Biomacromolecules 2004, 5, 97-105
97
Size and Structure Characterization of Ethylhydroxyethyl Cellulose by the Combination of Field-Flow Fractionation with Other Techniques. Investigation of Ultralarge Components Mats Andersson,§ Bengt Wittgren,† Herje Schagerlo¨f,‡ Dane Momcilovic,§ and Karl-Gustav Wahlund*,§ Department of Technical Analytical Chemistry, Center for Chemistry and Chemical Engineering, Lund University, Box 124, SE-221 00 Lund, Sweden, AstraZeneca R&D Mo¨lndal, SE-431 83 Mo¨lndal, Sweden, and Department of Biochemistry, Center for Chemistry and Chemical Engineering, Lund University, Box 124, SE-221 00 Lund, Sweden Received June 17, 2003; Revised Manuscript Received October 8, 2003
Ethylhydroxyethyl cellulose (EHEC) of three different viscosity classes (EHEC I, II, and III) was analyzed by programmed cross-flow asymmetrical flow field-flow fractionation coupled to multiangle light scattering and refractive index detectors to determine their size and molar mass distribution. Two size populations were detected in the two lower viscosity classes, EHEC I and II, one high molar mass and one ultrahigh molar mass (UHM). The two covered molar masses from 104 up to 109 g‚mol-1. The highest viscosity class EHEC III was less size-dispersed covering molar masses from 5 × 105 to 5 × 107 g‚mol-1. Filtering of the EHEC II solution removed small amounts of compact UHM material. Enzyme treatments were performed on EHEC II to further characterize it. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry and anion ion-exchange chromatography coupled to pulsed amperometric detection showed that the UHM component contained EHEC. Introduction Cellulose derivatives are a group of polymers used in a wide range of applications from thickening additives in products for the construction and paint industry to film formers for pharmaceutical dosage forms.1 Polymer properties in technical applications are often governed by molecular or aggregate size, and their molar mass (M) and molar mass distribution (MD) is therefore of importance to determine and control. It is not uncommon to find that cellulose derivatives also contain aggregates, and there have been a number of articles published on the subject.2-5 Because aggregates may affect the technological functionality of the finished product, it is of interest to detect their presence both in product quality assurance and in product optimization. However, the determination of the molar mass distribution of cellulose derivatives is not always an easy analytical task. The cellulose derivatives are often both very polydisperse and large in molecular size. If they contain aggregates, it puts further high demands on the analytical techniques. One method that has been used for separating such samples is asymmetrical flow field-flow fractionation (FFF). It is a size separation technique with which it is possible to fractionate large molecules and colloids over a very wide size range.6-9 When it is used in combination with multiangle light scattering (MALS) and refractive index (RI) detection, * Corresponding author. E-mail:
[email protected]. Fax: +46-46-222 4525. § Department of Technical Analytical Chemistry. † AstraZeneca R&D Mo ¨ lndal. ‡ Department of Biochemistry.
the molar mass of the eluted fractions can be directly determined and thereby the polymer molar mass distribution.10 During the past decade, the combination flow FFFMALS-RI has been used for the molar mass analysis of a wide range of polymers such as dextrans, pullulans, and derivatives of starch and cellulose.11-14 In a previous study, an ethylhydroxyethyl cellulose (EHEC) derivative was successfully separated by asymmetrical flow FFF and an ultrahigh molar mass (UHM) component of unknown origin detected.5 There were, however, difficulties to fractionate the entire polydisperse sample with one constant cross-flow field with satisfactory resolution for all polymer sizes. Analyzing a sample at multiple cross-flow rates, each optimized for different size fractions, may circumvent this drawback although it is more time-consuming and makes calculations of sample polydispersity and complete molar mass distributions more difficult. A potentially better way is to use a programmed field, where the cross-flow is decreasing with elution time. This technique has previously been successfully applied in the analysis of mixtures of proteins of widely different sizes using asymmetrical flow FFF,15 but few programmed field separations of polymers have been published.16 In this study, the molar mass distribution of three EHEC samples of different viscosity grades was to be determined, together with an investigation to elucidate whether UHM components were present also in these EHEC samples. For the purpose of enabling a more efficient fractionation of the entire samples, a programmed cross-flow field was used.
10.1021/bm030051z CCC: $27.50 © 2004 American Chemical Society Published on Web 11/27/2003
98
Biomacromolecules, Vol. 5, No. 1, 2004
Experimental Section Chemicals. The ethylhydroxylethyl cellulose (EHEC) samples (AkzoNobel Surface Chemistry, Stenungssund, Sweden) are of three different viscosity classes and in the text referred to as EHEC I, II, and III. The manufacturer reported the viscosity of EHEC I to be 3 × 102 mPa‚s in a 2% solution, that of EHEC II to be 5 × 103 mPa‚s in a 2% solution, and that of EHEC III to be 5 × 103 mPa‚s in a 1% solution as measured by a Brookfield LV viscometer. The degree of ethyl substitution, DSethyl, for the EHEC I, II, and III was 0.8, 0.9, and 0.9, while the molar degree of substitution of ethylene oxide, MSEO, was 1.7, 2.3, and 2.3 as reported by the manufacturer. The sodium chloride (Merck, Darmstadt, Germany), sodium azide and trifluoroacetic acid (Merck) were all of pa grade purity. The methanol was of HPLC grade (Merck). All water used was deionized and purified prior to use (Milli-Q, Millipore, Bedford, MA). Sample Preparation. The EHEC samples were obtained as a dry powder and were reported by the manufacturer to contain small amounts of salt. Therefore a desalting procedure was performed. The samples were dissolved in distilled water to a concentration of 0.5-0.1% (w/w) and were kept in a refrigerator during two weeks to ensure complete dissolution. The sample volume (100 mL) was thereafter dialyzed using a dialysis tubing with cutoff 12-14 kDa (Spectrapor4, rec. no. 132703, Spectrum Laboratories, Rancho Dominguez, CA), which was submerged into a 10 L bottle filled with distilled and purified water having a conductivity >18 MΩ‚cm-1 for a period of 2 weeks during which time the water was changed daily. The samples were thereafter assumed to be sufficiently desalted and were freeze-dried to obtain dry substances that were stored in dark bottles in a desiccant. These desalted samples were prepared by weight to a concentration of 0.1-0.02% and left to dissolve for at least 2 weeks before analysis. Sample solvent and carrier for the FFF-MALS-RI was 10 mM NaCl with 0.002% NaN3 added to prevent bacterial growth. Filtering. Prefiltering of the EHEC II sample solution was performed by one of four different types of syringe filters, 1.0 µm pore-size polyvinyldifluoride (PVDF) hydrophilic with a 13 mm diameter (cat. no. 6777-1310, Whatman Int. Kent, U.K.), 1.0 µm pore-size polytetrafluorethylene (PTFE) hydrophobic with a 13 mm diameter (cat. no. 6784-1310, Whatman Int.), 0.45 µm pore-size regenerated cellulose with a 25 mm diameter (ref. no. 463050, Schleicher & Schuell, Dassel, Germany), or 0.5 µm pore-size PTFE hydrophobic with a 25 mm diameter (cat. no. SLSR025NS, Millipore Corp., Bedford, MA). The PTFE filters were conditioned with 2 mL of methanol. All filters were flushed with 2 × 10 mL of solvent before use. The 0.05% (w/w) polymer solutions were very slowly filtered through the syringe filters using a wide (10 mL) syringe to reduce the risk of shear degradation of the polymer during the filtering. The first drops of polymer solution (∼1 mL) from the filter went to waste. Enrichment and Acid Depolymerization of Polymer on Filter. A 0.05% (w/w) solution of EHEC II dissolved in pure water was first filtered through a precolumn filter unit with a 2 µm pore-size polyetherketone (PEEK) frit (A‚355 and
Andersson et al.
A‚700, Upchurch Scientific, Oak Harbor, WA) to remove large insoluble components such as fibers. The sample solution was thereafter filtered through a 0.5 µm pore-size 25 mm PTFE syringe filter by a syringe pump at a flow rate of 100 µL‚min-1 to minimize the risk for shear degradation. No larger sample volume than 1.5 mL was sieved through a filter, after which the filter was exchanged to avoid excessive resistance due to blockage. Four filters were used in this way to filter a total of 6 mL of sample. Of the sample solution passing the filter, 1 mL was collected for acidic hydrolysis. A solution of 2 M trifluoroacetic acid (TFA) was then injected onto the 0.5 µm PTFE filters, after which the filters were incubated in an oven for 20 h at 90 °C to depolymerize the sample stuck on the filters. Each filter was thereafter rinsed twice with 1.5 mL of water and then with 1.5 mL of methanol to extract the hydrolyzed components. The solvent was evaporated by heating in a thermal block to 50 °C under nitrogen gas to prevent oxidation and speed up the evaporation. The sample solution passing the filter was diluted with 99% TFA to a concentration of 2 M TFA. The solution was then incubated at 90° for 20 h, and the solvent was evaporated as above. Enzymatic Treatment. To a 0.05% EHEC II solution was added a cellulose-depolymerizing enzyme, endo-glucanase Cel 12a (EG III) from Trichodermia reesei, to a final enzyme concentration of 1 µM. The enzyme was a kind gift from Dr. Michael Ward, Genencore, CA. The solution was then incubated for 12 h at room temperature. To deactivate the enzyme prior to analysis, the sample vial was submerged in a water bath (∼90-95 °C) for 5 min and then left to cool to room temperature before analysis. Field-Flow Fractionation-Multiangle Light Scattering-Refractive Index Detection (FFF-MALS-RI). The FFF instrument used for the fractionations was an Eclipse F asymmetrical flow FFF instrument connected to a Dawn DSP multiangle light scattering (MALS) detector and an Optilab DSP differential refractive index (DRI) detector both measuring at 632.8 nm (all Wyatt Technology, Santa Barbara, CA). A pump with in-line vacuum degasser and autosampler (1100 series, Agilent Technologies, Palo Alto, CA) delivered the carrier flow and handled sample injection onto the FFF channel. The channel had a nominal thickness of 350 µm, and the ultrafiltration membrane forming the accumulation wall was made of regenerated cellulose with a cutoff of 10 kDa (C010F, Nadir filtration, Wuppertal, Germany). The sample injection onto the channel was commenced at a flow rate of 0.50 mL‚min-1 during 10 min to thoroughly rinse the sample loop. The sample volume injected onto the channel was 50, 200, or 900 µL. The sample concentration was 0.10% for EHEC I, 0.05% for EHEC II, and 0.02% for EHEC III (w/w). During the injection, a focusing flow of 1.5 mL‚min-1 was applied. After injection, a relaxation/ focusing flow of 1.5 mL‚min-1 was maintained for 5 min before elution was started. A programmed cross-flow in four steps was used for the analysis. The first step used a crossflow rate, Fc, of 1.5 mL‚min-1, which was decreased linearly to 0.5 mL‚min-1 in 3 min. In the second step, the crossflow rate was decreased from 0.5 to 0.1 mL‚min-1 during 5 min. In the third step, the cross-flow rate was decreased
Size and Structure Characterization of EHEC
linearly from 0.1 to 0 mL‚min-1 during 20 min. In the fourth step, the cross-flow rate was maintained at 0 mL‚min-1 for 5 min to ensure that the entire sample had been eluted. The outlet flow rate was kept constant at 1.0 mL‚min-1. Processing of light scattering data was made by the Astra software (Wyatt Technology). The weight-average molar mass (Mw) and z-average rms radius (rrms,z) were obtained by Debye plots using Berry’s method 17 fitting a line to the data obtained at low scattering angles (26°-69°). The lowest scattering angle data (14°) was not included, because the data was considered too imprecise. A dn/dc value of 0.138 was determined for the EHEC III and was used for all EHEC samples because they were assumed to be similar in structure. The term containing the second virial coefficient was assumed to be negligible. Recovery was calculated from integration of the RI signal (known dn/dc). Anion-Exchange Chromatography-Pulsed Amperometric Detection (AEC-PAD). The analysis of released monomers after acidic depolymerization was made with anion-exchange chromatography and pulsed amperometric detection (AEC-PAD) (Dionex, Sunnyvale, CA). The AEC-PAD instruments consisted of an AS50 auto sampler, a GP40 gradient pump, a Carbopac Pa10 guard and analytical column, and an ED40 electrochemical detector. The flow rate was 1.0 mL‚min-1 using water as eluent. The sample volume loaded onto the column was 50 µL. A postcolumn pump (Dionex) with 600 mM NaOH as eluent enabled detection of the monomers. Identification of the glucose peak in the chromatogram from the depolymerized sample was performed by comparing with the elution time from the analysis of a glucose standard. Matrix-Assisted Laser Desorption/Ionization Time-ofFlight Mass Spectrometry (MALDI-TOF-MS). MALDITOF-MS was performed using a Perseptive Voyager-DE STR (Applied Biosystems, Framingham, MA) instrument according to Momcilovic et al.18 with slight modifications. Mass spectra were acquired in reflector mode, and the lag time was 150 ns plus instrument offset. The detector activation gate was set at m/z 175, the guide wire voltage was 0.001% of the acceleration voltage, and the spectra were acquired in positive ion mode. All mass spectra were accumulated from 100 laser shots. 2,5-Dihydroxybenzoic acid (DHB) was used as matrix. The DHB was dissolved to 10 g‚L-1 and the depolymerized EHEC to ∼1 g‚L-1 in ethanol. The matrix and analyte solutions were vortexed, mixed 4:1 (v/v), and then vortexed again. A portion (0.5 µL) of the mixture was deposited on a MALDI sample plate and allowed to dry in ambient atmosphere. Results and Discussion The MALS and RI fractograms from the analysis of the three viscosity classes of EHEC are presented in Figure 1. All three are eluted in broad time intervals, from ∼1 to 27 min, indicating that they indeed are polydisperse in size. The RI traces are monomodal for all three samples, while the MALS detects two not completely resolved populations in EHEC I and II. The first population is mainly eluted in the time interval 1-16 min for EHEC I and 1-18 min for EHEC
Biomacromolecules, Vol. 5, No. 1, 2004 99
Figure 1. Flow FFF of three different viscosity classes of EHEC: (I) 0.1% EHEC I; (II) 0.05% EHEC II; (III) 0.01% EHEC III. Full line represents MALS 90° signal fractograms; dashed line represents RI signal fractograms; circles represent log molar mass. Sample volume was 200 µL for EHEC I and II and 900 µL for EHEC III. The two pairs of vertical dashed lines indicate the molar mass range averaged for the construction of the Kratky plot in Figure 5.
II with the second population eluted at 16-25 and 18-25 min, respectively. The second population gives a comparably high MALS signal but no corresponding increase in the RI signal. Hence the component must be present in much lower concentration than the first population but have a much higher molar mass. The late elution is also an indication that the size is comparably large. In the MALS fractogram from EHEC III, there is no indication of a second size population. The molar masses of the eluted fractions (Figure 1), calculated from the MALS-RI data, increased continuously with elution time as is expected in a normal-mode FFF separation. This confirms that the EHEC I and II are extremely polydisperse, having molar masses from ∼3 × 104 to 109 g‚mol-1 including the second population. The EHEC III has a more narrow distribution with molar masses from ∼5 × 105 to 5 × 107 g‚mol-1. It was not possible to calculate any reliable molar mass data for the very first (tr < 3 min for EHEC I and II and 21 min for EHEC I and >24 min for EHEC II and III) eluted fractions for any of the samples. For the first eluted fractions, the MALS signal-to-noise ratio, and therefore the precision, was too low, whereas for the last fractions, the RI signal-to-noise ratio was too low. Hence, the polydispersity of the different samples (Table 1) are likely underestimated. It is remarkable that EHEC I and II contain fractions with ultrahigh molar masses (UHM) of 107 g‚mol-1 and above. This would not be expected in polymers based on cellulose because the literature1 suggests that the highest known weight-average molar mass, Mw, of molecularly dissolved cellulose is ∼2 × 106 g‚mol-1, namely, that from cotton linters (whereas wood cellulose has even lower Mw). The molar mass plots for EHEC I and II coincide in the elution time interval 4-13 min, whereas EHEC III (7-13 min) deviates considerably showing notably higher molar masses. This indicates that these size fractions of EHEC I and II have a similar conformation and structure because the relationship between molar mass and elution time is the same. The quite different behavior of EHEC III shows that it has a different structure, namely, that of a denser molecule. More information on conformation is obtained by the conformation plots of log rrms versus log M presented in Figure 2. Again, EHEC I and II show large similarity, whereas EHEC III deviates. For EHEC I and II, there is a sudden change in the slope at a molar mass of about 2 × 106 g‚mol-1, indicating a change in the conformation. The slope 0.5 for the data below this value corresponds to those of flexible chains, whereas the slope 0.2 above it indicates a very compact structure. For EHEC III, there is a constant conformation with a slope of 0.37, which indicates a structure corresponding to a collapsed coil. This is in line with the finding above that EHEC III behaves as a denser coil than the others. The EHEC I and II are based on cellulose from wood pulp, while EHEC III is based on cellulose from cotton linters. One may therefore speculate that the observation of an UHM component in EHEC I and II but not in EHEC III may be connected to the origin of the cellulose. Aggregates of very high molar mass, which were shown to be partly made up of hemicellulose (∼5-10% xylan), have been found in cellulose acetates based on cellulose from wood pulps.25,26 Such aggregates were not found in cellulose acetate based on cotton linters, which are low in hemicellulose content. The Mw and rrms,z of EHEC I were 1.4 × 106 g‚mol-1 and 3.6 × 102 nm, respectively, whereas for EHEC II, they were
Figure 2. Conformation plots of EHEC obtained from the same experiments as in Figure 1: (I) EHEC I; (II) EHEC II; (III) EHEC III. Filled circles indicate data selected for linear fits to obtain the white lines having the indicated slope values. The data confirms the conclusion from the molar mass plots (see above), that is, the EHEC III is more compact.
1.3 × 106 g‚mol-1 and 2.4 × 102 nm when taken across the complete eluted sample (Table 1). The EHEC III has a considerably higher Mw of 5.8 × 106 g‚mol-1 and an rrms,z of 2.0 × 102 nm. The UHM component has a large impact on the obtained Mw, rrms,z, and polydispersity index, Mw/Mn, as can be seen in Table 1 when the fraction containing the UHM component was excluded from the calculations of the averages. The trend in Mw and rrms,z when the fraction containing the UHM component is excluded follows what is expected from the viscosity grading, that is, a higher viscosity grade polymer would have a higher Mw and rrms,z (i.e., larger size). The recovery calculations show that all of the polymers are eluted. The slightly higher than 100% recovery (102%) of EHEC I and II is within the expected uncertainty. Furthermore, the Mw and rrms,z for the EHEC I and II including the ultrahigh molar mass population is to be regarded as approximate because the M and rrms for the entire ultrahigh molar mass population cannot be reliably
Size and Structure Characterization of EHEC
Figure 3. Fractograms of EHEC II after different filtration pretreatments: (top) RI signal fractograms; (bottom) MALS 90° signal fractograms; (circles) log molar mass; (A) unfiltered; (B) filtered by 1.0 µm pore-size PVDF filter; (C) filtered by 0.45 µm cellulose nitrate filter; (D) filtered by 1.0 µm PTFE filter; (E) filtered by 0.5 µm PTFE filter. Sample volume was 900 µL. Polymer concentration was 0.05%.
determined (for reasons discussed above) and included into the average calculations. EHEC II-Filtration-Ultrahigh Molar Mass Component. The EHEC II was studied in more detail to obtain a more accurate estimate of the molar mass distribution, especially for the early and late eluted fractions in which the molar mass data were imprecise. For this purpose, the sample amount loaded onto the separation channel was increased 4.5 times to 450 µg to increase the signal-to-noise ratio. A second measure to improve the precision of the molar mass determination was to filter the sample solution prior to analysis to remove large contaminants and impurities. It must be noted that the increase in sample amount and the filtering of the sample can have adverse effects on the results. The increased sample amount might lead to loss of resolution due to channel overloading or sample aggregation. In general, samples should be analyzed at the lowest possible load without compromising the signal-to-noise ratio. Filtering may remove not only impurities but also the sample itself if the filter is poorly chosen. To ensure that no significant amount of polymer is lost, the recovery should be measured and the fractograms compared with those from the analysis of an unfiltered sample. Nevertheless, fractionations using a higher load of filtered sample were made, and the resulting fractograms and molar mass plots are presented in Figure 3. The higher sample load gave an increased MALS signalto-noise ratio making it possible to determine the molar mass for the earliest eluted fraction at 3 min to be ∼2 × 104 g‚mol-1 (Figure 3A). This is close to the 10 kDa molecular weight cutoff of the accumulation wall membrane, which
Biomacromolecules, Vol. 5, No. 1, 2004 101
sets the lower size-limit for the analysis. Also the dialysis membrane used for desalting of the sample had a molecular weight cutoff of 12-14 kDa, and the molar mass is thereby close to the lowest molar masses that can be expected to be present in the sample after dialysis. Additionally, the RI signal-to-noise ratio has increased so that the molar mass of the last eluted fraction at 25 min can be determined to be ∼109 g‚mol-1. This shows the extreme polydispersity of the sample, spanning almost 5 magnitudes in molar mass, and the ability of the programmed field-flow FFF technique to separate it in one analysis. The molar mass plot shows an irregular behavior at 12-15 min where M does not increase continuously with elution time. This may be a sign of deterioration of the size separation due to the higher sample load because it was not observed during the analysis at lower sample load. The peak in the MALS fractogram at 22-27 min is very sharp. This is a sign that part of the very large material eluted in this interval may be eluted, at least partially, in hyperlayer mode. Filtering of the sample solution prior to analysis was made by four different filters of various pore size or chemical composition or both. Two hydrophilic filters were used, one made of polyvinyldifluoride (PVDF) with a pore size of 1.0 µm and the other of regenerated cellulose with a pore size of 0.45 µm. Two hydrophobic poly(tetrafluoroethylene) (PTFE) filters with pore sizes 0.5 and 1.0 µm were also used. The MALS 90° signal in Figure 3 (curves B-E) shows a much lower response for filtered sample solutions. This indicates that the filters removed some of the material present in the polymer solutions. The largest reductions in MALS response were for the later eluted fractions, as expected, because this is where the largest components were eluted, while the earliest eluted fractions were unaffected by the filtering. For the last eluted fractions, the MALS response from filtered sample solutions was less than 33% of that of the unfiltered. However, the decrease was not as large in RI response. In the case of the sample solution filtered by the 1.0 µm pore-size filters (B, D), the RI response was almost unaffected, -2%, while for the 0.45 µm pore-size cellulose nitrate and 0.5 µm pore-size PTFE filters (C, E), the total RI response was reduced by 8% and 10%, respectively. This indicates that the filters removed some kind of UHM component present in low amount. Filtering of the polymer solution had an even more striking effect on the lightscattering response at low scattering angles. For an unfiltered sample, there was a much stronger response than for a sample that was filtered. This is further indication of the removal of ultralarge components by filtering. Filtering by the 1.0 and 0.5 µm PTFE filter reduced the MALS response more than that by the 1.0 and 0.45 µm hydrophilic filters did. Evidently the removal of material by the filters is not governed by filter pore size alone but also by the filter material. The reduction of the MALS response for the late-eluted fractions, of course, has a large impact on the molar masses obtained. This is seen in the molar mass plots for the filtered samples, where the highest molar masses detected were ∼3 × 107 g‚mol-1 for the sample filtered by the PVDF filter (B). For the filtered samples, the molar mass increased
102
Biomacromolecules, Vol. 5, No. 1, 2004
Andersson et al.
Table 2. Weight-Average Molar Mass, Polydispersity Index, z-Average Root-Mean-Square Radius, and Recovery Obtained after Different Filtration Procedures of EHEC II sample filtration none 1.0 µm PVDF 0.45 µm cellulose nitrate 1.0 µm PTFE 0.5 µm PTFE
Mw × 10-6 rrms,z × 10-2 recovery -1 (g‚mol ) Mw/Mn (nm) (%) 40 1.6 0.94 0.65 0.61
>100 5.4 3.0 2.6 2.2
4 1.4 1.1 1.0 1.0
99 97 91 97 89
continuously, and the irregularities seen in the M plot at 1215 min (Figure 3) for the unfiltered sample were not present. This indicates that the irregularity was somehow connected to the UHM material removed by the filters. The effect of filtering on the estimated Mw, polydispersity index (Mw/Mn), and sample recovery are presented in Table 2. For the unfiltered sample, a Mw of 4 × 107 g‚mol-1, and a rrms,z of 4 × 102 nm was obtained. From the fractograms having the lowest MALS response (sample solution filtered through 0.5 µm pore-size PTFE filter, E), the Mw is much less, only 0.61 × 106 g‚mol-1. As a result of the lowered Mw, the polydispersity index decreased accordingly. The average molar mass and polydispersity obtained for the unfiltered sample in Table 2 are much higher than the values given in Table 1 obtained at a lower sample load. The reason is that it was possible to determine the molar mass for the ultrahigh size population more accurately when using the higher sample load. Thereby the molar mass data for the entire ultrahigh size population could be included in the calculation of the average molar mass, which therefore obtained a higher value. Another possibility that may explain part of the higher obtained Mw is that the higher sample load may lead to concentration-induced aggregation in the channel. Structure and Conformation. The conformation plots (Figure 4) obtained from the analysis of EHEC II filtered through 1.0 µm PVDF filter and 0.5 µm PTFE filter show that the slope in both equals 0.68 for the fractions of
