Referee Level Particle Standard Preparation

This article is a reprint of a the Poster Presentation at the 2013 PDA Visual Inspection Forum.  The document is provided as basic information on the manufacture of Referee Level Particle Standards (RLPS™) used for training and qualification of inspectors.  The paper discusses the steps necessary for the creation of standards that are verified as “clean” prior to the particle seeding procedure and then certifying that they are “clean” and container only a single seed particle.

This full document can be downloaded from the link provided but it is necessary to receive a key to open the document.  You can request the key by calling the Phoenix Imaging office or applying for one via email.

The complete Referee Level Particle Standard Preparation document.

Excerpts from the Poster Presentation follow:

Background

The Referee Level Particle Standards (RLPS™) can be used for both Human Baseline Determination and qualification of automated inspection equipment.  The standard RLPS™ set consist of 100 containers (36 seeded and 64 clean).  Additional samples may be added to the RLPS™ set to include more common manufacturing defects.  However, only the spherical samples should be considered for Human Baseline Determination.  Of the seeded samples 25 of them are in the “Must Reject Zone”.  The size of the contamination is dependent on both the container shape and size and most importantly the intensity of the light in the inspection environment.  The minimum detectable size of a particle within a container in the “Must Reject Zone” (that is considered 70.7% probability of rejection) is directly related to the light intensity of the inspection environment.  For example, at 550 foot-candle (5920 Lux) the normalized human inspector with 20/20 near vision correction is able to detect a 95µm diameter spherical particle 70.7% of the time.  If a lower intensity illumination is used in the inspection volume then the human inspector will see the same size spherical particle with a lower probability of detection.  The researcher must perform a Human Baseline Determination in order to determine the exact capability of their human inspection staff.  This Human Baseline Determination can be achieved with the RLPS™ set and a well construction inspection environment with consistent illumination (please refer to our MIB-50, MIB-90 or MIB-100 inspection booths for additional information).

RLPS™ Preparation

 

RLPS Preparation Cycle Figure 1 Process Flow Chart

RLPS Preparation Cycle Figure 1 Process Flow Chart

Container and Stopper Preparation:  Figure 1 below is a flow chart that represents the complete procedure for manufacturing the RLPS™ sets.  The diagram represents the sequence of steps necessary to manufacture a RLPS™ set.  The steps vary in the amount of time necessary to complete and each must be completed before moving on to the next step.  We process water through an elaborate process of filtering, RO, filter to 1µm, DI, filter to 0.5µm, distillation and final purification through 0.1µm filtration.  After the entire the water is measured not less than 0.5 micro-ohm.   The most import step is “Rinse Cycle“.  This step insures that the container is processed thoroughly to insure that the standard will provide years of service.  If not performed correctly the container and its contents will become turbid after several months making the product worthless.  It can take up to a dozen or more rinse cycle before the container is “clean”.  Any short cut in the “Rinse Cycle” will results in an inferior product that will be unable to perform its function in a year or two.

After completing the “Rinse Cycle” the product is ready for “Final Fill“.  Filled to the customer specified fill volume the product must be inspected before it can proceed to “Seeding”. Referee Level particle Standards are manufactured to the highest quality standards and are inspected with >10,000 foot-candle lighting to insure that no sub-visible particles greater the several microns are present in the solution.  We use a special instrument designed and built by Phoenix Imaging called the ParticleScope™ that is capable of identifying the presence of any particles.  If the container is accepted as “clean”, it is moved on to the seeding operation.

The “Sample Storage (Unseeded) and Seeding Cycle” is where containers are either used as part of a set or rest for extended periods.  The seeding procedure is a time intensive manual operation.  It begins with the selection of seeding materials that “Best Fits” the customers requirements.  A number of factor come into play here because the goal of the RLPS™ set is to have a number of seeds extending through the “Gray Zone” or the samples that an inspector would isolate 30% to 70% of the time.  Particles in the “Reject Zone” are easy to obtain and most inspector will have little difficulty identifying them as reject containers.  In order to develop the “Human Response Curve” (HRC) correctly the particle standards must be present along the entire curve.  The HRC is related to the Probability of Rejection (PR) at a specified light intensity within the inspection volume.  If all of the sample in a challenge set occur in the “Reject Zone”, then the true shape of the curve will not be identified and an incorrect lower limit of the “Reject Zone” will be assigned.

Phoenix Imaging has multiple levels for the RLPS™ sets.  Each level is associated with the distribution of particle sizes in the RLPS™ set, Level I is for small containers with the center of the distribution around 75µm diameters and Level VIII has a center around 150µm diameters.  It is import to note that a complete RLPS™ or Challenge set includes a wide range of seed particles of various materials.  However, only spherical seed particles should be used to determine the shape of the HRC and for Baseline determination.  This is because spherical seeds only require a single dimension in order to size correctly.  Particles of irregular shapes have a different dimension depending on their orientation and therefore have a different Probability of Rejection depending on the orientation during inspection.  A list of sample seed particles can be found in this website under RLPS Sample Seed Particle List.  This has been a fact argued many time with my associate, Julius Z. Knapp, and the his studies have verified that spherical particles generate consistent and reproducible results.  It is ok to include other particle shapes in the challenge set to help the inspector experience diversification but only use the data from the spherical particle for the Baseline Determination.

The seed particle are measured optically in air using a high resolution microscope and image processing system.  The particles are positioned on a precision laser encoded micrometer stage with NIST traceable certification an having an accuracy of better than 1.0µm.  The image processing system was designed and built by Phoenix Imaging and is capable of measuring particles to better than 1µm, but we report the particle only to ±1.0 µm in diameter or the particles proximal length.  Each particle is photographed and recorded as part of the Certificate of Analysis (COA).  When a customer receives a RLPS™ set from Phoenix Imaging they will receive a CDROM with images of each seed particle for reference.  Each seed receive a unique ID code and this is transferred to the container as a unique serial number laser etched onto the closure.  We also laser etch a 2-D code on to the closure so that a customer may use a bar code reader to quickly identify the container in the Accept and Reject collections after testing.  The containers are inspected after seeding to insure that it was not accidentally contaminated and that a single seed particle is present.  Phoenix Imaging provides the unique benefit that is not available from any other manufacturer by using the ParticleScope™ Instrument to verify that only one seed particle of the approximate size and shape is in each container.

This is a short description of the Referee Level Particle Standard Preparation but the full version is available for download.