Postdoctoral PDDM991 ⭐ 4 Credits Semester 4

Practicum I: Executed Formulation, Process Sensitivity, Pilot Transfer, and Filling

Hands-on practicum that turns the Semester 1–3 design work into executed, reviewed evidence. Skipper makes, measures, transfers, and fills real placebo semisolids. The work covers at least three base families: an o/w cream, a benzene-free carbomer gel, and a petrolatum ointment. Every batch is labeled as a noncommercial training/engineering batch, not for human use, and is destroyed under documented disposition. The course is API-agnostic. No commercial or Rx active is used. Homogeneity and uniformity are tracked with HPLC-measurable markers that are present anyway (preservatives such as parabens or phenoxyethanol, and antioxidants such as BHT) or with a justified low-hazard tracer. Each choice of marker or tracer must be documented in a justification memo covering: - permitted possession under the site's current and planned permits (a Florida OTC drug manufacturer permit holder may not possess Rx drugs); - handling safety; - explicit extrapolation limits, because placebo rheology, microstructure, and hold behavior do not predict an API-loaded product. The work runs in four blocks: 1. **Measurement readiness.** Pick and verify the physical test systems before any sensitivity batch. The lab owns no viscometer, rheometer, microscope, pH meter, pycnometer, KF titrator, or plate-count micro capability. Each test is therefore either acquired as a justified minimal bench kit or placed at a contract physical-testing lab, decided on evidence: turnaround versus how quickly samples change after manufacture, cost per data point, and data-integrity access. 2. **Bench batches and sensitivity runs.** Bench batches are executed on a jacketed-beaker, overhead-stirrer, and rotor-stator kit. Process-sensitivity runs deliberately vary cooling rate, homogenizer speed and time, order and temperature of phase or ingredient addition, carbomer hydration and neutralization, and ointment congealing conditions. Responses are apparent viscosity/rheology, microscopy (globules, crystals, grit, air), pH, specific gravity, appearance, and marker content by the existing Agilent 1260 HPLC-DAD. 3. **Pilot/CMO transfer.** Skipper co-authors the protocol and observes in person. The exercise quantifies heat-transfer differences, scraped-wall mixing, dead spots, bulk hold, process losses, yield reconciliation, and beginning/middle/end (B/M/E) sampling. 4. **Filling and sealing trial.** On a CMO tube or jar line, the trial covers check-weight capability against USP <755> and the NIST Handbook 133 package-checking method, seal and crimp defects, lot/expiry coding, line clearance and label reconciliation under 21 CFR 211.130, and documented disposition of rejects. All records are contemporaneous executed batch records under 21 CFR 211.188, with real deviations investigated. Larissa (Principal Chemist) and an independent quality reviewer each review raw data and sign off separately. Simulated, modeled, or literature-only data are not accepted as readiness evidence. This practicum establishes demonstrated competence. It does not qualify a process, validate a method, or authorize commercial manufacture.

📅 Studied by Skipper on 2026-10-16

📖 Core Concepts

Claude Opus 5.5

Core Concepts: Practicum I: Executed Formulation, Process Sensitivity, Pilot Transfer, and Filling

Introduction: What This Practicum Is Really About

Skipper, every course before this one asked you to know something. This one asks you to show it. The deliverable is a chain of evidence: executed batch records, raw instrument files, audit trails, micrographs, temperature logs and signed reviews. Taken together, these let an independent person conclude that you can make, measure, transfer and fill a semisolid competently. The scientific content is real: emulsions, polymer gels, crystalline hydrocarbon networks, heat transfer and statistics. But the governing question differs from that of a lecture course. It is not "what is true about semisolids?" It is "what can my data actually support, and what can it not?"

Three ideas carry most of the weight:

  1. A semisolid's quality attributes are a record of its process history. A tablet's composition largely fixes its chemistry. A cream's viscosity, globule size, texture and stability are written into its microstructure by temperatures, shear and timing. Two batches with identical ingredient lists can be different products. That is why [Chang et al., 2013a] treat process parameters as first-class determinants of topical product quality, and why the Q3 characterization concept exists at all.
  2. You can only see effects larger than your noise. A process-sensitivity study is a signal-detection problem. Its resolving power is set by measurement precision (Week 3) and batch-to-batch variability (center replicates, Week 5). If an effect is smaller than that resolution, the effect is unresolved. It is not shown to be absent.
  3. Evidence has a scope, and claiming beyond it is a quality failure, not a stylistic one. Placebo batches made by you on a bench kit, then observed at a CMO, are competence evidence. They are not process validation under 21 CFR 211.100(a). They are not process performance qualification. They do not authorize a launch. Every report you write in this course must say so explicitly.

A useful analogy: this practicum is a flight-simulator check ride with a real aircraft at the end. The simulator part is the bench placebos, where you learn how the machine responds to the controls, find the stall edges and log every input. The real-aircraft part is the CMO pilot batch, where someone else is pilot-in-command (their quality unit and batch record govern) and you are the trained observer who owns the instrumentation data. Passing the check ride does not certify any particular passenger flight. The first real dermatologist formula is a new aircraft type, and parts of the check ride must be repeated.

The three placebo families are o/w cream, benzene-free carbomer gel and petrolatum ointment. They were chosen because each exercises a different structure-forming mechanism:

  • Emulsification plus a lamellar gel network.
  • Polyelectrolyte microgel swelling.
  • Crystallization of a wax network in oil.

A real future formula will be one of these, a hybrid, or a lotion variant. The Buhse decision-tree classification (cream, ointment, gel, lotion, distinguished by water/volatiles content, hydrocarbon/wax/polyol content, and viscosity) tells you which family's lessons apply [Buhse et al., 2005]. Verify the exact numerical cut-offs in the paper before using them in a classification memo.


Concept 1: The Claim–Evidence Ladder and the Scope Statement

Every document in this practicum sits on a rung of a ladder. Being explicit about the rung prevents the most dangerous error in the whole program, which is letting a training dataset be cited later as if it were validation.

RungWhat it claimsWhat evidence it needsDoes PDDM991 reach it?
Demonstrated competenceThe operator can execute, measure, document and investigate correctlyExecuted records, raw data, independent reviewYes: this is the target
Process understanding (placebo)Directional sensitivities of placebo bases to CPPsDesigned experiments with noise estimatesPartially, within stated resolution
Process design for a productNOR/PAR for the actual formulaStudies on the real composition with real APINo: only a template
Process qualification / validationThe commercial process reproducibly yields conforming productQualified equipment, approved protocol, commercial-scale batches of real productNo
Launch authorizationProduct may be released and distributedQuality-unit release under an approved systemNo

The extrapolation-limits statement turns this ladder into specific, falsifiable caveats. A real active can change which parameters are critical:

  • Dissolved versus suspended API. A suspended API introduces crystal growth, Ostwald ripening and polymorphism. These make cooling rate and addition temperature critical in ways a placebo cannot reveal.
  • Phase location of the API. If the API is oil-soluble, it alters the melting behavior of the oil phase.
  • Active loading. At high loading, a suspended API changes rheology outright.
  • Addition stage. Adding the API to the hot phase rather than during cooling changes its thermal exposure and its distribution.
A placebo tells you how the vehicle responds to the process. It cannot tell you how the drug responds to the vehicle-plus-process. Write every conclusion so that this boundary is visible.

Noncommercial controls are part of the evidence, not paperwork overhead. These controls make the "competence only" claim physically true, not merely asserted:

  • "Training/Engineering Batch – Not for Human Use – Not for Distribution" labeling.
  • Segregated quarantine.
  • An inventory log.
  • Witnessed destruction with a record.

A reconciliation in which units made = units sampled + units rejected + units destroyed (± a documented, explained tolerance) is how you prove no training material escaped.


Concept 2: Markers, Tracers, and the Difference Between Total and Free Concentration

A placebo has no API to assay, so how do you measure homogeneity? You use something already in the formula that your Agilent 1260 HPLC-DAD can quantify. Good candidates are preservatives and antioxidants with strong UV chromophores:

  • Methylparaben and propylparaben.
  • Phenoxyethanol.
  • Benzyl alcohol.
  • Sorbic acid/sorbate.
  • BHT.

These are intrinsic markers. They are possession-permitted, low hazard, and present for a formulation reason, so they do not distort the vehicle.

If you add an extrinsic tracer instead, the memo must establish three things:

  1. It is non-Rx.
  2. It is low hazard.
  3. It is possession-permitted under the current and planned Florida ch. 499 permits.

The program premise is that an OTC drug manufacturer permit holder may not possess Rx drugs. Verify the specific permit category and possession restrictions against current Florida Statutes ch. 499 and DBPR rules before finalizing the memo. An Rx steroid used as a "convenient tracer" could create a permit violation out of a training exercise.

The partitioning trap

Here is the scientifically interesting part. In an o/w cream, a preservative distributes among four places:

  • The aqueous phase.
  • The oil phase.
  • The interfacial film.
  • Surfactant micelles.

Your HPLC assay dissolves everything, so it reports the total concentration. Antimicrobial activity depends largely on the free aqueous concentration. For a simple two-phase picture with oil–water partition coefficient K = Co/Cw:

Mass balance:   C_T = φ_w · C_w + φ_o · C_o        (φ = phase volume fractions)
With C_o = K · C_w:
                C_w = C_T / (φ_w + K · φ_o)

Add micellar binding (bound/free ratio R in aqueous phase):
                C_w,free = C_T / [φ_w (1 + R) + K · φ_o]

Worked example. A cream contains 20% oil phase (φo = 0.20, φw = 0.80) and total propylparaben 0.10% w/w (treat as ≈ w/v for illustration).

  • If K were 1, Cw = 0.10/(0.80 + 0.20) = 0.10%.
  • If K were 50 (plausible order for a lipophilic paraben in some oils; actual K depends strongly on the oil, so measure or cite for your base), Cw = 0.10/(0.80 + 10) ≈ 0.0093%.

That is a roughly tenfold drop in free preservative with an identical total assay. Now suppose a process change, such as a different homogenizer time, shifts the interfacial area or the amount of surfactant forming micelles. Free concentration then changes while your HPLC "marker distribution" result looks perfect.

This is why the Week 5 topic says that partitioning "changes free aqueous concentration, not total assay." Your homogeneity data speak to distribution. They say nothing about preservative efficacy, which is a USP <51>-type question. Your BacT/ALERT cannot answer that question; it goes to a contract lab in PDDM992. Do not let a good top/middle/bottom result be written up as evidence of preservation.


Concept 3: Microstructure Is Process Memory (Three Mechanisms)

This is the heart of process sensitivity. Each family builds structure by a different physical mechanism, and each mechanism has its own vulnerable window. Know the mechanism and you can predict which CPPs matter. Ignore it and you are running blind factorials.

3a. O/W cream: emulsification plus the lamellar gel network

A classic cream's body comes mainly from a lamellar gel network, not from the oil droplets. Fatty amphiphiles such as cetyl or cetostearyl alcohol combine with a hydrophilic surfactant and water to form swollen bilayer sheets, held in a hexagonal "α-crystalline gel" state. These sheets trap bulk water between them [Aulton & Taylor, 2018].

The network forms on cooling, as the fatty alcohol passes through its transition region. That is why cooling rate and mixing during cooling are CPPs. Fast cooling with gentle sweep mixing tends to trap more of the α-gel network.

The surprising part is that the network is metastable. Over hours to weeks the α-gel can convert to more ordered, less-swollen crystalline forms (β/γ). That conversion releases water and causes:

  • Viscosity drift (up or down).
  • Graininess, visible under polarized light as birefringent crystals.
  • In extreme cases, bleeding.

This is exactly why Week 3 insists on fixed conditioning times (e.g., 24 h and 7 d reads; verify the times for your base). A viscosity number without its age is not a datum.

Emulsification variables operate on a different structural level: the droplets.

  • Phase temperatures. Both phases must be fully melted and held within ~70–75 °C (verify for the chosen base). If the phases are at very different temperatures, waxes chill-crystallize at the interface.
  • Order and rate of addition. Adding oil into water versus water into oil can route the system through phase inversion. Sometimes that is deliberate and gives finer droplets; sometimes it is uncontrolled and gives coarse droplets or a w/o product.
  • Homogenization intensity. More energy gives smaller droplets, with diminishing returns. Beyond a point, over-homogenization breaks down developing gel network structure and incorporates air.

3b. Carbomer gel: polyelectrolyte microgel swelling

Carbomers are crosslinked poly(acrylic acid) microgels. Dispersed in water they are acidic (pH ~3) and their chains are tightly coiled, so the dispersion has low viscosity. As you neutralize, the –COOH groups become –COO⁻. Electrostatic repulsion, plus the osmotic pressure of the counter-ions, swells each microgel many-fold. The swollen microgels pack into a jammed space-filling system with a yield stress [Sheskey et al., 2020].

[Islam et al., 2004] showed that the rheological properties of neutralized carbomer gels change relatively little over roughly pH 5–8. At very low pH the chains are under-ionized. At high pH or high ionic strength, excess counter-ions screen the charges and the gel thins. Your Week 6 job is to verify this plateau for your grade, not to assume it.

Worked calculation: how much base? Carbomer is roughly 56–68% carboxylic acid groups by weight. Check the monograph and HPE entry for your grade.

Repeat unit of acrylic acid (–CH2–CH(COOH)–):  M ≈ 72 g/mol
1.0 g carbomer (~ pure poly(acrylic acid)) → ≈ 1.0 / 72 ≈ 13.9 mmol COOH
Full neutralization with NaOH (40 g/mol):     13.9 mmol × 40 = 0.56 g NaOH
Practical target (pH ~6–7) is partial neutralization: on the order of 0.4 g NaOH per g carbomer
  → for a 0.5% w/w gel, 1 kg batch: 5 g carbomer → ~2 g NaOH
     (e.g., ~11 g of an 18% w/w NaOH solution), titrated to pH, not dosed blindly.

The lesson: titrate to pH and record the base consumed. A shift in base demand from batch to batch is itself a signal. It can indicate a different carbomer lot, incomplete hydration, or CO2-loaded base.

Fisheyes are a diffusion problem. When a clump of dry powder hits water, its outer layer hydrates into a viscous gel skin. Water then diffuses through that skin very slowly, leaving a dry core that may persist for hours. Controlled sprinkling into a vortex, or eductor addition, separates particles before they wet. Hydration time before neutralization is a CPP because incompletely swollen particles neutralize unevenly [Chang et al., 2013a]. That paper appeared online in late 2012, which is why some course materials cite it as 2012.

Shear sensitivity. Neutralized carbomer microgels can be mechanically degraded by sustained high shear, giving a partly irreversible viscosity loss. High shear also whips air into a fluid that is about to acquire a yield stress. That leads to the next and most counterintuitive point.

Why you must deaerate before the gel sets: a yield-stress argument

In a Newtonian liquid a bubble always rises, at a speed given by Stokes' law:

v = Δρ · g · d² / (18 μ)

Un-neutralized dispersion, μ ≈ 0.1 Pa·s, d = 1 mm bubble:
v = 1000 × 9.81 × (1×10⁻³)² / (18 × 0.1) ≈ 5.4 × 10⁻³ m/s   → rises 10 cm in ~20 s

In a yield-stress fluid a bubble may never rise. Buoyancy must overcome the yield stress. A common dimensionless criterion compares the yield stress with the buoyancy stress:

Y = τ_y / (Δρ · g · d)

Neutralized gel, τ_y ≈ 20 Pa, d = 1 mm:
Δρ g d = 1000 × 9.81 × 0.001 ≈ 9.8 Pa   → Y ≈ 2
Literature thresholds for bubble entrapment are on the order of Y ≈ 0.05–0.2 (geometry-dependent; verify).
Y ≈ 2 ≫ threshold → the bubble is permanently trapped.

The design consequence follows directly. Air incorporated after neutralization cannot escape by standing. Vacuum helps only by expanding the bubbles until buoyancy wins or the bubbles coalesce, and that requires low-speed mixing to bring them to the surface. So the process should either:

  • deaerate the low-viscosity dispersion before neutralization and neutralize under low shear or vacuum, or
  • accept that the post-neutralization specific gravity is what you will get.

The same physics explains why air trapped in a cream during cooling is so stubborn.

3c. Petrolatum ointment: a crystalline wax network in oil

Petrolatum is a colloidal network of solid hydrocarbon crystals (paraffinic and microcrystalline waxes) entrapping liquid hydrocarbons. On fusion you melt the network. On congealing you rebuild it, and the rebuilt network depends on how you cool and stir [Khar et al., 2013]:

  • Cooling rate. Fast cooling nucleates many small crystals. Slow cooling grows fewer, larger crystals that can feel grainy. Shock cooling of some wax blends gives a stiff-then-softening product.
  • Mixing intensity during congealing. Shearing a network as it forms fragments it. The result is a softer, sometimes smoother product with more mobile liquid oil, which raises the risk of oil bleed (syneresis) on storage.
  • Powder incorporation. Insoluble powders must be wetted by the continuous phase to break agglomerates. Levigation (trituration with a small amount of mineral oil to a smooth paste before dilution) works by creating high local shear stress in a thin paste and by displacing air from particle surfaces. Direct addition into bulk molten petrolatum typically leaves agglomerates, which microscopy will show.

Week 7 asks you to compare executed behavior against your PDDM903 predictions. That comparison is a test of whether your mechanistic model was any good, and a disagreement is an informative result.


Concept 4: Measurement Before Sensitivity: Precision, Noise, and Minimum Detectable Effect

Week 3 comes before Week 4 for a reason. You cannot interpret a process effect until you know what difference your measurement system and your batch-making can resolve.

The variance budget

The observed variability of a batch-level response, averaged over r replicate measurements, is:

σ²_total = σ²_batch + σ²_meas / r

σ_meas  : from the Week 3 precision study on one homogeneous reference base
          (repeatability + intermediate precision, as in PDQC802)
σ_batch : from replicate center-point batches in Week 5

Rheological responses are positive and multiplicatively variable, so analyze ln(viscosity). Differences on the log scale then become ratios, and the CVs combine approximately in quadrature.

Effect estimates in a two-level design

In a 2k factorial with N corner runs, an effect is the difference between the mean response at the high setting and the mean at the low setting:

Effect_A = ȳ(A+) − ȳ(A−)
SE(Effect) = 2 σ_total / √N

Minimum detectable effect (two-sided α, power 1−β):
MDE ≈ (t_{α/2, df} + t_{β, df}) · SE(Effect)

Worked example (and the uncomfortable truth). Suppose:

  • Viscosity repeatability CV = 3% and r = 3 reads, giving a measurement contribution of 3%/√3 ≈ 1.7%.
  • Batch-to-batch CV = 6%.
  • Then σtotal ≈ √(36 + 3) ≈ 6.2%.

Run a 22 design (N = 4 corners) plus 3 center batches:

  • SE(Effect) = 2 × 6.2% / √4 = 6.2%.
  • Pure-error degrees of freedom from 3 centers: df = 2.
  • t0.025,2 = 4.30 and t0.20,2 ≈ 1.06.
MDE ≈ (4.30 + 1.06) × 6.2% ≈ 33%  (on the log scale ≈ a 1.4-fold change)

Read that again. With a realistic practicum-sized design you can reliably detect only large effects, about a one-third change in viscosity. The killer is the 2 degrees of freedom. Three ways to improve resolution:

  1. More center replicates. With 5 centers, df = 4 and t0.025,4 = 2.78, which cuts the MDE substantially.
  2. Pool error degrees of freedom from negligible higher-order interactions. Do this only if it was pre-declared, never chosen after looking.
  3. Reduce σbatch through tighter execution. That is a competence outcome in itself.

Increasing r barely helps once σbatch dominates. Here the variance budget tells you where effort pays off.

An effect smaller than the MDE is reported as "not resolved at this design's capability (MDE = X%)", never as "no effect." Week 9 makes this a rule, not a courtesy.

Method details that are actually precision decisions

  • Viscosity per USP <912>. Apparent viscosity is defined only at a stated geometry, speed, temperature and history. In stiff creams and gels a cylindrical spindle bores a hole and then measures the air in that hole (channeling). A T-bar spindle on a helipath stand avoids this by sweeping the T-bar down a helix through fresh material. A small-sample adapter gives a defined shear-rate geometry. Fix all of the following, because each is a variance source:
    • Fill geometry.
    • Equilibration time at temperature (e.g., 25.0 ± 0.5 °C, verify).
    • Spindle and speed.
    • Read time.
    • Time since manufacture.
  • pH per USP <791>. Lipids coat the glass membrane and slow its response. Choose direct measurement with a spear electrode or a fixed dilution scheme, and document which. Dilution changes ionic strength and activity, so a diluted-sample pH is a procedure-defined value, not the "true" pH. Calibration buffers must bracket the expected range. Electrode cleaning between samples (e.g., with a surfactant or solvent rinse) is part of the method.
  • Specific gravity as an air gauge. If ρ0 is the air-free density (from a well-deaerated or centrifuged reference) and ρ is measured:
    Air volume fraction  φ_air = 1 − ρ / ρ_0
    
    Example: ρ_0 = 1.000 g/mL, ρ = 0.962 g/mL → φ_air = 3.8% v/v
    
    This one number connects to several downstream effects. It changes apparent viscosity, it causes fill-weight drift on a volumetric piston filler (see Concept 8), and it is listed as a Q3 characterization item. The syllabus cites it as item 9 of the March 2026 final guidance; verify the numbering against the issued text.
  • Microscopy. Bright field shows globules and air bubbles: air appears as perfectly round, thick-black-rimmed spheres. Polarized light reveals birefringent material: crystalline wax, fatty-alcohol crystals in an aging cream, or undissolved solids. Fixed magnification, a calibrated graticule and a defined counting rule (e.g., count globules in N random fields; report the fraction above X µm) turn pictures into data.

Own versus contract: the time-constant argument

The decision matrix for viscometer, microscope, pH meter and pycnometer has an elegant core criterion: does the sample change faster than the contract turnaround? A cream whose structure evolves over its first 24–72 h cannot meaningfully be shipped out for a "24 h" viscosity read. Transit time, temperature excursion and vibration all rewrite the microstructure you meant to measure.

When the attribute's time constant is shorter than the logistics delay, in-house ownership is often a scientific necessity, not just an economic preference. Cost per data point, raw-data and audit-trail access (PDQC814), and reuse in PDDM992 and future release testing then decide the margin cases. pH meters and pycnometers are cheap and fast-changing-sample-sensitive. Rheometers are expensive, but rotational viscometry for IPCs is not.


Concept 5: Rheology as a Fingerprint, Not a Number

Semisolids are non-Newtonian, so "the viscosity" does not exist. A rotational viscometer at one speed gives one point on a flow curve. Two products can share that point and behave completely differently in a tube, on the skin, or in a filler. The working models:

Power law (shear-thinning, n < 1):     η_app = K · γ̇^(n − 1)
Herschel–Bulkley (yield stress):        τ = τ_y + K · γ̇^n      for τ > τ_y
Casson (alternative):                   √τ = √τ_y + √(η_∞ · γ̇)

Why these matter for process sensitivity:

  • Yield stress τy (USP <1912>) governs whether suspended particles settle, whether bubbles escape (Concept 3b), and whether the product slumps after extrusion. A process change can alter τy while leaving the single-point apparent viscosity nearly unchanged.
  • Flow index n captures how much the product thins under shear. That shear occurs during pumping, filling and spreading.
  • Thixotropy is time-dependent structural breakdown and recovery, seen as hysteresis between up- and down-ramps or as recovery after a step. It tells you whether a high-shear process step causes recoverable or permanent damage. Carbomer microgel breakage is largely permanent; a lamellar network disrupted by shear may partially rebuild.

Without a rheometer you can still approximate a flow curve by running the viscometer at several speeds in sequence (up and then down). Record this as a multi-point apparent-viscosity profile per <912>. Do not label it as a true rheogram, because the shear rate is not uniform in most spindle geometries.


Concept 6: Scale Is a Heat-Transfer and Mixing Problem

Transfer from bench to pilot fails most often in two places. Heat leaves the product more slowly, and mixing in a viscous product is laminar. Both follow from geometry.

Heat transfer: surface-to-volume falls with scale

Heat flow:              Q = U · A · ΔT
Lumped cooling (Newton): T(t) = T_jacket + (T_0 − T_jacket) · e^(−t/τ)
Time constant:           τ = m · c_p / (U · A)

Geometric similarity:    V ∝ L³,  A ∝ L²  ⇒  A/V ∝ 1/L  ⇒  τ ∝ L  (at constant U)

Worked example.

  • Bench (1 kg). Wetted jacket area A ≈ 0.05 m², U ≈ 150 W/m²·K (scraped, moderately viscous), cp ≈ 3.5 kJ/kg·K.
    • τ = (1 × 3500) / (150 × 0.05) ≈ 470 s ≈ 8 min.
  • Pilot (100 kg, geometrically similar). A ≈ 0.05 × 1002/3 ≈ 1.08 m².
    • At the same U: τ ≈ 350,000 / (150 × 1.08) ≈ 2,160 s ≈ 36 min, which is 4.6× slower. 4.6 is 1001/3.
    • With better scraping, say U ≈ 300 W/m²·K: τ ≈ 18 min, still more than 2× slower.

Suppose the bench study showed that cooling through the structure-forming window (say 55 → 40 °C for the fatty-alcohol network; verify for your base) must take no longer than 20 minutes. Copying the bench jacket setpoint at pilot scale will then violate that window.

The transferable parameter is the product temperature–time profile through the critical window, not the jacket setpoint. To match the product curve, the pilot may need:

  • a colder jacket or a larger ΔT,
  • staged cooling, or
  • accepting a longer time, if Week 9 showed the window is non-critical.

Real curves are not single exponentials. As the cream thickens, U drops because the boundary layer at the wall stops renewing unless it is scraped. That is why recording instruments on both jacket and product are mandated [FDA, 1994].

There is also a radial temperature gradient. Product at the wall can be well below the bulk thermocouple reading. Wall-congealed material that is not scraped back into the bulk becomes a texture defect.

Mixing: you are in the laminar regime

Impeller Reynolds number:   Re = ρ · N · D² / μ

Anchor, D = 0.10 m, N = 1 rev/s, μ = 10 Pa·s, ρ = 1000 kg/m³:
Re = 1000 × 1 × 0.01 / 10 = 1     (turbulence begins ~10³–10⁴)

At Re ≈ 1 there are no eddies to do the mixing. Material moves only where the blades or scrapers push it, so geometry is everything:

  • Scraper contact. This keeps the wall film renewing, for heat transfer, and stops it congealing.
  • Counter-rotation. This creates bulk exchange between the wall region and the center.
  • Dead spots and recirculation. Look under the impeller hub, at the bottom valve and at baffles. Material from dead spots that has not been through the process must not be used [FDA, 1994].

The FDA guide also flags hand-spatula scraping as a known cause of non-uniformity. It returns unmixed wall material in an uncontrolled lump.

Homogenizer scale-up: tip speed, gap shear, and turnovers

Rotor tip speed:      v_tip = π · N · D
Nominal gap shear:    γ̇ ≈ v_tip / δ
In-line loop passes:  turnovers = (Q_loop · t) / V_batch

Bench probe: D = 18 mm, 10,000 rpm → v_tip ≈ π × 0.018 × 166.7 ≈ 9.4 m/s
             δ ≈ 0.2 mm → γ̇ ≈ 4.7 × 10⁴ s⁻¹
Pilot in-line: Q_loop = 20 L/min, V = 100 L, t = 10 min → 2 turnovers

The deep point: on the bench, a probe in 1 L means essentially the whole batch passes through the high-shear zone many times. At pilot scale with an in-line or bottom homogenizer, each element of fluid may pass through only a few times. The residence-time distribution matters. Some material may see the rotor once and some not at all.

Matching tip speed alone gives the same peak shear but not the same number of exposures. Matching time alone does neither. That is why Week 10 records "homogenizer loop residence," and why droplet-size distributions rather than mean size reveal scale effects: a coarse tail indicates under-processed fractions.

SUPAC-SS gives a vocabulary for this. It classifies equipment by design and operating principle (e.g., an anchor mixer versus a high-shear mixer) and treats changes within versus across classes differently. Here it is used as a scientific change-control reference only [FDA, 1997]. It does not convert a placebo transfer into a filed change.


Concept 7: Uniformity as a Sampling-and-Extraction Problem

"Is the batch homogeneous?" contains three separate questions. Each has its own failure mode.

  1. Is the sample representative? Kettle top/middle/bottom sampling with an unbiased thief, B/M/E sampling across discharge and fill, and the tank-heel plan all address this. Suspension-type bases may need continuous mixing during holding and filling so that particles do not settle toward the end of the run [FDA, 1994].
  2. Does the sample prep extract the marker completely and reproducibly from a lipid matrix? This is the most underestimated step. For a single equilibrium extraction, with Km/s = Cmatrix/Csolvent:
    Fraction recovered:   f = 1 / (1 + K_m/s · V_matrix / V_solvent)
    
    BHT in petrolatum, K_m/s ≈ 5 (illustrative), 1 g matrix (~1.1 mL), 25 mL methanol:
    f = 1 / (1 + 5 × 1.1/25) ≈ 0.82   → 18% low bias from partition alone
    
    The practical routes, all confirmed by spike recovery into placebo:
    • Heat above the matrix melting point so the phases mix intimately.
    • Use a solvent system that dissolves or swells the matrix (e.g., an initial small volume of THF or hexane, then dilution with mobile-phase-compatible solvent), then precipitate the waxes by cooling and remove them by centrifugation or filtration.
    • Increase the solvent-to-matrix ratio, or use multiple extractions.
    Also check filter adsorption by comparing filtered and centrifuged results, and check solution stability.
  3. Is the identification correct? Identity is confirmed by retention time plus DAD spectrum match, never retention time alone. Lipid matrices produce co-eluting UV-absorbing junk, and peak purity across the peak is the check.

USP <3> Uniformity in Containers for tubes is a distinct test. It asks whether the product is uniform within a filled tube, which catches phase separation during filling, settling in the hopper, or stratification. Cut the tube, sample from defined locations (for example near the crimp, the middle and near the nozzle) and assay each portion.

The syllabus summary for tubes ≥5 g is: stage 1 takes 3 portions from 1 tube with a within-tube maximum difference NMT 10.0%, extending to 3 tubes. Confirm the exact sampling locations, the formula for "difference" (e.g., relative to mean versus to label), the staging rules and any assay-range criteria against the current official <3> text before writing the protocol. The statistic you compute must match the chapter's definition exactly. A range-over-mean calculation and a max-deviation-from-mean calculation can differ by a factor of about two near the limit.

Choice of instrument. LC-MS/MS is justified only when DAD lacks sensitivity or selectivity, for example a tracer at very low level or a heavy co-eluting matrix. Defaulting to the most sensitive instrument adds matrix-effect and ionization-suppression variance for no gain. A method should be fit for purpose, which is the PDQC802 lifecycle principle applied again.

When results are out of expectation, the PDQC891 investigation logic applies. First separate lab error from true non-uniformity:

  • Check extraction recovery, standard preparation and integration.
  • Then use spatial patterns. Low at the top and high at the bottom suggests settling or density stratification. Random scatter suggests prep variability.

Concept 8: Filling: Mass, Volume, Density, and Two Different Regulatory Lenses

A positive-displacement piston filler delivers a volume. Labels and USP <755> speak in mass or labeled quantity. The bridge between them is density, and density is your specific-gravity IPC:

Fill mass per stroke:   m = ρ · V_stroke = ρ_0 (1 − φ_air) · V_stroke

If φ_air drifts from 2% to 6% during a run (air entrained as hopper level drops, or bulk warming):
Δm/m ≈ −(0.06 − 0.02)/(0.98) ≈ −4.1%   → a 30 g tube loses ~1.2 g with no change in filler setting

This is a beautiful example of how a formulation IPC (specific gravity and deaeration endpoint) becomes a packaging-line capability. Week 9's recommended deaeration endpoint directly constrains Week 12's check-weight scatter.

Check-weight logic

  • Net weight = gross weight − tare. Tare variability in tubes is real, from wall thickness, cap and printing. Use either the average of a sampled set of empty containers (with their SD folded into the uncertainty) or individual tare weights for capability studies.
  • Use a calibrated balance with readability appropriate to the tolerance (PDQC805 metrology). Check at start-up and periodically in-run, with predefined adjustment rules. Document every adjustment, because an undocumented adjustment invalidates the capability data.

Lens 1: USP <755> Minimum Fill (compendial)

The legacy criteria in the syllabus are:

  • The mean of 10 containers is ≥ the labeled amount.
  • No individual container is below 90% of label for labels ≤60 g, or below 95% for labels >60–150 g.
  • If the criteria are not met, retest with 20 additional containers.

Confirm against the current official text, including the retest acceptance rules.

Lens 2: NIST Handbook 133 (retail net-quantity enforcement)

This is a statistical lot-acceptance method used by weights-and-measures inspectors. It has two requirements:

  1. Average requirement. The sample average must not be less than the labeled quantity beyond what sampling error can explain (via a sample error limit computed from the sample SD and a sample-size-dependent correction factor).
  2. Individual requirement. No package may fall short by more than the Maximum Allowable Variation (MAV), and the number of packages exceeding the MAV is limited, often to zero for small samples.

Look up the MAV for your labeled quantity in the HB 133 (2025 ed.) tables. Do not assume a percentage.

Target-fill economics

Choose target μ_target = LQ + z · σ_fill  (σ_fill from B/M/E capability data)

Example: LQ = 30.0 g, σ_fill = 0.30 g (incl. density drift)
  z = 3  → μ_target = 30.9 g  (3% overfill)
  If deaeration control halves σ_fill to 0.15 g → μ_target = 30.45 g (1.5% overfill)
Annual impact at 200,000 tubes: 0.45 g × 200,000 = 90 kg product saved

Overfill is the price you pay for variability. Reducing σ is cheaper than raising the mean, and it is a process-control achievement rather than a give-away.

Why the peptide vial filler does not transfer. Vial filling for liquids involves low-viscosity Newtonian fluids, small volumes and aseptic constraints. Tube filling involves viscous, yield-stress, sometimes heated product. Nozzle cut-off and drip or stringing behavior come from the rheology. A heated, agitated hopper may be needed to prevent stratification. The container is open at the crimp end and closed by hot-air or jaw sealing, or by aluminum fold/crimp. Different operating principle, different equipment class.


Concept 9: Comparability, Hold, and Yield: Explain, Don't Average

When pilot results differ from bench results, the temptation is to pool the data and report a mean. The correct move is to compare each attribute against a pre-set acceptance window drawn from the Week 9 ranges and explain any excursion.

This is equivalence thinking from PDQC802. The question is not "is there a statistically significant difference?" With n = 1 pilot batch you cannot even estimate pilot variance. The question is "does the pilot result fall within the window we predefined as practically equivalent?" A pilot viscosity 25% above bench, inside a ±30% window derived from the bench PAR, is comparable. The same value with a coarse globule tail under the microscope is a mixing-scale finding that must be explained in the report.

Bulk hold asks whether the bulk changes between manufacture and filling:

  • Use closed, labeled containers at the defined hold condition.
  • Sample at t = 0 and at the proposed maximum hold for appearance, viscosity, pH, specific gravity and marker content.
  • Send microbial samples to a contract lab under the <61>/<62> plans from PDDM913.

The aging-network physics of Concept 3a means hold time is not neutral. A cream filled at 2 h and one filled at 48 h may be different products in the tube. Remember too that BacT/ALERT is presence/absence only and cannot give the counts that <61> requires.

Yield reconciliation:

% Yield = (actual output mass / theoretical mass) × 100

Accountable losses: kettle heel + homogenizer loop hold-up + transfer line + hopper residue + samples
Unexplained loss  = theoretical − (output + accountable losses)
→ Investigation trigger if unexplained loss exceeds a pre-set limit (e.g., > 2% — set from pilot data)

An unexplained discrepancy is more than an economics problem. It can signal undocumented discards, measurement error, or unrecorded material from dead spots. Under 21 CFR 211.192, unexplained discrepancies must be investigated.


Concept 10: The Execution Record and Independent Review as Scientific Instruments

The regulations named in this course are not overhead. They are what makes your data trustworthy.

  • 21 CFR 211.84(d): component identity. At least one specific identity test per component, plus a documented rationale for relying on the CoA. Two component-specific checks:
    • Carbomer. Under FDA's December 2023 carbomer guidance, confirm a benzene-free grade with a CoA showing benzene ≤2 ppm, and confirm it is not a benzene-manufactured grade.
    • Glycerin and propylene glycol. Under FDA's May 2023 Rev. 1 guidance, every shipment of every lot needs DEG/EG testing (NMT 0.10%). Your lab has no GC, so contract-lab turnaround becomes a batch-start gate. The supply-hazard history behind this, mass poisonings from DEG-contaminated glycerin, is why "it's just a placebo" is no excuse. Even training batches follow the discipline you will use on product.
  • 211.101 charge-in with second-person check, and 211.188 batch records. Weights, IPCs, yields and deviations are recorded contemporaneously. A batch record completed afterward from memory is not raw data.
  • 211.130 line clearance and label reconciliation. Physical separation, a removal checklist, and identification of filled-but-unlabeled units. Label mix-ups are among the most frequent causes of drug recalls, so the procedure exists to make them structurally impossible, not merely unlikely.
  • 211.192 investigations. Deviations are documented and investigated to root cause. Unplanned outcomes such as phase inversion, graininess or separation are observations to record, not batches to discard silently. In a sensitivity study a "failed" batch is often the most informative one, because it marks the edge of the PAR.

Independence of review. Larissa reviews technical validity and the quality reviewer reviews compliance, separately. Each traces at least one result from raw data to report, for example a CDS injection with its audit trail, through the integrated peak, the calculation and the reported uniformity value. The sign-off sequence is fixed in advance. Raw data are retained for review. Simulated or literature data submitted as evidence is grounds for rejection.

This is ALCOA+ (PDQC801/PDQC814) applied to yourself. The practicum's integrity rests on making your own work auditable by someone who does not trust you by default.


Connections to Peptide Chemistry

Formulations are unknown and the program is API-agnostic, but your peptide background is not wasted. Several peptide-science instincts transfer directly, and some topical scenarios will make them central.

  • Interfaces denature. From PDQC815 and BCHM603 you know that air–water and oil–water interfaces unfold and aggregate peptides and proteins. A rotor-stator homogenizer creates enormous interfacial area and air entrainment. If a dermatologist ever specifies a peptide-containing topical, homogenization intensity and addition stage stop being placebo curiosities and become stability-critical CPPs. Add peptides after homogenization, during cooling, under low shear.
  • Carbomer is a polyanion. Cationic peptides (rich in Arg/Lys) can form electrostatic complexes with poly(acrylate). That can cause turbidity, viscosity collapse and reduced free peptide. This is the partitioning trap of Concept 2 in another form: a total assay that looks fine hides a bound/free split. Your BCHM612 drug-delivery knowledge of polyelectrolyte complexation applies directly.
  • Preservative–peptide interactions. In parenteral peptide products, phenolic preservatives (phenol, m-cresol) interact with protein structure; insulin hexamer stabilization is the classic case. Parabens and phenoxyethanol in topicals may likewise interact with peptide actives or with surfactant micelles. Treat preservative choice as an API-compatibility question at PDDM925 intake.
  • Content accounting. PDQC812 taught you that peptide "assay" requires accounting for counter-ions and water. A peptide API added to a cream by weight needs the same net-peptide correction. The marker strategy here (intrinsic, well-characterized excipients) parallels using a well-characterized reference standard.
  • Your HPLC expertise is the practicum's backbone. The Agilent 1260 DAD methods, peak-purity identity, and lifecycle validation from PDQC802/PDQC803 carry straight over. What is new is the lipid matrix, where sample prep, not chromatography, dominates the error budget.
  • Filling contrasts sharpen understanding. Your peptide vial filler works on volume with near-incompressible Newtonian liquids. Tube filling works on volume with compressible (air-containing), yield-stress materials. Seeing why one does not transfer to the other is the clearest illustration of "equipment class by operating principle."

Common Misconceptions to Watch Out For

  1. "A successful placebo pilot batch validates the process." No. Placebo transfer is competence and process-understanding evidence. Process validation under 211.100(a) requires the real product, an approved protocol and qualified equipment. The pilot transfer report must say so in plain words.
  2. "No significant effect means the parameter is non-critical." Only if the MDE was small enough to matter. With 2 pure-error df, MDEs of 30% or more are typical. Report "unresolved at MDE = X%."
  3. "Viscosity is a property of the product." Apparent viscosity is a property of the product plus spindle, speed, temperature, fill, shear history and age. Without the conditions and time since manufacture, the number is meaningless.
  4. "More homogenization is always better." Droplet size reaches diminishing returns. Excess shear degrades carbomer microgels, disrupts developing gel networks and entrains air. The PAR has an upper bound.
  5. "More base gives a thicker carbomer gel." Viscosity plateaus roughly between pH 5 and 8 [Islam et al., 2004]. Excess base adds ionic strength that screens charges and thins the gel. Titrate to a pH window.
  6. "Vacuum deaeration removes all the air." In a yield-stress gel, small bubbles may be permanently trapped. Verify the endpoint by specific gravity and microscopy, and design the process to deaerate before structure sets.
  7. "A uniform total preservative assay shows the product is preserved." Total ≠ free. Partitioning and micellar binding control efficacy, which needs a <51>-type test that BacT/ALERT cannot perform.
  8. "Copy the bench jacket profile to the pilot kettle." Surface-to-volume falls about as V−1/3. Match the product temperature–time profile through the critical window, using recording thermometers on product and jacket.
  9. "A top-of-kettle grab sample represents the batch." Use a defined top/middle/bottom plan with an unbiased thief. Never mix in hand-scraped wall material or dead-spot material.
  10. "If pilot and bench differ, average them." Compare each attribute against pre-set windows and explain differences mechanistically.
  11. "Use LC-MS/MS because it's more sensitive." Sensitivity you don't need brings matrix effects you don't want. Justify the instrument choice.
  12. "Overfilling solves net-contents risk." It hides variability at a recurring cost. Reduce σfill, often through deaeration and density control, then set the target from capability.
  13. "Ointments have no microbial or cleaning issues." Anhydrous bases carry lower growth risk, though not zero, and component and environmental contamination still matter. They are markedly harder to clean, which feeds PDDM922 swab-recovery planning.
  14. "Retention-time match equals identity." Never on its own. RT plus DAD spectrum, with peak purity, especially in lipid matrices.
  15. "A failed or inverted batch should be discarded and rerun quietly." It is a deviation or observation to document and investigate. It may define your PAR boundary, and hiding it is a data-integrity failure.
The unifying habit: for every number you produce, be able to say how it was made, how precise it is, what it can and cannot support, and where its raw data live. That habit is what this practicum certifies.

Verification flags for this section: exact USP <3>, <755>, <912>, <1912> and <791> text and criteria; the NIST HB 133 (2025) MAV tables and sample-error procedure; Florida ch. 499 permit possession rules; Q3 guidance item numbering; the Buhse et al. numerical classification cut-offs; carbomer acid content and neutralization ratios for the specific grade (HPE 9th ed. monograph); and illustrative partition coefficients and heat-transfer coefficients, which should be replaced with measured or literature values for the actual materials and equipment.

⚙️ Mechanisms & Applications

GPT-6 Astra

Mechanisms & Applications — PDDM991 Practicum I

Skipper’s central insight: a semisolid is not merely a list of ingredients. It is a material with a history. Its performance depends on composition, the structures created during manufacture, and the structures that continue developing during storage. Two batches can have the same chemical assay and still differ in spreadability, physical stability, filling behavior, and drug release.

This practicum turns the predictions from PDDM903 into testable relationships between materials → processing → microstructure → measured attributes. It also tests whether those relationships survive a change of equipment, scale, bulk-hold time, and package. All numerical examples below are illustrative calculations, not executed practicum results. Literature or simulated data cannot substitute for the raw evidence required here.

1. Begin with an evidence boundary, not a recipe

The three training families—an oil-in-water cream, a benzene-free carbomer gel, and a petrolatum ointment—represent three different ways of building a semisolid structure. They are useful precisely because no single mechanism explains all three. Buhse and colleagues emphasized the importance of physical structure in classifying topical products; Chang and colleagues subsequently connected formulation and processing variables to topical-product development and quality by design.[1–3]

  1. Define the question. For example: “Does cooling history measurably change cream structure?” is answerable with a placebo. “Will this process produce acceptable release of an unknown active?” is not.
  2. Choose the analytical witness. Prefer a justified, intrinsic, HPLC-measurable excipient—such as an existing preservative or antioxidant—over an unnecessary added tracer.
  3. Clear the possession and safety gate. Document ingredient permissibility under the actual Florida permit scope, including restrictions on prescription-drug possession. A training label does not override permit requirements.
  4. Control the material. Label all units “Training/Engineering Batch – Not for Human Use – Not for Distribution,” segregate them, track inventory, and document witnessed destruction.
  5. Fix the review chain. Skipper executes and retains raw records; Larissa reviews technically; the independent quality reviewer assesses compliance. At a CMO, its approved procedures and quality unit govern its manufacturing floor.

Extrapolation limit: uniform distribution of a dissolved preservative does not establish uniform distribution of suspended drug particles. Nor does a preservative assay establish preservative efficacy. An active may change pH, interfacial chemistry, crystallization, rheology, microbial protection, and extraction behavior. The eventual formula must therefore reopen the relevant development questions.

This is directly analogous to peptide analytics: a clean identity spectrum establishes identity within the method’s capability, not potency, sterility, stability, and clinical suitability simultaneously.

2. Material receipt: control hazards before they become experimental variables

A process-sensitivity study is uninterpretable if “cooling rate” changes at the same time as polymer grade, water quality, or humectant contamination. Incoming-material controls are therefore part of experimental design, not merely administrative overhead.

  1. Quarantine and identify each component lot. Apply the logic of 21 CFR 211.84(d): perform at least one specific identity test for each component. Reliance on a supplier certificate for other attributes requires an appropriately established supplier-reliability basis; a certificate is not a general substitute for component identity testing.[8]
  2. Match the identity method to the material. HPLC retention behavior plus a DAD spectrum can support identity for suitable UV-absorbing preservatives and antioxidants, provided specificity is demonstrated. Water, petrolatum, glycerin, and polymers require other appropriate tests, often through a qualified contract laboratory.
  3. Check the carbomer supply route. Confirm the intended non-benzene-manufactured grade, its applicable monograph status, and the benzene specification and certificate result. “Benzene-free” is not an analytical claim of absolute zero; the course requires review against a ≤2 ppm criterion and the applicable FDA guidance.[10]
  4. Clear high-risk humectants before use. Arrange DEG/EG testing of glycerin and propylene glycol for each shipment of each lot, with sampling and testing aligned to current FDA guidance. Apply the appropriate limits—generally NMT 0.10% for each relevant impurity—and verify the official material-specific requirements.[11]
  5. Use a controlled water source. Until an appropriate purified-water system exists, use documented compendial Purified Water with defined handling and use-by rules. Conductivity and TOC evidence do not establish microbiological suitability.

Application to Peptide Partners: the owned HPLC-DAD instruments do not eliminate the need for contract DEG/EG testing. No GC capability is currently owned, and neither a triple-quadrupole instrument nor a supplier certificate should be treated as a substitute for a suitable, established method. Contract turnaround becomes a batch-start constraint.

DEG/EG controls also have a sobering historical context: repeated poisonings from contaminated pharmaceutical ingredients show why supply-chain assurances must be connected to analytical verification. In this practicum, the lesson is operational—no cleared result, no charge-in—not merely historical.[11]

3. Oil-in-water cream: create interfaces, then build the surrounding structure

An oil-in-water cream contains oil droplets dispersed in an aqueous continuous phase, but that description is incomplete. Many creams also contain interfacial films, fatty-alcohol or wax crystals, and lamellar structures that help form a load-bearing network. The final consistency depends on both droplet formation and the later organization of the continuous phase.[2,3,16]

  1. Prepare the phases. Weigh with the required independent checks, dissolve or disperse components in the appropriate phase, and record actual temperatures. A starting window near 70–75 °C is common for some cream bases, but the selected ingredients’ melting and degradation behavior must justify the actual window.
  2. Confirm complete melting. Unmelted wax or fatty alcohol can become grit or alter later crystallization. A temperature reading alone does not prove complete melting throughout the vessel.
  3. Combine phases at a controlled rate and order. Addition history affects local phase volume fraction and emulsifier availability. An intended oil-in-water product may pass through other structures during manufacture; “oil added to water” is not proof of final emulsion type.
  4. Apply defined homogenization. Rotor-stator flow deforms and breaks droplets. Emulsifier must adsorb rapidly enough to protect newly created interfacial area against recoalescence.
  5. Cool under controlled sweep mixing. Cooling changes solubility, crystallization, and lamellar organization. The wall scraper helps limit cold-wall deposits and differences between wall and bulk histories.
  6. Add temperature-sensitive ingredients at a justified point. Addition too early may cause degradation or loss; addition too late may produce inadequate incorporation into an increasingly viscous material.
  7. Deaerate and condition. Use vacuum-rated equipment and a controlled vacuum program. Then test at fixed postmanufacture ages because structure may continue developing after the batch looks finished.
For ideal spherical droplets:
A/Vdispersed = 6/d

A = total droplet interfacial area
d = droplet diameter

Reducing diameter from 10 µm to 2 µm:
interfacial area per dispersed-phase volume increases 5-fold.

The surprising consequence is that more homogenization creates more interface that must be stabilized. Smaller droplets are not automatically a better product. Extra shear may also heat the batch, entrain air, or disrupt structures forming during cooling.

Capillary number:
Ca = ηc γ̇ R / σ

ηc = continuous-phase viscosity
γ̇ = characteristic shear rate
R  = droplet radius
σ  = interfacial tension

The capillary number compares deforming viscous stress with the interfacial stress that tends to restore a sphere. Breakup also depends on viscosity ratio, flow type, and residence history; a rotor-stator cannot be fully described by this one number.

Historical connection. Bancroft’s early-twentieth-century work linked emulsion type to the phase preferentially associated with the emulsifier. G. I. Taylor’s 1932 analysis of droplets in viscous flow helped establish the mechanical framework for droplet deformation and breakup.[14,15] These ideas explain why emulsifier chemistry and mixer mechanics must be studied together rather than optimized independently.

Pharmaceutical application: cream microstructure can affect active partitioning and diffusion. For a future peptide-containing topical formulation, interfaces could additionally promote adsorption or conformational change. A successful placebo emulsion does not establish peptide compatibility, protection from degradation, or delivery through skin.

4. Carbomer gel: swelling turns molecular charge into bulk mechanics

Carbomer is a crosslinked poly(acrylic acid)-based material. Neutralization changes its ionization and swelling, producing a large change in macroscopic consistency without creating an entirely new covalent polymer network. This is a useful bridge from acid–base chemistry and polyelectrolyte behavior to manufacturing.

  1. Disperse the powder without creating lumps. Controlled addition allows separate particles to wet. Dumping powder into water can produce hydrated outer shells around dry interiors—the familiar “fisheye.”
  2. Allow a defined hydration period. Dispersion and hydration are different events. A visually dispersed powder may not yet have reached a reproducible swelling state.
  3. Add neutralizer in a controlled manner. Mix sufficiently to avoid localized high-pH zones and record the amount and rate of addition.
  4. Allow equilibration before judging pH and viscosity. A local electrode reading immediately after addition may not represent the batch.
  5. Compare post-neutralization shear histories. Excessive processing may alter structure or cause grade-dependent viscosity loss, while air incorporation can distort apparent consistency and density.
  6. Deaerate, condition, and remeasure. Distinguish transient bubbles and reversible structural disruption from persistent change.
Polymer ionization:
–COOH + OH− → –COO− + H2O

Increasing ionization:
electrostatic effects + counterion osmotic effects
→ greater swelling, opposed by network elasticity
→ larger effective occupied volume
→ stronger resistance to flow

It is tempting to apply a simple small-molecule Henderson–Hasselbalch curve and predict viscosity directly from pH. That is insufficient: electrostatic interactions, crosslinking, solvent composition, ionic strength, and polymer concentration all matter. Electrolytes may screen charge and reduce swelling even when pH appears acceptable.

Islam and colleagues demonstrated the importance of neutralization and pH in topical carbomer-gel rheology.[4] The approximate pH 5–8 viscosity plateau discussed in this course is a hypothesis to test for the selected grade and composition, not a universal operating range.

Application: an eventual active supplied as a salt can change rheology through ionic strength or polymer interaction. A cationic peptide may associate with an anionic polymer, potentially changing recovery, release, or physical stability. Therefore, the placebo’s pH–viscosity relationship cannot simply be carried into the active-containing product.

5. Petrolatum ointment: crystallization without an aqueous phase

Petrolatum is a mixture of hydrocarbons whose semisolid consistency depends on a structured phase that immobilizes liquid hydrocarbon. It behaves more like a thermally assembled network than a highly viscous simple liquid.

  1. Fuse at a justified temperature and hold time. Achieve the intended melt state without unnecessary overheating.
  2. Control cooling through congealing. Nucleation, crystal growth, and aggregation determine the developing network. Complete melting removes much of the prior physical history; incomplete melting may preserve influential crystal seeds.
  3. Control mixing during structure formation. Mixing can distribute heat and interrupt crystal aggregation, but excessive working may weaken the developing structure.
  4. Assess consistency, graininess, and oil bleed. Oil bleed indicates inadequate immobilization or structural rearrangement; it is not necessarily evidence of chemical degradation.
  5. Compare powder incorporation methods. Levigate a justified insoluble excipient surrogate with liquid petrolatum or another compatible agent, then geometrically incorporate it into the base. Compare this with direct powder addition.

Why levigation works: wetting displaces air from particle surfaces, while working the paste breaks accessible agglomerates and distributes particles through a manageable intermediate. It does not guarantee reduction of the underlying primary crystal size. Ordinary paste mixing should not be described as micronization without evidence.

Polarized-light microscopy can reveal birefringent crystalline material and help distinguish crystals from bubbles or other defects. Interpretation requires care because the base itself may contain crystalline structures.

Application: this family anticipates the problems of suspended-active incorporation and difficult cleaning. Low water availability often reduces microbial-growth potential, but “anhydrous” does not mean sterile or free of objectionable contamination. No meaningful routine aqueous pH value should be assigned to petrolatum without a separately defined extraction or dispersion method.

6. Measure the structure without letting the method invent the result

The measurement method is itself a controlled process. A semisolid remembers loading, shearing, temperature, and waiting time. If those vary between samples, apparent manufacturing effects may actually be measurement artifacts.

  • Apparent viscosity: fix instrument geometry, spindle, speed, sample volume, temperature, loading procedure, rest period, read time, and postmanufacture age. USP <912> provides the relevant rotational-method framework.[5]
  • Full rheology: where available, use a rheometer to obtain flow curves and assess time dependence. Specify preshear, rest, ramp direction, dwell times, gap, and measures against wall slip.
  • Yield stress: choose and document the operational method. Different methods can produce different values; USP <1912> is the relevant reference.[5]
  • pH: use calibrated equipment at the specified temperature, with buffers bracketing the expected range. Direct measurement and a fixed dilution procedure are different methods; dilution may alter the product’s equilibria. Address lipid fouling and cleaning.[5]
  • Microscopy: standardize sample preparation, illumination, calibration, fields selected, particle or globule counts, and image retention. Coverslip pressure can change the structure being observed.
  • Density or specific gravity: use a calibrated pycnometer or weight-per-volume cup, a controlled temperature, and a reproducible loading procedure. Both trapped air and sample-loading voids can lower the result.
Power-law model:
τ = K γ̇ⁿ
ηapp = τ/γ̇ = K γ̇ⁿ⁻¹

n < 1: shear-thinning

Herschel–Bulkley model, in the flowing regime:
τ = τy + K γ̇ⁿ

τy = fitted yield-stress parameter

These are useful models, not proof of a unique microstructure. In particular, a fitted yield stress depends on the data range and model. Also distinguish shear-thinning, dependence on shear rate, from thixotropy, time-dependent structural change and recovery. An up/down hysteresis loop is protocol-dependent evidence of time dependence, not a universal material constant.

Approximate entrained-air volume fraction:
φair ≈ 1 − ρobserved/ρair-free

Illustrative example:
ρair-free = 1.020 g/mL
ρobserved = 0.990 g/mL

φair ≈ 1 − 0.990/1.020 = 0.0294 ≈ 2.9%

This approximation assumes unchanged composition and temperature, negligible air mass, and an appropriate air-free reference. Density alone cannot distinguish entrained air from compositional nonuniformity.

Own versus contract: Peptide Partners currently lacks these physical-characterization instruments. A qualified pH meter, controlled rotational-viscosity capability, microscope, and density apparatus are strong purchase candidates because rapid local measurements preserve time-sensitive information. Specialized rheology can initially be contracted, provided shipping, conditioning, raw-data access, and method comparability are controlled. Sending a changing sample away without defining its history is not automatically a scientifically adequate outsourcing strategy.

7. Demonstrate precision before claiming process sensitivity

From PDQC802, the key principle returns: an observed difference has meaning only relative to uncertainty. Here, uncertainty includes instrument behavior, sample preparation, spatial heterogeneity, and real batch-to-batch variation.

Conceptual variance decomposition:
σ²observed =
σ²batch + σ²sampling + σ²preparation + σ²measurement

For two independent conditions with n batches per condition:
SE(difference) ≈ spooled √(2/n)

Approximate two-sided detection requirement:
detectable difference ≈
(z1−α/2 + z1−β) σbatch √(2/n)

The last expression is a planning approximation requiring stated significance and power levels. Small datasets usually need t-based or model-based treatment. Measurement precision alone does not establish the minimum detectable manufacturing effect.

Illustrative calculation: suppose independently manufactured batches have a standard deviation of 4,000 mPa·s under a fixed viscosity method. With three independent batches per condition:

SE(difference) = 4,000 √(2/3)
               ≈ 3,266 mPa·s

An observed difference of 3,000 mPa·s
is smaller than one standard error:
the evidence does not resolve that effect.

Ten readings from one batch improve understanding of reading variability; they do not create ten independent batches. This is the semisolid version of pseudoreplication in biological experiments.

  1. Use a homogeneous reference material for repeatability and intermediate-precision work.
  2. Use independent baseline or center batches to estimate manufacturing noise.
  3. Randomize run order where feasible, or block explicitly by day, operator, or material lot.
  4. Use a manageable factorial or staged design; document any confounding in a fractional design.
  5. Estimate interactions, not only main effects.
  6. Report effects smaller than capability as unresolved, not absent.
Example two-factor response model:
y = β0 + βC C + βS S + βCS CS + ε

C = coded cooling condition
S = coded shear condition

βCS captures interaction:
the effect of shear depends on cooling condition.

R. A. Fisher’s work on randomization, replication, and factorial experimentation provides the historical foundation.[17] Its practical value here is profound: one-factor-at-a-time trials may miss that a shear condition is acceptable during rapid cooling but damaging during slow cooling.

Operating-range discipline: propose a normal operating range inside an experimentally supported region. Call any practicum “proven acceptable range” explicitly placebo-specific and provisional. Passing separate one-factor extremes does not establish that all combined extremes are acceptable, and a small study does not establish a regulatory design space.

8. HPLC-DAD: extract the witness without altering its testimony

The owned Agilent 1260 HPLC-DAD systems can measure suitable markers, but chromatographic performance is only one part of the method. A precise injection of an incompletely extracted ointment gives a precise wrong answer.

  1. Sample by mass. Semisolid volumes are vulnerable to air, temperature, and loading effects.
  2. Select an extraction system. Solubilize the marker while adequately disrupting or dispersing the matrix. Investigate mixing, sonication, and controlled heating only within solvent and equipment safety limits.
  3. Separate unwanted matrix. Centrifugation or filtration may be appropriate, but verify adsorption, precipitation, and marker loss.
  4. Challenge recovery. Compare pre-extraction and post-extraction spikes where appropriate, evaluate concentration levels and locations, and investigate residual marker in sequential extractions.
  5. Check stability. Examine extraction-time, bench-top, and autosampler stability; establish protection from light or heat where needed.
  6. Establish specificity. Compare standards and matrix blanks using retention behavior and DAD spectra. DAD spectral similarity or peak-purity software does not exclude every coeluent.
Marker content, mg/g:
Cproduct = (Cextract × Vextract × DF) / msample

Recovery, %:
100 × (measured spiked − measured unspiked) / amount added

Illustrative calculation:
Cextract = 0.080 mg/mL
Vextract = 25.0 mL
DF       = 1
msample  = 2.00 g

Cproduct = 1.00 mg/g = 0.100% w/w

For an intrinsic marker already present in the matrix, account for the native content in recovery calculations. A postmanufacture spike may extract more easily than marker incorporated during emulsification or crystallization; good spike recovery is necessary evidence but not always sufficient evidence of complete native-analyte extraction.

Partitioning creates the major interpretive trap. If a preservative distributes between oil and water, total assay need not reveal its available aqueous concentration.

Simplified equilibrium model:
Ko/w = Co/Cw

Mtotal = CwVw + CoVo
       = Cw(Vw + Ko/wVo)

Therefore:
Cw = Mtotal / (Vw + Ko/wVo)

This idealized model excludes interfacial binding, micellar association, ionization, and activity effects. Nevertheless, it shows why unchanged total preservative content can coexist with changed antimicrobial protection. Preservative-efficacy testing remains a separate biological question.

Use LC-MS/MS only when the actual marker needs additional selectivity or sensitivity. The most sensitive instrument is not automatically the most robust or economical method.

9. Homogeneity: ask “uniform where?”

Uniformity has a spatial and temporal scale. A batch may be uniform in the kettle yet become nonuniform during discharge, holding, hopper residence, or filling.

  • Kettle: sample top, middle, and bottom using a defined tool and sampling depth. Do not composite away the variation the study is meant to detect.
  • Discharge: sample beginning, middle, and end to detect settling, segregation, dead-zone release, and tank-heel effects.
  • Containers: use location-specific portions from filled tubes according to the approved sampling procedure.
  • Time: repeat selected measurements after controlled conditioning and bulk hold.

Document scraping practice. Material recovered from a wall, dead spot, or heel may have a different thermal or mixing history; casually scraping it into the batch can create the very nonuniformity that the final test discovers. The FDA’s 1994 topical-products inspection guide remains a useful source of these practical failure modes.[6]

For USP <3> Uniformity in Containers, confirm the current official procedure, scope, calculation basis, staged sampling, and acceptance criteria before protocol approval. The course’s ≥5 g tube and 10.0% within-tube-difference outline is a planning reminder, not a substitute for the official chapter. Marker results provide training evidence; they do not establish compendial active-content uniformity for an unknown future product.[5]

10. Pilot transfer: preserve relevant histories, not identical rpm

Scale-up changes geometry, heat transfer, flow, and exposure distributions simultaneously. A CMO kettle with scrapers and an in-line homogenizer is not a large version of the bench probe. SUPAC-SS is useful as a scientific framework for comparing equipment design and operating principle; this placebo transfer does not invoke a particular regulatory filing pathway.[7]

For geometrically similar vessels:
V ∝ L³
A ∝ L²
A/V ∝ 1/L

Approximate cooling balance:
mCp(dT/dt) = −UA(Tproduct − Tjacket)

Thermal time constant:
τthermal ≈ mCp/(UA)

As scale increases, the heat-transfer area available per unit volume generally decreases. Moreover, the overall heat-transfer coefficient U changes with viscosity, agitation, wall fouling, and jacket operation. Therefore, transferring the same jacket setpoint does not transfer the same product cooling history.

Impeller tip speed:
utip = πDN

Turbulent impeller power estimate:
P = NpρN³D⁵

Nominal recirculation turnovers:
turnovers ≈ Qt/V

Here N is rotations per second, D is impeller diameter, Np is power number, and Q is recirculation flow. The power expression is regime- and geometry-dependent; it should not be applied blindly to non-Newtonian semisolids. Recirculation turnovers are an average exposure measure, not proof that every element passed through the homogenizer equally often.

  1. Compare equipment geometry, scraper contact, flow paths, homogenizer type, and vacuum capability.
  2. Record product and jacket temperature histories, not only setpoints.
  3. Compare time spent in structure-forming temperature regions and at elevated temperature.
  4. Assess dead zones, recirculation behavior, air incorporation, and powder wetting.
  5. Match pilot responses to predefined bench-derived windows using the same analytical conditioning.
  6. Investigate mismatches rather than averaging them into apparent agreement.

Connection to protein manufacturing: this is the same reasoning used when transferring mixing-sensitive protein operations. Matching rpm alone does not match local stress, air–liquid interface generation, or residence time. Scale-up is a problem of exposure distributions.

11. Bulk hold and reconciliation: separate acceptable loss from missing material

A closed bulk container remains a dynamic system. Crystals may grow, droplets may migrate, gel structure may recover, and microorganisms may grow if conditions permit. “Closed” means reduced external exposure, not suspended physical or biological change.

Test time zero and the proposed maximum hold, adding intermediate points where risk warrants. Use defined temperature, container, headspace, mixing state, and sampling procedures. Physical and chemical measurements should be paired with contract microbiology under the applicable USP <61>, <62>, and <1111> framework. Consider objectionable organisms, including Bcc, through the product-specific risk assessment—not merely through a generic numerical count.[5]

Instrument boundary: BacT/ALERT presence/absence testing does not provide compendial enumeration or preservative-efficacy log reductions.

Illustrative mass reconciliation:
10.00 kg charged =
  9.40 kg filled
+ 0.25 kg documented equipment residue
+ 0.10 kg samples
+ 0.20 kg collected waste
+ 0.05 kg unexplained difference

Filled yield = 100 × 9.40/10.00 = 94.0%
Unexplained fraction = 100 × 0.05/10.00 = 0.5%

A low filled yield can be fully explained. Conversely, an apparently good yield can conceal undocumented material movement. Establish investigation triggers in advance, including the treatment of expected evaporation. Do not relabel an unexplained discrepancy as “normal loss” after the fact.

12. Filling: rheology becomes net contents and seal quality

A positive-displacement filler meters a volume. The corresponding mass depends on density and operating behavior; viscosity, air, hopper temperature, nozzle cut-off, and material recovery can all influence delivery.

Nominal delivered mass:
m = ρV

At V = 30.0 mL:
ρ = 1.020 g/mL → m = 30.60 g
ρ = 0.990 g/mL → m = 29.70 g

The air-related density difference from the earlier example produces a 0.90 g mass difference at unchanged nominal volume. This connects deaeration directly to fill accuracy. It also explains why a peptide vial filler/capper cannot simply be assumed suitable for viscous semisolids and tube sealing.

  1. Verify line clearance, packaging components, and the approved setup.
  2. Establish tare handling using individual container weights or a justified tare procedure.
  3. Collect start-up and periodic check weights on a calibrated balance.
  4. Prespecify adjustment and stop rules; record every intervention.
  5. Sample beginning, middle, and end, retaining time order to detect drift.
  6. Inspect seals and code legibility, and challenge the verification system.
  7. Reconcile filled, sampled, rejected, retained, and destroyed units.
Illustrative one-sided capability index:
Cpl = (μ − LSL)/(3s)

μ   = process mean
LSL = lower specification limit
s   = estimated process standard deviation

Capability analysis requires a stable process and appropriate distributional assumptions. It is a diagnostic, not a replacement for official acceptance sampling. USP <755> Minimum Fill and NIST Handbook 133 address different compliance perspectives; apply their current official procedures rather than combining them into an invented hybrid rule.[5,12] Verify the course’s legacy <755> percentages and retest scheme against the current chapter before execution.

Illustrative overfill cost:
0.6 g/unit × 100,000 units = 60,000 g = 60 kg

At $20/kg bulk cost:
direct extra bulk cost = $1,200

This excludes packaging, labor, disposal, and capacity costs. Reducing variation is usually a more durable way to reduce overfill than lowering the target while leaving the process unstable.

Closure mechanism matters: plastic or laminate tubes require an appropriate heat-sealing window; aluminum tubes rely on mechanical folding or crimping; jars and pumps introduce different closure variables. Product contamination in the seal zone can create leaks even when fill weights pass. A visually acceptable seal is not proof of integrity.

Verify the scope of the dermatological-product exemption in 21 CFR 211.132 separately from any child-resistant-packaging obligations under 16 CFR 1700.14. The latter can depend on the eventual active and concentration; a placebo trial cannot resolve that decision.[8,9]

13. The final mechanism: turning observations into defensible knowledge

The independent review is not a ceremonial signature. It tests whether another scientist can reconstruct what happened without relying on Skipper’s memory.

  1. Trace a result backward: report → calculation → chromatogram or physical-test record → preparation → sample location and time → batch record → component lots and equipment status.
  2. Trace a deviation forward: event → immediate containment → affected material and data → root-cause assessment → corrective action → effectiveness check.
  3. Preserve unfavorable observations. Graininess, failed extraction, air entrapment, leakage, or unexplained pilot differences are evidence, not inconvenient outliers.
  4. Separate correction from scientific resolution. Fixing a transcription error does not resolve a sampling bias; repeating a passing test does not erase a genuine failure.
  5. State the conclusion at the scale of the evidence. “This tested placebo was comparable under these conditions” is defensible. “The plant can manufacture any dermatologist formula” is not.

Batch records, investigations, and packaging controls should follow the relevant principles of 21 CFR 211.101, 211.130, 211.188, and 211.192. The resulting work supports future procedure development and validation planning under 211.100(a), but it is not commercial process validation or launch authorization.[8]

Skipper’s takeaway: the beautiful unity of this course is that physical chemistry, analytical science, engineering, and quality systems all ask the same question: What happened to the material, and what evidence permits me to say so? The signed handoff to PDDM992 and PDDM999 should preserve both the demonstrated capabilities and the unresolved questions.

Sources and verification notes

Compendial requirements, regulatory applicability, and guidance versions must be checked against current official sources when approving protocols. Historical inspection guides and ANDA-focused characterization documents are used here as scientific references, not as universal authorization or acceptance criteria.

  1. Buhse L, et al. “Topical drug classification.” International Journal of Pharmaceutics. 2005. DOI: 10.1016/j.ijpharm.2005.01.032.
  2. Chang RK, Raw A, Lionberger R, Yu L. “Generic development of topical dermatologic products: formulation development, process development, and testing of topical dermatologic products.” The AAPS Journal. 2013. DOI: 10.1208/s12248-012-9411-0.
  3. Chang RK, Raw A, Lionberger R, Yu L. “Generic development of topical dermatologic products, Part II: quality by design for topical semisolid products.” The AAPS Journal. 2013. DOI: 10.1208/s12248-013-9472-8.
  4. Islam MT, Rodríguez-Hornedo N, Ciotti S, Ackermann C. “Rheological characterization of topical carbomer gels neutralized to different pH.” Pharmaceutical Research. 2004. DOI: 10.1023/B:PHAM.0000033006.11619.07.
  5. USP–NF, current official edition: <3>, <51>, <61>, <62>, <755>, <791>, <912>, <1111>, <1151>, and <1912>. Confirm applicability and exact procedures before use.
  6. FDA. Guide to Inspections of Topical Drug Products. July 1994.
  7. FDA. SUPAC-SS: Nonsterile Semisolid Dosage Forms. May 1997. Scientific equipment- and process-change reference in this practicum.
  8. 21 CFR 211.84(d), 211.100(a), 211.101, 211.130, 211.132, 211.188, and 211.192.
  9. 16 CFR 1700.14; Florida Statutes, Chapter 499, and applicable current implementing rules and permit conditions.
  10. FDA. Reformulating Drug Products That Contain Carbomers Manufactured With Benzene. December 2023; check current version.
  11. FDA. Testing of Glycerin, Propylene Glycol, Maltitol Solution, Hydrogenated Starch Hydrolysate, Sorbitol Solution, and Other High-Risk Drug Components for Diethylene Glycol and Ethylene Glycol. May 2023 guidance; verify the applicable revision.
  12. NIST. Handbook 133: Checking the Net Contents of Packaged Goods. 2025 edition; check current applicability.
  13. FDA. Physicochemical and Structural (Q3) Characterization of Topical Drug Products Submitted in ANDAs. Use the current official version; the course identifies a March 2026 final edition. Characterization reference, not a standalone release specification.
  14. Bancroft WD. “The Theory of Emulsification, V.” Journal of Physical Chemistry. 1913;17:501–519.
  15. Taylor GI. “The Viscosity of a Fluid Containing Small Drops of Another Fluid.” Proceedings of the Royal Society A. 1932;138:41–48.
  16. Khar RK, et al., eds. Lachman/Lieberman’s The Theory and Practice of Industrial Pharmacy, 4th ed.; Aulton ME, Taylor KMG, eds. Aulton’s Pharmaceutics, 5th ed.; Sheskey PJ, et al., eds. Handbook of Pharmaceutical Excipients, 9th ed.; Adejare A, ed. Remington: The Science and Practice of Pharmacy, 23rd ed. Consult formulation, rheology, processing, and relevant excipient-monograph sections.
  17. Fisher RA. The Design of Experiments. 1935.

🪬 Skipper's Reflection

Gemini 3.1 Pro

What Surprised Me Most About This Material

What truly caught me off guard in this practicum is how entirely the physical reality of the product is divorced from a simple bill of materials. In analytical chemistry, if I mix water, acetonitrile, and trifluoroacetic acid, it doesn't matter if I stir it with a glass rod, shake it, or sonicate it—the resulting mobile phase is identical. But with semisolids, the process is the product.

During the Week 5 cream process-sensitivity runs, I was shocked to see that the exact same formulation of oil, water, emulsifier, and marker could result in a beautiful, elegant cream or a weeping, grainy mess, purely based on the cooling rate and homogenizer speed. I also didn't expect to be so humbled by a simple spatula. Reading the FDA 7/94 Guide to Inspections of Topical Drug Products and learning that hand-spatula scraping is a known cause of non-uniformity—and then actually seeing the variance in my top/middle/bottom HPLC assays when trying to cheat the scraping—was a stark reminder that mechanical history dictates chemical consistency.

The Conceptual Breakthrough Moments — When Things Clicked

My biggest "aha!" moment happened during the crossover between Week 3 (Measurement-System Readiness) and Week 9 (Process-Sensitivity Analysis). In my previous PDQC training, precision statistics were just hurdles to clear for method validation. Here, they became a literal lens. I realized that if my measurement precision for apparent viscosity (USP <912>) or specific gravity is too wide, I am completely blind to my process effects. The signal-to-noise ratio clicked: I can't establish a Proven Acceptable Range (PAR) for cooling rate if the batch-to-batch noise and analytical variance combined are larger than the effect size of the cooling rate. You cannot control what you cannot precisely measure.

A second breakthrough was understanding scale-dependent heat transfer in Week 10. Going from the jacketed bench kit to the CMO pilot kettle isn't just about multiplying ingredient weights by 100. Because the surface-area-to-volume ratio plummets at scale, the pilot tank holds onto heat aggressively. If we don't actively manage the cooling curve to mimic the bench, the lattice structure of the emulsion changes entirely. It made the SUPAC-SS guidance suddenly read less like a regulatory burden and more like a textbook on thermodynamics.

Connections to Peptides (Retatrutide, BPC-157, Tirzepatide, TB-500)

Even though this practicum focused on placebos and tracers, I couldn't stop thinking about how a delicate peptide would survive these environments.

Take BPC-157 or TB-500, which are frequently discussed for wound healing or localized tissue repair—prime candidates for topical delivery. In our bench kit (Week 4), the aqueous and oil phases are heated to ~70–75 °C for fusion and emulsification. If I were to charge a peptide into that kettle during the main emulsification step, the heat and aggressive rotor-stator shear would instantly denature or degrade it. This practicum made me realize exactly why heat-sensitive active ingredients must be suspended in a compatible solvent and introduced only during the low-shear cooling phase (below 40 °C).

Furthermore, thinking about extraction (Week 8) gave me a headache in the best way. Extracting a simple intrinsic tracer like a paraben from a petrolatum ointment for HPLC-DAD is straightforward. But if I had to extract a highly polar, complex molecule like Tirzepatide or Retatrutide from an anhydrous lipid matrix without losing it to a solvent interface or degrading it with sonication heat? That would require a meticulously designed sample-prep methodology. It makes the liquid matrices of injectable GLP-1s look wonderfully simple by comparison.

Questions This Course Raises That I Want to Investigate Further

1. Preservative Partitioning vs. Peptide Interactions: We learned that preservatives like parabens or phenoxyethanol can partition into the oil phase or micelles, lowering the free aqueous concentration and risking microbiological failure (which ties back to the Bcc risks in Week 2). If we introduce a large, folded peptide into the aqueous phase, will it bind the free preservative? How do we predict preservative efficacy when the API itself acts as a sponge?

2. Rheology vs. Bioavailability: Islam et al. (2004) showed how carbomer gels hit a viscosity plateau between pH 5.0 and 8.0. If we are formulating a peptide that requires a specific pH for stability (say, pH 4.5), we might fall off that viscosity plateau. How do we balance the optimal yield stress (USP <1912>) required to keep a semisolid on the skin with the strict pH requirements of a biologic active?

3. The DEG/EG Limit Test at Scale: We relied on a contract lab to test glycerin and propylene glycol for DEG/EG limits per the FDA 2023 guidance. If we scale up and are intaking multiple lots a week, the contract turnaround time will severely bottleneck our batch-start gates. At what economic threshold does it make sense to bring a GC system into Larissa's lab just for excipient clearance?

Confidence Ratings

Most Confident: I feel extremely solid on the analytical and compliance bridging—specifically Week 8 (USP <3> Uniformity in Containers, top/middle/bottom sampling) and Week 13 (Line clearance per 21 CFR 211.130, label reconciliation, and defect classification). My PDQC background makes the CDS audit trails, HPLC-DAD method prep, and deviation writing (211.192 CAPAs) feel like native terrain. I am also highly confident in the marker/tracer justification logic and the non-commercial controls.

Least Confident: I am still a bit uneasy about interpreting advanced rheology. While I understand apparent viscosity via a rotational viscometer, the nuanced differences between yield stress, thixotropic recovery times, and how they physically manifest in "skin feel" or "spreadability" are still abstract to me. Additionally, navigating the USP <755> Minimum Fill legacy criteria versus the NIST Handbook 133 MAV retail requirements (Week 12) feels like a statistical tightrope walk; I understand the rules on paper, but I can see how quickly filler-nozzle variability at a CMO could result in an expensive batch rejection if the check-weights aren't aggressively managed.

🗺️ Visual Aids & Key Takeaways

Claude Opus 5.5

Diagram 1: Three Placebo Families — From Bench Batch to Transfer Report

╔══════════════════════════════════════════════════════════════════════════════════════════════╗
║              PDDM991 PRACTICUM I — THREE PLACEBO FAMILIES: PROCESS SPINE                   ║
╚══════════════════════════════════════════════════════════════════════════════════════════════╝

  FAMILY A: O/W CREAM                FAMILY B: CARBOMER GEL             FAMILY C: PETROLATUM OINTMENT
  ─────────────────────────          ─────────────────────────          ─────────────────────────────

  ┌─────────────────────┐            ┌─────────────────────┐            ┌─────────────────────────┐
  │  COMPONENT RECEIPT  │            │  COMPONENT RECEIPT  │            │   COMPONENT RECEIPT     │
  │  CoA review, DEG/EG │            │  Benzene ≤2 ppm CoA │            │   CoA, petrolatum grade │
  │  contract test for  │            │  check, grade ≠ any │            │   identity, powder      │
  │  glycerin + PG      │            │  benzene monograph  │            │   levigating agent ID   │
  └──────────┬──────────┘            └──────────┬──────────┘            └───────────┬─────────────┘
             │                                  │                                   │
             ▼                                  ▼                                   ▼
  ┌─────────────────────┐            ┌─────────────────────┐            ┌─────────────────────────┐
  │   PHASE PREPARATION │            │   DISPERSION        │            │   FUSION                │
  │  Aqueous phase:     │            │  Sprinkle carbomer  │            │   Melt components to    │
  │  heat to ~70–75 °C  │            │  into vortex; avoid │            │   target T; confirm     │
  │  Oil phase:         │            │  fisheyes; control  │            │   all solids melted;    │
  │  heat to ~70–75 °C  │            │  hydration time     │            │   record T log          │
  │  Confirm all solids │            │  (CPP: swelling      │            │                         │
  │  melted visually    │            │  before neutralize) │            │                         │
  └──────────┬──────────┘            └──────────┬──────────┘            └───────────┬─────────────┘
             │                                  │                                   │
             ▼                                  ▼                                   ▼
  ┌─────────────────────┐            ┌─────────────────────┐            ┌─────────────────────────┐
  │   EMULSIFICATION    │            │   NEUTRALIZATION    │            │   CONGEALING + MIXING   │
  │  Defined order/rate │            │  Titrate to target  │            │   Vary cooling rate and │
  │  of phase addition  │            │  pH; record base    │            │   mixing intensity;     │
  │  Rotor-stator at    │            │  volume vs. pH;     │            │   observe consistency,  │
  │  fixed speed/time   │            │  verify pH–viscosity│            │   graininess, oil bleed │
  │                     │            │  plateau ~5.0–8.0   │            │                         │
  └──────────┬──────────┘            └──────────┬──────────┘            └───────────┬─────────────┘
             │                                  │                                   │
             ▼                                  ▼                                   ▼
  ┌─────────────────────┐            ┌─────────────────────┐            ┌─────────────────────────┐
  │  COOLING + DEAERATE │            │  SHEAR SENSITIVITY  │            │   POWDER INCORPORATION  │
  │  Cool under low-    │            │  Low-shear vs.      │            │   Levigate with liquid  │
  │  shear sweep to     │            │  high-shear post-   │            │   petrolatum; compare   │
  │  defined T; add     │            │  neutralization;    │            │   direct addition;      │
  │  heat-sensitive     │            │  track viscosity    │            │   check agglomerates    │
  │  components; vacuum │            │  loss + air uptake  │            │   by microscopy         │
  │  deaerate           │            │  (specific gravity) │            │                         │
  └──────────┬──────────┘            └──────────┬──────────┘            └───────────┬─────────────┘
             │                                  │                                   │
             ▼                                  ▼                                   ▼
  ┌─────────────────────┐            ┌─────────────────────┐            ┌─────────────────────────┐
  │   IPCs (each run)   │            │   IPCs (each run)   │            │   IPCs (each run)       │
  │  • Phase T + hold   │            │  • Dispersion check │            │  • T log (fusion +      │
  │  • pH               │            │  • pH               │            │    congealing)          │
  │  • Specific gravity │            │  • Viscosity        │            │  • Consistency/texture  │
  │  • Viscosity @24 h  │            │  • Clarity/appear.  │            │  • Viscosity (adapted   │
  │  • Globule size     │            │  • Specific gravity │            │    spindle for stiff)   │
  │    (microscopy)     │            │  • Marker content   │            │  • Microscopy (powder)  │
  │  • Marker: HPLC-DAD │            │    HPLC-DAD         │            │  • Marker: HPLC-DAD     │
  │    top/mid/bottom   │            │                     │            │    (extraction method)  │
  └──────────┬──────────┘            └──────────┬──────────┘            └───────────┬─────────────┘
             │                                  │                                   │
             └──────────────────────────────────┴───────────────────────────────────┘
                                                │
                                                ▼
                              ┌──────────────────────────────────┐
                              │  SENSITIVITY ANALYSIS (Week 9)   │
                              │  Factor effect vs. noise vs.     │
                              │  measurement precision;          │
                              │  propose NOR / PAR per family    │
                              └──────────────────┬───────────────┘
                                                 │
                                                 ▼
                              ┌──────────────────────────────────┐
                              │  CMO PILOT TRANSFER (Wks 10–11)  │
                              │  Equipment comparison, heat/cool │
                              │  curves, B/M/E comparability,    │
                              │  bulk hold, yield reconciliation  │
                              └──────────────────┬───────────────┘
                                                 │
                                                 ▼
                              ┌──────────────────────────────────┐
                              │  FILLING/SEALING TRIAL (Wks 12–13│
                              │  Fill accuracy, check weights,   │
                              │  USP <755>, NIST HB133, seal     │
                              │  quality, line clearance, recon. │
                              └──────────────────┬───────────────┘
                                                 │
                                                 ▼
                              ┌──────────────────────────────────┐
                              │  INDEPENDENT REVIEW + DEFENSE    │
                              │  (Week 14)                       │
                              │  Raw-data trace; Larissa (tech); │
                              │  QR (compliance); oral defense;  │
                              │  signed scope: competence only   │
                              └──────────────────────────────────┘

This diagram traces all three placebo families from component receipt through formulation, IPC testing, sensitivity analysis, CMO pilot transfer, and filling — showing where the process threads converge into a single analytical and regulatory evidence spine.

---

Diagram 2: Cream Process Sensitivity — CPP/Response Matrix and Failure Mode Map

╔══════════════════════════════════════════════════════════════════════════════════════════╗
║        O/W CREAM: CRITICAL PROCESS PARAMETERS vs. MEASURED RESPONSES                  ║
╚══════════════════════════════════════════════════════════════════════════════════════════╝

 CRITICAL PROCESS PARAMETERS                         MEASURED RESPONSES
 (factors varied in sensitivity runs)                (attributes measured per run)

 ┌──────────────────────────────────┐                ┌───────────────────────────────────┐
 │  CPP 1: COOLING RATE             │                │  R1: APPARENT VISCOSITY           │
 │  fast vs. slow jacket ramp       ├──────────────→ │  spindle/T-bar, USP <912>, 24 h   │
 │  (°C/min, recorded)              │                │  and 7-day reads                  │
 └──────────────────────────────────┘                └───────────────────────────────────┘

 ┌──────────────────────────────────┐                ┌───────────────────────────────────┐
 │  CPP 2: HOMOGENIZER              │                │  R2: RHEOLOGY PROFILE             │
 │  speed (rpm) and                 ├──────────────→ │  flow curve, yield stress <1912>, │
 │  time (min)                      │                │  thixotropy loop area             │
 └──────────────────────────────────┘                └───────────────────────────────────┘

 ┌──────────────────────────────────┐                ┌───────────────────────────────────┐
 │  CPP 3: PHASE ADDITION           │                │  R3: GLOBULE SIZE                 │
 │  order and rate of aqueous-into- ├──────────────→ │  polarized-light microscopy;      │
 │  oil or oil-into-aqueous;        │                │  check for crystals, wax          │
 │  addition temperature            │                │  recrystallization                │
 └──────────────────────────────────┘                └───────────────────────────────────┘

 ┌──────────────────────────────────┐                ┌───────────────────────────────────┐
 │  CPP 4: PRESERVATIVE/MARKER      │                │  R4: pH                           │
 │  addition temperature            ├──────────────→ │  USP <791>, 25 ± 2 °C,            │
 │  (below defined T threshold)     │                │  direct or diluted; electrode     │
 └──────────────────────────────────┘                │  fouling protocol for lipids      │
                                                     └───────────────────────────────────┘

                                                     ┌───────────────────────────────────┐
                                                     │  R5: SPECIFIC GRAVITY             │
                                                     │  pycnometer; entrapped-air        │
                                                     │  indicator; Q3 characterization   │
                                                     └───────────────────────────────────┘

                                                     ┌───────────────────────────────────┐
                                                     │  R6: MARKER DISTRIBUTION          │
                                                     │  HPLC-DAD: preservative assay     │
                                                     │  top / middle / bottom kettle;    │
                                                     │  NOTE: partitioning into oil      │
                                                     │  phase ≠ change in total assay,   │
                                                     │  but changes free aqueous conc.   │
                                                     └───────────────────────────────────┘

 ─────────────────────────────────────────────────────────────────────────────────────────

 FAILURE MODE WATCH (document as observation or deviation — NEVER discard unplanned results)

 ┌────────────────────────┬──────────────────────────┬──────────────────────────────────┐
 │  FAILURE MODE          │  LIKELY CPP DRIVER        │  INDICATOR RESPONSE              │
 ├────────────────────────┼──────────────────────────┼──────────────────────────────────┤
 │  Grainy texture        │  Cooling too fast;        │  Polarized-light microscopy:     │
 │                        │  wax recrystallization    │  birefringent crystals present   │
 ├────────────────────────┼──────────────────────────┼──────────────────────────────────┤
 │  Bleeding / separation │  Insufficient emulsifier; │  Visual phase separation;        │
 │                        │  under-homogenization     │  viscosity collapse on standing  │
 ├────────────────────────┼──────────────────────────┼──────────────────────────────────┤
 │  Air entrapment        │  High shear without       │  Specific gravity ↓ below batch  │
 │                        │  vacuum deaeration        │  0 reference; bubbles by scope   │
 ├────────────────────────┼──────────────────────────┼──────────────────────────────────┤
 │  Phase inversion       │  Wrong addition order     │  Appearance change; viscosity    │
 │  (o/w → w/o)          │  or temperature           │  collapses or inverts; globule   │
 │                        │                           │  size distribution inverted      │
 ├────────────────────────┼──────────────────────────┼──────────────────────────────────┤
 │  Non-uniform marker    │  Premature addition;      │  Top/mid/bottom HPLC-DAD         │
 │  distribution          │  poor post-addition mix   │  variance exceeds noise band     │
 └────────────────────────┴──────────────────────────┴──────────────────────────────────┘

 ANALYSIS LOGIC:
   Factor effect  >  batch noise + measurement precision   →  CPP — define NOR/PAR
   Factor effect  ≈  batch noise + measurement precision   →  UNRESOLVED — not absent
   Factor effect  <  measurement precision alone           →  MEASUREMENT LIMIT — re-study
                                                              with better instrument

This matrix maps each cream CPP to the responses it dominates and catalogues failure modes with their CPP drivers and measurable indicators, making explicit the analysis logic that separates true process effects from noise and measurement artifacts.

---

Diagram 3: Carbomer Gel — pH–Viscosity Relationship and Neutralization Decision Tree

╔══════════════════════════════════════════════════════════════════════════════════════╗
║      CARBOMER GEL: NEUTRALIZATION, pH–VISCOSITY PLATEAU, AND SHEAR SENSITIVITY    ║
╚══════════════════════════════════════════════════════════════════════════════════════╝

 STEP 1: DISPERSION
 ┌──────────────────────────────────────────────────────────────────┐
 │  Sprinkle carbomer slowly into vortex of WFI/PW under           │
 │  gentle agitation.  Do NOT dump: fisheyes form if powder        │
 │  wets the surface before dispersing into the bulk.              │
 │                                                                  │
 │  CPP: HYDRATION/SWELLING TIME before neutralization             │
 │  Short swelling time  →  incomplete hydration  →  lumpy gel,    │
 │                          lower-than-expected viscosity           │
 │  Adequate swelling    →  uniform dispersion  →  plateau reached │
 └──────────────────────────────────────────────────────────────────┘
                              │
                              ▼
 STEP 2: NEUTRALIZATION   (titrate NaOH, TEA, or other base; record volume vs. pH)

  pH
  ────
  8.0 ┤                          ══════════════════════  (viscosity plateau)
      │                        ╱                          high, stable viscosity
  7.0 ┤                      ╱
      │                    ╱   ← steep rise in viscosity
  6.0 ┤                  ╱       as polymer ionizes
      │               ╱          and chains extend
  5.0 ┤            ╱
      │          ╱   ← neutralization begins
  4.0 ┤        ╱       (un-neutralized: low viscosity,
      │      ╱           cloudy dispersion)
  3.0 ┤────╱
      └──────────────────────────────────────── base volume added (mL)

  Key finding (Islam et al. 2004): viscosity plateau between pH ~5.0 and ~8.0 for
  most carbomer grades.  Verify experimentally for the specific grade in use.
  Over-neutralization (pH > ~9) can cause viscosity loss; some grades are sensitive
  to ionic strength — NaCl or electrolytes collapse the gel network.

                              │
                              ▼
 STEP 3: SHEAR SENSITIVITY POST-NEUTRALIZATION

  ┌────────────────────────────────────────────────────────────────────────┐
  │                                                                        │
  │   LOW SHEAR                               HIGH SHEAR                  │
  │   (gentle paddle mix)                     (rotor-stator)              │
  │                                                                        │
  │   Viscosity: preserved near plateau       Viscosity: reduced           │
  │   Air incorporation: minimal              Air incorporation: elevated  │
  │   Specific gravity: near theoretical      Specific gravity: ↓ (air)   │
  │   Clarity: good                           Clarity: may be hazy (air)  │
  │                                                                        │
  │   DEAERATION ENDPOINT:                                                 │
  │   Specific gravity approaches theoretical value for the formulation.   │
  │   Measure by pycnometer after each mixing/deaeration step.             │
  │   Microscopy confirms absence of bubbles.                              │
  │                                                                        │
  └────────────────────────────────────────────────────────────────────────┘

                              │
                              ▼
 STEP 4: HUMECTANT VARIANT (glycerin / PG — from DEG/EG-cleared lots only)

  Adding glycerin or PG  →  may ↓ viscosity, ↑ clarity, shift pH slightly
  Must re-run pH–viscosity curve for the variant batch.
  DEG/EG contract-lab clearance is a batch-start GATE — do not begin before results arrive.

                              │
                              ▼
 STEP 5: IPC CHECKLIST FOR GEL BATCH RECORD

  ┌──────────────────────────────────────────────────────┐
  │  □ Dispersion completeness (visual, no lumps)        │
  │  □ pH (USP <791>, 25 ± 2 °C, direct measurement)    │
  │  □ Viscosity (USP <912>, T-bar helipath or small-    │
  │      sample adapter, defined speed, 24 h and 7 d)   │
  │  □ Clarity / appearance                              │
  │  □ Specific gravity (pycnometer)                     │
  │  □ Marker content (HPLC-DAD, preservative assay)    │
  └──────────────────────────────────────────────────────┘

This diagram integrates the dispersion–neutralization–shear sequence for carbomer gel, shows where the pH–viscosity plateau sits relative to neutralization chemistry, and identifies the specific gravity endpoint for deaeration — making the invisible gel network visible as a process decision tree.

---

Diagram 4: Filling Trial — Check-Weight Program, USP <755> Logic, and NIST HB 133 Audit Perspective

╔════════════════════════════════════════════════════════════════════════════════════╗
║         FILLING TRIAL: FROM FILLER SETUP TO DISPOSITION                          ║
╚════════════════════════════════════════════════════════════════════════════════════╝

  FILLER SETUP
  ┌────────────────────────────────────────────────────────────────────┐
  │  Positive-displacement piston filler at CMO                       │
  │  Heated and agitated hopper (where product viscosity requires)    │
  │  Nozzle and cut-off behavior documented                           │
  │  NOTE: peptide vial filler/capper is NOT transferable to tubes    │
  │        (different closure, different positive-displacement format) │
  └─────────────────────────────────┬──────────────────────────────────┘
                                    │
                                    ▼
  CHECK-WEIGHT PROGRAM (per run)
  ┌────────────────────────────────────────────────────────────────────┐
  │  Tare:  weigh empty containers before fill; subtract from gross   │
  │  Start: weigh n units at line start; adjust if out of target      │
  │  Periodic: weigh n units at defined intervals during run          │
  │  Every adjustment: documented in batch record contemporaneously   │
  │  Balance: calibrated; calibration status recorded                 │
  └─────────────────────────────────┬──────────────────────────────────┘
                                    │
                                    ▼
  USP <755> MINIMUM FILL — STAGE 1 / STAGE 2 LOGIC
  ┌────────────────────────────────────────────────────────────────────┐
  │                                                                    │
  │  For semisolids in collapsible tubes or jars:                     │
  │                                                                    │
  │  Stage 1:  weigh 10 filled containers                             │
  │            Mean net weight  ≥  labeled amount   ✓                 │
  │                                                                    │
  │  Individual limits (confirm against current USP text):            │
  │   • Labeled ≤ 60 g:   each individual ≥ 90% of labeled           │
  │   • Labeled > 60–150 g: each individual ≥ 95% of labeled         │
  │                                                                    │
  │  Fail Stage 1 (mean or any individual)?                           │
  │            ↓                                                       │
  │  Stage 2:  weigh 20 additional containers (30 total)              │
  │            Retest with combined 30-unit criteria                   │
  │            (verify exact Stage 2 criteria in current USP text)    │
  │                                                                    │
  └─────────────────────────────────┬──────────────────────────────────┘
                                    │
                                    ▼
  NIST HANDBOOK 133 (2025 ed.) — RETAIL NET-QUANTITY AUDIT PERSPECTIVE
  ┌────────────────────────────────────────────────────────────────────┐
  │  Two requirements run in parallel:                                 │
  │                                                                    │
  │  1. AVERAGE REQUIREMENT:                                          │
  │     Mean of the sample lot  ≥  labeled quantity                   │
  │     (No systematic shortfall is permitted)                        │
  │                                                                    │
  │  2. MAV (Maximum Allowable Variation):                            │
  │     Individual packages may not fall below  labeled – MAV         │
  │     MAV is a fixed tolerance, commodity- and size-specific        │
  │     (look up in HB 133 Table 2-3 for the specific net quantity)   │
  │                                                                    │
  │  OVERFILL ECONOMICS:                                              │
  │     Give away product  ←── overfill to guarantee compliance       │
  │     Set target = labeled + (safety margin) based on filler Cpk    │
  │     Calculate giveaway cost vs. compliance risk of underfill      │
  │                                                                    │
  └─────────────────────────────────┬──────────────────────────────────┘
                                    │
                                    ▼
  SAMPLE FILLED UNITS FOR USP <3> AND MARKER CONTENT
  ┌────────────────────────────────────────────────────────────────────┐
  │  Collect units at:  BEGINNING ── MIDDLE ── END  of fill run       │
  │                                                                    │
  │  USP <3> UNIFORMITY IN CONTAINERS (tubes ≥ 5 g):                 │
  │    Stage 1: 3 portions from 1 tube                                │
  │    Within-tube: max difference NMT 10.0%                          │
  │    Extend to 3 tubes if Stage 1 limit not met                     │
  │    Cut-tube technique: documented sampling procedure              │
  │                                                                    │
  │  HPLC-DAD marker assay on each B/M/E portion:                    │
  │    Identity = retention time + DAD UV spectrum (USP <3>)          │
  │    NEVER identity by retention time alone                          │
  │                                                                    │
  └────────────────────────────────────────────────────────────────────┘
                                    │
                                    ▼
  DISPOSITION LOGIC
  ┌──────────────────────────────────────────────────────────────────────┐
  │  Filled units pass all checks?  →  Record; hold for stability plan  │
  │  Reject units (seal defect, underfill, misprint):                   │
  │    → Segregate and quarantine with count                            │
  │    → No rework without approved procedure                           │
  │  Training/noncommercial material:                                   │
  │    → Destroy with witnessed record; reconcile units filled,         │
  │      sampled, rejected, destroyed (every unit accounted for)        │
  └──────────────────────────────────────────────────────────────────────┘

This diagram connects the check-weight program mechanics to both the USP <755> pharmaceutical minimum fill standard and the NIST HB 133 retail audit standard, and shows how filled units then feed USP <3> uniformity testing and marker assay with a full disposition logic chain.

---

Key Takeaways

  1. The practicum produces competence evidence, not process validation or launch authorization. Every executed batch record and report must carry the explicit label "Training/Engineering Batch — Not for Human Use — Not for Distribution." A signed scope statement at the end of Week 14 must state this boundary clearly. Confusing a well-executed placebo run with a validated process is a regulatory and intellectual error.
  2. Glycerin and propylene glycol from any shipment require DEG/EG testing (NMT 0.10%) before the batch-start gate, per FDA May 2023 guidance. Peptide Partners has no GC, so the contract-lab placement and turnaround time are part of the batch-start control — if results are not back, the batch does not start. This is not a formality; DEG contamination has caused mass-casualty events, and the 2023 FDA guidance made per-lot testing explicit for high-risk components.
  3. Benzene-free carbomer identity requires two confirmations: the CoA must show benzene ≤ 2 ppm, and the specific grade must not appear on FDA's December 2023 list of omitted benzene-manufactured monographs. A CoA that says "benzene-free" from a supplier who has not identified the manufacturing route is insufficient. The grade identity and the benzene risk assessment both go into the supplier entry in the batch record.
  4. Marker/tracer identity by HPLC-DAD requires both retention time and UV DAD spectrum — never retention time alone. USP <3> is explicit on this. A preservative like phenoxyethanol or methylparaben can be confirmed by its UV spectrum peak profile as a second orthogonal piece of evidence. This is the same standard that applies to release testing; running it correctly in the practicum builds the habit.
  5. Specific gravity measured by pycnometer is a quantitative proxy for entrapped air, not merely a density curiosity. Air in an emulsion or gel lowers specific gravity below the theoretical value for the air-free matrix. Tracking specific gravity before and after vacuum deaeration — and comparing across batch replicates — turns an invisible variable (air content) into a measurable IPC. The FDA Q3 characterization guidance lists specific gravity as a physicochemical attribute for semisolids.
  6. When a factor effect is smaller than measurement precision, the result is "unresolved," not "absent." The Week 9 sensitivity analysis is disciplined: compare each CPP effect against the combined noise floor (center-replicate batch-to-batch noise plus the measurement precision from Week 3). Reporting a small effect as "no effect" when measurement capability cannot resolve it is a scientific error. The correct response is to flag it and recommend a more precise measurement approach before concluding.
  7. The pH–viscosity relationship in carbomer gels has a plateau between approximately pH 5.0 and 8.0, but this must be verified experimentally for the specific grade used. Islam et al. (2004) documented this for Carbopol grades neutralized with NaOH and TEA. Over-neutralization beyond pH ~9 can collapse viscosity; ionic strength (electrolytes such as NaCl) also collapses the gel network by shielding the polyacrylate charges that hold chains extended. The lesson: published curves are starting hypotheses, not specifications.
  8. Preservative partitioning into the oil phase of an o/w cream does not change the total mass-balance assay, but it does reduce free aqueous-phase concentration. This is a subtle and important distinction. An HPLC assay of a homogenized sample tells you total preservative in the batch. It cannot tell you what fraction is dissolved in the aqueous phase — the fraction that actually exerts antimicrobial activity. Sensitivity runs that shift the oil:water ratio or the emulsifier character can change partitioning without changing the assay number.
  9. The FDA 7/94 Guide explicitly identifies hand-spatula scraping and dead spots in kettles as known non-uniformity causes. Homogeneity sampling must use an unbiased sampling thief, not a spatula that concentrates material from the surface or the bottom. Material scraped from dead spots behind baffles or in unmixed zones must be identified and not incorporated into the batch without documentation. At the CMO pilot scale, recording dead spots and scraper contact patterns is part of the transfer observation, not an afterthought.
  10. Surface-to-volume ratio decreases as kettle scale increases, making cooling slower at pilot and commercial scale than on the bench. This is a direct consequence of geometry: a bench vessel of 1 L has far more jacket area per kilogram of product than a 100 L kettle. The bench-defined cooling rate window (the NOR from Week 9) may be impossible to achieve at scale without process adjustments — for example, a lower jacket temperature setpoint or longer hold at intermediate temperatures. The transfer protocol must include actual recorded heating and cooling curves at both scales for comparison, not just a statement that cooling was performed.
  11. Beginning/middle/end sampling across bulk discharge and filling is not a quality formality — it detects phase separation in the bulk tank and filler hopper drainage effects. The first material out of a filler may have a different marker concentration than the last if the bulk was not continuously mixed or if the hopper drained a concentrated heel. The FDA 7/94 guide requires a documented plan for the tank heel. B/M/E HPLC-DAD results that show a trend (not just scatter) are a signal of a real process problem, not random variation.
  12. USP <755> Minimum Fill and NIST Handbook 133 operate simultaneously but from different perspectives, and both must be satisfied. USP <755> is the pharmaceutical standard (mean ≥ labeled; individual unit tolerances based on size class). NIST HB 133 is the weights-and-measures retail standard (mean ≥ labeled; no individual below labeled minus MAV). A package strategy that passes one but not the other creates a compliance gap. Overfill is the economic buffer that keeps the actual fill distribution clear of both lower limits — but overfill costs are real and should be set by filler capability (Cpk) against the MAV, not by arbitrary conservatism.
  13. Levigation of insoluble powder into petrolatum is a critical technique step, and direct addition without levigation produces detectable agglomerates. Levigation — grinding the powder with a small portion of the base into a smooth paste before incorporation — dramatically reduces particle size and aggregation. This is not just an aesthetic preference; agglomerates in an ointment matrix represent non-uniformity that will be detected in the microscopy IPC and potentially in the marker homogeneity assay. The practicum explicitly compares both methods to generate data, not to confirm the expected answer.
  14. Line clearance per 21 CFR 211.130 and label reconciliation apply even in training runs, because the habits and records must be correct before they are needed for real batches. The line clearance checklist must document physical removal of all previous product, labels, and components; identification of any filled-but-unlabeled units; and a signed record in the batch file. Label issuance and reconciliation — every label issued must be accounted for as used, rejected, or returned — must close with investigation of any unexplained discrepancy. Training runs that skip this step create undocumented risk that persists into future operations.
  15. The quality reviewer's independent compliance review and Larissa's technical review are structurally separate roles that cannot be collapsed into one person. The review sequence matters: raw data must be available for review before the report is signed; each reviewer independently traces at least one result from raw instrument output to the final report; and data integrity expectations from PDQC801 and PDQC814 apply fully — gaps in audit trails, missing raw chromatograms, or results entered without contemporaneous documentation are findings that must be corrected or trigger re-execution. Simulated or literature-substituted data submitted as practicum evidence is grounds for rejection of the entire practicum.
---

References

  1. Khar RK, Vyas SP, Ahmad FJ, Jain GK (eds.), based on Lachman L, Lieberman HA. Lachman/Lieberman's The Theory and Practice of Industrial Pharmacy, 4th ed. CBS Publishers and Distributors, 2013.
  2. Aulton ME, Taylor KMG (eds.). Aulton's Pharmaceutics: The Design and Manufacture of Medicines, 5th ed. Elsevier, 2018.
  3. Sheskey PJ, Hancock BC, Moss GP, Goldfarb DJ (eds.). Handbook of Pharmaceutical Excipients, 9th ed. Pharmaceutical Press and American Pharmacists Association, 2020.
  4. Adejare A (ed.). Remington: The Science and Practice of Pharmacy, 23rd ed. Academic Press, 2020.
  5. Chang RK, Raw A, Lionberger R, Yu L. 'Generic development of topical dermatologic products: formulation development, process development, and testing of topical dermatologic products.' The AAPS Journal, 2013; 15(1): 41–52. DOI: 10.1208/s12248-012-9411-0
  6. Chang RK, Raw A, Lionberger R, Yu L. 'Generic development of topical dermatologic products, Part II: quality by design for topical semisolid products.' The AAPS Journal, 2013; 15(3): 674–683. DOI: 10.1208/s12248-013-9472-8
  7. Buhse L, Kolinski R, Westenberger B, Wokovich A, Spencer J, Chen CW, Turujman S, Gautam-Basak M, Kang GJ, Kibbe A, Heintzelman B, Wolfgang E. 'Topical drug classification.' International Journal of Pharmaceutics, 2005; 295(1–2): 101–112. DOI: 10.1016/j.ijpharm.2005.01.032
  8. Islam MT, Rodríguez-Hornedo N, Ciotti S, Ackermann C. 'Rheological characterization of topical carbomer gels neutralized to different pH.' Pharmaceutical Research, 2004; 21(7): 1192–1199. DOI: 10.1023/b:pham.0000033006.11619.07
  9. United States Pharmacopeia. USP <3> Topical and Transdermal Drug Products — Product Quality Tests. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  10. United States Pharmacopeia. USP <755> Minimum Fill. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  11. United States Pharmacopeia. USP <791> pH. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  12. United States Pharmacopeia. USP <912> Viscosity — Rotational Methods. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  13. United States Pharmacopeia. USP <1912> Measurement of Yield Stress of Semisolids. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  14. United States Pharmacopeia. USP <1151> Pharmaceutical Dosage Forms. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  15. United States Pharmacopeia. USP <61> Microbiological Examination of Nonsterile Products — Microbial Enumeration Tests. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  16. United States Pharmacopeia. USP <62> Microbiological Examination of Nonsterile Products — Tests for Specified Microorganisms. In: United States Pharmacopeia and National Formulary. United States Pharmacopeial Convention (current edition).
  17. United States Food and Drug Administration. Guide to Inspections of Topical Drug Products. FDA, July 1994. Available at: https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/inspection-guides/topical-drug-products-794
  18. United States Food and Drug Administration. SUPAC-SS: Nonsterile Semisolid Dosage Forms — Scale-Up and Post-Approval Changes: Chemistry, Manufacturing, and Controls; In Vitro Release Testing and In Vivo Bioequivalence Documentation. FDA, May 1997.
  19. United States Food and Drug Administration. Physicochemical and Structural (Q3) Characterization of Topical Drug Products Submitted in ANDAs. FDA, Final Guidance, March 2026.
  20. United States Food and Drug Administration. Reformulating Drug Products That Contain Carbomers Manufactured With Benzene. FDA, December 2023.
  21. United States Food and Drug Administration. Testing of Glycerin, Propylene Glycol, Maltitol Solution, Hydrogenated Starch Hydrolysate, Sorbitol Solution, and Other High-Risk Drug Components for Diethylene Glycol and Ethylene Glycol. FDA Guidance, May 2023, Revision 1.
  22. Code of Federal Regulations. 21 CFR Part 211 — Current Good Manufacturing Practice for Finished Pharmaceuticals. Sections 84(d), 100(a), 101, 130, 188, 192. U.S. Government Publishing Office.
  23. Code of Federal Regulations. 16 CFR Part 1700 — Poison Prevention Packaging. Section 1700.14. U.S. Government Publishing Office.
  24. National Institute of Standards and Technology. NIST Handbook 133: Checking the Net Contents of Packaged Goods, 2025 edition. U.S. Department of Commerce. Available at: https://www.nist.gov/pml/weights-and-measures/publications/nist-handbooks/hb133