Preparative TREF

A somewhat less sophisticated TREF system for fractionating low-density polyethylenes was described by Shirayama, Okada and Kita [1]. In their system (Fig. 2) a larger quantity of polymer (4 g) was fractionated. It was coated onto sea sand (1400 g) by slow-cooling a hot xylene solution in the presence of the packing and added to the 7 x 38 cm column. Fractionation was carried out stepwise over the temperature range 50-80 °C using a temperature controlled oil bath to provide the constant temperatures. Up to 10 fractions were recovered over this temperature range. 

The studies of Shirayama et al. [1] were instrumental in demonstrating the practical application of TREF and encouraged the further development of the technique for polyethylene analysis. This lead directly to the development of a system which was large scale, robust and continuous. This type of system, has 

Fall-out from the considerable activity in the development of liquid chromatography systems has shown itself in some of the latest experimental variations for conducting preparative TREF. In addition to the use of valves, tubing, fittings and pumps normally associated with LC systems, progratmned ovens have also 

Polymer Sample Size (g) Solvent Column Dimensions (cm) Packing Material Cooling Rate K/hour Heating Rate K/hour Fraction Analysis Ref. 

LDPE 4 Xylene 38x7 Sand Slow Stepwise IR, IV [1] 

---

Aust et al. [128] have used the molar mass fractionation first on a medium density polyethylene, and Faldi and Soares [129] the composition fractionation first on an LLDPE resin. One should choose the fractionation technique that results in the most discriminated fractions [80] in the first step. The most general approach is to use preparative TREF fractionation because the CCD is usually more discriminating than the MMD in complex polyolefins. 

Because of the significant experimental differences involved, a discussion of techniques is best separated into two parts, one concerning preparative TREF and the other relating to the operation of analytical TREF. For the purpose of this review we define preparative as a finctionation in which fractions are recovered for further evaluation and, thus, relatively large samples (1-10 g) are separated. 

Fig. 1. Glass preparative TREF apparatus used by Wijga, van Schooten and Boerma [6] Reprinted with permission, Hiithig Wepf Verlag Basel...

Fig. 4. Relationship between elution temperature and methyl content for preparative TREF fractions from both high and low pressure polyethylenes. Reprinted with permission [8]...

Fig. 5. Schematic diagram of the preparative TREF system (a) oven (b) three-way valve (c) column (70-mm i.d., 1000 mm) (d) thermoelectric couple (e) preheating column (/) stop valve (g) relief valve (h) pressure gauge (i) pump (/) filter (fe) solvent tank (f) solution tank (m) constant-pressure valve (n) Nj inlet, (Usami, Gotoh and Takayama [10]) Reprinted with permission copyright 1986 Amer. Chem. Soc,...

As the data in Fig. 7 suggests there is certainly room for improvement in terms of the resolution achieved by preparative TREF. It is anticipated that future developments in the preparative TREF area will concentrate on strategies for achieving greater fractionation efficiency. 

Table 1 summarized the various preparative TREF systems described in the literature. 

Fig. 16. Comparison of elution fractionation temperatures for linear PE fractions with the calibration curve obtained using preparative TREF fractions from branched PEs [9]. Reprinted with permission...

Kelusky et al. also indicated that by use of a semi preparative TREF system one could achieve an effective characterization of PE/EPDM terpolymer and PE/polyisobutylene blends. The approach for this type of analysis of blends, in which one of the components was non-crystalline, was to first collect the PIB or EPDM by elution at 30 °C followed by elution of the PE component as an analytical TREF or it could be eluted off, precipitated and recovered for subsequent analysis by SEC or NMR. They point out that the TREF separation of this type of mixture is superior to the usual solvent extraction methods which invariably remove some highly branched or low molecular weight PE in addition to the PIB or EPDM. One would anticipate that such a separation scheme would be effective for other crystalline/amorphous polymer blends such as impact polypropylenes which contain ethylene-propylene rubber. 

Nakatio and Goto used a highly automated system speciaUy designed for cross fractionation in which a SEC is interfaced with a semi-preparative TREF. This has been described in the analytical TREF section above. They demonstrated the principle of complete cross fractionation for analysing a polyethylene blend. The results are shown in Fig. 40 and earlier in Fig. 12 and 13. 

The initial use of analytical TREF was to characterize polyethylene fractions [120]. Mirabella and Ford used TREF to fractionate LDPE and LLDPE for further characterization using SEC, X-ray C-NMR, DSC and viscosity measurements [121]. These investigators determined that the melting behavior of LLDPE correlated well with a multimodal SCB distribution and the distribution became narrower with increasing molecular weight. Mingozzi and collaborators [122] have applied both analytical and preparative TREF to the analysis of tacticity distribution in polypropylene. The results showed that both methods could be used successfully analytical TREF gave faster qualitative results on the polymer microstructure, while preparative TREF, with subsequent analysis of the fractions, could yield detailed quantitative tacticity information. Hazlitt has described an automatic TREF instrument to measure SCB distributions for ethylene/ a-olefin copolymers [123]. 

The authors conducted a similar investigation of precatalysts 36 and 31 using TiBA and trityl tetrakis(pentafluor-ophenyl)borate as the cocatalyst. They concluded that this material contained no fraction that could be characterized as blocky. It was therefore proposed that reversible chain transfer occurred only with MAO or TMA and not with TiBA. This stands in contrast to the work of Chien et al. and Przybyla and Fink vida supra), who claim reversible chain transfer with TiBA in similar catalyst systems. Lieber and Brintzinger also investigated a mixture of isospecific 36 and syndiospedfic 33 in attempts to prepare iPP/sPP BCPs. Extraction of such similar polymers was acknowledged to be difficult and even preparative TREF was only partially successful. 

To support this hypothesis, the OBC sample can be fractionated by a TREF experiment. Preparative TREF fractionation of the OBC, followed by evaluation of the comonomer content by C NMR, reveals the data shown in Figure 29. For traditional RCPs produced with the same comonomer type, a distinct relationship between the elution temperature and comonomer content is observed, indicating that regardless of the catalyst nature, the fractions from these RCPs have a statistically random distribution of comonomer. Historically, Wild has demonstrated that the peak elution temperature direaly relates to the degree of SCB in a copolymer. Thus, according to this behavior, each molecule present in a polymer blend will dissolve and elute according to its comonomer content. The results are expected to follow a calibration line in which TREF behavior can be predicted, which is labeled Random copolymer line in Figure 29. 

---