FTD and MMD. The analysis of an octylphenoxy-terminated PEO with respect to FTD and MMD has been demonstrated by Adrian et al. 225 A separation of the sample with respect to the terminal groups could be achieved by LC/CC on an RP-18 stationary phase and a critical eluent composition of methanol water 86 : 14 (v/v). All peaks in the chromatogram could be identified by MALDI/TOF MS as pure fractions of different functionality, proving that the separation followed the chemical structure of the endgroups. The combination of LC/CC with SEC in a 2-D chromatography experiment resulted in the contour plot shown in Figure At the abscissa, the retention volume of the SEC runs (second dimension) is given, whereas the ordinate gives the retention volume of the LC/CC (first dimension). Relative concentrations are mapped to a color code on a log scale to make small quantities visible.

42 40 POLYMERS AND RUBBERS Endgroups (by LC-CC) [-O-] n HO-OH HO-COOH HOOC-COOH Alk-OH Cycles Alk-Alk SEC volume (ml) Relative concentration Figure 42 Contour map of the 2-D LC/CC and SEC separation of the aliphatic polyester with acid number of about 5; functionality (on y-axis) was calibrated by model compounds. The contour plot (cf. Figure 43) clearly reveals five spots corresponding to the five functionality fractions, fraction 2 being the main fraction containing the a- octylphenoxy-w-hydroxy oligomers. In addition, a,wdi(octylphenoxy) oligomer fractions and fractions having butylphenoxy endgroups are identified. The 2-D experiment yielded separation with respect to functionality and molar mass, and FTD and MMD could be determined quantitatively. For calculating FTD, the relative concentration of each functionality fraction must be determined. These concentrations are equivalent to the volume of each peak in the contour plot. With the appropriate software this can be done easily. Determination of the MMD for each fraction was possible after calibrating chromatograph 2 with PEO calibration standards. Calculation of the MMD could then be achieved in the usual way, taking one chromatogram for each functionality fraction, preferably from the region of the highest peak intensity. In similar approaches other polyalkylene oxides have been analysed by 2-D chromatography. Murphy et al. 234 separated PEGs and Brij type surfactants according to chemical composition and molar mass by RP HPLC versus SEC. The analysis of methacryloyl-terminated PEOs by LC/CC versus SEC was described by Krüger et al. 233 The functionality-type separation was conducted on an RP system at a critical eluent composition of ACN water 43 : 57 (v/v). The functionality fractions, including PEG, a-methoxy-w-hydroxy, a-methoxy-wmethacryloyloxy, and a,w-di(methacryloyloxy) PEO, were identified by MALDI/TOF MS. Finally, a technical C 13,C 15 -alkoxy-terminated PEO was analysed by Pasch and Trathnigg using LC/CC vs SEC Two-dimensional Combination of Different High Performance Liquid Chromatography Modes The deformulation of alcohol ethoxylates, which are an important class of nonionic surfactants, by 2-D chromatography has been published by Murphy et al. 236 This class of products possesses a polydispersity of

43 COUPLED LIQUID CHROMATOGRAPHIC TECHNIQUES IN MOLECULAR CHARACTERIZATION O O n Elution volume HPLC (ml) O O n HO O n HO O n Elution volume GPC (ml) Figure 43 2-D separation of octylphenoxy-terminated PEO shown as a contour plot; the labels indicate the molecular structure verified by MALDI/MS. methylene groups and a distribution of ethylene glycol units in the PEG segment of the nonionic surfactant. The authors established chromatographic methods for both heterogeneities independently and then combined them later for the 2-D investigations. They used an RP system with a 3 µm C-18 column with 100 Å pore width to separate the alcohol alkyl chains using an isocratic mobile phase (methanol water (95 : 5)) (see Figure 44). The PEG segments were separated on a normal phase system comprising a 3 µm nonmodified silica with 70 Å pores run with a concave water ACN gradient (see Figure 45). Both chromatograms show high resolution separations for the alcohol groups or the PEG MMD, without any indication of any other unresolved property. The combination of both techniques in a 2-D experiment revealed the complex nature of the surfactant. The authors applied the gradient normal phase liquid chromatography (NPLC) separation of PEG as the first dimension with the reversed phase liquid chromatography (RPLC) as the second dimension using ELSD as a highly sensitive means of detection. Figure 46 shows the 2-D contour map which displays both the alkyl chain separation on the x-axis and the ethylene oxide chain length UV Time (min) Figure 44 RP separation of FAE non-ionic surfactant for ethylene oxide content determination. on the y-axis. The separation is not truly orthogonal, which the authors attribute to the concave gradient in the NPLC separation. This separation clearly demonstrates the increased peak capacity of 2-D experiments. The high resolution separation in NPLC showed very high peak capacities of about 15, whereas the RPLC experiment resulted in an extremely good separation into the four