Particle Size Analysis: Laser Diffraction vs DLS

Laser diffraction vs DLS is usually settled by glancing at a size-range chart, which is exactly why the two results so rarely reconcile. The published ranges overlap across roughly a decade and a half of particle size, both instruments return a clean distribution in minutes, and both report a number that looks like "the" particle size. Yet a D50 from one and a Z-average from the other can differ by tens of percent on the same batch, and neither instrument is malfunctioning.

The reason is that particle size analysis by light scattering is never a direct measurement. Each technique observes a different physical signal, inverts it through a different model, and weights the resulting population differently. Choosing between them is therefore less about which range fits the sample and more about which failure mode the sample, and the decision riding on the result, can tolerate.

What Each Technique Actually Measures

How particle size is measured determines what the reported number means. Dynamic light scattering (DLS) measures the rate of intensity fluctuations in scattered light and converts it to a translational diffusion coefficient. Stokes–Einstein then converts diffusion into a hydrodynamic diameter: the size of a sphere that diffuses at the same rate in the same medium. That diameter includes everything that moves with the particle, such as solvation layers, adsorbed polymer, ligand shells, and protein coronas. For a sterically stabilized nanoparticle, the hydrodynamic diameter can legitimately exceed the core size seen by electron microscopy by a wide margin.

Laser diffraction measures the angular distribution of scattered light and inverts it into a distribution of volume-equivalent spheres, using either Mie theory or the Fraunhofer approximation. Laser diffraction is a form of static light scattering (SLS): it resolves size from time-averaged angular intensity rather than from how that intensity fluctuates. It says nothing about diffusion, and it does not see a solvation shell. It sees optical contrast.

The real difference in DLS vs laser diffraction is weighting, and that is where most of the disagreement originates. In the Rayleigh regime, scattered intensity scales with roughly the sixth power of diameter, so a single 50 nm particle scatters about as much light as a million 5 nm particles. Laser diffraction is volume-weighted, which scales with the third power: one 10 µm particle carries the same weight as a million 100 nm particles. The two techniques are therefore amplifying different ends of the same distribution before any model is applied.

"When laser diffraction and DLS disagree on the same sample, neither instrument is necessarily wrong. They are answering different physical questions, and the gap between them is information."

Converting a DLS intensity distribution to volume weighting in order to force agreement with laser diffraction does not reconcile the two. It propagates the refractive index assumption and the inversion error of the DLS fit into a new number that looks more comparable than it is.

Laser Diffraction vs DLS: Why Measurement Range Is the Wrong First Filter

Range charts present the upper and lower limits of each technique as optical properties. In practice, the limits that matter are set by other physics.

The DLS ceiling is a sedimentation limit, not a detection limit. Once particles settle through the scattering volume on the timescale of the measurement, the correlation function no longer reflects Brownian motion alone. The usable DLS ceiling therefore sits in the low-micron range and moves with particle density: the heavier the material, the earlier sedimentation intrudes. A 2 µm polymer latex and a 2 µm tungsten powder are not equally measurable, even though both sit inside the nominal range.

Laser diffraction's weak point is at the fine end and in its dependence on the optical model. Mie inversion requires the complex refractive index of the particle and the real refractive index of the dispersant. An incorrect imaginary component, which is common for absorbing or mixed-phase materials, shifts the fines tail without producing an obvious fit failure. The Fraunhofer approximation avoids the need for optical constants but misreports transparent and sub-micron particles, often inflating the fine fraction or producing artifact peaks.

The practical consequence is visible in how analytical service specifications are written. The nanoparticle size and zeta potential (DLS) analytical service is scoped at roughly 1 nm to 3 µm, while the laser diffraction particle size analytical service covers 0.1 µm to 1,000 µm. Both require the refractive index to be supplied up front, and dry-method laser diffraction particle size analysis also requires density. Those requirements show that the reported size is only as good as the physical inputs behind it.

Between roughly 0.1 and 3 µm, both methods will return plausible, internally consistent, and mutually incompatible results. That overlap zone is where range-based selection fails most often, because it offers no reason to prefer either method.

Sample Concentration and Its Two Opposite Failure Windows

DLS particle size analysis fails at both ends of the concentration axis, and the two failures look different.

At high concentration, photons scatter more than once before reaching the detector. Multiple scattering decorrelates the signal faster than single-particle diffusion would, so the apparent size drifts downward. Particle interactions add a second bias, because in concentrated or charge-stabilized suspensions the measured diffusion is collective rather than self-diffusion. Backscatter geometries shorten the optical path and reduce the first effect, but they do not remove the second.

At ultra-low concentration, the number of particles in the scattering volume fluctuates. Wang and colleagues showed that these number fluctuations add a non-Gaussian term to the intensity autocorrelation function, so standard analysis models, which assume Gaussian statistics, recover an inaccurate distribution. The failure is often mistaken for the presence of large particles or dust, prompting further dilution that makes it worse.

Laser diffraction has a narrower but more visible obscuration window. Too little sample gives a signal-to-noise problem in the outer detectors that carry the fines information. Too much sample gives multiple scattering that the inversion misreads as a shift toward finer sizes.

The variable most often overlooked is that dilution can change the sample itself. Proteins with reversible self-association shift their oligomer distribution on dilution. Emulsions stabilized near a surfactant threshold can coalesce or ripen when diluted into an unmatched continuous phase. Partially soluble fines dissolve. In each case, the concentration chosen to satisfy the instrument has produced a sample that no longer represents the process. The measurement concentration is a sample-preparation decision with its own consequences, not an instrument setting.

Polydispersity Sensitivity and the Limits of PDI

Extracting a size distribution from a DLS correlation function is an ill-posed inversion. As Salazar and co-authors note in their work on DLS analysis methods, the polydisperse case is a Fredholm integral equation of the first kind, where small amounts of noise produce large changes in the recovered distribution. Cumulant analysis sidesteps the problem by fitting a single population, reporting a Z-average and a polydispersity index. Regularization methods allow multimodal output but introduce smoothing choices that shape the answer.

How to interpret DLS results? Read the Z-average for what it is: the intensity-weighted harmonic mean hydrodynamic diameter, calculated as though the particles formed a single, narrowly distributed population. The PDI is the normalized second cumulant of that fit, a relative variance of the assumed decay-rate distribution. It describes a model, not the sample.

That distinction matters more than most QC workflows acknowledge. A 2021 NIST study by Farkas and Kramar, examining how well DLS polydispersity descriptors represent actual distributional quality, found them unreliable and misleading for both monodisperse reference materials and broad-distribution biomedical nanoparticles. The authors also observed that the choice of weighting and mean is often made ad hoc to bring DLS results into agreement with other methods. A PDI below some threshold is therefore not evidence of a narrow distribution, and a PDI above it is not proof of a broad one.

"A polydispersity index describes the width of an assumed distribution, not the one actually in the cuvette. Treating it as a specification turns a fitting parameter into a release criterion."

The same intensity weighting that makes DLS a poor descriptor of a distribution makes it an exceptionally sensitive detector of trace large species. In biologic formulation and stability work, that sensitivity is the point: an early rise in Z-average often signals aggregation long before a volume-weighted method registers any change. In nanoparticle QC, the same behavior means a handful of aggregates can overwhelm the signal from the primary population. Batch DLS also has limited resolving power: populations that are not well separated in size tend to merge into a single broadened peak rather than resolve as distinct modes.

Laser diffraction fails in the opposite direction. Volume weighting suppresses the fines population, so a meaningful number fraction of sub-micron particles in a coarse powder can be nearly invisible in the D10. The D10 is also the value most exposed to refractive index error. A specification written on D10 is therefore written on the least robust number in the report.

Dispersion State: Are You Measuring Particles or Agglomerates?

Neither technique distinguishes a primary particle from a well-bonded agglomerate. Both report whatever units are present in the measurement volume, so the dispersion protocol decides which of those units the report describes.

In wet laser diffraction, sonication energy, time, and dispersant chemistry determine whether soft agglomerates are broken, left intact, or, if energy is excessive, whether primaries are fractured. In dry laser diffraction, dispersion air pressure plays the same role. The only defensible method-development step is a titration: measure at increasing dispersion energy or pressure until the distribution plateaus, then set the method just past the onset of that plateau. Any method set below the plateau is reporting a dispersion-dependent size, and any method set well beyond it risks reporting a milling result. The relationship between how dispersion energy and shear translate into deagglomeration is the same one that governs lab-scale processing, which is why the energy input needs to be controlled rather than improvised. Probe-type ultrasonic homogenizers deliver far more localized energy than a bath, so the two are not interchangeable within the same method.

"In both techniques, the reported size is partly a property of the dispersion protocol. Change the sonication energy, and you have changed the specification."

DLS adds a further trap. Pre-filtering to protect the measurement from dust also removes the upper tail of the real distribution, including the aggregate population that stability studies are designed to detect. Filtering is appropriate when the goal is the primary population and inappropriate when the goal is aggregation, and the method record should state which applies.

A Particle Size Analysis Decision Framework for Powder QC and Nanoparticles

The more useful selection criterion is the question the measurement has to answer, not the nominal size of the sample.

For battery active materials, ceramic feedstocks, and pharmaceutical APIs, laser diffraction is the natural release method because packing, flow, and dissolution depend on the volume- and surface-weighted distribution rather than on diffusion. For suspensions of nanoparticles & nano powder materials, the answer depends on what the size is used for. Hydrodynamic size governs colloidal behavior and transport, but a supplier certificate may be reporting a primary crystallite or BET-equivalent size that neither light-scattering method measures directly. When both methods are needed, each result should be reported on its own weighting basis rather than converted into agreement.

Reading Disagreement Between Methods as Data

When the two methods diverge, the direction of the gap usually points to the cause.

A laser diffraction D50 well above the DLS Z-average usually indicates agglomerates or coarse material that sedimented out of the DLS measurement volume or was filtered before measurement. The DLS result is describing a subpopulation.

A DLS result well above the laser diffraction result usually points to one of two causes. The first is a trace large population amplified by intensity weighting. The second is a thick solvation, polymer, or ligand layer that contributes to hydrodynamic size but not to optical contrast. Measuring at a second concentration and inspecting the intensity distribution, rather than the Z-average alone, usually separates the two.

For a well-characterized material, the offset between the two methods is often stable. Once established, that offset becomes a baseline, and a change in it is frequently the earliest indication that something in the process or supply has shifted: surface treatment, milling energy, drying history, or raw material lot. Labs that measure both techniques routinely and track the gap gain a diagnostic that neither method provides alone.

Key Takeaways

• Range overlap is the least informative part of the comparison. Weighting, optical-model inputs, and the physics that set each technique's real limits determine which method is appropriate. The published range cannot.

• Sample preparation manufactures the result more often than the instrument does. Dilution, dispersion energy, and filtration each change what is actually being measured, and all three belong in the method record alongside the reported size.

• DLS is a better detector than descriptor. Its intensity weighting and fitted PDI make it unreliable for describing a broad distribution, but those same properties make it the more sensitive early warning for aggregation.

• Disagreement between methods is a measurable variable. A stable cross-method offset works as a process fingerprint, and a change in that offset is often the first sign of process drift.

Final Thoughts

Laser diffraction and DLS are complementary rather than interchangeable. Each is precise within its own physics and misleading when its output is read as a universal particle size. The most reliable programs treat technique selection as a question of which number the decision depends on, and document the preparation conditions that produced it.

Choosing between laser diffraction and DLS ultimately depends on what the measurement has to defend: a release specification, a stability claim, or evidence of a process change. MSE Analytical Services supports both techniques for teams that need orthogonal sizing data without committing to new instrumentation. For non-standard sample configurations or material specifications, explore our customization capabilities. To discuss method selection with our technical team, contact us. You can also follow ongoing application insights on LinkedIn or visit the MSE Supplies homepage.

References

1. Farkas, N., & Kramar, J. A. (2021). Dynamic light scattering distributions by any means. Journal of Nanoparticle Research, 23(5). https://doi.org/10.1007/s11051-021-05220-6

2. Wang, M., Shen, J., Thomas, J. C., Mu, T., Liu, W., Wang, Y., Pan, J., Wang, Q., & Liu, K. (2021). Particle size measurement using dynamic light scattering at Ultra-Low concentration accounting for particle number fluctuations. Materials, 14(19), 5683. https://doi.org/10.3390/ma14195683

3. Salazar, M., Srivastav, H., Srivastava, A., & Srivastava, S. (2023). A user-friendly graphical user interface for dynamic light scattering data analysis. Soft Matter, 19(34), 6535–6544. https://doi.org/10.1039/d3sm00469d