BETTER THAN SEC’s Paul S. Russo Louisiana State University Texas Polymer Center Freeport, TX October 31, 2001

Obligatory Equation SEC = GPC = GFC Size Exclusion Chromatography Gel Permeation Chromatography Gel Filtration Chromatography

GPC Solvent flow carries molecules from left to right; big ones come out first while small ones get caught in the pores. It is thought that particle volume controls the order of elution. But what about shape ?

Simple SEC degas pump injector DRI VeVe log 10 M c c c

Osmometry: Real Science Semipermeable membrane: stops polymers, passes solvent. h  V = n R T n = g/M c = g/V  = c R T

Light Scattering: Osmometer without the membrane 100,000  x c 

LS adds optical effects  Size q = 0 in phase I s maximum q > 0 out of phase, I s goes down

SEC/MALLS degas pump injector DRI MALLS

SEC/MALLS Scattering angle VeVe Scattered intensity

Scattering Envelope for a Single Slice

SEC/RALS/VIS degas pump injector DRI LS90 o  P   viscometer

Universal Calibration Grubisic, Rempp & Benoit, JPS Pt. B, 5, 753 (1967) One of of the most important Papers in polymer science. Imagine the work involved! 6 pages long w/ 2 figures. Selected for JPS 50 th Anniv. Issue.

Universal Calibration Equations [  ] = KM a [  ] A  M A = [  ] S  M S = f (V e ) Mark-Houwink Relation Universal Calibration A = analyte; S = standard Combine to get these two equations, useful only if universal calibration works!

Objectives Use  -helical rodlike homopolypeptides to test validity of universal calibration in GPC. Can GPC/Multi-angle Light Scattering arbitrate between disparate estimates of stiffness from dozens of previous attempts by other methods?

Severe test of universal calibration: compare rods & coils Combine M’s from GPC/MALLS with [  ]’s from literature Mark-Houwink relations. Strategy Ld Hydrodynamic volume

Polymers Used [NH-CHR-C] x O R = (CH 2 ) 2 COCH 2 [CH-CH] x R = (CH 2 ) 2 CO(CH 2 )CH 3 Polystyrene (expanded random coil) Solvent: THF = tetrahydrofuran Homopolypeptides (semiflexible rods) PBLG = poly(benzylglutamate) Solvent: DMF=dimethylformamide PBLG = poly(stearylglutamate) Solvent: THF = tetrahydrofuran

Mark-Houwink Relations [  ] = 0.011·M w for PS [  ] = 1.26·10 -5 ·M w 1.29 for PSLG [  ] = ·M w 1.35 for PBLG

Polystyrene Standards: the Usual

Polypeptide Samples Were Reasonably Monodisperse NCA-ring opening was used to make these samples. Most were just isolated and used; a few were fractionated.

Universal Calibration Works for These Rods and Coils

2 nd Virial Coefficient Equations A 2, = M 2 A 2 /N a  = kT(1 + A 2, + …) A 2, = dL 2 /4 Osmotic pressure in number density concentration ( ) units Relationship to the “normal” 2 nd virial coefficient for conc. in mass per volume units. Onsager 2 nd virial coefficient for rods (L= length, d = dia.) R g for rods

2 nd Virial Coefficient (Excluded Volume Limit) is Another Universal Descriptor

Persistence Length a p from R g Persistence length is the projection of an infinitely long chain on a tangent line drawn from one end. a p =  for true rod.

Persistence Length of Helical Polypeptides is “Very High” What the biggest polymers in our sample would look like at this a p

SEC/MALLS in the Hands of a Real Expert a p  15 nm Much less than PBLG Macromolecules, 29, (1996)

Conclusions The new power of SEC/Something Else experiments is very real. SEC is now a method that even the most jaded physical chemist should embrace. For example, our results favor higher rather than lower values for PBLG persistence length. This helps to settle about 30 years of uncertainty. Universal calibration works well for semiflexible rods spanning the usual size range, even when the rods are quite rigid. So, SEC is good enough for physical measurements, but is it still good enough for polymer analysis?

They were young when GPC was.

Small Subset of GPC Spare Parts To say nothing of unions, adapters, ferrules, tubing (low pressure and high pressure), filters and their internal parts, frits, degassers, injector spare parts, solvent inlet manifold parts, columns, pre-columns, pressure transducers, sapphire plunger, and on it goes…

Other SEC Deficiencies 0.05 M salt at 10 am, 0.1 M salt at 2 pm? 45 o C at 8 am and 50 o C at noon? Non-size exclusion mechanisms: binding. Big, bulky and slow (typically 30 minutes/sample). Temperature/harsh solvents no fun. You learn nothing by calibrating.

Must we separate ‘em to size ‘em? Your local constabulary probably doesn’t think so. I-85N at Shallowford Rd. Sat. 1/27/01 4 pm

Sizing by Dynamic Light Scattering—a 1970’s advance in measuring motion, driven by need to measure sizes, esp. for small particles. t IsIs It’s fluctuations again, but now fluctuations over time! DLS diffusion coefficient, inversely proportional to size. Large, slow molecules Small, fast molecules

Molecular Weight Distribution by DLS/Inverse Laplace Transform--B.Chu, C. Wu, &c. Where: G(  ) ~ cMP(qR g )  = q 2 D  q 2 kT/(6  R h ) R h = XR g g(t) log 10 t ILT  q 2 D G(  ) CALIBRATE MAP M c log 10 M log 10 D 

Hot Ben Chu / Chi Wu Example MWD of PTFE Special solvents at ~330 o C Macromolecules, 21, (1988) This only “works” because of that wide, wide M distribution. Main problem with DLS/Laplace inversion is poor resolution. Things kinda go to pot at low M, too. Some assumptions have to be made to do this.

Reptation: inspired enormous advances in measuring polymer speed…and predicts More favorable results for polymers in a matrix. T here once was a theorist from France Who wondered how molecules dance. "They're like snakes," he observed, "as they follow a curve, the large ones can hardly advance."* D ~ M -2 deGennes More generally, we could write D ~ M -  where  increases as entanglements strengthen *With apologies to Walter Stockmayer

Matrix Diffusion/Inverse Laplace Transformation Goal: Increase magnitude of  Difficult in DLS because matrix scatters, except special cases. Difficult anyway: dust-free matrix not fun! Still nothing you can do about visibility of small scatterers DOSY not much better Replace DLS with FPR. Selectivity supplied by dye. Matrix = same polymer as analyzed, except no label. No compatibility problems. G(  ) ~ c (sidechain labeling) G(  ) ~ n (end-labeling) log 10 M log 10 D Stretching  Solution:   Matrix:  

Painting Molecules* Makes Life Easier *R. S. Stein Small Angle Neutron Scattering ForcedRayleighScattering Fluorescence Photobleaching Recovery Index-matched DLS match solvent & polymer refractive index can't do in aqueous systems Depolarized DLS works for optically anisotropic probes works for most matrix polymers

Fluorescence Photobleaching Recovery 1. An intense laser pulse photobleaches a striped pattern in the fluorescently tagged sample. 2. A decaying sine wave is produced by moving the illumination pattern over the pattern written into the solution. 3. An exponential decay is produced by monitoring the amplitude of the decaying sine wave. Fitting this curve produces  from which D can be calculated.

FPR for Pullulan (a polysaccharide) Probe Diffusion: Polymer physicsCalibration: polymer analysis

FPR Chromatogram  Indicates targeted M. Sure this is easy. Also easy for GPC. But not for DLS or DOSY!

Separation Results Separation Results Pullulan M = 50/50 mix of 11,800 and 380,000

Better Resolution “Soon”?  Indicates targeted M. Improvement in resolution is observed at lower concentrations due to a more viscous characteristic. A compatibility problem is seen though at higher concentrations.

Simulation of FPR Results (Most Desirable Situation)

Examples of Separation Results from Simulation Data  Indicates targeted M.

Ultimate Goal: A Black Box for MWD Matrix FPR Easily Maintained Accurate Precise Simple Concept Expedient Easy System Switch Basic Info Obtained Miniaturizable Detect Large Masses Labeling Required GPC Accurate Simple Concept Miniaturizable No Labeling Required Broad Distributions Pumps Parts Press for MWD DLS Form Factor Index Matching Long Acquisition for Multiangle Experiments Precise Accurate DOSY Easy System Switch Precise Accurate Obtain Basic Info Labeling Required

Conclusions For a limited number of cases, this could really work. We may not always need leaking pumps and large parts bins for polymer characterization. What is good about GPC (straight GPC) is the simple concept; Matrix FPR keeps that—just replaces V e with D.

Thank you! LSULSU

Better than SEC’s Monday, January 29, 2001 Physical Info from SEC Elena Temyanko Holly Ricks N$F Replacing SEC Garrett Doucet David Neau Wieslaw Stryjewski

History of this Talk Used first at Georgia Tech, mods made after Same modifications to the USC talk, which is designed to be a little shorter & simpler The changes affect mostly the early parts of the diffusion part, near deGennes and Chu Used at Dow--Freeport