Dynamic light scattering and sub-micron peptide assemblies: what the autocorrelation function can and cannot report

A reconstituted peptide solution can look perfectly clear and still contain a population of assemblies that no routine chromatographic assay will register. Soluble oligomers and sub-micron clusters fall into a gap between techniques: they are too small for visual inspection and light obscuration, which operate at micrometer scales, and too large or too labile to survive many size-exclusion or reversed-phase separations intact. Dynamic light scattering (DLS) occupies this gap, which is the reason it appears so often in characterization packages for aggregation-prone sequences.

This post reviews how a DLS measurement is built, why its intensity-weighted output behaves the way it does, how to read the polydispersity index, and where the technique stops being informative. The goal is not to promote the method but to describe what a reported hydrodynamic diameter does and does not mean.

How the measurement works

A DLS instrument illuminates a sample with a coherent laser and records the scattered light at a fixed angle, commonly 90, 173, or a forward angle depending on the design. Particles in solution undergo Brownian motion, so the phase relationships among light scattered by different particles fluctuate continuously. Smaller particles diffuse faster and produce rapid fluctuations; larger particles diffuse slowly and produce slow ones. The instrument computes an autocorrelation function of the scattered intensity as a function of delay time, and the decay rate of that function is related to the translational diffusion coefficient.

The Stokes-Einstein relationship converts the diffusion coefficient into a hydrodynamic diameter, defined as the diameter of a hypothetical sphere with the same diffusion behavior as the particle. Two features of this definition matter for interpretation. First, the result depends on solvent viscosity and temperature, so both must be known and controlled; a few degrees of error in sample temperature shifts the reported size measurably. Second, the hydrodynamic diameter includes the solvation shell and any associated counterions, so it is systematically larger than a diameter inferred from crystallography or electron microscopy. A monomeric peptide of a few kilodaltons is typically reported at around a nanometer or somewhat above, which is near the lower practical limit of many instruments.

Intensity weighting and the dominance of large species

The most common source of misreading in DLS reports is the weighting of the size distribution. Scattered intensity from a particle in the Rayleigh regime scales with the sixth power of its diameter. A particle ten times larger scatters roughly a million times more light. The primary result of the experiment is therefore an intensity-weighted distribution, in which a tiny number-fraction of large assemblies can dominate the signal.

Software packages convert intensity distributions into volume or number distributions using assumed optical properties and a model of particle shape. These conversions are mathematically legitimate but amplify noise and model error, and they should be treated as illustrative rather than quantitative. A report that states a mostly monomeric population by number while the intensity distribution shows a second peak in the hundreds of nanometers is describing the same sample two ways; the intensity view is the more direct reflection of what was measured. Reputable characterization notes state which weighting is being quoted and, ideally, show the raw correlation function so a reader can judge data quality independently.

This sensitivity cuts both ways. It makes DLS an effective detector of trace aggregates, since a very small mass fraction of large species is enough to shift the cumulant mean noticeably. It also makes the technique vulnerable to contamination: a single dust particle or a shed fragment from a poorly rinsed cuvette can produce an apparent aggregate peak. Filtration of buffers through appropriately small membranes, careful cuvette handling, and replicate measurements are standard protections, and a peak that appears in one acquisition but not the next is generally treated with suspicion rather than reported.

Polydispersity and the cumulant analysis

Two numbers dominate summary reports: the Z-average diameter and the polydispersity index (PDI). Both come from a cumulant fit to the log of the autocorrelation function. The Z-average is an intensity-weighted harmonic mean size, and the PDI is a dimensionless measure of the width of the distribution, derived from the second cumulant.

Commonly cited conventions treat very low PDI values as indicating a near-monodisperse sample, while values approaching the upper range of the scale indicate a broad or multimodal distribution for which the cumulant model is no longer adequate. For a small peptide in dilute solution, the signal is weak and the PDI is often poorly defined; a reported PDI in that case reflects instrument noise at least as much as sample properties. The cumulant approach assumes a single, roughly symmetric population, so a sample containing monomer plus a minor population of large clusters can return a moderate Z-average that corresponds to no physical species present. Distribution-fitting algorithms, such as regularized inversions of the correlation function, can resolve multiple populations, but their resolution is limited; populations closer than roughly a factor of three to five in size are generally not separated reliably.

Sample variables that change the answer

Because the method is non-invasive and requires no label or column, it is tempting to treat the result as a property of the peptide. In practice it is a property of the peptide in a specific solution at a specific concentration and temperature. Ionic strength changes electrostatic screening and therefore both the extent of self-association and the apparent size through the electrical double layer. Concentration matters directly: at higher concentrations, interparticle interactions alter the apparent diffusion coefficient, and the measurement may report a collective rather than a single-particle diffusion. Studies of self-assembling sequences frequently describe a critical concentration above which a distinct large population appears, and DLS is among the techniques used to locate that threshold.

Time is a further variable. Assembly kinetics for some sequences are slow, and a sample measured immediately after reconstitution can differ from one measured after hours at controlled temperature. Time-resolved or repeated measurements of the same cuvette can therefore be more informative than a single snapshot, because the trajectory of the Z-average and the scattered count rate distinguishes a stable solution from one that is evolving. Count rate itself is worth recording; a steady rise at constant instrument settings points to growth of scattering species even when the size estimate is noisy.

Dilution deserves caution. Diluting a sample to improve optical conditions can dissociate reversible assemblies, so a measurement made after dilution may understate the aggregation present in the original solution. Conversely, concentrating a solution may induce assemblies that were not present at the working concentration. Methods sections that specify how the sample was prepared for the measurement are more useful than those reporting only the size.

Interpretive limits and orthogonal pairing

DLS does not identify what an assembly is made of. A signal at a few hundred nanometers could be peptide aggregates, excipient-derived particles, silicone oil droplets, or particulate from a container surface; the technique cannot distinguish among them. It also provides no information about chemical identity, covalent versus non-covalent association, or whether assemblies are reversible. For these questions, laboratories typically pair it with orthogonal methods: size-exclusion chromatography with light-scattering detection for resolved populations, analytical ultracentrifugation for solution-phase sedimentation behavior without a column, nanoparticle tracking analysis for number-based sub-micron counts, and flow imaging or light obscuration for larger particulates.

Agreement among methods is more persuasive than any single result, and disagreement is itself informative. A sample that appears clean by size-exclusion chromatography yet shows a large-species population by DLS may contain assemblies that dissociate on the column or are filtered out by it. Treating the techniques as complementary views of a size continuum, rather than competing answers, is the more defensible reading.

Synthesis

DLS is best understood as a sensitive detector of large species and a poor quantifier of them. Its intensity weighting, its dependence on solution conditions, and its inability to assign identity define what a reported number means. Documentation that states the weighting, shows the correlation data, specifies sample preparation and temperature, and pairs the result with at least one orthogonal measurement allows a reader to evaluate the finding on its merits. These materials are discussed here for research and analytical-methods purposes only; nothing in this post describes or supports use outside laboratory investigation.