A Z-average of “180 nm” and a PDI of “0.25” mean nothing on their own. The Z-average is an intensity-weighted cumulants mean, not the average particle diameter a microscope would show you, and a PDI in the 0.1-0.3 range is normal for a great many real formulations, not a red flag. Most misreads of a dynamic light scattering (DLS) report come from treating these two numbers as if they were a simple physical measurement instead of the output of a specific mathematical fit to a correlation function. This guide covers what each number actually represents, how to read the polydispersity index, and why the intensity, volume and number distributions on the same report can tell three different stories about the same sample.
What DLS Actually Measures
Dynamic light scattering (also called photon correlation spectroscopy, PCS) measures the rate at which the intensity of laser light scattered by particles in suspension fluctuates over time. Small particles diffuse faster under Brownian motion than large ones, so the scattered-light intensity at a fixed detector angle fluctuates faster for a suspension of small particles than for large ones. The instrument builds a time autocorrelation function of these intensity fluctuations, fits it to extract a diffusion coefficient, and converts that diffusion coefficient to a hydrodynamic diameter using the Stokes-Einstein equation, which also requires the solvent viscosity and temperature as inputs.
“Hydrodynamic diameter” is the diameter of a hypothetical hard sphere that would diffuse at the same rate as the particle actually being measured, including any solvation shell, surface structure or attached molecules moving with it. It is routinely somewhat larger than the “dry” particle diameter reported by electron microscopy for that reason, and the two are not directly interchangeable.
Z-Average: What It Is and Isn’t
The Z-average size and the polydispersity index (PDI) both come from a single calculation called cumulants analysis, applied to the correlation function under ISO 22412 (Particle size analysis — Dynamic light scattering). Cumulants analysis fits the correlation function’s decay to a single exponential plus correction terms, producing:
- Z-average diameter — an intensity-weighted harmonic mean size. It is a single, robust, standard value suitable for comparing batches under fixed, identical measurement conditions, but it is not a number-average or a volume-average, and it should never be reported as “the particle size” without qualification.
- PDI — a dimensionless width parameter derived from the second-order term of the same fit, describing how broad the size distribution is around the Z-average.
Because the Z-average is intensity-weighted, it is disproportionately pulled toward larger particles (see the worked example below). It is the correct value to quote for routine QC comparisons of nominally similar samples measured the same way. It is not the correct value to compare against a number-average size from a different technique, and it is not meaningful for a sample that the correlation function shows to be multimodal — for those, the intensity, volume or number distribution (the separate peak table on the report) is the more honest description.
Reading the Polydispersity Index (PDI)
PDI runs from 0 (theoretically perfectly monodisperse) upward, with no fixed upper bound, though cumulants analysis itself becomes unreliable at high PDI. The bands below reflect general guidance used across DLS instrument documentation (e.g. Malvern Panalytical’s technical notes on DLS terminology); treat them as a reading aid, not a hard pass/fail specification — the “acceptable” PDI for a given sample depends entirely on what that sample actually is (a purified virus-like particle and a crude liposome prep do not share a target band).
| PDI | What it typically indicates | Practical read |
|---|---|---|
| < 0.05 | Highly monodisperse | Rare outside certified monodisperse latex/polystyrene size standards used to check instrument performance; if a real biological or formulation sample reads this low, treat it as worth double-checking rather than assuming exceptional quality. |
| 0.05-0.1 | Narrow, near-monodisperse | Typical of well-controlled, single-population nanoparticle or protein preparations. |
| 0.1-0.3 | Moderately polydisperse | Common and often perfectly acceptable for many real formulations (e.g. many liposomal or polymeric nanoparticle preps); the cumulants Z-average and PDI are still meaningful here. |
| 0.3-0.5 | Broad / polydisperse | Cumulants Z-average is still calculable but describes an increasingly heterogeneous population; check the intensity distribution for multiple peaks before trusting a single Z-average number. |
| > 0.5-0.7 | Very broad | The single-exponential cumulants fit is a poor description of the underlying population; report the distribution (peaks), not just Z-average/PDI. |
| > 0.7 | Not a reliable cumulants fit | The sample is likely multimodal, aggregating, or contains dust/debris. Most DLS software will flag results in this range as low-confidence; consider filtration, centrifugation to remove aggregates/dust, or a technique better suited to broad or multimodal distributions. |
Intensity vs. Volume vs. Number Distributions — A Worked Example
Every DLS report that shows a distribution (rather than just Z-average/PDI) actually derives three distributions from the same raw correlation data, and they can look dramatically different for the identical sample. This is the single most common source of confusion in reading a DLS report, and it follows directly from the physics of light scattering.
Under the Rayleigh approximation (valid for particles small relative to the laser wavelength), the intensity of light scattered by a particle scales with the sixth power of its diameter (I ∝ d⁶). Volume scales with the third power (V ∝ d³). Number, by definition, does not scale with size at all — it just counts particles. The consequence: a small number of large particles or aggregates can dominate the intensity distribution while being essentially invisible in the number distribution.
Illustrative worked example (a constructed scenario to show the arithmetic, not a specific instrument reading from any real study): suppose a sample by particle count is 99% 20 nm particles and 1% 500 nm aggregates. Compare the scattered-light intensity contributed by one 500 nm particle to one 20 nm particle using the d⁶ relationship:
(500 / 20)⁶ = 25⁶ = 25 × 25 × 25 × 25 × 25 × 25 ≈ 2.44 × 10⁸
So a single 500 nm particle scatters roughly 244 million times more light than a single 20 nm particle. Even at only 1% of the particles by number, the 500 nm population can contribute effectively all of the measured scattered intensity, producing an intensity distribution dominated by a peak the sample is, by particle count, almost entirely not made of. Converted to a volume distribution (d³ scaling), that same 1% aggregate population still shows up but far less dominantly — 25³ = 15,625 times the volume-per-particle of the small population, roughly a 99:1 ratio in volume terms once the 99:1 number ratio is factored in, i.e. the aggregate volume becomes comparable to, not overwhelming of, the main population’s volume. Converted to a number distribution, the 500 nm peak nearly disappears, correctly showing that the sample is overwhelmingly made of 20 nm particles by count.
The practical reading rule: intensity distribution is the most sensitive to trace aggregates/dust and the least representative of “how many particles are what size”; number distribution is the most representative of population composition by count but the least sensitive (and the noisiest) for detecting a small aggregate population; volume distribution sits between the two and is often the most relevant for applications where mass/dose matters (e.g. drug delivery nanoparticle formulations). Most DLS software derives volume and number distributions from the intensity distribution using Mie theory and the sample’s refractive index — treat these derived distributions as good for identifying whether a second population exists and roughly how significant it is, not as a precise particle count.
Key DLS Parameters and What Each One Trades Off
| Parameter | What it controls | Trade-off |
|---|---|---|
| Measurement (scattering) angle | Sensitivity to different particle sizes and to multiple scattering | Backscatter detection (commonly ~173°) reduces multiple scattering and stray light in concentrated or turbid samples and lets measurement close to the cuvette wall, minimizing path length through the sample; 90° detection is more traditional and sensitive to smaller particles in dilute, clean samples but more affected by dust and multiple scattering. |
| Laser attenuation / count rate | How much light reaches the detector | Too little attenuation (too much light) saturates the detector; too much attenuation (too little light, e.g. a very dilute or weakly scattering sample) gives a noisy, low-count-rate correlation function. Modern instruments auto-adjust this, but a manual override is sometimes needed for very dilute biological samples. |
| Temperature equilibration time | Accuracy of the viscosity value used in the Stokes-Einstein calculation | Viscosity is temperature-dependent and enters the size calculation directly; too short an equilibration time (thermal gradients, convection currents in the cuvette) inflates apparent size and PDI. Longer equilibration costs throughput. |
| Number of runs / measurement duration | Signal averaging, precision of the correlation function | More/longer runs reduce noise and improve reproducibility, especially for weakly scattering or low-concentration samples, at the cost of instrument time. |
| Sample concentration | Signal strength vs. multiple scattering | Too dilute gives a weak, noisy signal; too concentrated causes multiple scattering (photons scattered more than once before detection), which biases the correlation function and apparent size. Most methods specify a validated concentration range or a dilution step. |
| Cuvette type (disposable vs. quartz) | Optical path, minimum sample volume, cleanliness | Disposable plastic cuvettes are convenient and avoid cross-contamination but have a fixed geometry and can scratch; quartz cuvettes give better optical quality and can be used with smaller-volume low-scattering samples but require rigorous cleaning between uses to avoid carryover. |
How to Read a Full DLS Report, Field by Field
- Z-average and PDI — the cumulants-fit headline numbers; see the sections above. Always read PDI before trusting a single Z-average value.
- Intensity, volume and number distribution peaks — usually reported as up to three peaks each with a mean/mode size and a %intensity, %volume or %number contribution. A report showing one dominant intensity peak at, say, 98% and a small second peak at 2% intensity does not mean the sample is 98% one size by count — convert your reading to the distribution type relevant to your question, per the worked example above.
- Correlation function (correlogram) plot — the raw decay curve the whole analysis is fit from. A smooth, single-exponential-looking decay supports a clean monomodal fit; a decay with a shoulder or a second, slower-decaying component is visual evidence of a second, larger population even before looking at the numeric peaks.
- Fit quality / residuals — most software reports how well the cumulants fit matched the measured correlation function. A poor fit (high residuals) means the reported Z-average/PDI should be treated with caution regardless of what the numbers say.
- Count rate (kcps) — the raw photon count rate at the detector. Useful for confirming the sample is neither too dilute (low, noisy count rate) nor too concentrated (very high count rate, risk of multiple scattering); also a fast way to spot a sample that was accidentally not loaded or is far more dilute than intended.
- Attenuation index / measurement position — instrument settings the software chose (or that were set manually) to bring the count rate into range; large swings between replicate measurements on the same sample can flag inconsistent sample handling (e.g. settling, bubbles) between runs.
- Y-intercept / intercept value — a data-quality indicator on some instruments; a low intercept generally signals a poor signal-to-noise correlation function (e.g. too little scattering, stray light, or a dirty cuvette) rather than a real sample property.
Troubleshooting: Common DLS Report Problems
| What you see | Likely cause(s) | What to try |
|---|---|---|
| PDI consistently > 0.5-0.7 on a sample expected to be uniform | Dust/debris contamination, sample aggregation, incomplete filtration, insufficient equilibration | Filter the diluent and, where the sample tolerates it, the sample itself (check pore size against expected particle size first); centrifuge briefly to pellet dust/large aggregates before measuring; re-check cuvette cleanliness. |
| Z-average drifts upward across repeat measurements of the same sample | Progressive aggregation over time, sample settling, or contamination building up during handling | Reduce time between preparation and measurement; check formulation stability separately from the DLS measurement itself; measure fresh replicate cuvettes rather than repeatedly re-measuring one. |
| Very low or erratic count rate | Sample too dilute; weak scatterer (small particles, low refractive index contrast with solvent); air bubble in the beam path; wrong cuvette placement | Increase concentration within the technique’s validated range; check for bubbles; re-seat the cuvette; increase measurement duration or number of runs to average out noise. |
| Very high, saturating count rate | Sample too concentrated; multiple scattering | Dilute the sample and re-measure; confirm the diluted result is stable across a small dilution series (a size that changes with concentration is itself informative — it can indicate multiple scattering or, in some biological samples, real concentration-dependent self-association). |
| Multiple, inconsistent peaks between replicate measurements | Rare large aggregates or dust particles moving randomly through the small scattering volume | Increase the number of runs/measurements to average out the effect of rare transient events; filter or centrifuge; report the volume or number distribution alongside intensity so a reviewer can judge how significant the aggregate population actually is. |
| Z-average much larger than the size seen by electron microscopy on the same sample | Hydrodynamic diameter genuinely includes a solvation/hydration shell and any surface coating (not a discrepancy to “fix”); or a trace aggregate population is inflating the intensity-weighted Z-average | Check whether the number or volume distribution mode size is closer to the EM value (aggregate effect) or whether the whole distribution is uniformly shifted (real hydrodynamic-vs-dry-diameter difference); these have different explanations and different fixes. |
Which Number Should You Actually Report?
For routine batch-to-batch QC comparisons under a fixed, validated method, Z-average and PDI together are usually the right, standardized pair to report — they are reproducible, ISO-defined, and comparable between runs on the same instrument and method. For characterizing a sample that the correlation function or peak table shows to be multimodal, or for any claim about what fraction of the sample by count or by mass is a given size, report the relevant distribution (number, volume or intensity, chosen for the question being asked) rather than a single Z-average figure, and say explicitly which distribution type is being quoted — “180 nm” without saying whether that is Z-average, an intensity-peak mode, or a volume-peak mode is not a complete result.
Sample Preparation Notes
Because DLS is exquisitely sensitive to rare large particles (per the d⁶ scaling above), sample preparation has an outsized effect on the result compared to many other analytical techniques. Diluent should be filtered (commonly through a 0.02-0.2 µm filter, chosen relative to the expected particle size so the filter does not remove the analyte itself). Cuvettes and pipette tips should be handled to minimize introducing dust. Where dilution is required to bring concentration into a instrument’s working range, a short dilution series is often worth running to confirm the reported size is stable and not an artifact of concentration-dependent multiple scattering or aggregation. See our guides on serial dilution technique and molarity and solution calculations for the arithmetic behind building an appropriate dilution series, and analytical balance calibration if formulation components are being weighed in as part of sample prep.
Frequently Asked Questions
Is a lower PDI always better?
Not necessarily. A lower PDI means a narrower size distribution, which is usually desirable for a formulation meant to be a single, uniform population. But PDI is a description of the sample’s actual homogeneity, not a quality score independent of what the sample is supposed to be — a deliberately mixed or broadly-distributed system will (correctly) show a higher PDI, and that is not a measurement failure.
Can DLS distinguish two populations that are close in size?
Cumulants analysis (which produces Z-average/PDI) assumes a single, roughly log-normal population and struggles to resolve two populations that are within about a factor of 3 in size or that differ greatly in relative amount. Multimodal fitting algorithms (distribution analysis, sometimes marketed as “multiple narrow modes” or similar) can do better at resolving close or minor populations, but even these have practical resolution limits — a technique with fundamentally different physics (e.g. size-exclusion chromatography, analytical ultracentrifugation, or electron microscopy) is often needed to confirm a suspected close-lying second population.
Why does my DLS size not match the size from electron microscopy?
DLS reports a hydrodynamic diameter in solution, including any solvation shell or surface layer moving with the particle; electron microscopy typically measures a dry, often shrunk or flattened particle in vacuum. Some difference between the two is expected and not itself evidence of an error in either measurement — see the troubleshooting table above for how to tell that expected difference apart from an aggregation artifact.
What is a “good” PDI for a nanoparticle drug delivery formulation?
This varies by application and is set by the formulation’s own quality requirements, not by DLS itself; a PDI below roughly 0.2-0.3 is commonly used as an internal batch-acceptance criterion in nanomedicine formulation work, but there is no universal regulatory or physical threshold that applies to every particle type, and the appropriate cutoff should be defined and justified for the specific product.
Related Lab Technique and Instrument Guides
DLS is one of several particle- and molecule-characterization techniques in routine lab use. See also:
- Flow Cytometry: How It Works, and How to Read Dot Plots and Histograms — another light-scattering-based sizing and counting technique, using forward/side scatter on individual cells or particles rather than an ensemble correlation function.
- Confocal Microscopy: Principle, Setup, and When to Use It — direct imaging as a complement to (or check on) DLS’s indirect, ensemble-averaged size measurement.
- Powder Diffraction File (PDF): ICDD’s Reference Database for XRD Phase Identification — a different particle-characterization technique (crystal structure/phase, not size).
- UV-Vis Spectrophotometer Basics and Spectrophotometer Calibration — related optical/spectroscopic bench techniques and instrument QC.
- HPLC: Columns, Mobile Phases, and a Peak-Problem Troubleshooting Table, Gas Chromatography and LC-MS Explained — separation and identification techniques often used alongside particle sizing in formulation and materials characterization workflows.
- Thin-Layer Chromatography: How to Run a Plate and Calculate Rf — a lower-instrumentation separation technique for the same kind of bench workflow.
- Analytical Balance Calibration, Serial Dilution Technique and Molarity and Solution Calculations for the Lab — sample-preparation fundamentals that directly affect DLS result quality, per the sample preparation notes above.
