Granule Size Distribution: Measurement, Specifications and Process Control

Granule Size Distribution is the relative amount of granules within defined size ranges in a specified sample, expressed on a stated measurement basis. This is an operational definition: it describes how a sample is apportioned across sizes, not a legal definition or an acceptance limit.

Identify the tested material as drug substance, intermediate granulate or finished granule product, and name the manufacturing stage. For example, label an intermediate result as before or after milling after granulation, rather than simply “particle size.” Keep that identity attached to comparisons and any particle size gate.

Use this comparison to keep neighbouring measurement questions separate; the D-value conventions are defined once in the methods section below.

Distinguish interval percentages within individual ranges from cumulative percentages below or above a boundary, and label the cumulative direction; the full reporting context is listed in the methods section.

Why the result matters is product-specific. In general terms, the fines fraction and the coarse tail affect how granules flow and pack into a die, how evenly they fill by volume, and how they consolidate under compression; a distribution that shifts toward fines often accompanies flow and weight-variation problems, while an excess of coarse granules can affect content uniformity in small tablets. These are reasons to monitor the distribution, not demonstrated relationships for any particular product; ICH Q6A asks for development evidence before a size test becomes a specification.

Treat a granule-size result as a candidate explanatory variable, not a diagnosis. A proposed product-specific investigation should ask three separate questions: do distribution changes accompany feeding or flow variation, compression-performance changes, or variation in the composition of sampled material? Evaluate fines, coarse fractions and shifts in the central portion separately.

For the composition question, design the comparison around stratified blend sampling. Keep variation between sampled portions distinct from the finished-unit question addressed by content uniformity. Record the identity and timing of each paired sample so that size and downstream results describe comparable material.

A development hypothesis proposes a relationship worth testing. A demonstrated relationship needs repeatable, product-specific evidence under defined conditions, with competing explanations examined. An approved acceptance criterion is a separately authorised decision boundary—not a correlation promoted into a limit. Do not choose “narrower” or “coarser” as the improvement target in advance; define the downstream response sought and test whether the proposed size change supports it.

Use the following as a proposed method-documentation checklist, not a regulatory requirement. Before comparing methods, define the intended comparison: agreement in selected size fractions, percentile estimates, or sensitivity to process changes. Set comparison criteria beforehand; do not assume that matching instrument labels or nominal size ranges establishes interchangeable results.

A proposed sieve-analysis record

Record initial sample mass in grams, sieve-opening identifiers in micrometres, net retained mass for every interval, net material mass in the pan, test conditions and total recovered mass. For this proposed calculation, recovered mass equals all retained masses plus pan material mass; interval percentage equals 100 × interval mass / recovered mass. Report recovery separately as 100 × recovered mass / initial sample mass. Keep the denominator visible in the sieve analysis record, rather than reporting an unexplained percentage.

A proposed laser-diffraction record

Laser diffraction works differently from sieving. The instrument passes a dispersed sample through a light beam, records the angular pattern of scattered light, and converts that pattern into a size distribution through an optical model; the output is usually volume-based and expressed as equivalent spherical diameters. Two consequences follow for granules. The result depends on the optical model and dispersion conditions chosen, so those settings belong with every result; and because irregular, porous granules are reported as equivalent spheres, a laser-diffraction D50 and a sieve-based median for the same lot are not expected to match numerically.

Record dispersion conditions (dry or wet, dispersion pressure or medium), instrument method and version, optical assumptions, reporting basis, reported size range and the full distribution, not only summary D-values. A change in dispersion energy can break agglomerates or even fracture friable granules, so the same lot can produce a finer result under harsher dispersion; treat that as a method question before reading it as a process change.

For the proposed cumulative-undersize convention, D10, D50 and D90 are the sizes at or below which 10%, 50% and 90% of the stated distribution basis lie. Label each set mass-based, volume-based or number-based, retain units, and document interpolation and any range truncation. Preserve these labels in LIMS, not only on the instrument printout.

Finally, state whether the target is granules in their sampled state or material after a defined preparation treatment. Treat that choice as part of the method-comparison design, alongside sampling and instrument settings.

A conditional recommendation for new drug substances

ICH Q6A 3.3.1(b) is guidance addressing particle size for new drug substances. Its stated material and product scope matters when applying it to a granule-size specification:

"For some new drug substances intended for use in solid or suspension drug products, particle size can have a significant effect on dissolution rates, bioavailability, and / or stability."

The conditional recommendation in that same paragraph is: "In such instances, testing for particle size distribution should be carried out using an appropriate procedure, and acceptance criteria should be provided."

The paragraph then states: "Decision tree #3 provides additional guidance on when particle size testing should be considered." Consult Decision tree #3 in the guideline for its decision sequence; the paragraph's reference is not a substitute for working through that sequence. See also the broader ICH Q6 specification framework.

Use development evidence to justify the control

ICH Q6A 2.4 says experience and data accumulated during development should form the basis for setting specifications. Its particle-size bullet relates test placement to product performance: testing "may be performed as an in-process test, or may be performed as a release test, depending on its relevance to product performance." The specification rationale should therefore explain the performance relevance, rather than merely identify a convenient sampling stage.

Proposed specification record: keep three limit types distinct

As a practical design proposal, keep the result’s measurement context (defined in the methods section) together with three distinct limit types:

• Development ranges, labelled as observations or study ranges.
• Operating alerts, labelled as prompts for investigation or process attention.
• Acceptance criteria, labelled as approved decision rules with their justification.

Start with the material’s regulatory role and intended market, not the word “granule.” Use the following scope checks to keep specification guidance, legal requirements and change assessments separate.

A separate drug-product context appears in ICH Q6A 3.3.2.1: its introductory paragraph identifies tests applicable to tablets, coated and uncoated, and hard capsules. It then says one or more of those tests may also apply to soft capsules and granules. Keep that dosage-form discussion separate from the earlier particle-size specification question.

As an applicability-check practice, record the product category, jurisdiction and reason for using each reference. The 21 CFR 211 overview can orient the U.S. assessment; base legal conclusions on the current regulation.

The sequence below is a proposed operating example, not an ICH decision tree or a list of statutory record requirements. Assign an owner to each handoff so the result remains connected to the material and decision it supports.

1. Process owner → sampler: Begin with the intended decision, such as assessing readiness for the next operation. Identify the material, batch and process stage. Define the sampling point, sample identifier and relationship to the batch before selecting the test execution.
2. Sampler → laboratory receiving owner: Record collection time, collector identity, sample handling and storage conditions, and any observations during collection. At the sample login handoff, reconcile the container identifier with the request and acknowledge receipt against the same batch reference.
3. Laboratory receiving owner → analyst: Assign the test and record sample preparation, method identifier and version, instrument or sieve identifiers, and analyst identity. Preserve recorded observations separately from subsequent calculations; do not replace an observation with an interpreted result.
4. Analyst → technical reviewer: Declare size units, measurement units and reporting basis. Apply the method’s corrections or conversions, calculate the reported fractions or distribution statistics, and apply its rounding convention. Preserve inputs and calculation steps. Compare the result with the identified, applicable specification version—not an unversioned spreadsheet limit.
5. Technical reviewer → authorised disposition owner: Review traceability, execution, calculations and the specification comparison. Record the reviewer’s assessment separately from both observations and calculated results. Route exceptions to authorised people for evaluation and disposition; a passing calculation alone is not a release decision.

The disposition owner records the decision and communicates it to the next operation’s owner. Link the sample, result, review and disposition within the batch execution history, preserving who accepted each handoff and when.

These hypothetical examples separate the calculation from the acceptance decision. All limits are invented teaching limits, not regulatory requirements or recommended specifications.

Case one: calculate interval mass fractions

Assume a 200.0 g sample, complete recovery and a mass-based distribution. Assign retained masses of 12.0 g, 148.0 g and 40.0 g respectively to three non-overlapping intervals: less than 125 µm, 125 µm to less than 500 µm, and at least 500 µm. Material at exactly 125 µm belongs in the middle interval; material at exactly 500 µm belongs in the coarse interval.

Calculate each percentage as interval mass ÷ 200.0 g × 100. Apply hypothetical criteria of no more than 8.0% below 125 µm and no more than 15.0% at or above 500 µm.

The recovered mass equals the starting mass, making the example's mass balance explicit; see in-process reconciliation. The fine-fraction criterion is met. The coarse fraction exceeds its criterion by 20.0 − 15.0 = 5.0 percentage points. This conclusion addresses only the two stated criteria.

Case two: the same median, different tails

Assume two hypothetical distributions use the same method, reporting basis and reporting convention. Both have a reported D50 of 300 µm. This shared median is stipulated, not inferred from the broad intervals below. Batch A reports 5% below 125 µm and 10% at or above 500 µm; Batch B reports 15% and 25%, respectively.

Batch A meets both hypothetical tail limits; Batch B meets neither. Their identical reported median does not answer this acceptance question because it does not specify either tail percentage. The middle fractions follow by subtraction, but do not locate the median within that interval.

Case three: from interval fractions to cumulative undersize

Reuse the case one fractions: 6.0% below 125 µm, 74.0% from 125 µm to below 500 µm and 20.0% at or above 500 µm. Cumulative undersize adds intervals from the fine end: 6.0% below 125 µm, then 6.0 + 74.0 = 80.0% below 500 µm, and 100.0% once the top interval is included. Cumulative oversize runs the other way: 20.0% at or above 500 µm and 94.0% at or above 125 µm. The two curves carry the same information, so a report must say which one it shows; a limit written as “not more than 20% above 500 µm” is an oversize statement and cannot be read off an undersize curve without conversion.

With only three intervals, D-values cannot be read precisely: D50 lies somewhere inside the 125–500 µm interval, and any single number would depend on an interpolation assumption that should be stated. That is one practical reason to choose sieve sets with enough intervals around the sizes the specification cares about.

Neither example establishes flow, compression or dissolution performance, or a product-release decision. Those conclusions are outside these calculations.

Use a proposed before-and-after comparison within change control, rather than treating a matching median as evidence that a manufacturing change is acceptable. This is a technical comparison framework, not a regulatory classification.

• Align material identity, sampling stage, method version and reporting basis before comparing distribution results. Link those results to the downstream responses already identified in development.
• Record changed operating conditions, including previous and proposed settings with units. Explain why the comparison batches were chosen and how they represent the conditions being assessed.
• Use a batch comparison overlay to examine the reported distribution, including its tails, alongside the selected downstream responses—not just the median.

Separate the process-change comparison from any measurement-method change. If both change together, identify which results share a common method and reporting basis and remain directly interpretable. Treat cross-method differences as unresolved until investigated; where feasible, propose testing representative retained samples by both methods to help distinguish measurement effects from process effects.

Where a post-approval change to an immediate-release product is involved, check the current FDA SUPAC-IR guidance separately. Keep the granule-size comparison distinct from regulatory handling: this article assigns no change category, testing package or filing requirement.

The reviewer must resolve three questions: Did the result change? Is the comparison valid? What does the product-specific evidence mean?

Use a proposed, vendor-neutral data model for laboratory-to-manufacturing integration: link the sample and batch identifiers to the method version, source run identifier, size-fraction records, applicable specification version and human review record. This is a design proposal, not a software capability or regulatory requirement.

Carry the reporting conventions from the methods section with every transferred distribution, and keep original results distinct from recalculated or corrected versions, each linked to its predecessor and correction reason.

Use these proposed troubleshooting checks to frame review, not as a regulatory investigation checklist. Acceptance thresholds and investigation decisions remain procedure-specific.

Link decisions to the batch execution history; use the applicable out-of-spec handling procedure when relevant.

These questions test the control strategy as a whole; the execution workflow above covers the routine run.

• Does each limit trace to development evidence or a documented rationale?
• Can a reviewer reproduce every reported fraction from recorded masses or the instrument file?
• Is a method change separable from a process change in the trend record?
• Is the disposition owner named for out-of-limit results?

These ownership assignments are proposed responsibilities, not legally prescribed job titles.

For limited-data situations, ICH Q6A 2.5 advises that “the initially approved tests and acceptance criteria should be reviewed as more information is collected, with a view towards possible modification.”

Set this data-accumulation review timing within the product-specific procedure. Treat a changed trend as a reason to assess the evidence—not as permission to revise limits. Route any proposed revision through change control, with the rationale and approval decision recorded.