I am regularly asked how long a measurement takes on one of our Zetasizer instruments. I get this question even more frequently when people are asking about our Zetasizer APS (auto plate sampler) dynamic light scattering instrument, which analyzes samples that have been prepared on either 96- or 384-well plates.
It is an intriguing question, and not one with a black and white or definitive answer. Strangely, it often makes me wonder what the person is really asking.
Let me explain
When making light scattering measurements, there are numerous factors that come into play. Just a few of these factors include: the length of time defined for a single run (or auto-correlation function acquisition), the number of repeat runs specified to average together or include in the analysis for a single measurement, and then how many repeat measurements are made to establish repeatability on each sample.
As any scientist will tell you, the more data you have the better you can establish statistical significance and hence have greater confidence in the result. In light scattering measurements, as with most other things in life, there are trade-offs involved.
The fastest car on the road is typically not also the safest or most reliable car. Cars are designed with different purposes in mind and they deliver different levels of performance, handling, and safety as a result. Different purposes, different designs.
Back to light scattering
Designing a light scattering experiment is no different than designing just about anything else. One has to first think about the outcome or performance level they desire first and then design the experiment to deliver that outcome. Speed often comes through sacrifice of other things.
The person conducting a light scattering experiment could define a single measurement to consist of a single correlation curve of 1 second duration and not make any repeat measurements. That would make analysis of a single sample very fast (indeed, only 1 second) but it might not be a reliable measurement. And without repeated measurement, there simply is not enough data to establish a high confidence level in the result nor is there any indication of the repeatability error associated with a single sample to then try to establish any sort of statistical significance to differences seen between two or more samples.
In other words, if I don’t know the error associated with the measurement of a single sample then I can have no idea if a different sample with a different result is actually statistically and reliably different. Speed is gained but quality of and confidence in the result is lost.
On the other hand, I could design an experiment with very long acquisition times, many repeat runs and many repeat measurements, but at some point I’ll reach the point of diminishing returns and the time required to collect the data will lead to minimal or negligible increases in statistical reliability.
There is a balance between speed and data quality that must be considered.
When designing a light scattering experiment, keep these things in mind. You should strive to collect measurements that consist of multiple autocorrelation acquisitions of significant enough duration to capture enough of the scattering information for an accurate and reliable result. This can usually be achieved within a few minutes per sample. Even then, though, you ideally should repeat the process several times on the same sample to establish same-sample repeatability.
So the next time you feel the urge to ask, “How fast can I make a DLS measurement?”, stop and think about what quality of data you are comfortable with. Good science should not be sacrificed for speed nor should you make decisions based upon different perceived results that might not even be statically significant.