Development of an Improved Fluidization Segregation Tester for Use with Pharmaceutical Powders

Understanding the segregation potential of pharmaceutical dry powders and granules is an important aspect of solid oral-dosage form development. Powder segregation demixes a homogenous powder blend. Further processing of this heterogeneous mixture can result in final-product quality deficiencies such as variable dosage-form potency, variable tablet or capsule fill weight, variable tablet hardness, and nonuniform appearance. Segregation also can affect manufacturing process robustness through erratic or unstable powder flow, under- or overcompression, variation in tablet-core tensile strength, and unacceptable blend uniformity (1).

AUTHORS

Fluidization segregation is a common segregation mechanism for pharmaceutical dry-powder and granule systems (1). It can occur during unit operations such as blender-to-bin transfers, bin discharge, pneumatic conveyance, bin-to-tablet press transfer, and fluidization in a fluid bed. The fluidization segregation mechanism is illustrated in Figure 1 using the example of a bin discharge unit operation. Before discharge, the particles in the bin are arranged homogeneously (see Figure 1a). During discharge, the particles are entrained in a counterflow air stream (see Figure 1b). Smaller particles and less-dense particles will be carried higher in the air stream. They also have a lower terminal velocity and will settle at a slower rate compared with larger particles. These particles also can be deflected easily by air turbulence, further increasing settling time. This results in a layer of small particles on top of a bed of larger particles or a particle-size gradient in the settled pile (see Figure 1c).

Figure 1

The extent of fluidization segregation is dependent upon a combination of material properties, process equipment characteristics, and process conditions. Material-selection guidelines that can help minimize fluidization segregation risk have been discussed, but it is not always possible or desirable to modify the properties of mixture components because of material limitations and regulatory considerations. The particle factors often cited in the guidelines include particle size, particle-size distribution, particle shape, and true density for which, in general, segregation risk is minimized when the properties of a mixture's individual components are most similar. Cohesion also plays an important role where sticking together of individual and dissimilar particles can minimize segregation by limiting demixing. Methods exist to measure this property, but the threshold value where it has an effect on segregation is not well understood. Equipment characteristics such as chute angles, material of construction, and powder-drop heights also can cause segregation to occur (2). This multitude of factors with potential interaction terms makes it difficult to predict a priori the tendency of a given powder blend to segregate, although the guidelines give a good starting point for formulation and process design.

A test method that gauges fluidization segregation potential in the early stages of solid oral-dosage form development would be a valuable tool for the development scientist. The method must have the following characteristics:

• use a minimal amount of material because a finite quantity of the active pharmaceutical ingredient (API) is typically available during early drug product development;

• be rapid to allow the screening of numerous formulation options;

• be reproducible to ensure a meaningful comparison of test results.

One test method that is standardized through ASTM International is the Jenike and Johanson (J&J, Tyngsborough, MA) fluidization segregation tester (FST) (3, 4). The tester fluidizes a powder sample in a column of air, allows the particles to settle in the column, and has a mechanism to retrieve top, middle, and bottom samples of settled powder. These samples can then be analyzed to determine if there is a property gradient within the column. The operation of this equipment is described as an ASTM standard practice (4). Though the test equipment reproduces the fluidization mechanism, it has several disadvantages which include a limited data set of three samples per test, the need to riffle each sample before analysis, operator dependence on setting the airflow needed to achieve fluidization, and the use of a relatively large amount of material compared with what is available during the early stages of development.

Other researchers have reported alternative experimental methods to determine fluidization segregation tendencies of pharmaceutical powder–granule mixtures. Their approaches have either increased resource requirements or are confounded with other segregation mechanisms. An example of the former is a variation of the J&J FST described by Wormsbecker et al. (5). They carried out the fluidization experiment in a modified fluid bed, and a core sampler was developed to remove samples that enabled the evaluation of both axial and radial segregation. It provides a more thorough analysis of the segregation pattern through an increase in resources, material, and time (5). An example of the latter approach is provided by Liss et al. through the measurement of segregation caused by a vertical drop through a pipe. Fluidization is only one of the segregation mechanisms operating in this system, and the researchers have not provided a means to analyze the results to identify the individual causal effects (6). In a similar vein, Johanson Indices are claimed to be predictive of a powder mixture's fluidization potential, which can lead to segregation during a transfer operation. This segregation measurement is indirect and must be validated by a more direct test (7, 8).

Theoretical approaches have been explored as a means to predict fluidization segregation tendencies, but these approaches are not yet suitable for routine use in formulation development projects. Wu and Baeyens used a combination of experimentation and calculation to develop a predictive approach based on a calculation of the fluidization mixing index from experimental data. They showed some success with binary mixtures of noncohesive, spherical particles, but more work would be needed to demonstrate the utility of the experimental approach and model with the more complicated mixtures typically encountered in pharmaceutical systems (9). Abatzoglou and Simard developed a mathematical tool for understanding segregation for bin-discharge operations. Fluidization is a element of the model, but the model does not allow the contribution of fluidization to be examined separately from other components (10). Theoretical modeling of particle fluidization has been extended to understanding and predicting particle segregation. These studies are important for advancing knowledge and developing future technologies (11, 12).

Analysis of available test methodology indicated the J&J FST tester with the ASTM method is most applicable for routine formulation testing. The tester can be improved by making it material sparing, more efficient, and more reproducible. The primary objective of this study is to use the J&J FST and ASTM method as a basis to develop an improved tester dubbed the fluidization material-sparing segregation tester (FMSST) and an improved test protocol called the sawtooth method.

The FMSST was developed and fabricated, and then tested with representative pharmaceutical formulations. The tests compared the FST and FMSST using the ASTM method, compared the ASTM and sawtooth methods using the FMSST, explored the importance of sample preparation with the FMSST, and examined the reproducibility and robustness of the FMSST when run with the sawtooth method. The tests were intended to show the utility of the FMSST with the sawtooth method as a routine means to determine a formulations' segregation potential during early development when limited quantities of material are available. This information can be used as a criteria for selecting formulations and manufacturing processes for clinical trials supplies and commercial products. These tests were carried out on an ad hoc basis and were not intended to provide a rigorous evaluation of the tester's capabilities or a comparison with the FST–ASTM system. Instead, the work described herein was designed to show how the FMSST with the sawtooth method can be used in place of the FST–ASTM method. Advantages include reduced material requirements, a more controlled test, and the suitability for the early stages of solid oral-dosage form development.

Materials and methods

Instrument development. The new fluidization tester was designed to provide a routine test that would rank a formulation's potential to undergo fluidization segregation using a minimal amount of material. The tester design was based on the J&J FST instrument with improvements that would decrease test-material requirements, minimize operator bias, increase control and reproducibility, and eliminate the need for a setup run before the actual test. Specific requirements to meet these goals were:

• The tester must reproduce the fluidization segregation mechanism accurately while using less than 50 g of material per test. The final version has a test volume of 20 mL, which corresponds to 6–10 g of material with a density of 0.3–0.5 g/cc.

• Reproducibility must be attained through automated, user-friendly software controls to set airflow rates and to monitor pressure changes. These must allow the user to reach and not to overshoot flow rates for complete fluidization.

• The controls must allow the conditions to be repeated, thereby minimizing operator bias. External test influences (e.g., humidity, temperature, static, vibration) should also be controlled easily by the operator and be reproducible.

• The tester must work with both powder and granulation mixtures because both are commonly used in the development of pharmaceutical solid oral-dosage forms. That is, the tester must fluidize and subdivide powder with particle sizes in the range 1–2000 μm and with a wide range of cohesive tendencies.

• Samples removed from the tester at the end of a test must be sized to be approximately a unit dose (~100–1000 mg) needed for analysis thus avoiding subsampling and rifling.

• The tester must contain powder to minimize operator exposure to hazardous materials and avoid the need for the enclosure of the test instrument inside a secondary containment chamber.

• The test must be visible to the user, thereby allowing correlation between fluidization behavior, gas flow, and pressure, and observed physical changes.

Figure 2

Instrument design. The FMSST basic design elements are shown in Figure 2, and a picture of the actual test system (including the airflow controller, computer, and FMSST) is shown in Figure 3. The primary elements of the FMSST include:

• a base to support a sample carousel and test column;

• a fluidization test column bored out of acrylic;

• a means to introduce air into the fluidization test column;

• an expansion chamber and high-efficiency particulate air (HEPA) filter assembly to retain elutriated particles;

• a fill-chamber assembly to introduce the sample to the fluidization test column;

• a sampling disk assembly to recover the samples from the test column;

• a sample vial carousel to receive collected samples.

Figure 3

The base is made from aluminum (subsequently anodized), which was chosen for its low cost, low weight, and mechanical stability.

The fluidization test column is 16 mm in diameter and is filled to a height of approximately 95 mm. Acrylic was chosen to allow for viewing the material in the column to observe the fluidization behavior.

Air is introduced into the bottom of the fluidization test column through a 5-μm sintered metal diffuser. The diffuser provides the distribution of the air across the column.

The fill chamber is constructed from an acetal resin (Delrin, DuPont, Wilmington, DE) and can be removed for filling with test material in a powder containment hood. One end of the fill chamber is sealed with a ball valve, and the other end is closed off with an acrylic vial fitted with an o-ring seal. The acrylic vial typically is used to introduce the powder into the fill chamber. The swivel-top rotation places the fill chamber above the fluidization test column, and the test powder falls into the column when the ball valve is opened. Once the test powder has been added to the fluidization test column, the swivel-top rotation places the expansion chamber–HEPA filter assembly above the fluidization test column. The placement of the HEPA filter on top of the expansion chamber reduces operator exposure to the sample powder during the test.

Test procedure. A test is run by flowing air through the bottom of the fluidization test column. A computer controls the airflow rate and test profile (discussed later in this article) with a custom software program and records airflow rates and column pressure drops during the test. The test method has two parts: an initial test of the background pressure drop through the assembled tester without material and the fluidization of the material in the tester. The pressure drop of the empty tester is subtracted from the gross pressure reading during the test to provide the net pressure drop through the material of interest.

At the conclusion of the test, the unit-dose samples are removed one at a time from the bottom of the column by manipulating the sampling disk. The movement of the handle shifts the disk from the bottom of the column to the top of a glass sample vial and the sample falls into the vial. The disk is moved back to the bottom of the column to receive the next sample from the bottom of the powder column. The sample-vial carousel also is indexed one position to place a clean, empty vial in place to receive the next sample. The tester is sized to hold 19 mL (9.5 g at 0.5 g/cc) of powder–granule mixture that will give 15–17 ~1.2 mL samples (~590 mg at 0.5 g/cc) that need not be further subdivided before analysis by HPLC assay, dry-powder laser-diffraction particle-size measurement, or any other suitable analysis methods. The sampling-disk dimensions can be modified to yield smaller or larger sample sizes. The sample-vial carousel has a cover that shields the operator from powder exposure. When all samples have been retrieved, the sample vial carousel is detached from the base and transferred to a powder-containment hood, the cover is removed, and the sample vials are capped. The samples then can be analyzed by the preferred method.

Figure 4

An example of a segregated material within the test column before sample retrieval is shown in Figure 4.

Method development. A flow-rate ramp-up method based on a sawtooth profile was initially conceived as a means to precondition the material before conducting a test through the ASTM standard practice. Ultimately, the sawtooth profile alone was used as the fluidization profile for the segregation test. The sawtooth profile systematically increases the flow rate in a controlled manner until the pressure drop in the fluidization column indicates that powder fluidization has been achieved. This profile allows the evaluation of the fluidization behavior and decision-making between cycles, thereby ensuring the test sample has been sufficiently fluidized but not overfluidized. The ramp profile makes it easier to fluidize materials that are difficult to fluidize and reduces the likelihood of gas channeling.

Figure 5

The FMSST sawtooth profile is shown in Figure 5. For each step, the flow rate is increased from zero to the target value over a period of 10 s, held at the target value for 5 s, and ramped down to zero over a period of 10 s. A time delay of 5 s occurs between each step to allow powder settling before the next step. Simultaneous monitoring of flow rate and net pressure drop gives an actual profile for each step (see Figure 6). Fluidization is achieved when the slope of the net pressure drop versus flow rate curve becomes horizontal. The first stage of the test is considered complete when complete fluidization has been achieved with three consecutive steps in the sawtooth sequence. At the end of the last step, the time for the flow rate to decrease to zero is increased to 60 s to slow the settling rate, simulating what may occur in a larger-scale (e.g., manufacturing) process step. Shorthand designation for a test is X*Y%, A–B–C, D in which X is the number of steps, Y is the percent increase between steps, A is the ramp-up time, B is the hold time, C is the step ramp downtime between steps, and D is the ramp downtime after the final step. Operation of the method with various powder mixtures is detailed in the Results and Discussion sections.

Table I: Formulation 1.

Tests were carried out with formulations to demonstrate the viability of the FMSST instrument with the sawtooth protocol, the reproducibility of the tester, and the sensitivity of the sawtooth profile to the percent increase between steps and the magnitude of the airflow in the last step of the sawtooth profile. Viability of the FMSST with the sawtooth profile was demonstrated by a direct comparison between the FST and the FMSST instruments (using the ASTM protocol with both instruments) and by comparing the ASTM and sawtooth profiles on the FMSST instrument. The reproducibility of the FMSST–sawtooth profile was examined by replicate testing different formulations. The tester's sensitivity to changes in the sawtooth profile was examined by multiple tests of a difficult-to-fluidize formulation when the step size and maximum airflow were sequentially increased from run to run.

Table II: Formulation 2.

Materials. Five formulations were chosen for use in the method-development studies (Tables I–V). Their selection was based on known or suspected fluidization-segregation behavior according to previous manufacturing history, particle-size differences, API cohesivity, or all three factors:

• Formulation 1 is a placebo blend with a large amount of sucrose that can readily separate from other components primarily on the basis of particle-size differences.

• Formulation 2 is a dry blend that has the potential to segregate in a tablet-manufacturing process because of the API's small particle size and cohesivity.

• Formulation 3 is a wet-granulated product with a wide particle-size distribution and a nonuniform distribution of API among the granule particle-size range.

• Formulation 4 contains a small particle, free-flowing API that might be expected to segregate. Five different lots of this formulation were used in testing for which different API lots were used in each of the formulations.

• Formulation 5 is very cohesive and may experience significant flow problems during manufacturing. It is also challenging to fluidize.

Table III: Formulation 3.

Sample analysis. Samples from the FMSST were analyzed for API assay of the active ingredient using high-performance liquid chromatography or mixture particle size using a dry-powder laser-diffraction method. For each FMSST experiment, the entire set of samples was used for either assay or particle-size analysis. Samples taken from a run were not further subdivided, but the samples were used in either the assay or particle-size analysis. In examples for which both assay and particle size results are reported, two separate experimental runs were carried out: one to generate assay samples and one to generate particle-size samples.

Table IV: Formulation 4.

For all FMSST experiments, the sample number increases with the sample's vertical displacement from the bottom of the tester (e.g., Sample 1 is from the bottom of the tester column and is the first removed, and the highest sample number is at the top of the tester column and is the last sample removed).

Table V: Formulation 5.

Samples from the FST were analyzed for mixture particle size using a dry-powder laser-diffraction method. For each FST experiment, a large (15–20 g) subsample was collected from the top, middle, and bottom sections of the tester. These subsamples were further subdivided using a spin-riffling technique to produce subgram samples, and these multiple samples collected as a result of riffling were used for particle-size analysis.

Figure 6

Particle-size measurements were carried out with a laser-diffraction sensor system (HELOS, Sympatec, Clausthal-Zellerfeld, Germany) equipped with a dry-powder feeder system (VIBRI, RODOS, Sympatec) or a laser-diffraction particle-size analyzer (Malvern 2000 MasterSizer, Malvern Instruments, Worcestershire, UK) with a dry-powder dispersion module (Scirocco, Malvern Instruments). The entire sample was used for each measurement. Measurement conditions for the Sympatec laser-diffraction sensor system were:

VIBRI feedrate = 100%

primary pressure = 3 bar

secondary pressure = 86 mbar.

The dispersion pressure for the Malvern system was 0.5 bar. Data analysis was conducted with the vendors' proprietary software using the Fraunhofer or Mie model, and reported values are median particle size as calculated by the software.

HPLC methods for API assay previously were developed by others and were modified as needed for the FMSST sample size.

Figure 7

Results

Method development. Comparison of FST and FMSST with ASTM test profile. Comparison of the FMSST with the FST using the ASTM fluidization test profile (see Figure 7) was run with Formulation 1 (see Table I). Particle-size measurement (Malvern 2000) was used as the discerning sample analysis test. The ASTM test profile determined for this mixture in the FMSST was:

ramp time to high rate = 2 s

high rate = 32% (of 2000 standard cm³ /min maximum flow rate)

high rate hold time = 3 s

ramp time to low rate = 1 s

low rate = 14%

low ramp hold time = 120 s

ramp time to zero flow = 30 s.

The FST has a larger cross-sectional area, so the flow rates were increased proportionally. Sixteen samples were removed from the FMSST and analyzed for particle size (see Figure 8, "FMSST original sample").

Figure 8

A clear segregation trend was observed. The bottom four samples had a large particle size (d50 = ca. 400 μm) with a slight and continuous size decrease as sample number increases to 4. A sharp slope decrease between Samples 5 and 8 shows a significant decrease in d50 from ca. 400 μm to less than 50 μm. This does not change further with samples from higher positions in the tester.

As per ASTM protocol, three FST samples (top, middle, and bottom) were analyzed from the test following a riffling step to subdivide the FST samples into analyzable subsamples. As a rough approximation of the FMSST sampling profile, the bottom FST sample is represented in Figure 8 as equivalent to FMSST Samples 1–5, the middle FST sample as equivalent to FMSST Samples 6–10, and the top FST sample as equivalent to FMSST Samples 11–16. The bars representing the FST data line up well with the FMSST curve, thus indicating, to a first approximation, the two test methods give equivalent results. The FMSST analysis yielded an increased resolution of the segregation trend compared with the FST by sampling at regular intervals along the length of the test column rather than taking a composite sample from the bottom, middle, and top thirds.

Comparison of ASTM and sawtooth profile with FMSST. Two experiments were conducted with the FMSST instrument to determine if the sawtooth profile, which is capable of fluidizing a wider range of materials than the ASTM profile, is sufficient to also segregate the sample. Initially, the sawtooth profile was conceived as a method to precondition the material before the ASTM profile. Nonetheless, it was later deduced that the sawtooth profile alone could be used as the segregation test method. Formulation 1 was run through an 8*10%, 10–5–10, 10 sawtooth sequence followed by the ASTM profile (Figure 7). In a second experiment, Formulation 1 was run through a 8*10%, 10–5–10, 60 sawtooth sequence only. Particle-size plots of the 16 samples retrieved from the two tests (see Figure 9) have the same trend except for a slight difference in the middle Samples 6, 7, and 8. This difference is not considered significant within the intent of the test because both experiments show the same tendency towards fluidization segregation. This trend is the same as the initial experiment described previously, in which the ASTM profile was used with the FMSST instrument.

Figure 9

Effect of sample order addition. This experiment tested whether the powder-loading method is important. A new sample of Formulation 1 was run on the FMSST using the ASTM parameters described previously (see Figure 7). The 16 samples were removed and added back to the tester one at a time in reverse order of their removal, thereby creating a column with fine particles at the bottom and coarse particles at the top. The test was run again with the ASTM profile; 16 samples were extracted and analyzed for particle size (see Figure 8, "FMSST reverse loaded"). The curve shape is similar to that of the original sample. Exceptions are a slight shift in the break point toward smaller particle size and the point where the curve reaches the minimum particle size. The slight difference between runs is attributed to differences among the starting samples. The similarity of the curves indicates powder loading is not critical to the test (i.e., one does not have to ensure they have a homogenous mixture before running the test). It also confirms the sample has been fluidized, because without the entrainment of the fine particles in an airstream, they would have not moved to the top of the column.

Method reproducibility. The reproducibility of the sawtooth method was evaluated by replicate testing Formulations 2, 3, 4, and 5 (Tables II–V). The X*Y%, 10–5–10, 60 sawtooth pattern was used for each formulation with X and Y dependent upon powder mixture fluidization behavior.

Six test runs were carried out with Formulation 2: three of the sample sets were analyzed for particle size and three were assayed for wt/wt% API (see Figure 10). The API in these three formulations is significantly smaller than the excipients and this formulation can potentially segregate during large-scale manufacturing operations. If the fluidization-segregation mechanism occurs, the particle size should decrease and the assay should increase from the bottom to the top of the column (i.e., sample number increases). The three particle-size curves show a continuous particle-size decrease as the sample number increases. Except for an apparent outlier (Sample 13, which was traced to an analytical artifact) in curve 2–1, all three curves are very similar. The three assay curves show a continuous increase as sample number increases and they are indistinguishable from each other.

Figure 10

Six Formulation 3 samples were tested to demonstrate the reproducibility of the sawtooth method. Three of the sample sets were analyzed for particle size, and three were assayed for wt/wt% API. The particle-size and assay data show the same trend: an increase in value from Sample 1 that reaches a plateau between Samples 3 and 5 and is followed by a steady decrease with some leveling out at the tail end (see Figure 11). The three assay curves can be superimposed, thereby indicating the reproducibility of the method: minor differences can be attributed to slight variance among the three samples. The same behavior is seen among the three particle-size curves. The steady decrease in assay and particle size from the plateau in the sample 3–5 region confirms that fluidization segregation occurs with this sample. As indicated in the Materials section, this mixture contains granules from a wet-granulation process in which the API is mainly associated with the larger granules.

The lower assay and particle-size values for the bottom three Formulation 3 samples were unexpected. Dahl and Hrenya also observed this trend in their discrete-particle simulations of gas–solid fluid beds (11). They attributed this trend to stagnant areas at the bottom of the fluid bed resulted in incomplete fluidization of the mixture and therefore incomplete segregation in this area of the tester. This phenomena probably also occurred with Formulation 3. Dahl and Hrenya also showed this effect is dependent upon characteristics of a mixture's particle-size distribution which could explain why it occurs with Formulation 3 but is not observed with all mixtures used in this study. The effect is not fully understood at present and warrants further study.

Figure 11

Five variations of Formulation 4 (Samples A–E) were tested, for which the differences were the API lot and particle-size distribution. The median particle-size range for the five samples was 182–195 μm. In all cases, the API particle size was considerably smaller than that of the major excipients. In this case, the data indicate that the API particle-size differences did not have a significant effect on fluidization segregation because all five particle curves can be superimposed to show minor differences (see Figure 12). The particle-size trend for samples removed from the bottom of the tester is similar to that of Formulation 3, for which the sample particle size increases from Sample 1 to Sample 3. This effect is followed by a sharp decrease in particle size to Sample 8, after which no further sample particle change is seen. This plateau after the pronounced particle-size decrease was also seen with Formulation 1 (see Figures 8 and 9). The particle-size data clearly show a fluidization segregation trend with the larger particles at the bottom of the tester and small particles at the top half of the tester.

Figure 12

Sample E was run a second time for an API assay test. The assay trend mirrored the particle-size trend where the assay was low at the bottom of the tester before undergoing a pronounced increase between Samples 7 and 9, after which no significant increase occurred. This result was consistent with the particle-size results because the API is significantly smaller than the excipients and would be expected to be found at the top of the test column after the occurrence of fluidization segregation. Note that the assay data does not show evidence of stagnation at the bottom of the tester, which could be caused by the small fraction of large particles that contain API.

As a final investigation of the test method consistency, two different lots of Formulation 5 that were manufactured at different facilities (using an identical process with slight differences in manufacturing equipment) were compared. A single test was run for each lot and the X, Y sawtooth pattern settings were the same (7 and 8%, respectively). The samples were analyzed for particle size only. Both lots show the same fluidization segregation trend: particle stagnation at the bottom of the tester (indicated by an increase in particle size from Sample 1 to Sample 3), followed by a steady particle-size decrease that levels out by Sample 10 (see Figure 13). There are particle-size differences between the two samples during the particle-size decrease portion of the curve. The differences probably are caused by particle-size distribution variations between the two samples, although the test may also distinguish differences arising because of variations in manufacturing sites. From a practical perspective, this difference is not considered significant because both lots met product-release specifications.

Figure 13

Method robustness. The sawtooth profile's robustness was evaluated by testing the sensitivity of a segregation profile to the size of the sawtooth ramp increase and the final airflow rate. Formulation 3, a difficult-to-fluidize material, was chosen for this test as a worst-case scenario. Five tests were run. For each test, the number of steps remained constant (X = 7) while the percent change between steps (Y value) increased from test to test (Y = 2, 3, 5, 7, and 9%, respectively for the five tests). The flow rates at the final step for these tests are 14, 21, 35, 49, and 63%, respectively. The X and Y values for the previous test on this material (see Figure 11) were 7 steps and 8%, which results in a flow rate at the final step of 56%. Thus, the range for this study represented over- and underfluidized conditions.

Figure 14 shows particle-size trends for each test. At the two lowest peak flow rates, 14 and 21%, no noticeable segregation occurs because the particle size versus sample curves are relatively flat with an oscillation that is probably indicative of test, sampling, and measurement variations. At the higher flow rates (Tests 3, 4, and 5), a clear segregation trend occurs with larger-sized particles in the bottom samples followed by a steady decrease in size as sampling proceeds from bottom to top of the tester. As with previous tests on Formulation 3 (see Figure 11), particle size increases from the bottom sample (Sample 1) to Sample 4, and then steadily decreases. For Test 3, the measured particle size for the first three samples was significantly less than that of Samples 4 and 5. Although the overall similarity of the plots for Tests 3, 4, and 5 shows that slight under- or overfluidization is not a large concern, the differences in the first four points indicates that subtle variations can be missed if underfluidization is too great.

Figure 14

Discussion

Pharmaceutical powder mixtures tested in the FMSST tester with either the ASTM or the sawtooth test profile showed the same segregation tendency when compared with testing in the FST with the ASTM test profile. A material that underwent fluidization segregation was identified by the particle size or assay sample profile. In the particle-size profile of a segregated material, particle size would decrease as the sample number increased (i.e., as the sample location moved from the bottom to the top of the tester, where the smaller particles were contained in the upper regions of the test column). In the powder mixtures used in this study, the API had a smaller particle size than the majority of the other components, as is typical with most pharmaceutical powder mixtures. This results in a higher concentration of API in the upper regions of the test column so the API assay should increase as sample number increases. This difference was seen clearly with several samples. The particle-size and assay sample patterns are not easily predictable because some samples showed gradual changes while others changed more abruptly. The relative differences between bottom and top samples tend to be greater in the particle-size profiles than the assay profiles. A particle-size span is generally greater than actual assay differences, which would be expected to be significantly less than the theoretical maximum of 100%. The correlation of observed particle size and assay changes to expected performance in a unit operation has yet to be determined and remains an open area of research.

The effectiveness of the sawtooth profile in fluidizing a mixture in the FMSST was demonstrated in the experiment in which a segregated sample was reverse loaded in the tester before a test run. The resulting test repeated the initial segregation test, which conclusively showed the sample was completely fluidized and mixed during the test.

Reproducibility of the FMSST was shown by replicate testing Formulations 2 and 3 with both particle-size analysis and API assay as the discerning tests. In both cases, the particle-size and assay results showed the same segregation trend and provided a confirmation of the conclusions' validity.

Two other reproducibility tests also gave a qualitative indication of method robustness toward slight, but expected changes in component particle sizes. In the first, five different lots of Formulation 4, for which API particle varied slightly in each, were tested. The test samples were analyzed for particle size. The results showed test reproducibility robustness toward slight changes in component particle sizes. An additional run of one Formulation 4 preparations was analyzed for API assay, and the results confirmed the conclusions reached from the particle-size analysis. The other reproducibility test compared the two lots of Formulation 5 prepared at different manufacturing facilities. The same fluidization-segregation trend occurred for both formulations despite differences in material sources, process equipment, and manufacturing location. (The test also may be detecting subtle differences between sites.) Again, this indicated the test's repeatability and robustness.

FYI

The sawtooth method requires an initial estimate of the airflow rate needed to fluidize the sample. Although models are available to provide this estimate and databases of similar materials that can be used to provide an initial estimate, they only give rough approximations, and different values will be found during an actual experiment. Because this leads to the possibility of slight under- or overfluidization, a test was run with a difficult-to-fluidize material to determine the effect this can have on fluidization segregation test results. The result of this study with Formulation 3 clearly showed that slight under- or overfluidization will not affect the conclusions drawn from the test.

Conclusions

This study demonstrated the development of a fluidization material-sparing segregation tester (FMSST) and an accompanying test procedure that uses less than 25 mL of material per test. The sawtooth test profile used with the FMSST was shown to be comparable with the FST instrument with the ASTM profile for testing fluidization segregation potential. Nonetheless, the FMSST, while using less material, yields additional data on segregation and fluidization behavior that the FST–ASTM profile does not provide. The sawtooth profile is preferred because it provides a more-controlled approach toward the fluidized state, while making it easier to fluidize difficult-to-fluidize materials.

Note

A patent application has been filed that covers various aspects of the tester design and test method.

Acknowledgements

Angela Kong and Padma Narayan are thanked for carrying out and analyzing the FST tests. Nathan Krakue's tireless efforts to modify existing HPLC assay methods for the FMSST sample sizes and to carry out the FMSST assay tests were greatly appreciated. Craig Bentham provided us with the Formulation 4 samples and arranged for API assay testing. Tom Troxel and Tom Baxter provided valuable guidance during the design of the tester, the design of the experiments and, the interpretation of experimental results. Kris Dermody is thanked for providing the two lots of Formulation 5 from Pfizer manufacturing facilities.

David B. Hedden was an associate reseach fellow at Pfizer (Ann Arbor, MI) and is currently the director of pharmaceutical sciences at deCODE chemistry (Woodridge, IL). Dean L. Brone was a senior principal scientist at Pfizer Inc. (Ann Arbor, MI) and is currently the associate director of technical services with Sepracor (Marlborough MA). Scott Clement is a senior project engineer at Jenike & Johanson (San Luis Obispo, CA). Mike McCall is a project engineer at Jenike & Johanson, Inc. (San Luis Obispo, CA). Angela Olsofsky was an associate scientist at Pfizer Inc. (Ann Arbor, MI) and is currently located in Thousand Oaks, CA. Phenil J. Patel is a senior principal scientist at Pfizer Inc. (Ann Arbor, MI). James Prescott* is a senior consultant at Jenike & Johanson, Inc., 400 Business Park Dr., Tyngsborough, MA 01879, tel. 978.649. 3300, fax 978.649.3399, mail08@jenike.com Bruno C. Hancock is a research fellow at Pfizer Inc. (Groton, CT).

*To whom all correspondence should be addressed.

Submitted: June 14, 2006. Accepted: Aug. 31, 2006.

Keywords: powder, granulation, segregation, fluidization, tester.

References

1. Z.T. Chowan, "Segregation of Particulate Solids, Part I," Pharm. Technol. 19 (5), 56–70 (1995).

2. J.K. Prescott and R.J. Hossfeld, "Maintaining Product Uniformity and Uninterrupted Flow to Direct-Compression Tableting Presses," Pharm. Technol. 18 (6), 99–114 (1994).

3. J.K. Prescott, S.A. Clement, and J.W. Carson, "Fluidization Segregation Tester," US Patent No. 6,487,921, Oct. 26, 2000.

4. "Standard Practice for Measuring Fluidization Segregation Tendencies of Powders," D6941-03 (2003), ASTM International.

5. M. Wormsbecker et al., "Segregation by Size Difference in a Conical Fluidized Bed of Pharmaceutical Granules," Powder Technol. 153 (2), 72–80 (2005).

6. E.D. Liss et al., "Segregation of Powders during Gravity Flow through Vertical Pipes," Pharm. Technol. 28 (2), 78–96 (2004).

7. "How the Johanson Indices Predict Segregation Effects," Binside Scoop 2, (1994).

8. J. Johanson, "Predicting Segregation of Bimodal Particle Mixtures Using the Flow Properties of Bulk Solids," Pharm. Technol. 20 (5), 46–57 (1996).

9. S.Y. Wu and J. Baeyens, "Segregation by Size Difference in Gas Fluidized Beds," Powder Technol. 98 (2), 139–150 (1998).

10. N. Abatzglou and J.-S. Simard, "Prediction of Segregation Tendency Occurrence in Dry Particulate Pharmaceutical Mixtures: Development of a Mathematical Tool Adapted for Granular Systems Application," Pharm. Devel. and Technol. 10 (1), 59–70 (2005).

11. S.R. Dahl and C.M. Hrenya, "Size Segregation in Gas–Solid Fluidized Beds with Continuous Size Distributions," Chem. Eng. Sci. 60, 6658–6673 (2005).

12. S. Cooper and C.J. Coronella, "CFD Simulations of Particle Mixing in a Binary Fluidized Bed," Powder Technol. 153 (1), 27–36 (2005).