GPC Calibration Converter

How this works

Most GPC/SEC systems are calibrated with narrow polystyrene (or PMMA) standards, so the molecular weight your software reports is a "PS equivalent" value, the molecular weight a polystyrene chain would need to elute at the same retention time. Your actual polymer almost never has the same relationship between chain length and hydrodynamic size as polystyrene, so that number is only a true molecular weight if your sample happens to be polystyrene too.

The universal calibration principle says two polymers that elute at the same retention volume have the same hydrodynamic volume, which is proportional to [η]·M. Since intrinsic viscosity follows the Mark Houwink equation [η] = K·M^α, you can convert between the calibration standard's apparent molecular weight and your polymer's true molecular weight if you know both polymers' K and α values in the same solvent.

The conversion itself doesn't care what columns produced the number, but it is only as good as the calibration behind it: the apparent molecular weight must come from the linear region of your column bank, not from a peak crowding the exclusion or total-permeation limit. If you're choosing columns, or wondering whether adding a second column is worth it, see the interactive column-bank builder on the peak interpretation page.

Conversion calculator

A worked example, and how big the error is

Take the page's defaults: a PMMA sample run against polystyrene standards in THF, with K_PS = 1.14 × 10⁻⁴ dL/g, α_PS = 0.716, K_PMMA = 0.80 × 10⁻⁴ dL/g and α_PMMA = 0.70. Feeding a reported peak of 45,500 g/mol through the expression above returns about 62,000 g/mol. The GPC was under-reporting by a third.

Two things are worth taking from that table. The correction is large – a third of the reported value, comfortably bigger than most of the effects people try to interpret from a GPC trace. And it is not a constant: because the two α values differ, the ratio drifts with molecular weight, from 1.34 at 10 kg/mol to 1.38 at 250 kg/mol. You cannot correct a distribution by scaling it with one factor; the low and high ends move by different amounts.

Dispersity is the exception, and it is worth knowing which way. Both moments pass through the same transform, so the ratio between them survives it in closed form: Đ_true = Đ_apparent^{(1 + α_std) / (1 + α_sample)}. Neither K appears. With the numbers above the exponent is 1.009, so a reported Đ of 1.30 becomes 1.303 – a shift no one could see, against molecular weights that moved by a third. It only becomes visible when the two α values are far apart and the sample is broad: at α_sample = 0.60 a Đ of 5.0 would convert to 5.6. So a molecular weight from a mismatched calibration needs converting and a dispersity usually does not, and since K drops out of that expression entirely, the parameter people most often disagree about cannot touch Đ at all. The integration window is another matter entirely – there Đ is the most fragile number on the report, which the trace analyzer shows on your own data.

This is the usual explanation when a controlled polymerisation appears to miss its target. If you designed for 60 kg/mol of PMMA and the instrument says 45 kg/mol, nothing has gone wrong – you are comparing a real molecular weight against a polystyrene-equivalent one. Convert first, then judge the chemistry, and see the calculator for how the target was set.

When this conversion does not rescue the number

• Branching. A branched chain is more compact than a linear one of the same molecular weight, so it elutes late and reads low – and universal calibration cannot see the difference, because it only knows hydrodynamic volume. Mark–Houwink constants are measured on linear chains, so applying them to a branched sample corrects one error while leaving a larger one in place. Long-chain branching needs a light-scattering or viscometry detector, not a better calibration.
• Constants from the wrong conditions. K and α are specific to a polymer in a solvent at a temperature. Values measured in THF at 30 °C do not apply to a DMF run at 50 °C, and using them will introduce an error of its own. The two sets also have to come from the same solvent as each other, not merely from the same paper. The picker above enforces this: choose two reference polymers characterised in different eluents and the converter refuses rather than returning a number, because there is no approximation on offer – universal calibration equates hydrodynamic volume between chains in the same liquid, and across two liquids there is no equality to work from.
• Copolymers do not have Mark–Houwink constants. A statistical copolymer's hydrodynamic behaviour lies somewhere between its two homopolymers and shifts with composition, so there is no single pair of constants to look up. Treat any converted value for a copolymer as an estimate bracketed by the two homopolymer results.
• Interaction with the column is not a calibration problem. If your polymer adsorbs to the packing or is charged and unshielded, it is not eluting purely by size and universal calibration does not apply at all. Suspect this when a polar or ionic sample gives an implausibly low molecular weight or a tailing peak – the fix is salt in the eluent or a different column chemistry, covered on the peak interpretation page.
• The peak has to be inside the calibrated range. A conversion applied to a peak sitting at the exclusion limit propagates a meaningless input into a precise-looking output.

Reference Mark Houwink parameters

Representative literature values, typically THF near room temperature unless noted. Polystyrene's constants are very well established; the others vary more across sources depending on tacticity, microstructure, and exact conditions. Always use your own column supplier's or a primary literature source's values for quantitative work.

This conversion assumes the universal calibration principle holds and that the GPC was calibrated with narrow, well characterized standards spanning the relevant molecular weight range. It does not account for branching, which changes a polymer's hydrodynamic volume independent of its true molecular weight, so branched samples will still read inaccurately even after this correction.