Screening Suppliers for Potential USP Particle Counts Failures Using FIM Analysis

USP particle counts tests such as USP <788> and USP <789> do not require flow imaging microscopy (FIM) analysis, however, when it comes to providing a quick and accurate screening of products that failed a USP particle counts test, we find that FIM analysis is invaluable. It is also helpful to screen different suppliers and vendors ahead of time when selecting new packaging components and containers that come into contact with the product. We refer to this type of screening as an FIM rinse study because empty bottles or packaging containers are rinsed within our cleanroom facility with milli-Q water and the effluent is run through the FlowCam.

The objective of this FIM rinse study was to analyze the effluent from each vendor, and compare the particle counts between vendors, then use the morphology information in the dataset to classify the particles into one of four groups (air bubbles, fibers, protein, and other). These categories were not predetermined. They were selected based on what particles are observed in the samples.

We rinsed three sets of five empty 50 mL plastic bottles with particle free water (PFW) within our cleanroom facility and then tested each sample for particulate using FIM. Each set of five was a similar plastic bottle from three different vendors. The containers were rinsed by filling one of the five containers with approximately 30 mL of water, agitating the container, pouring the effluent from one container to the next, and collecting the final effluent for FIM analysis.

The particle counts were binned into the two common size ranges typically used for USP particle counts testing (1) greater than 10 µm, (2) greater than 25 µm. All length measurements were reported in equivalent spherical diameter (ESD).

Table 1 shows that none of these vendors would necessarily fail a USP particle counts test, but it looks like the bottles from Vendors B and C had the most particles. These are the raw numbers without any corrections for particle types observed.

We can use the particle classifications to find the percentage of each particle type for particles in these size ranges. This helps clients understand the nature of the particles that may be causing USP <788> or USP <789> failures.

The particles from Vendor A were 46% other particle types overall and they made up 50% of the particles with ESD > 10 µm and 100% of the particles with ESD > 25 µm. Few air bubbles were observed in this sample.

Overall, the particles from Vendor B were mostly classified as oil-like, fiber-like, and other particles, but in the larger size ranges we see an increasing fraction of fiber-like and protein-like particles and a decreasing fraction of other particle types. Some air bubbles were also observed in this sample.

The bottles from Vendor C had high particle counts, but the classification results show that the majority of the particles observed were actually air bubbles. We can correct the counts data by excluding air bubbles from the counts in each size range.

Air bubbles plague other particle sizing and counting methods because they cannot be distinguished from actual particles, but FIM provides a way of distinguishing and labelling them in the dataset.

This changes the narrative slightly. Before, it looked like Vendor A had the fewest particles and Vendors B and C were not as clean. Now, it appears that Vendors A and C are particle free and Vendor B has higher particle contamination.

We can also remove the air bubbles from the classification pie charts and show the true particle composition for each sample. This can be very useful when investigating an actual USP particle counts test failure. Not only can we identify and correct for potential artifacts, but we can also understand what kinds of particles are causing the failures if there are multiple particle types in the sample.

Vendor A did not have many air bubbles, so the results are very similar.

Vendor B had a few air bubbles but most of the counts were from true particles.

Since the raw particle counts from Vendor C were mostly air bubbles, this vendor’s bottles were actually the most particle free.

In summary, FIM is an excellent method for producing accurate particle counts and concentrations for fluid samples, and the morphological information can be used to sort each particle into general classification categories such as air bubbles, oil droplets, fibers, protein, glass, and other. In this way, FIM can be used as a powerful screening tool for possible contamination sources and investigations.

Do you have a project you need to discuss? Speak to a scientist.