The Gold Standard of Polymer Science | Branching and the Molar-Mass Ambiguity Behind One SEC Peak

Size-exclusion chromatography, or SEC, separates dissolved molecules through their access to pores. Under suitable size-exclusion conditions, the retention behaviour reflects hydrodynamic size. It is not a direct sorting rule that assigns one molar mass to every retention time.

Molar mass describes mass per amount of molecules. Architecture describes how their connected segments are arranged: linear chains, stars, other branched structures and networks can differ even at the same molar mass.

Intrinsic viscosity, written [η], describes the limiting contribution to solution viscosity per concentration as concentration approaches zero, with the solvent and temperature specified. It is not the viscosity of the bulk melt.

We will compare soluble chains of the same chemistry under the same solvent and temperature. Their architecture can differ. Aggregation, adsorption, degradation, detector noise and incomplete recovery are excluded initially so that the ambiguity can be seen without those additional complications.

Use the ideal universal-calibration coordinate H = [η]M. Under the applicable SEC conditions, chains with equal H are treated as co-eluting. H has units of volume per mole when [η] is volume per mass and M is mass per mole.

For the linear reference, assume [η]_linear(M) = K M^a. This is a Mark–Houwink-type relation over a stated range, not a law that supplies the same K and a for every polymer, solvent or temperature.

For a branched chain at the same molar mass, define g′(M) = [η]_branched(M)/[η]_linear(M). Thus H_branched = g′(M) K M^(a+1). The prime distinguishes this viscosity ratio from other contraction factors based on radius of gyration.

A lower g′ can describe lower intrinsic viscosity at fixed M. It does not by itself count branch points. Architecture-to-branch-count relations need their own model and evidence. We will specify g′ as part of a fictional construction rather than infer it from the peak alone.

Set a = 1 and K = 5 × 10⁻⁶ in compatible units, so that [η] is in decilitres per gram when M is entered in grams per mole. This exponent is chosen for transparent arithmetic; it is not a claim about a particular experimental polymer.

At M_L = 100,000 g mol⁻¹, the linear intrinsic viscosity is K M_L = 0.500 dL g⁻¹. Its calibration coordinate is H_L = 0.500 × 100,000 = 50,000 dL mol⁻¹.

At M = 200,000 g mol⁻¹, the linear reference would instead have [η] = 1.000 dL g⁻¹ and H = 200,000 dL mol⁻¹. Without a change in architecture, these two masses do not share H.

Keep that comparison visible. The next step will change both mass and contraction, allowing one effect to compensate the other. Treating the mass increase alone as the complete retention prediction would discard the reason architecture was introduced.

Choose a branched chain with M_B = 200,000 g mol⁻¹ and g′ = 0.25 at that mass. Its intrinsic viscosity is 0.25 × 1.000 = 0.250 dL g⁻¹.

Its H is 0.250 × 200,000 = 50,000 dL mol⁻¹, exactly matching the lighter linear chain. In the ideal model, they co-elute even though one has twice the molar mass.

A conventional linear-reference mass calibration would assign 100,000 g mol⁻¹ to both at this retention. That assignment is correct for the constructed linear chain and low by a factor of two for the constructed branched chain.

The calculation does not assert that all branched polymers are underestimated by two. The factor follows from these chosen values. It demonstrates non-uniqueness: one measured retention coordinate can be compatible with more than one mass–architecture pair.

For a general exponent a in the model, co-elution with a linear-reference mass M_app gives K M_app^(a+1) = g′(M_B) K M_B^(a+1).

If g′ at the relevant mass is known and positive, and a + 1 is nonzero, this becomes M_B/M_app = [1/g′]^(1/(a+1)). With a = 1 and g′ = 0.25, the factor is the square root of four, or two.

If a were 0.5 while the same ratio g′ applied, the factor would be 4^(2/3), approximately 2.520. Using 1/g′ itself would give four and would be wrong for both examples.

More generally g′ can change with M. The equation may then require solving with a specified g′(M) relation. A single constant factor across a full chromatogram needs evidence; a correction established at one point does not automatically calibrate an entire distribution.

Mix equal masses of the 100,000 linear chains and the 200,000 branched chains. Their mass fractions are w_L = w_B = 0.5. Assume ideal dilute additivity for viscosity, identical appropriate concentration-detector response and the same H for both species.

The mixture intrinsic viscosity is the mass-weighted value: [η]_mix = 0.5 × 0.500 + 0.5 × 0.250 = 0.375 dL g⁻¹. A narrow observed slice does not make the two true masses equal.

The true number-average molar mass is M_n = 1/(0.5/100,000 + 0.5/200,000) = 133,333.3 g mol⁻¹. The mass average is M_w = 0.5 × 100,000 + 0.5 × 200,000 = 150,000 g mol⁻¹.

Dispersity is M_w/M_n = 1.125. The single retention coordinate has concealed a nonuniform molar-mass population. This is intrinsic co-elution in the construction, before ordinary instrumental broadening is added.

If the slice is treated as having one mass, a universal-calibration conversion would calculate H/[η]_mix = 50,000/0.375 = 133,333.3 g mol⁻¹.

Here that equals M_n, not M_w. The reason can be derived rather than guessed. Every species in this constructed slice obeys [η]_i M_i = H, so [η]_mix = Σw_i[η]_i = H Σ(w_i/M_i).

Therefore H/[η]_mix = 1/Σ(w_i/M_i). It is the number average of the co-eluting population under these particular ideal assumptions.

This identity is not permission to ignore detector response, mixing behaviour or nonideal separation in real data. Its purpose is narrower: treating every narrow slice as monodisperse can collapse different averages even when the viscosity measurement itself is exact.

Now replace the mixture with a single-mass population at M = 133,333.3 g mol⁻¹. The linear reference at this mass has [η] = 0.6666667 dL g⁻¹.

Choose g′ = 0.5625. Then the constructed chains have [η] = 0.375 dL g⁻¹ and H = 50,000 dL mol⁻¹. At the same total concentration they match the mixture's ideal retention and intrinsic-viscosity signals.

For this second population M_n = M_w = 133,333.3 g mol⁻¹ and dispersity is one. The first population instead has M_w = 150,000 and dispersity 1.125.

Thus those two exact signals do not uniquely determine the within-slice mass distribution. The alternative g′ is an assumed possible model value; the exercise does not claim a specific synthesis yields it. Additional evidence is needed to decide which population describes a real specimen.

Under a suitable calibrated dilute-solution scattering measurement with correct concentration and optical parameters, static light scattering supplies a mass-weighted molar-mass response. In this ideal comparison it can distinguish 150,000 from 133,333.3 g mol⁻¹.

That is additional information, not an automatically error-free answer. Refractive-index increment, aggregation, angular dependence, detector alignment and recovery still need attention. The detector-response guide explains why a wrong concentration conversion can affect more than one stage.

A branching interpretation also needs a suitable linear reference of matching chemistry and experimental conditions. Changes in solvent quality, composition or association can change conformation without being the architecture change originally proposed.

NIST's SEC chapter and viscometric-detection review provide the measurement background. For an actual experiment, report what each detector establishes, what assumptions connect its signal to a quantity, and which alternatives remain after combining the evidence.

Keep a = 1 and the same K. A linear reference at 80,000 g mol⁻¹ has [η] = 0.400 dL g⁻¹ and H = 32,000 dL mol⁻¹.

Choose a branched species with g′ = 0.64. Its matching mass is 80,000/√0.64 = 100,000 g mol⁻¹. Its intrinsic viscosity is 0.64 × 0.500 = 0.320, giving the same H of 32,000.

An equal-mass mixture has [η]_mix = 0.360 dL g⁻¹. Its number average is 1/(0.5/80,000 + 0.5/100,000) = 88,888.9 g mol⁻¹. Its mass average is 90,000, so dispersity is 1.0125.

Check the inferred single slice mass: 32,000/0.360 = 88,888.9. A very small hidden dispersity still differs from exactly one. Whether an actual instrument can resolve that difference depends on precision and the complete measurement model, not this arithmetic alone.

Does a narrow SEC peak prove a narrow mass distribution?

No. Species with different masses and architectures can co-elute. Peak width also includes instrument and column effects.

Is g′ a direct branch-count reading?

No. It is a viscosity ratio at matching mass and conditions. Converting it into branch number needs a model for architecture and other relevant evidence.

Does universal calibration always give the wrong mass?

No. It is useful when its assumptions and inputs hold. Our counterexample tests the extra assumption that an elution slice contains only one molar mass.

Does adding light scattering settle every structure question?

No. It adds a different mass-sensitive observation under its own conditions. Architecture, composition, aggregation and unresolved mixtures can still need further investigation.

A useful report separates measured retention, concentration and viscosity signals from inferred molar masses and architecture. It states the calibration, solvent, temperature, population represented and detector-response assumptions.

When a distribution is computed slice by slice, ask why each slice can be treated as sufficiently uniform for the intended average. A sharp-looking chromatogram is evidence about separation in its measurement coordinate, not a proof of chemical uniformity.

For transfer, change only the architecture while holding chemistry fixed, as this guide did. Then change only the concentration-response factor, as the previous guide did. Explaining why these are different failures shows more understanding than applying the same correction to both.

The next useful step is an evidence comparison: identify two proposed populations, predict what each instrument would observe, and choose an independent measurement on which they disagree. That is how a plausible polymer story becomes a testable one.

Sources and scope

The reference sources support SEC, universal calibration, viscometry and architecture-sensitive interpretation. All K values, contraction ratios, constructed populations and worked arithmetic here are fictional original examples. They are not reference-material values or an experimental result.

• André Striegel, NIST: Size-exclusion chromatography, including universal calibration and multiple detection
• NIST: Viscometric Detection in Size-Exclusion Chromatography: Principles and Select Applications

Separate calibration, response and population averages

• The Gold Standard of Polymer Science | Detector Response Factors and Mass Fractions in Copolymer SEC
• The Gold Standard of Polymer Science | Size-Exclusion Calibration and Distorted Molar-Mass Distributions
• The Gold Standard of Polymer Science | Matching Molar-Mass Averages, Different Chain Populations

Follow architecture, networks and recovery

• The Gold Standard of Polymer Science | Extractable Additives and Incomplete Mass Recovery
• The Gold Standard of Polymer Science | Sol and Gel Fractions with Unequal Unit Masses
• The Gold Standard of Polymer Science | A Finite Number Average Beside a Diverging Cluster Average
• The Gold Standard of Polymer Science | Branching, Gel Points and Network Formation
• The Gold Standard of Polymer Science | Step-Growth Imbalance and the Ceiling on Average Chain Length
• The Gold Standard of Polymer Science | Chain Scission, Molecule Counts and the Fall in Number-Average Molar Mass

Learning that travels

• The Gold Standard of Education | Learning That Transfers into Life
• Explore How X Works
• Return to the Polymer Science subject root