Milling After Granulation

Granulation — whether wet (high-shear or fluid-bed) or dry (roller compaction) — never produces a particle size distribution (PSD) that is ready for direct compression. Wet high-shear granulation produces a mixture of oversize agglomerates, on-size granules and fines that needs wet-milling before drying (to reduce drying time and improve uniformity) or dry-milling after drying (the more common option in modern production). Fluid-bed granulation produces a tighter PSD but still benefits from dry-milling to break the occasional oversize. Roller compaction produces ribbons that are useless without a paired mill — the mill is what converts ribbon into a flowing granule.

The mill therefore sits between two CQAs: the granulation step delivers an input PSD that the mill has to convert into an output PSD targeted at the compression operation. The compression operation has a sweet spot — typically D10 ≥ 50 µm (avoid sticky fines), D50 in a defined band (predict die fill), D90 ≤ 800 µm (avoid feeder cheating and weight variation), fines fraction < 75 µm under 20 % (avoid capping and lamination). The mill is the last opportunity to shape PSD before compression; if the mill is wrong, the press is wrong, and the only recovery is to mill again.

• Oscillating granulator: gentle, narrow PSD output, lower throughput — historical default for wet milling.
• Cone mill (Comil): medium shear, very broad applicability, integrated screen + impeller — the modern workhorse for both wet and dry milling.
• Hammer mill: high shear, high throughput, risk of fines and heating — used where throughput dominates or where the granule is unusually tough.
• Screen aperture sets the coarse cut of the output PSD; impeller speed sets the fines fraction.
• Screen type (round, grater, square) shapes the particle morphology and the fines tail.

Of these, screen aperture and impeller speed do most of the work. A 1.0 mm screen at 1,500 rpm and a 1.5 mm screen at 1,000 rpm can give very different PSDs from the same input even though the operator regards them as 'similar settings'. PPQ has to lock both — and the type of screen, not just aperture, because a 1.0 mm grater and a 1.0 mm round-hole are not interchangeable.

Screen wear is the silent CPP. A round-hole screen run at high impeller speed loses material slowly off the upstream face of each aperture; over a few campaigns the effective aperture grows by 5–15 % and the output PSD shifts coarser. By the time the mill output looks 'wrong' on the trend chart, the screen is months past its real life. The defence is to track mill cycles per screen, not calendar time, and to PM by validated cycles.

Wet milling (post-granulation, pre-drying) is rarely needed in modern production but still common in legacy processes. The wet granule is sticky; the mill has to break agglomerates without smearing them into a paste. Settings: large aperture (typically 1.5–2.5 mm), slow impeller (300–800 rpm), round impeller arm, generally a single pass.

Dry milling (post-drying, pre-blend or pre-compression) is the standard. The granule is brittle; the mill has to size-reduce by breakage at the weak points of the granule. Settings: smaller aperture (typically 0.6–1.2 mm), faster impeller (1,000–2,500 rpm), knife or square-arm impeller, possibly two passes for tough material. A common 483 finding is a recipe that documents only 'milling step' without separating wet and dry passes, with corresponding lack of distinct parameter sets and PSD evidence for each.

1. Pre-mill inspection — screen for cracks, dents, missing apertures; impeller for wear and balance; feed throat for residual material. Photograph the screen face as evidence.
2. Confirm screen and impeller match the recipe — interlock the equipment ID with the recipe phase if possible.
3. Verify mill cycle counter; if PM is due, do not start.
4. Start the mill empty, run at recipe speed for 30 seconds to confirm no abnormal vibration or noise.
5. Begin feed at the validated throughput; monitor motor current — high current indicates blinding or overload.
6. Take post-mill PSD sample per the validated sampling plan (location, count, timing).
7. Run PSD by sieve or laser diffraction against the spec; release-or-rework decision is at the mill, not deferred to compression.
8. On completion, clean and inspect screen; if any aperture damage is found, record and quarantine the affected output for assessment.

Throughput is the under-monitored CPP. A cone mill rated at 200 kg/h, fed at 350 kg/h to catch up on a schedule, generates more heat, smears the granule against the screen, and produces a coarser fines tail than the validated condition. Modern mills capture motor current as a live signal; a step-change in current correlates with throughput excursions and is a useful CPV trend.

The mill output PSD propagates into three downstream CQAs in predictable ways:

• Blend uniformity (BUE) — fines fraction sets segregation tendency. Above ~25 % under 75 µm, the blend de-mixes during transfer and BUE fails the second sampling pull.
• Compression force at target weight — D50 sets die-fill behaviour. A 50 µm shift in D50 changes die fill by 1–2 % at constant turret speed, which the press feeds back as a force variation.
• Tablet hardness and friability — fines fraction sets the air content of the die at compression; higher fines = more air entrapment = higher capping and lamination risk.

The implication is that the mill's output PSD must be specified, tested and trended every batch, not just 'checked occasionally'. The supplier-quality argument that 'the press is fine, so the mill is fine' is wrong in both directions: the press can be marginal because the mill is on the edge of spec, and the press can pass on day 1 of a campaign while screen wear is silently moving the mill toward failure by day 30.

1. Skipping pre-mill screen inspection — a torn screen lets oversize through and produces a bimodal output PSD that passes blend uniformity but fails compression force.
2. Using one screen size for two products with different input granulation PSDs — without PSD bridging data, the regulator treats the second product as untransferred.
3. Hammer mill on a heat-sensitive product without temperature monitoring — degrades the API silently and gives a passing PSD with a failing assay.
4. Re-using the wet-mill setting for dry milling (or vice versa) — wrong breakage mechanism, wrong PSD.
5. No post-mill PSD sample — assuming the screen choice is sufficient evidence of output PSD. PSD is the CQA; the screen is a CPP.
6. Screen replacement tracked by calendar, not by validated cycle count — silent drift across long campaigns.
7. Mill PSD trended only on the COA, not on a CPV dashboard — the trend toward spec is invisible until a batch fails.
8. Mill installed without an interlocked recipe — operator picks the screen out of a drawer, no equipment-ID-to-recipe match recorded.
9. Two passes through the mill not documented as separate phases — record shows 'milled' once when in fact the granule was milled twice with different settings.
10. Cleaning verification overlooked at screen and impeller — cross-contamination from previous product carries forward.

Cone-mill scale-up is one of the better-behaved unit operations. Tip speed (the linear velocity of the impeller arm tip past the screen) is the key invariant — a small mill at 2,500 rpm with a 100 mm impeller has a tip speed of 13 m/s; a production mill with a 250 mm impeller hits the same tip speed at 1,000 rpm. Maintain tip speed and the breakage regime is largely preserved. Throughput per unit screen area is the second invariant; scaling load proportionally to screen area keeps residence-time distribution similar.

Tech-transfer documentation should capture: screen aperture, screen type, impeller type and tip speed (not rpm), throughput per unit screen area, validated PSD acceptance window with PPQ evidence, and the screen-PM cycle-count basis. Hammer mills are harder to scale because tip speeds reach 80–100 m/s and the breakage regime is closer to impact than to attrition; transfer data are best validated at the receiving site rather than assumed from a different mill geometry.

• Solid-dose pharma — Comil is the dominant choice for both wet and dry milling; hammer mill reserved for tough granulations.
• Roller compaction — paired mill (often integrated into the compactor frame) is a system component, not a separate operation.
• Nutraceutical — same mill technology with additional constraints from palatability (avoid heat-driven flavour shifts).
• Veterinary medicated articles — milling preserves medicated active integrity; cold-mill setups common.
• Agrochemical (water-dispersible granules) — analogous mills shape granule for dispersion behaviour rather than tabletting.
• Detergent and household-product granules — similar equipment, very different spec drivers (dissolution rate, dust generation).

• Equipment-ID-to-recipe match enforced — wrong screen or wrong impeller blocks the phase.
• Photographic pre-mill screen inspection captured into the batch record.
• Cycle counter on every screen and impeller; PM driven by validated cycles, not calendar.
• Wet-mill and dry-mill are distinct recipe phases with distinct parameter sets.
• Throughput limit per validated cycle enforced via feeder integration; overload alarms.
• Post-mill PSD result posts against the spec; OOS opens the standard investigation with full equipment, screen and operator traceability.
• PSD trend per product per mill on a CPV dashboard with control limits derived from PPQ data.