Drug-polymer miscibility, glass transition kinetics, and molecular-mobility-based prediction of recrystallization risk for amorphous solid dispersions
A crystalline solid sits in a deep thermodynamic energy well: the ordered lattice is stabilized by optimized intermolecular packing, and dissolution requires paying the full lattice free energy to break that order. An amorphous solid has no long-range order — molecules are frozen in a random, liquid-like arrangement — and therefore sits at a substantially higher free energy state. This directly translates to higher apparent solubility and faster dissolution (often 10–100× versus the crystalline form), but the same excess free energy that drives the solubility advantage also thermodynamically drives the material back toward the stable crystalline state — the central engineering tension of amorphous solid dispersion (ASD) technology.
Free energy relationships driving amorphous solubility advantage:
• Solubility ratio (amorphous/crystalline): RT·ln(C_amorphous/C_crystalline) ≈ ΔG_fusion − T·ΔS_config, related to the free energy difference between the amorphous and crystalline states at the storage/dissolution temperature • This "spring" of excess free energy can transiently push solution concentration well above the crystalline equilibrium solubility — the amorphous form acts as a supersaturation-generating "spring," while a well-chosen polymer acts as a "parachute" slowing the fall back to equilibrium (the widely used spring-and-parachute conceptual model, Guzmán et al. 2007)
Why a pure amorphous API alone is rarely viable: • A neat amorphous API, without a stabilizing polymer, typically has a glass transition temperature (Tg) close to or below room/body temperature and very high molecular mobility, recrystallizing within hours to days of manufacture — far too fast for a practical shelf-life product • A polymer carrier serves multiple stabilizing roles simultaneously: (1) raises the mixture Tg well above storage temperature via the Gordon-Taylor relationship, freezing out molecular mobility; (2) forms hydrogen bonds or other specific interactions with the API that disrupt the API's own intermolecular packing motif, raising the nucleation barrier; (3) increases the local viscosity around each drug molecule, slowing diffusion-limited nucleation and growth even above Tg
Gordon-Taylor equation for mixture Tg: Tg_mix = (w1·Tg1 + K·w2·Tg2) / (w1 + K·w2)
where w1, w2 are weight fractions of drug and polymer, Tg1, Tg2 their pure-component glass transitions, and K a constant related to the density and thermal expansion coefficients of each component — used to predict how drug loading will affect the dispersion's Tg before manufacturing
The amorphous solid dispersion (ASD) strategy, therefore, is not simply "make the drug amorphous" but rather "make a single-phase, molecularly mixed amorphous drug-polymer system with sufficiently high Tg and sufficiently strong drug-polymer interaction that recrystallization is kinetically suppressed over the intended shelf life" — the engineering problem addressed by the manufacturing, miscibility, and stability-prediction stages that follow.
Two manufacturing technologies dominate industrial ASD production, each converting a physical blend of crystalline drug and polymer into a single, kinetically trapped amorphous phase by a different route: hot-melt extrusion (HME) applies heat and shear to melt-mix the components without solvent, while spray-drying rapidly evaporates solvent from a co-dissolved feed solution, freezing the molecularly mixed state before phase separation or recrystallization can occur.
Hot-melt extrusion (HME):
• Twin-screw extruder conveys, melts, and intensively mixes a physical blend (or pre-blend) of crystalline drug and polymer through heated barrel zones, typically 120–180°C, using co-rotating intermeshing screws that provide both distributive and dispersive mixing • Screw design (kneading blocks, mixing elements) and residence time (typically 1–5 minutes) must be sufficient to achieve molecular-level mixing (true solid solution) without exceeding drug degradation temperature — a critical process parameter balancing melt viscosity reduction against thermal stability • Requires the API to be either melt-processable below its degradation temperature, or "melt-mediated dissolved" into the molten polymer below the pure API melting point (since a sufficiently high polymer concentration and processing temperature can dissolve crystalline API directly into the polymer melt without fully melting the drug crystal lattice first — analogous to eutectic-type melting-point depression) • Solvent-free process — no residual solvent risk, no ICH Q3C solvent-class considerations, and continuous manufacturing compatible (in-line PAT, real-time release testing), an increasingly important advantage under modern continuous manufacturing regulatory frameworks • Common polymers: Soluplus (polyvinyl caprolactam-polyvinyl acetate-polyethylene glycol graft copolymer, designed specifically for HME), Kollidon VA64 (PVP-VA), Eudragit grades, HPMCAS (higher processing temperature required due to higher Tg)
Spray-drying:
• Drug and polymer are co-dissolved in a common (or solvent-blend) organic/aqueous solvent system, then atomized into fine droplets (typically 10–100 µm) into a heated drying chamber • Solvent evaporates in well under one second per droplet — this extremely rapid kinetic quench is what traps the molecularly mixed amorphous state before the system can phase-separate or nucleate crystalline drug, making spray-drying particularly effective for drugs and polymers with only moderate intrinsic miscibility • Critical process parameters: inlet/outlet temperature (balancing solvent removal against thermal exposure), feed solution concentration and viscosity, atomization method (two-fluid nozzle, pressure nozzle, ultrasonic), and atomization gas flow rate — together determining particle size, residual solvent, and bulk density • Requires a solvent system in which BOTH drug and polymer are adequately co-soluble at process-relevant concentration — can be a significant formulation development challenge for chemically dissimilar drug-polymer pairs, sometimes requiring solvent blends (e.g., dichloromethane/methanol, acetone/water) • Residual solvent must be controlled to ICH Q3C limits via a secondary drying (tray or vacuum) step; typical spray-dried intermediate residual solvent 0.5–3% w/w before secondary drying, <0.5% after • Common polymers: HPMCAS (hydroxypropyl methylcellulose acetate succinate, multiple substitution grades — L, M, H — offering different pH-dependent dissolution triggers), PVP/PVP-VA, HPMC
Platform selection drivers: • HME favored for: thermally stable APIs, solvent-sensitive/solvent-free manufacturing preference, continuous manufacturing integration, lower cost at scale • Spray-drying favored for: thermally labile APIs (lower effective process temperature via solvent evaporative cooling), higher-Tg polymers difficult to melt-process (e.g., HPMCAS-H), and cases requiring fine control of particle morphology/size for downstream tableting or capsule filling
Not every drug-polymer combination forms a single, homogeneous amorphous phase at every drug loading — beyond a critical concentration, the mixture can phase-separate into drug-rich and polymer-rich amorphous domains, and drug-rich domains have dramatically reduced Tg and much higher local mobility, effectively eliminating the stabilization benefit and accelerating recrystallization. Flory-Huggins solution theory, adapted from polymer science, provides the quantitative framework for predicting miscibility and the maximum drug loading that remains a true single-phase solid solution.
Flory-Huggins theory of polymer solution thermodynamics, applied to drug-polymer amorphous mixtures:
Free energy of mixing per lattice site: ΔG_mix/RT = (φ_drug/n_drug)·ln(φ_drug) + (φ_polymer/n_polymer)·ln(φ_polymer) + χ·φ_drug·φ_polymer
• φ_drug, φ_polymer: volume fractions of drug and polymer • n_drug, n_polymer: degree of polymerization (n_drug ≈ 1 for a small-molecule API; n_polymer >>1, often several hundred to thousand for a high-MW polymer) • χ (chi): the Flory-Huggins interaction parameter, capturing the net energetic favorability (χ<0, strongly favorable, e.g. hydrogen bonding) or unfavorability (χ>0) of drug-polymer contacts relative to drug-drug and polymer-polymer contacts
Interpretation: • The combinatorial entropy terms (first two terms) are ALWAYS negative (favor mixing) but are numerically small because n_polymer is large (a long polymer chain loses much less translational entropy per mixing event than would two small molecules) — meaning drug-polymer mixing is entropically much LESS favorable than typical small-molecule solution mixing, and the SIGN AND MAGNITUDE OF χ dominates the miscibility outcome • χ < 0: mixing is thermodynamically favorable at essentially all compositions — fully miscible system, drug loading limited mainly by Tg/mobility considerations (Stage 4) rather than by phase separation • 0 < χ < ~2: partially miscible — a binodal/spinodal phase diagram exists with a critical drug loading above which the system phase-separates into drug-rich and polymer-rich domains; below the critical loading, single-phase mixing is thermodynamically stable • χ > ~2: largely immiscible at any practically useful drug loading — this drug-polymer pair is a poor ASD combination regardless of manufacturing method
Estimating χ without extensive experimentation:
1. Hansen solubility parameter (HSP) approach: χ ≈ V_site·(δ_drug − δ_polymer)²/(RT), where V_site is a reference molar volume and δ are the total Hansen solubility parameters of drug and polymer — a fast first-pass estimate (same HSP framework used for cocrystal coformer screening, see companion topic) showing good (R² 0.7–0.9) correlation with experimentally measured miscibility across published ASD literature datasets
2. Melting point depression method (Marsac, Taylor et al.): physical mixtures of drug + polymer at varying ratios are analyzed by DSC; the drug melting point depression as a function of polymer volume fraction is fit to the Flory-Huggins/Nishi-Wang equation to extract χ directly from experimental thermal data — more rigorous than pure HSP estimation but requires DSC access to a range of physical mixtures
3. Solvent-casting/annealing screening: films cast from drug-polymer-solvent solutions at multiple loadings are annealed above Tg and monitored (PLM, PXRD) for the onset of recrystallization or visible phase separation — an empirical determination of the practical maximum stable loading, complementing but not replacing the thermodynamic χ estimate
Practical formulation impact: for a moderately miscible system (χ ≈ 0.5–1), maximum single-phase drug loading is typically in the 20–40% w/w range — informing dose/pill-burden trade-offs early in ASD formulation development, since exceeding the miscibility limit does not just risk instability but produces an ASD that is ALREADY phase-separated (and therefore already partially recrystallization-prone) immediately after manufacture.
Even a thermodynamically miscible, single-phase ASD is not automatically stable indefinitely — recrystallization is a KINETIC process, and its rate is governed by molecular mobility, which in turn depends steeply on how far the storage temperature sits below the glass transition temperature (Tg). Predicting real-world shelf-life from short-duration accelerated stability studies requires understanding how α-relaxation time scales with temperature via the Vogel-Tammann-Fulcher (VTF) and Williams-Landel-Ferry (WLF) frameworks, and translating that mobility into a practical recrystallization-onset prediction.
Molecular mobility below and above Tg:
• Below Tg, a glass is in a non-equilibrium, kinetically arrested state — large-scale segmental (α-relaxation) motion is frozen on practical timescales, but smaller-scale local (β-relaxation, Johari-Goldstein) motions persist and can still, over long timescales (months to years), permit enough molecular reorganization for nucleation to slowly occur, especially near Tg • Above Tg, α-relaxation time τ drops sharply and follows non-Arrhenius (super-Arrhenius) temperature dependence, well described by the Vogel-Tammann-Fulcher (VTF) equation:
τ(T) = τ0·exp[DT0/(T−T0)]
where T0 is the Vogel temperature (typically Tg−50°C) and D is the "fragility" parameter (small D = "fragile" liquid, mobility rises very steeply as T approaches Tg from above; large D = "strong" liquid, more gradual)
• Equivalently expressed via the Williams-Landel-Ferry (WLF) equation, widely used in the pharmaceutical mobility literature:
log(aT) = −C1(T−Tg) / [C2+(T−Tg)]
with "universal" constants C1≈17.4, C2≈51.6 K providing a reasonable first approximation near Tg, though compound-specific fitting is more accurate
The practical "Tg − 50°C" storage rule: • A widely used (though approximate) formulation heuristic: storing an ASD at least 50°C below its Tg places the system deep in the VTF/WLF regime where α-relaxation times become extremely long (years to decades), providing a substantial kinetic stability margin • This rule is a useful first screen but is NOT a substitute for actual stability data, since fragility (D), drug-polymer specific interactions, and residual nuclei from imperfect manufacturing all modulate the real recrystallization onset time independent of the Tg gap alone
Water as a plasticizer — the dominant real-world stability risk: • Nearly all pharmaceutical polymers (HPMCAS, PVP-VA, HPMC) are hygroscopic and absorb atmospheric moisture; sorbed water acts as a potent plasticizer, lowering the mixture Tg substantially (10–50°C depression is common at moderate-to-high RH) via the same Gordon-Taylor relationship used for drug-polymer Tg mixing • This is why ASD products require moisture-protective packaging (desiccant, aluminum-aluminum blister) and why accelerated stability testing is run under both temperature AND humidity stress (ICH Q1A(R2) standard condition: 40°C/75%RH for 6 months, correlated empirically to ~2 years at long-term 25°C/60%RH storage for many oral solid dosage forms)
Accelerated stability protocol for recrystallization risk: • Samples stored at multiple T/RH combinations (e.g., 25°C/60%RH, 40°C/75%RH, and an open-dish 40°C/75%RH stress condition without protective packaging) are periodically pulled and analyzed by PXRD (crystallinity onset, detection limit typically 1–2% crystalline content) and modulated DSC (mDSC, detecting subtle Tg depression or small recrystallization exotherms before PXRD-detectable crystallinity develops) • Time-to-onset-of-crystallinity at each accelerated condition is fit to an Arrhenius or VTF temperature/humidity model to extrapolate a predicted room-temperature shelf life — analogous in spirit to accelerated stability extrapolation used throughout solid-dosage pharmaceutics, but with recrystallization kinetics (not chemical degradation kinetics) as the tracked failure mode
Several commercially successful ASD products demonstrate that Flory-Huggins-guided polymer selection, HME or spray-dry manufacturing, and mobility-based accelerated stability prediction translate reliably into robust, shelf-stable oral dosage forms with a durable bioavailability advantage over the crystalline API — provided the formulation science in Stages 2–4 is executed rigorously and packaging/storage conditions are matched to the predicted stability margin.
A representative marketed ASD case study (HME-manufactured protease-inhibitor combination product, PVP-VA-type carrier, historically among the earliest and most commercially significant HME-based ASD products, alongside earlier spray-dried triazole antifungal ASD products using HPMC-type carriers):
Formulation development sequence, mirroring Stages 1–4: 1. Crystalline drug substance exhibited very low aqueous solubility and strongly food-dependent, variable oral bioavailability in its earlier crystalline soft-gelatin capsule formulation — the classic BCS Class IV starting problem 2. Polymer screening using Flory-Huggins/HSP-guided miscibility assessment identified a PVP-VA-type copolymer as thermodynamically compatible at the required clinical drug loading, balancing miscibility against melt-processability for HME 3. Twin-screw HME process development established processing temperature and screw configuration achieving complete amorphous conversion (PXRD halo, no residual crystalline peaks) with acceptable residence time and no significant thermal degradation 4. Accelerated (40°C/75%RH) and long-term (25°C/60%RH) stability studies, tracking PXRD crystallinity and mDSC Tg, confirmed the ASD remained amorphous (<1–2% crystallinity, at or near the PXRD detection limit) through the full proposed shelf life when packaged in moisture-protective (aluminum-aluminum blister or induction-sealed HDPE bottle with desiccant) configurations
Clinical/commercial outcome: • The HME tablet formulation achieved substantially improved, more consistent oral bioavailability and reduced food effect versus the earlier crystalline-suspension-based soft-gelatin capsule formulation, and improved room-temperature storage stability versus a refrigerated predecessor formulation — a combination of solubility AND stability improvement directly attributable to the ASD approach • Demonstrates that the ASD platform, when the underlying drug-polymer thermodynamics (χ), manufacturing process control, and mobility-based shelf-life prediction are addressed rigorously and in the correct sequence, is not merely a laboratory curiosity but a validated, scalable, regulator-accepted commercial technology
General lessons for ASD formulation programs: • Polymer selection should be finalized using Flory-Huggins/HSP miscibility screening BEFORE extensive manufacturing process development — reformulating around a poorly miscible polymer late in development is costly • Manufacturing process (HME vs. spray-dry) selection should account for API thermal stability and drug-polymer common-solvent availability early, since switching platforms later often requires substantial re-optimization • Packaging strategy (desiccant, barrier film) must be matched to the predicted Tg depression under realistic humidity exposure — an ASD that is stable in a sealed aluminum blister may fail rapidly in a high-permeability bottle in a humid climate, so accelerated stability testing must include the ACTUAL intended packaging configuration, not just the bulk material
The single most important lesson from marketed ASD products: recrystallization risk is not a fixed property of the drug molecule but an engineered outcome of the drug-polymer-process-package system as a whole. A well-selected polymer (low χ, strong specific interactions), a well-controlled manufacturing process (complete amorphous conversion, no residual nuclei), a Tg comfortably above realistic in-use storage and humidity conditions, and moisture-protective packaging together create a kinetically stable product — while getting any one of these wrong, even with the other three correct, has caused real, costly late-stage and even post-market ASD stability failures across the industry.