A. Release mechanism and LoC

How it was shown

Indulkar and co-workers (2017) asked whether nifedipine dispersions dissolve by a drug- or a polymer-controlled route; they dissolved 10 and 20% dispersions with HPMC-E5 or PVPVA (150–250 µm) at 5 °C in 20 mL D2O at 300 rpm, once into plain D2O to 220 µg/mL drug and once into D2O presaturated with nifedipine at 110 µg/mL plus a further 110 µg/mL from the dispersion, following the drug by UV at 360 nm indulkar2017b P1, P7. Purohit and Taylor (2017) dissolved ritonavir films cast on vial walls in 18 mL of 100 mM pH 6.8 phosphate at 25 °C, followed the drug by UV (240 nm minus 400 nm) and the polymer by colorimetry of 1 µm-filtered aliquots (phenol–sulfuric acid for HPMCAS, potassium triiodide for PVP and PVPVA), and measured the free drug after ultracentrifugation; the film area was not controlled purohit2017b P14, P16, P19, P20. Bhardwaj and co-workers (2018) pressed 25 ± 5 mg of itraconazole (ITZ)–PVPVA at 500 psi in an 8 mm die, backed it with ethyl cellulose pressed at 1000 psi and sealed it with a silicone septum so that one face of about 0.50 cm² dissolved, in 70 mL 0.1 N HCl at 37 °C and 300 rpm; ITZ was read at 253 nm and PVPVA from the residual absorbance at 220 nm (quadratic calibration), n = 3, and rates were the 0–20 min slopes bhardwaj2018 P6, P7, E7. They expressed the ITZ rate relative to crystalline ITZ and divided both rates by the component's weight fraction bhardwaj2018 E2, E4. Indulkar and co-workers (2019) used Wood's apparatus with 100 mg compressed at 1600 lbs in an 8 mm die, rotated at 100 rpm in 100 mL 50 mM pH 6.8 phosphate at 37 °C, assayed unfiltered samples for ritonavir (HPLC, 210 nm) and PVPVA (size-exclusion chromatography, 205 nm), ran neat amorphous ritonavir and neat PVPVA tablets, and repeated the 10, 30 and 50% tablets in buffer presaturated with ritonavir at about 30 µg/mL indulkar2019a P2, P3, P4, P5. Que and co-workers (2019) used the same set-up at 1500 psi, mixing each drug sample with ethanol to dissolve drug-rich colloids and filtering the polymer sample through 0.2 µm, with a 250 mg neat PVPVA compact que2019 P3, P4, X13. Saboo and co-workers (2019) used 100 mg at 1500 psi in 100 mL 100 mM pH 6.8 phosphate at 100 rpm, took rates from the initial linear part of percent release normalized to each component's content, compared drug and polymer rates by t-test, and added powder dissolution at a nominal 100 µg/mL saboo2019 P9, P10, E4, P20, P19. Saboo and co-workers (2020) wrote the rate as A = kV/(SN) and assayed HPMC and HPMCAS by HPLC with evaporative light scattering and PVP, PVPVA and Eudragit S 100 (EUDS) by size-exclusion chromatography, with EUDS at pH 7.4 saboo2020a E3, P6, P7. The papers of 2020–2022 kept the 100 mg, 0.5 cm² compact but varied the read-out: Saboo and co-workers (2020) took rates from samples every minute over the first 5 min, each drug in its own medium saboo2020 P5, X5; Que and co-workers (2021) defined the limit as the last loading with simultaneous release and at least 80% of the drug in 90 min que2021 P9; Yang and co-workers (2021) compared drug and polymer rates by t-test at each loading and also dissolved powders, the PVPVA powders in medium holding 400 µg/mL HPMCAS yang2021 P17, P12, X18; Correa Soto and co-workers (2022) normalized by R = k/(Sx) without the medium volume and sampled to 120 min correasoto2022a E1, P11; and Hiew and Taylor (2022) dissolved into 250 mL and read the limit from drug release alone, without a stated criterion hiew2022b P8, X6. Indulkar and co-workers (2022) kept the 100 mg Wood's set-up with unfiltered samples so that drug held in colloids was counted, and set early PVPVA release of ternary tablets against a binary tablet reprinted from the 2019 paper, in amounts not normalized to the polymer content indulkar2022 P3, C7, P2, X6. Yang and co-workers (2022) printed the normalized rate as R = kV/(Ax) but reported only percent-release profiles, so no normalized rates are available from that paper yang2022a E1, X12. Bapat and co-workers (2023) kept the 100 mg, 0.5 cm² compact at 100 rpm and 37 °C but varied the medium (phosphate pH 6.8 at 3–51 mM/ΔpH buffer capacity, FaSSIF, FeSSIF, glycine and pH 7.5 and 9.0 buffers), assayed HPMCAS by HPLC with evaporative light scattering calibrated in each medium, and took rates from the initial linear part of each curve by R = kV/(Sx) bapat2023 P7, P8, P14, E1. Deac and co-workers (2023) added the drug content of the gel scraped from the compact at 10, 20 and 30 min, which they consider a more sensitive test of congruency than separate release rates because the drug gathers in the gel while release is incongruent (F25) deac2023a P4, C13, R3; their percent release is calculated on a 100 mg basis for compacts of 100–200 mg deac2023a X9.

What the data show

The neat references set the scale on which every dispersion is read. Melt-quenched ITZ dissolved at 51 ± 2 and crystalline ITZ at 2.8 ± 0.2 µg/min/cm², a ratio of about 18 that the authors equate with the amorphous-to-crystalline solubility ratio, which they did not measure bhardwaj2018 R4, R5, C45. Neat PVPVA dissolved at 10.4 mg/min/cm² in 0.1 N HCl at 300 rpm, 3.49 mg/min/cm² in pH 6.8 phosphate at 100 rpm, and ≈3.4 in the 2020 paper, where an extractor check of the plotted curve gives ≈1.7 bhardwaj2018 R9; saboo2019 R5; saboo2020a R13, X10 (synthesis: the three values come from different media, paddle speeds and compact masses and are not comparable). Amorphous nilvadipine dissolved at 0.002 mg/min/cm² and amorphous felodipine at ≈0.0008, neat amorphous ritonavir released essentially nothing in 120 min, and neat amorphous ledipasvir nothing detectable in 2 h saboo2019 R5; saboo2020a R13; indulkar2019a R1; que2019 R2. Against these, normalized drug and polymer rates of congruent dispersions were statistically equal (for example 2.13 against 2.35 mg/min/cm² for 10% nilvadipine, "ns"), and those of incongruent ones differed at p ≤ 0.001 to 0.0001 saboo2019 R5, R8. In ritonavir-saturated medium the 10% tablet released ≈90% by 30 min against ≈99.5% in plain buffer, the 30% tablet ≈40.5% by 120 min against ≈29.5%, and the 50% tablet 0 at every point up to 50 min against ≈0.44% in plain buffer indulkar2019a R12. Nifedipine dispersions in medium already at 110 µg/mL rose to ≈201 (10% HPMC-E5) and ≈211 µg/mL (10% PVPVA) within 5 and 1 min, and to ≈214–218 µg/mL at 60 min indulkar2017b R8, R9. Powders behaved like compacts: nilvadipine powders reached ≈89–91 µg/mL at 5–10% and ≈13.5–15 µg/mL at 15–20%, cilnidipine ≈96–99 up to 15% and ≈0.65–1.33 µg/mL at 20–25% saboo2019 R20, R21; ritonavir powders formed LLPS for the same combinations as films (data not shown) purohit2017b R19. In the papers of 2020–2022 the read-outs agreed where both were taken: at the limits of four phenolphthalein analogues drug and polymer rates differed with p = 0.022–0.67, and powders of six drugs showed the same cliff as compacts que2021 R21; yang2021 C41. The 5-min read-out of indomethacin methyl ester gave a limit of at least 25%, but over 30 min the ester released congruently only at 5% because crystals formed on the surface (F19) saboo2020 C14, R8.

Figure

Red (drug) and black (PVPVA) bars stand together and are marked "ns" up to 10% nilvadipine and 15% cilnidipine; one 5% step higher the drug bar collapses below the break of the y-axis while the polymer bar remains saboo2019 R8, R5, R6.

The 10% curves overlap in both media, while the 50% tablet (inset) shows a small rise in plain buffer and none in saturated buffer indulkar2019a R12.

Why

The presaturated-medium test rests on the dissolution equation of a single solid,

\frac{dm}{dt} = A\,\frac{D}{h}\,\left(C_s - C_b\right)

indulkar2017b E4, which for a drug-controlled surface with C_s = C_{amorph} = C_b gives no driving force indulkar2017b E5. The authors therefore read continued release into a saturated medium as polymer control ("clearly indicate"; "confirming") and zero release as a drug-rich surface "in agreement with" the Nernst–Brunner form dC/dt = (DA/Vh)(C_s - C) indulkar2017b C35; indulkar2019a C43, C42, E2. For the polymer-controlled case they write the drug rate as the polymer rate times the mass ratio, (dC/dt)_{drug} = (dC/dt)_{polymer}\, m_{drug}/m_{polymer}; at 10% ritonavir and 20 min, ≈67 mg PVPVA × 10/90 gives ≈7.4 mg against ≈7.6 mg ritonavir read from the plot indulkar2019a E3, R1, R2. Normalized by weight fraction, the same relation predicts equal drug and polymer rates, which is how congruency is read in the 2020 paper saboo2020a E2, E3. The constant face removes the particle-size and disintegration effects of powders (the authors' stated reason), although packed particle layers present a renewed surface during dissolution indulkar2019a C15; bhardwaj2018 C47, C64. Reading the amorphous-to-crystalline IDR ratio as a solubility ratio is carried over from polymorphs ("should be the same") bhardwaj2018 E3, C45. (synthesis: the drug concentration alone cannot separate the regimes, because a slow polymer-controlled dispersion and a drug-controlled one can give similar drug curves; only the polymer curve or the saturated-medium test shows which component sets the rate.)

Conditions and limits

Compacts of 25–100 mg at 37 °C in 0.1 N HCl (ITZ), 50 or 100 mM pH 6.8 phosphate, or pH 7.4 for EUDS; ritonavir films at 25 °C; nifedipine at 5 °C bhardwaj2018 P7; indulkar2019a P2; saboo2019 P9; saboo2020a X8; purohit2017b P14; indulkar2017b P7. The weight-fraction normalization assumes similar component densities saboo2020a C8. How the medium was presaturated, and whether the starting concentration was subtracted, is not stated indulkar2019a X8; the two nifedipine media received different amounts of dispersion indulkar2017b X10. Some percent releases exceed 100% without a stated basis indulkar2019a X12. The neat PVPVA compact of the ledipasvir study weighed 250 mg against 100 mg for the dispersions, and the compact mass of the neat polymers of the 2020 study is not stated que2019 X13; saboo2020a X10. High-loading HPMCAS and EUDS rates are slopes after a lag saboo2020a X9. In the ritonavir films polymer release was measured only at 10:90 and 50:50 purohit2017b X12. The later papers differ in medium volume, sampling window and limit criterion (5-min rates, at least 80% in 90 min, t-test of rates, drug release alone), so their limits compare only to the loading step saboo2020 P5; que2021 P9; yang2021 P17; hiew2022b P8, X6. In the 2022 ritonavir surfactant paper the axis of the early polymer-release plot is labelled mg/mL but the values behave as milligrams released per vessel indulkar2022 X5. From 2024 the PVP molecular-weight paper counted release as congruent when more than 80% of both components left and took the limit as the highest loading passing 80% of the drug, over 90–120 min, and the bedaquiline paper of 2025 dispersed powders without assaying the polymer deac2024a C36, X6; kaur2025 P11, X14.

Counterexamples and conflicts

None found against the test itself. Papers that measured only the drug inferred the regime: the spray-dried ritonavir study normalized the drug rate by area and loading only, a coated-tablet study did not measure PVPVA, and a fluorescence study read the regime from whether the amorphous solubility was reached li2020 E13; benson2025 X18; tres2018 X10. A congruency test over the first 5 min missed the later failure of a fast crystallizer (F19) saboo2020 X7.

Transfer check

Press compacts of the target dispersion and of the neat polymer and the neat amorphous drug of the same mass into the same die, assay the polymer in the same samples (size-exclusion chromatography, light-scattering detection or colorimetry), and compare normalized rates by a statistical test saboo2019 P10, P20; saboo2020a P7. Run the lowest and the target loading once more in medium presaturated at the amorphous solubility: release that continues is polymer-controlled, release that stops is drug-controlled indulkar2019a R12, C43, C42. Near an expected switch use loading steps finer than 5%, since every limit reported here is a bracket saboo2020a X11; indulkar2019a X7. Run the test for at least 30–60 min, since 5-min rates missed surface crystallization saboo2020 R8. For an enteric polymer, record the buffer capacity of the medium beside its pH, since the HPMCAS rate changed about sixfold between 3 and 51 mM/ΔpH at pH 6.8 (F23) bapat2023 R5. Scrape and assay the gel layer at 10–30 min: a rising drug content of the gel marks incongruent release even while percent release still matches the neat polymer deac2023a R3, C13.

History

Compacts of sulfathiazole–PVP had shown crystalline-drug-, amorphous-drug- and polymer-controlled regimes as loading fell (Simonelli 1969, cited), and normalized ketoconazole release had tracked polymer release at 20% loading (Chen 2016, cited) bhardwaj2018 R22, R23. In the library Indulkar and co-workers (2017) first used a drug-saturated medium to show polymer control, and Purohit and Taylor (2017) first assayed drug and polymer together, on films indulkar2017b C35; purohit2017b C52. Bhardwaj and co-workers (2018) introduced the constant-area compact with both components and loading-normalized rates bhardwaj2018 C65. Indulkar and co-workers (2019) combined both components, neat references and the presaturated medium in Wood's apparatus indulkar2019a C15, C16; Saboo and co-workers (2019) took over the surface-normalized method (builds_on) and added t-tests and powders, and Que and co-workers (2019) and Saboo and co-workers (2020) carried it to ledipasvir and to five polymers saboo2019 R8; que2019 P3; saboo2020a E3. The papers of 2020–2022 kept the method for indomethacin and its ester, four phenolphthalein analogues, felodipine with TPGS, loratadine at two temperatures, clopidogrel, atazanavir and lumefantrine salts, and added powder release of six drugs saboo2020 P5; que2021 P9; saboo2021 P6; yang2021 P15, P12; correasoto2022a P11; correasoto2022b P5; hiew2022b P8. The HPMCAS papers of 2024–2025 kept the constant-area compact with drug and polymer assayed for eight drugs and three grades, while the bedaquiline study of 2025 used powders and drug assay only (F32, F33, F34) bapat2024 P7; bapat2025a P6; kaur2025 P11.

How it was shown

The set-ups are those of F1. Indulkar and co-workers (2017) tested 10 and 20% nifedipine with HPMC-E5 or PVPVA at 5 °C indulkar2017b P1, P7. Purohit and Taylor (2017) followed ritonavir and polymer from 10:90 and 50:50 films of PVP K-90, PVPVA and HPMCAS-MF at a dose of 60 µg/mL drug (540 or 60 µg/mL polymer) purohit2017b P14, A1, A2. Bhardwaj and co-workers (2018) compared spray-dried and melt-extruded ITZ–PVPVA at 50, 25, 20 and 10% bhardwaj2018 A1, A2. Indulkar and co-workers (2019) tested ritonavir–PVPVA at 10, 20, 25, 30, 35, 40 and 50% indulkar2019a A1. Que and co-workers (2019) tested ledipasvir–PVPVA at 2, 5, 7.5, 10 and 20% que2019 A1. Saboo and co-workers tested nilvadipine at 5–20% and cilnidipine at 5–25% in PVPVA (2019) and felodipine in PVP, PVPVA and HPMC at 5–30% and in HPMCAS and EUDS at 10–50% (2020) saboo2019 A2; saboo2020a A2. Bapat and co-workers (2024) used the same constant-area compact for HPMCAS-MF dispersions of eight drugs at 5–60% with drug and polymer assayed, and Bapat and Taylor (2025) for 25% loratadine and cinnarizine in three HPMCAS grades (F32, F33) bapat2024 P7, E1; bapat2025a P6. Deac and co-workers (2024, polymer molecular weight) released clotrimazole from six PVP grades in pH 8.0 buffer and counted release as congruent when both components passed 80% deac2024a P5, C36.

What the data show

Ritonavir: the 10% tablet was essentially done by 30 min (drug ≈99.5%, PVPVA ≈94%), the 20% by 40 min and the 25% by 60–75 min; within this range release slowed with loading (≈77% against ≈46% of the drug at 20 min for 10 and 25%) indulkar2019a R3. In mass the 10% tablet released ≈67 mg PVPVA by 20 min against ≈46 mg for the neat PVPVA tablet indulkar2019a R2. Nilvadipine at 5 and 10% gave drug rates of 2.53 and 2.13 against polymer rates of 2.76 and 2.35 mg/min/cm², cilnidipine at 5, 10 and 15% 2.51, 2.26 and 2.18 against 2.74, 2.41 and 2.36, and every polymer rate from a dispersion was below the 3.49 of neat PVPVA (p ≤ 0.05 to 0.01) saboo2019 R5, R6, R8. Ledipasvir at 2 and 5% gave ≈54 and ≈50% drug with ≈55 and ≈49% polymer at 30 min and about 100% of both by 60 min que2019 R3, R4. ITZ at 10% gave 199 µg/min/cm² from both processes, with normalized rates of ≈2000 (drug) against ≈2265 µg/min/cm² (polymer) for the spray-dried compact and ≈555 against ≈810 at 20%, while the polymer left at about 0.22 (spray-dried) and 0.29 (extruded) of the rate of neat PVPVA bhardwaj2018 R4, R11, R12. Felodipine: PVP gave ≈3.3/≈3.3 (5%) and ≈3.8/≈4.1 mg/min/cm² (10%, drug/polymer), PVPVA ≈2.4/≈2.6 at 5 and 10% and ≈1.9/≈1.9 at 15%, HPMC ≈0.26/≈0.29 and ≈0.14/≈0.17, against neat PVP ≈3.6, PVPVA ≈3.4 and HPMC ≈0.37; HPMCAS stayed congruent at ≈0.82–0.93 up to 30% and EUDS at ≈0.07–0.16 up to 50% (see F11) saboo2020a R14, R13. Nifedipine dispersions released completely in under 5 min (PVPVA) and about 20 min (HPMC-E5) at 5 °C, also into drug-saturated water indulkar2017b R8, R9, C23. From 10:90 ritonavir films the polymer reached ≈500 of 540 µg/mL PVP by 10 min and ≈565 (PVPVA) and ≈550 µg/mL (HPMCAS) by about 5 min, while ritonavir reached ≈83, ≈116 and ≈113 µg/mL purohit2017b R15, R16, R17. (synthesis: the drug gain over the neat amorphous drug is 2.13/0.002 ≈ 1100-fold for 10% nilvadipine, 3.8/0.0008 ≈ 4800-fold for 10% felodipine–PVP and 199/51 ≈ 3.9-fold for 10% ITZ in 0.1 N HCl; the polymer from the dispersion ran at 2.35/3.49 = 0.67 (nilvadipine 10%), 2.36/3.49 = 0.68 (cilnidipine 15%) and ≈2.6/≈3.4 ≈ 0.76 (felodipine–PVPVA 10%) of neat PVPVA.) The papers of 2020–2022 add IND at 5 and 10% (drug and polymer about 3.44/3.47 and 3.14/3.19 mg/min/cm², against about 0.002 for amorphous IND) and INDester at 5–25% (about 3.3–4.2) saboo2020 R3; the four analogues at their limits (about 2.8–3.7 for drug and polymer) que2021 R21; felodipine with 5% TPGS at 10–30% (about 2.6–3.2) saboo2021 R9; loratadine–PVPVA at 20 and 30% (about 3.33/3.37 and 2.70/2.88) and loratadine–HPMCAS at 5–20% (about 0.40–0.61) yang2021 R21, R18; clopidogrel at 40% (about 4.5 for the drug and 5.2 for the polymer) correasoto2022a R18; ATZ at 5% (about 100% of both by 90 min) correasoto2022b R1; and lumefantrine free base at 5–35% (about 3.15–3.9 mg/min/cm² for the drug) hiew2022b R12. The 2023 HPMCAS study adds IND at 10% (drug and polymer about 0.12/0.10, 0.24/0.23, 0.55/0.54, 0.80/0.78 and 0.82/0.76 mg/min/cm² at 3, 5, 13, 26 and 51 mM/ΔpH, and 0.51/0.49 in FaSSIF) and INDester at 10% (about 0.13/0.11 to 0.57/0.59) bapat2023 R9, R13. Neat amorphous IND gave about 0.40 against 0.80 for IND from the dispersion, a comparison per unit of drug in the compact; per unit area the dispersion released about 0.08 mg IND/min/cm², about five times less than the neat amorphous drug bapat2023 R9, X14. In HPMCAS-MF the polymer left indomethacin dispersions at about 0.84 (5%) to 0.51 mg/min/cm² (60%) beside drug rates of 0.85 to 0.50, and indomethacin methyl ester dispersions at about 0.67 to 0.37 beside 0.66 to 0.39, against about 0.91 for the neat polymer; at 25% the ritonavir, loratadine, clotrimazole and cinnarizine dispersions released drug and polymer at about 0.40 and 0.43, 0.10 and 0.12, 0.05 and 0.08, and 0.03 and 0.05 bapat2024 R2, R3, R4. Loratadine dispersions in HPMCAS LF, MF and HF released drug and polymer at about 0.25 and 0.27, 0.10 and 0.12, and 0.01 and 0.02 at pH 6.8 bapat2025a R2. Clotrimazole–PVP released about 86–88% of the drug and 103% of the polymer (K25, 20%), 87–93% of both (K29/32, 30%), 96 and 101% (K60, 35%) and, at 120 min, 83 and 96% (K90, 38%), and neat PVP and the congruent dispersions released more slowly as molecular weight rose deac2024a R1, R3, C19.

Figure

The red (ritonavir) and blue (PVPVA) curves lie on top of each other and both reach about 100% by 30 min indulkar2019a R3.a–b.

Why

The authors' explanation is that at low loading there is too little drug to form an intact surface layer, so that the dissolving polymer brings the drug into solution ("is thought to occur") indulkar2019a C25. For ritonavir they add that congruent release "suggests" a drug–polymer interaction that persists during hydration and hinders amorphous–amorphous phase separation (AAPS), and that dissolution is faster than AAPS, keeping the interface composition constant indulkar2019a C36, C37. The classic two-component theory does not predict this: in Higuchi's diffusion model only one mixture ratio releases both components at once,

\frac{N_{drug}}{N_{polymer}} = \frac{D_{drug}\,C_{s(drug)}}{D_{polymer}\,C_{s(polymer)}}

saboo2020a E4, and with an amorphous solubility of 8 µg/mL, a PVPVA solubility of 46 mg/mL and a diffusion-coefficient ratio of 10 that ratio is about 1.7 × 10⁻³ for felodipine–PVPVA (printed 1.7 × 10⁻⁶) saboo2020a R27, X1. Congruent release over a range of polymer-rich compositions therefore contradicts the model, as Corrigan had noted (cited) indulkar2019a C23, C24; saboo2020a C36, C37. Saboo and co-workers (2020) go further: because the PVPVA limit varies from 5 to 25% with the drug and HPMCAS and EUDS stay congruent to 50%, the amount of drug alone "doesn't seem to hold true" as the cause, and water-induced phase separation at the interface must be considered (see F8) saboo2020a C38. The slower polymer in dispersions is put down "most likely" to the higher hydrophobicity of the formulation or to drug–polymer interactions, as seen before for ketoconazole (cited) saboo2019 C13; bhardwaj2018 C49, R23. The non-linear 10% ITZ profiles "might" reflect initial hydration and swelling of the polymer bhardwaj2018 C20. (synthesis: since the drug follows the polymer, the polymer's own dissolution rate in the medium caps polymer-controlled release; the gain over the amorphous drug is then large where the amorphous drug alone is very slow (nilvadipine, felodipine at pH 6.8) and small where it is relatively fast (ITZ in 0.1 N HCl).) Bapat and co-workers (2023) take the equal rates to mean that the step limiting polymer dissolution also limits drug release ("expected"), and generalize to HPMCAS dispersions of acidic and neutral drugs ("demonstrate"), although drug release was measured at 10% loading only bapat2023 C35, C49, X4. Bapat and co-workers read matching drug and polymer rates in HPMCAS as polymer control ("indicates"), and Deac and co-workers describe the release mechanism of a congruent dispersion as "similar to that of neat glassy polymers" bapat2024 C10; deac2024a C31. (synthesis: the HPMCAS-MF series of 2024 shows the coupling without a cliff up to 60% for drugs that do not bind the polymer, the clotrimazole–PVP series shows the polymer's own rate, set by its molecular weight, capping the coupled release (F29), and for drugs that complex HPMCAS the coupled rate falls with the polymer's to about zero at 25–30% (F32).)

Conditions and limits

Shown for PVPVA with six drugs, for PVP, HPMC and EUDS with felodipine, for HPMC-E5 with nifedipine and for HPMCAS with felodipine and ritonavir; non-disintegrating compacts at 37 °C, nifedipine at 5 °C, ritonavir films at 25 °C; drug loadings of 2–25% (to 50% for HPMCAS and EUDS) saboo2020a A2; indulkar2017b P7; purohit2017b P14. No disintegrating tablet with excipients was tested saboo2020a C53; bhardwaj2018 C63.

Counterexamples and conflicts

Congruent does not mean fast: 10% felodipine–HPMC released ≈0.14 mg/min/cm², about a sixth of 10% felodipine–HPMCAS (≈0.82) saboo2020a R14. At 25% ritonavir the drug trailed the polymer mid-release (≈60 against ≈73% at 30 min) although the tablet is classed as congruent indulkar2019a X7. From 50:50 ritonavir–PVPVA films the polymer took about 2 h to dissolve, against "less than 10 min" in the text purohit2017b X4. Outside the subcategory a peptide-enhancer–PVPVA dispersion released its polymer about twice as fast as the neat polymer, the opposite of the slowing seen here bapat2025b C13. In the papers of 2020–2022 the polymer left congruent compacts at about its neat rate or faster, up to about twice for clopidogrel (F15) que2021 R21; correasoto2022a R18. Congruent release at a limit raised by a surfactant was about half as fast as at lower loadings saboo2021 X3. In the slowest HPMCAS-HF systems of 2025 the drug left at about half the polymer's rate bapat2025a X6.

Transfer check

Measure the neat polymer and the neat amorphous drug in the same die and medium; compacts at 5 and 10% loading should give equal normalized drug and polymer rates (no significant difference) and continued release into presaturated medium saboo2019 R8; indulkar2019a R12. A drug whose neat amorphous rate is within about an order of magnitude of the polymer's (as for ITZ in acid) gains little from polymer control; one that is three orders slower gains most bhardwaj2018 R4; saboo2019 R7. Check the polymer fraction of its neat rate: values of 0.2–0.3 (ITZ) mean the drug is already slowing the polymer bhardwaj2018 R12, C49.

History

Simonelli and co-workers (1969, cited) described polymer-controlled release of sulfathiazole–PVP at very low loading indulkar2017b R14. In the ASD dissolution chapter Alonzo and co-workers (2011) proposed polymer control for 90:10 dispersions from particle disintegration, without measuring polymer release (Supersaturation from ASD dissolution F10) alonzo2011 C37, C38. Indulkar and co-workers (2017) showed it for nifedipine through release into saturated medium indulkar2017b C35. Bhardwaj and co-workers (2018) found the normalized drug and polymer rates converging at 10–20% ITZ bhardwaj2018 C51. Indulkar and co-workers (2019) named the regime congruent release and wrote the rate relation indulkar2019a C1, E3; Saboo and co-workers (2019) called it the "sweet spot" and showed equal rates by t-test; Saboo and co-workers (2020) extended it to five polymers saboo2019 C30; saboo2020a C16, C17. Que and co-workers (2021) found that at the limit the polymer controls release whatever the compound, so that drug release can be predicted from polymer release que2021 C39, C41, and Correa Soto and co-workers (2022) found the polymer released faster than from a neat compact (F15) correasoto2022a C30; the second half of the title therefore now holds for the papers of 2018–2020 only. Bapat and co-workers (2023) showed for HPMCAS that the polymer-paced drug rate follows the medium's buffer capacity (F23) bapat2023 C18. Bapat and co-workers (2024) extended the drug-and-polymer assay to eight drugs in HPMCAS-MF, Bapat and Taylor (2025) to three HPMCAS grades, and Deac and co-workers (2024) to six PVP grades, where the coupled rate followed the polymer's molecular weight bapat2024 C23; bapat2025a C13; deac2024a C19.

How it was shown

As in F1 and F2, with the loadings above the switch: ITZ at 20 (extruded), 25 and 50% bhardwaj2018 A1; ritonavir tablets at 30–50% and ritonavir films at 20–50% with PVP and PVPVA indulkar2019a A1; purohit2017b A2; ledipasvir at 7.5, 10 and 20% que2019 A1; nilvadipine at 15–20%, cilnidipine at 20–25% saboo2019 A2; felodipine at 15–30% in PVP, PVPVA and HPMC saboo2020a A2.

What the data show

ITZ: 53 ± 4 (spray-dried) and 47 ± 2 µg/min/cm² (extruded) at 50%, 64 ± 1 and 55 ± 1 at 25%, about 17–23 times crystalline ITZ against about 18 for melt-quenched ITZ; the extruded 20% compact (48.8 ± 0.9) stayed on this plateau bhardwaj2018 R4, R6. The loading-normalized rates show the polymer ahead: ≈395 against ≈100 µg/min/cm² at 50% and ≈880 against ≈250 at 25% for the spray-dried series, and after 20 min the 50:50 compacts had released about 5.0 (spray-dried) and 12.2 (extruded) mg/cm² PVPVA against about 1.0 mg/cm² ITZ bhardwaj2018 R11, R8. Ritonavir: at 40% the drug rose from ≈0.15% at 10 min to ≈2.95% at 180 min and PVPVA from ≈2.2 to ≈13.2%; at 50% ≈0.1 to ≈0.65% and ≈2.05 to ≈6.2%; in mass both stayed below ≈0.5 mg in 120 min, like the neat amorphous tablet indulkar2019a R3.k–n, R1. The 50% tablet released 0 at every point up to 50 min into saturated buffer against ≈0.155–0.44% in plain buffer indulkar2019a R12. Ledipasvir at 7.5 and 10% released no drug up to 60 min and at most ≈7% by 90 min, with ≈17 and ≈23% polymer; the 20% dispersion released no drug and ≈19% of its polymer by 90 min que2019 R3, R4, R11, X6. Nilvadipine at 15 and 20% gave 0.035 and 0.013 mg/min/cm² drug against 0.72 and 0.52 polymer, cilnidipine at 20 and 25% 0.006 and 0.002 against 0.24 and 0.21 saboo2019 R5, R6. Felodipine with PVP fell to ≈0.020, ≈0.010 and ≈0.005 mg/min/cm² at 15, 20 and 30% while PVP still dissolved at ≈1.7–1.9, and the medium stayed below 1 µg/mL against an amorphous solubility of 8 µg/mL saboo2020a R14, R6, R1. In 50:50 ritonavir films PVP reached ≈57 of 60 µg/mL within 5 min and PVPVA ≈65 near 120 min, while ritonavir reached ≈18 (PVP) and ≈9 µg/mL (PVPVA) by 240 min; the free drug of the non-LLPS PVP and PVPVA films was 10–25 µg/mL after 4 h purohit2017b R15, R16, R10. (synthesis: relative to the neat amorphous drug, the drug-controlled rates are 0.035/0.002 ≈ 17 and 0.013/0.002 ≈ 6.5 for nilvadipine and 0.020/0.0008 = 25 to 0.005/0.0008 ≈ 6 for felodipine–PVP, but 53/51 ≈ 1.0 and 64/51 ≈ 1.3 for ITZ.) The papers of 2020–2022 add IND at 15, 20 and 25% (drug about 0.04, 0.02 and 0.01, PVPVA about 0.72, 0.64 and 0.50 mg/min/cm²) saboo2020 R3; TIPP at 15 and 20% and Me-TIPP at 50% (about 0% drug and 13–18% PVPVA in 90 min) que2021 R15; miconazole–PVPVA at 50% (about 2% drug against 97% polymer in 60 min) and elbasvir–PVPVA at 50% (about 5 against 34%) yang2021 R13; ATZ at 10% (about 0.07 against 0.24 mg/min/cm²) correasoto2022b R3; clopidogrel at 60% (about 0.2 against 4.1) correasoto2022a R18; and the benzoate salt of lumefantrine at 30%, which released about 20% of its PVPVA before any drug hiew2022b R13. (synthesis: IND at 15–25% released 0.01/0.002 = 5 to 0.04/0.002 = 20 times faster than amorphous IND alone.) Miconazole at 20% in the enteric polymer HP-50 released about 8.6% of its drug by 60 min and 11.7% by 120 min while the polymer reached about 23.3 and 42.2%, which the authors call incongruent release qi2022 R14, C22. Ritonavir–PVPVA compacts of the 2022 imaging study released about 35% of the drug against 69% of the polymer by 90 min at 30%, and at most about 3% of the drug with about 14 and 10.5% of the polymer at 40 and 50%, while neat amorphous ritonavir released at most about 0.5% yang2022a R2, X1, X2. In 2023–2024 incongruent release showed the same pattern: clotrimazole–PVP K12 at 5% released about 107% of its polymer and 35% of its drug, K17 at 5% about 101–103% and 74–77%, K29/32 at 33% about 103% and 77%; ritonavir–PVPVA with Span 20 at 40% about 78% and 8%, with SDS at 40% about 70–79% and 48–54%, and with Tween 80 at 50% about 85% and 22% deac2024a R1; yang2023 R2, R3.

Figure

Points near the amorphous-ITZ value (about 18) at 50 and 75% polymer are drug-controlled; the spray-dried point at 80% polymer (about 40) and both points at 90% (about 72) lie far above it bhardwaj2018 R6.

Why

The authors' common reading is that the hydrophilic polymer dissolves faster than the drug, so the surface loses polymer and becomes drug-rich, and the drug then dissolves on its own bhardwaj2018 C51; indulkar2019a C39; saboo2019 C38. A drug-controlled surface dissolves by

\frac{dx}{dt} = C\,(S - x)

with S the amorphous solubility, so the rate falls to zero as the solution approaches S and the concentration cannot pass it ("can be rationalized") purohit2017b E3, C50. Indulkar and co-workers state the consequence for the colloid: LLPS "can only occur" when components that dissolve faster than the drug control release (see F4) indulkar2019a C48. Bhardwaj and co-workers call the drug-rich outer layer a "supposition", which also explains the curving 50:50 polymer profiles, the polymer having to pass through a thickening drug-rich layer bhardwaj2018 C52. That ritonavir at 40–50% released at about the rate of the neat drug "suggests" rapid AAPS and early surface enrichment indulkar2019a C39, and the gap of more than three orders between neat PVPVA and amorphous nilvadipine "explains" the size of the drop when the mechanism changes saboo2019 C43. The authors describe the drug-controlled rates as close to the amorphous drug saboo2019 C12; saboo2020a C16. (synthesis: the measured drug rates lie up to 25 times above a flat compact of neat amorphous drug; a drug-rich layer that is porous or still holds some polymer would give such values, but neither the area nor the composition of the layer was followed over time, see F7.) (synthesis: in the K12 dispersion the polymer left completely while the drug stayed behind as droplets agglomerated in the gel, a route to polymer-first release that needs no continuous drug-rich layer (F29) deac2024a R10, C26.)

Conditions and limits

PVPVA (all drugs), PVP and HPMC (felodipine, ritonavir films); compacts at 37 °C except the ritonavir films at 25 °C; ITZ in 0.1 N HCl, where the weak base is relatively soluble bhardwaj2018 M1. The zero release into saturated medium was shown for one loading of one drug, with the presaturation and baseline not described indulkar2019a X8.

Counterexamples and conflicts

High loading is not drug-controlled with every polymer: 50:50 ritonavir–HPMCAS films released drug with the polymer (≈32–34 µg/mL within 5 min) and formed LLPS, and felodipine–HPMCAS stayed congruent to 50% (see F11) purohit2017b R17, C38; saboo2020a R14. In the ASD dissolution chapter a 50:50 PVP danazol dispersion dissolved as fast as a 90:10 one jackson2015 X9. Praziquantel–HPMCAS showed a gradual decline with loading rather than a switch, read as polymer control polyzois2024 C17. Partly incongruent release of low-Tg drugs still passed the amorphous solubility: loratadine–PVPVA at 35% released about 23% of its drug by 90 min against an amorphous-solubility line of about 3.2%, and clopidogrel–PVPVA at 50% about 52% with the polymer about twice as fast (F4) yang2021 R19, C15; correasoto2022a R17, R18.

Transfer check

Compare the drug rate of the target loading with a compact of the neat amorphous drug in the same die: within about one order of magnitude, and with the polymer leaving several times faster, the dispersion is drug-controlled and will not exceed its amorphous solubility saboo2019 R5, X9; purohit2017b C50. Confirm with a run in medium presaturated at the amorphous solubility indulkar2019a R12, C42.

History

The amorphous-drug-controlled regime at intermediate loading goes back to sulfathiazole–PVP compacts (Simonelli 1969, cited) bhardwaj2018 R22. In the ASD dissolution chapter Alonzo and co-workers (2011) proposed a drug-rich layer at the dissolving surface of 50:50 dispersions, without observing it alonzo2011 C46. Purohit and Taylor (2017) first measured the polymer leaving 50:50 ritonavir films ahead of the drug, Bhardwaj and co-workers (2018) the amorphous-drug plateau of ITZ, and Indulkar and co-workers (2019) the zero release into saturated medium purohit2017b C36; bhardwaj2018 C48; indulkar2019a C42; Saboo and co-workers (2019, 2020) and Que and co-workers (2019) added nilvadipine, cilnidipine, felodipine and ledipasvir saboo2019 C12; saboo2020a C16; que2019 C31. Saboo and co-workers (2020) and Que and co-workers (2021) added hydrogen-bonding IND and phenols, Yang and co-workers (2021) six drugs in powders, and Hiew and Taylor (2022) lumefantrine salts, whose lag-phase compacts released polymer before drug saboo2020 C12; que2021 C22; yang2021 C13; hiew2022b C18. Deac and co-workers (2024) found polymer-first release already at 5% with short PVP chains and named its cause Failure Mode I (F29) deac2024a C38.

How it was shown

Saboo and co-workers (2020) dissolved 100 mg compacts (0.5 cm², 1500 psi, 100 rpm, 100 mL, 37 °C) of indomethacin (IND)–PVPVA in pH 2.1 buffer and of indomethacin methyl ester (INDester)–PVPVA in pH 6.8 buffer, each with HPMC-606 in the medium, sampled every minute for 5 min and normalized each component's rate by area and weight fraction, R = kV/(Sx) saboo2020 P5, E1, M5, M6. Que and co-workers (2021) dissolved compacts of four phthalide analogues at their limits in 50 mM phosphate buffer (pH 6.8; pH 2.5 for TIPP) and compared drug, polymer and neat-polymer rates by t-tests at 99% confidence que2021 P9, P10. Yang and co-workers (2021) dissolved loratadine compacts with PVPVA or HPMCAS-MF and the neat polymers at 37 and 5 °C, and polymer and drug from powders yang2021 P15, P12, R18, R21. Saboo and co-workers (2021) dissolved a drug-free PVPVA matrix containing 5% TPGS saboo2021 P7. Correa Soto and co-workers (2022) dissolved clopidogrel (CPD)–PVPVA compacts at 40, 50 and 60% with and without 5 g surfactant per 100 g, with neat PVPVA as reference, and normalized by R = k/(Sx); in the atazanavir (ATZ) study they dissolved neat PVPVA containing 10% SDS or a Span correasoto2022a P10, P11, E1; correasoto2022b R22. Indulkar and co-workers (2022) compared the PVPVA released in the first 20 min from 30:70 ritonavir–PVPVA tablets with 5 parts of SDS, Tween 80, TPGS, Span 20 or Span 85 per 100 parts of drug and polymer with that of a binary 25:75 tablet at the limit, taken from the 2019 paper indulkar2022 P1, P3, P2, R20. Qi and Taylor (2022) dissolved 100 mg compacts of the enteric polymer HPMCP (HP-50) alone and of its 20% miconazole dispersion in 100 mL 50 mM pH 6.8 phosphate at 37 °C and 100 rpm, with the polymer assayed colorimetrically qi2022 P10, P11, P12, A6. Bapat and co-workers (2023) compared HPMCAS released from 10 and 50% indomethacin and indomethacin methyl ester dispersions with neat HPMCAS-MF in pH 6.8 phosphate of 3 to 51 mM/ΔpH and measured the water uptake of the powders at 95% RH and 37 °C for 12 h bapat2023 A1, A3, P11.

What the data show

IND–PVPVA at 5 and 10% released drug and polymer at about 3.44/3.47 and 3.14/3.19 mg/min/cm² against about 3.4 for neat PVPVA in that medium, and INDester at 5–25% at about 3.3–4.2 against about 3.8 saboo2020 R3, X5. At their limits TIPP (10%), Me-TIPP (30%), PHPH (5%) and Me-PHPH (40%) released PVPVA at about 3.2, 3.2, 2.8 and 3.4 mg/min/cm² against about 3.0 for neat PVPVA (p = 0.37–0.48 against neat) que2021 R21. Loratadine–PVPVA compacts at 20 and 30% gave PVPVA at about 3.33 and 2.70 against about 3.45 for neat PVPVA; with HPMCAS the 5% compact released polymer faster than neat HPMCAS (about 0.61 against 0.47 mg/min/cm²) yang2021 R21, R18. CPD–PVPVA binary compacts gave PVPVA at about 5.2 (40%), 6.7 (50%) and 4.1 mg/min/cm² (60%) against about 2.6 and 2.7 for neat PVPVA in two panels; with SDS and CTAB the 40% compacts kept about 5.4 and 5.2, with TPGS and Tween 80 about 2.1 and 3.0 correasoto2022a R18. Neat PVPVA with 10% SDS dissolved at almost twice the neat rate, with Spans unchanged, and a drug-free PVPVA matrix with 5% TPGS released at about 3.2 against 3.4 correasoto2022b R22; saboo2021 R14. In 10% ATZ–PVPVA compacts with additives the PVPVA rate ran from about 1.7 (TPGS) to 5.1 mg/min/cm² (CTAB) against about 2.7 for neat PVPVA correasoto2022b R3. At high loading the drug held the polymer back: elbasvir–PVPVA powders at 50% released about 34% of their polymer in 60 min against about 97% for miconazole–PVPVA at 50% yang2021 R13. (synthesis: polymer rate from the dispersion over the neat rate: 3.19/3.4 ≈ 0.94 for 10% IND, 3.4/3.0 ≈ 1.1 for 40% Me-PHPH, 6.7/2.7 ≈ 2.5 for 50% CPD, against 2.35/3.49 = 0.67 for 10% nilvadipine in F2.) In the ritonavir surfactant tablets the SDS dispersion had released about 35.4 against 14.0 units of PVPVA for the binary tablet at the limit by 10 min and 48.8 against 26.2 by 15 min, while Span 20, Span 85 and TPGS followed the binary curve (about 12.7–15.0 at 10 min) and Tween 80 lagged (about 6.6) indulkar2022 R20, C34. The values are amounts not normalized to the larger polymer content of the 25:75 tablet, and the SDS ratio falls from about 2.5 at 10 min to about 1.2 at 20 min indulkar2022 X5, X6. With an enteric polymer the drug slowed the polymer: HP-50 left its 20% miconazole dispersion at about 23.3% by 60 min and 42.2% by 120 min against about 28.4 and 60.3% for neat HP-50 compacts in the same medium, a comparison the authors do not comment on qi2022 R14, R9. With HPMCAS both drugs slowed the polymer, most at high buffer capacity and high loading: at 51 mM/ΔpH HPMCAS left at about 1.01 mg/min/cm² neat, 0.76 and 0.58 from 10% IND and INDester and 0.58 and 0.45 from the 50% dispersions, while at 3 mM/ΔpH all lay between 0.04 and 0.16 bapat2023 R11, R12. In relative terms the slowing at 3 mM/ΔpH was not smaller (dispersion over neat about 0.25–0.69), and the INDester-over-IND ranking held only at 26–51 mM/ΔpH bapat2023 X3, X12. Water uptake after 12 h fell from about 6.2% (neat) to 5.1 and 4.6% (10% IND, INDester) and 3.9 and 2.4% (50%) bapat2023 R20. A plasticizer alone sped PVPVA: drug-free PVPVA compacts with 10% w/w glycerol tributyrate or tributyl citrate dissolved at about 4.8 and 6.1 against 2.7 mg/min/cm², and plasticized atazanavir dispersions at 20% released drug at about 4.6–5.1 with PVPVA at 3.8–4.2 (F26) correasoto2023 R13, R6. In 2023 MePHPH–PVPVA compacts released drug and polymer together at a total of about 4.4, 6.4 and 4.3 mg/min/cm² at 5, 20 and 40%, and 5% PHPH at about 5.5, against 3.7 for neat PVPVA, with the shortest gel residence time (1.9 min) at 20% MePHPH deac2023a R4, C20. PVPVA blends with 12.5 wt% surfactant dissolved at about 3.6 (neat), 3.8 (Span 85), 3.6 (Span 20), 5.7 (Tween 80) and 6.9 mg/min/cm² (SDS) yang2023 R5, X9. Neat PVP compacts released more slowly as molecular weight rose (profiles in an unsupplied SI) deac2024a R3. In HPMCAS-MF at 25% and pH 6.8 the polymer left at about 0.43 (ritonavir), 0.12 (loratadine), 0.08 (clotrimazole) and 0.05 mg/min/cm² (cinnarizine) against 0.90 for the neat polymer, and at about 0.70 (loratadine, pH 7.5) and 0.28 (clotrimazole, pH 9.0), where these drugs are un-ionized bapat2024 R4, R5. Neat HPMCAS LF, MF and HF dissolved at about 1.50, 0.90 and 0.45 mg/min/cm², and a 25% cinnarizine dispersion left HF too slowly to measure bapat2025a R1, R3.

Figure

In panels d–f the blue PVPVA bars of the binary dispersions stand above the neat-PVPVA bar at the right of each panel, while the TPGS and Tween 80 bars fall back to about its height correasoto2022a R18.

All bars lie near 3 mg/min/cm², with p-values for drug against polymer and for dispersion polymer against neat polymer que2021 R21.

Why

A glassy polymer dissolves by water ingress into a plasticized gel layer and chain disentanglement, which is faster above its Tg; the papers therefore expect the polymer rate to depend on the water the gel layer takes up and on the Tg of the hydrated matrix relative to the test temperature saboo2020 C54, C55; correasoto2022b C33, C36; yang2021 C30. Correa Soto and co-workers (2022) state that CPD lowers the dispersion Tg (data in the SI) and read the faster PVPVA as a plasticization effect, as 20 wt% PEO doubled the dissolution rate of PMMA ("is thought to") correasoto2022a R26, R30, C48. Yang and co-workers find it "feasible" that a little low-Tg loratadine plasticizes HPMCAS, while more low-polarity drug "would be anticipated" to reduce water entry and slow the polymer yang2021 C31, C32. Que and co-workers rationalize the opposite case: a drug–polymer hydrogen bond that persists in water occupies polymer solvation sites, cuts water uptake and slows disentanglement ("could explain"; see F17), whereas without specific interactions release is "agnostic" to the compound and drug release can be predicted from polymer release que2021 C44, C41. Hiew and Taylor (2022) quantify the water a glassy polymer needs before its chains disentangle with the critical polymer volume fraction,

v_p^{*} = \frac{1/\rho_p}{1/\rho_p + \frac{T_g - T}{\beta/\alpha_f}\cdot\frac{1}{\rho_s}}

hiew2022b E1, which for PVPVA at 37 °C gives 0.89, at least 11% water hiew2022b R21, C40. SDS, which interacts with PVPVA, sped the neat polymer, and the non-ionic surfactants that interacted with CPD slowed both components, which the authors call "likely unfavorable" correasoto2022b R22, C43; correasoto2022a C56, C57. (synthesis: the ceiling of polymer-controlled release in F2 is the rate of the polymer in the dispersion; a low-Tg drug or SDS can raise it above the neat rate and a non-ionic surfactant or a high-Tg drug at high loading can lower it, so the neat rate bounds the dispersion only roughly.) Indulkar and co-workers (2022) suggest that SDS "perhaps" speeds release by forming micellar aggregates on the PVPVA chains, a complex that "may" dissolve faster than the polymer alone, so that release outruns phase separation; SDS–PVPVA complexation was not measured in that paper indulkar2022 C33. They group the effect by charge: ionic SDS "may" act on the polymer and raise the overall release rate, the non-ionic surfactants on drug–polymer phase separation indulkar2022 C56, X11. Qi and Taylor (2022) explain the slower enteric polymer by a lipophilic drug lowering the water fraction of the gel layer that chain disentanglement needs, which the higher hydration of a pre-ionized polymer salt "likely" helps to offset (F24) qi2022 C40. Bapat and co-workers (2023) attribute the slower HPMCAS "most likely" to the drug lowering the amount and rate of water the polymer takes up, which leaves a more viscous gel from which the polymer escapes less easily; at low buffer capacity the polymer's solubility at the acidic interface still dominates, so the drug matters less there (F23) bapat2023 C46, C21. Gel hydration in the medium was not measured; the uptake came from humid air bapat2023 X9. Correa Soto and co-workers (2023) read the faster plasticized PVPVA through free-volume theory: a plasticizer raises the fractional free volume of the hydrated gel, chain mobility and so the disengagement rate (Devotta, cited), which they find "reasonable to suggest" as one mechanism; chain mobility was not measured (F26) correasoto2023 C40, C41, C39. Deac and co-workers (2023) propose that the congruent rate is set by water penetration and chain disentanglement, both faster up to 20% MePHPH because the gel is thinner and the low-Tg drug (onset 22.5 °C) plasticizes the polymer, and slower above it as the drug's lipophilicity lowers the driving force for water uptake ("may be due") deac2023a C73, C50, C51, C53. The residence times printed match the dry gel mass rather than the total gel weight deac2023a X3. Yang and co-workers suggest that faster water uptake, plasticization and, for SDS, a soluble SDS–PVPVA complex "could" explain the faster polymer, none measured on the blends yang2023 C49. For HPMCAS the 2024–2025 papers attribute the slowing by cationic drugs to an insoluble ion-pair complex that lowers the polymer's interfacial solubility (F32) bapat2024 C32; bapat2025a C34.

Conditions and limits

PVPVA except one HPMCAS series; compacts of about 100 mg at 37 °C; IND at pH 2.1 and TIPP at pH 2.5, so their neat-polymer references ran in other media saboo2020 X5; que2021 X5. How the rates of the analogue study were derived and how many runs lie behind them is not printed que2021 X6, X13. The CPD dispersion Tg is in an SI that was not supplied correasoto2022a R26. The ternary CPD and ATZ loadings are drug:polymer ratios with 5 g surfactant on top correasoto2022a X9; correasoto2022b X12. The surfactant blends held 12.5 wt% surfactant, more than the 9.1 wt% of the ternary dispersions of the same paper yang2023 X9.

Counterexamples and conflicts

In the 2018–2019 papers the polymer left congruent compacts more slowly than from a neat compact (0.22–0.29 for 10% itraconazole, 0.67–0.68 for the dihydropyridines) (F2) bhardwaj2018 R12; saboo2019 R5, R8. TPGS raised the felodipine limit although it did not speed PVPVA in a drug-free matrix, so its gain did not come from a faster polymer (F21) saboo2021 R14, C17, X15.

Transfer check

Dissolve a neat polymer compact in the same die and medium as the dispersion and compare the polymer rate of a low-loading compact with it: a ratio above 1 points to plasticization by the drug or an additive, one well below 1 to a drug or additive that slows the polymer correasoto2022a R18; saboo2019 R8. Repeat with each candidate surfactant, since non-ionic surfactants slowed PVPVA and SDS sped it correasoto2022a R18; correasoto2022b R22. For an enteric polymer, compare dispersion and neat polymer at the buffer capacity of use, since the drug's effect on HPMCAS grew with buffer capacity in absolute terms bapat2023 R12, C48.

History

Bhardwaj and co-workers (2018) and Saboo and co-workers (2019) found the polymer slowed in congruent dispersions, "most likely" by the hydrophobicity of the formulation or drug–polymer interactions bhardwaj2018 C49; saboo2019 C13. Saboo and co-workers (2020) and Que and co-workers (2021) found it at about the neat rate with indomethacin, its ester and four analogues saboo2020 R3; que2021 C39. Yang and co-workers (2021) saw a little loratadine speed up HPMCAS yang2021 C31. Correa Soto and co-workers (2022) were the first in the library to report PVPVA from a dispersion at about twice its neat rate ("Surprisingly") correasoto2022a C30, and a companion study showed SDS doubling neat PVPVA correasoto2022b R22. Hiew and Taylor (2022) brought in the critical polymer volume fraction hiew2022b E1. Indulkar and co-workers (2022) found the same SDS acceleration of PVPVA inside ritonavir dispersions, with Tween 80 slowing the polymer indulkar2022 C32, C34. Qi and Taylor (2022) extended the observation to an enteric polymer, which miconazole slowed qi2022 R14, R9. Bapat and co-workers (2023) found that the slowing of HPMCAS by a drug depends on the buffer capacity of the medium bapat2023 C21, C25. Correa Soto and co-workers (2023) sped neat PVPVA with plasticizers that are not surfactants correasoto2023 C18. Deac and co-workers (2023) showed the polymer-paced rate rising and then falling with loading below the limit deac2023a C22. Yang and co-workers (2023) measured surfactants in PVPVA blends, Deac and co-workers (2024) the effect of PVP molecular weight, and Bapat and co-workers (2024) and Bapat and Taylor (2025) the slowing of HPMCAS by weak bases and its dependence on grade yang2023 C6; deac2024a C19; bapat2024 C14; bapat2025a C12.

How it was shown

Qi and Taylor (2022) asked what limits the dissolution of hydroxypropylmethyl cellulose phthalate HP-50 (phthalic acid 1.61 mmol per gram, pKa about 4.2) at pH 6.8 qi2022 M1, C1. They compressed 100 mg into an 8 mm die (0.5 cm²) and rotated it at 100 rpm in 100 mL buffer at 37 °C, n = 3, assaying the polymer by the phenol–sulfuric acid method, in pH 6.8 phosphate with K, NH4, Cs, Na or tetrabutylammonium as cation, in 50 mM BIS-TRIS, Tris and TEA buffers (calculated pKa 6.5, 8.1 and 10.2, against 7.2 for phosphate), and in 50 and 5 mM phosphate, whose buffer capacities were measured as 29 and 3 mM/ΔpH qi2022 P10, P12, A1, A2, A4, M16, M17, M23, M24, M25. Bapat and co-workers (2023) used the same compact and speed for HPMCAS-MF (pKa 4.92 or 5.0 cited, about 18,000 Da), its triethylamine salt and 10 and 50% dispersions with indomethacin (IND, ionized at pH 6.8) or its methyl ester (INDester, neutral), assaying the polymer by HPLC with light-scattering detection calibrated in each medium bapat2023 M3, M4, A3, P7, P8. They varied the buffer capacity of pH 6.8 phosphate from 3 to 51 mM/ΔpH (5 to 100 mM), compared FaSSIF with its blank buffer, FeSSIF-V2 (pH 5.8), FeSSIF V1 (pH 5.0), phosphate at pH 7.5 and 9.0 and glycine at pH 9.0, and doubled the ionic strength with NaCl bapat2023 P14, A1, A2, P12. They imaged the gel layer of partly dissolved compacts with prodan by confocal microscopy, read the pH of gel scraped from compacts containing chlorophenol red or bromothymol blue against a calibration in solution, and ran the rotation speed from 100 to 700 rpm bapat2023 P9, P10, A4.

What the data show

HP-50: with five phosphate cations of 182–415 pm about 22.6–35.7% of the polymer dissolved in 60 min, the authors seeing little effect of cation size; with different buffer species at the same pH it was about 44.8% (BIS-TRIS), 28.4% (phosphate), 6.6% (Tris) and 1.8% (TEA) qi2022 R1, R2, C13. At 3 against 29 mM/ΔpH the rate fell from about 0.92 to 0.14 mg/min/cm², the "6-fold reduction" of the paper qi2022 R12, C4. HPMCAS-MF in pH 6.8 phosphate: about 0.16, 0.26, 0.59, 0.89 and 1.01 mg/min/cm² at 3, 5, 13, 26 and 51 mM/ΔpH, levelling at high buffer capacity ("may be approaching a plateau"); ionic strength 98 against 196 mM made no difference bapat2023 R5, C9, R6. FaSSIF (pH 6.5, 12 mM/ΔpH) gave about 0.56 and its blank buffer 0.57; FeSSIF-V2 (pH 5.8, 25 mM/ΔpH) about 0.58 against 0.89 in pH 6.8 phosphate of similar capacity; FeSSIF V1 (pH 5.0) dissolved no polymer bapat2023 R5, R7, R8. Phosphate at pH 7.5 (18 mM/ΔpH) gave about 1.09, phosphate at pH 9.0 (2 mM/ΔpH) about 0.67, glycine at pH 9.0 about 0.67 and 0.99 at 25 and 51 mM/ΔpH bapat2023 R5. A gel layer formed only from 13 mM/ΔpH, 3 ± 1 µm thick there and 19 ± 5 µm at 51 mM/ΔpH; at 5 mM/ΔpH the surface kept its grain boundaries bapat2023 R14. The gel pH read 5.6 ± 0.2, 6.0 ± 0.1 and 6.2 ± 0.2 in pH 6.8 phosphate of 13, 26 and 51 mM/ΔpH and 6.3 ± 0.1 in pH 7.5 phosphate, against rates of about 0.60, 0.90, 1.01 and 1.09; the pre-ionized salt read 6.7 ± 0.1 with about 1.62 (F24) bapat2023 R15, R16. The rate rose linearly with the square root of the rotation speed in every medium, to about 2.9–3.3 times the 100 rpm value at 700 rpm, without a plateau bapat2023 R18, R19. The drug followed: from the 10% IND dispersion IND and HPMCAS left at about 0.12/0.10 at 3, 0.55/0.54 at 13 and 0.80/0.78 mg/min/cm² at 26 mM/ΔpH and 0.51/0.49 in FaSSIF bapat2023 R9. (synthesis: 1.01/0.16 ≈ 6.3 for neat HPMCAS over 3–51 mM/ΔpH and 0.92/0.14 ≈ 6.6 for HP-50 over 3–29 mM/ΔpH; in the usual 50 mM phosphate (26 mM/ΔpH) HPMCAS ran 0.89/0.16 ≈ 5.6 times faster than at 3 mM/ΔpH.) Neat HPMCAS-MF released at about 0.91 in 50 mM and 1.00 mg/min/cm² in 100 mM pH 6.8 phosphate and the 25% loratadine dispersion at about 0.10 and 0.12 and 0.25 and 0.29 (drug and polymer), and neat HPMCAS-MF at about 0.63 in 50 mM pH 9.0 glycine buffer against 0.90 at pH 6.8 bapat2024 R6, R4. Neat HPMCAS LF, MF and HF dissolved at about 1.50, 0.90 and 0.45 mg/min/cm² at pH 6.8 and 1.73, 1.09 and 0.63 at pH 7.5 bapat2025a R1. Bedaquiline dispersions released about 17–19.5% at 10% in pH 6.5 buffer and 87.5–97% in FaSSIF-V1 kaur2025 R13, R15.

Figure

The grey pH 6.8 phosphate squares rise steeply from 3 to 26 mM/ΔpH and level off; FaSSIF and its blank buffer overlap at 12 mM/ΔpH; FeSSIF-V2 at pH 5.8 lies low and pH 7.5 phosphate highest bapat2023 R5.

Four HPMCAS points rise with gel pH from 5.6 to 6.3; the pre-ionized salt stands alone at gel pH 6.7 with the highest rate; the gel pH is an average over gel scraped from the compact bapat2023 R16, X5, X6.

Why

Both papers take the framework of Nguyen and Fogler (cited) for carboxylic polymers: a polymer insoluble at low pH must ionize before its chains dissolve, ionization releases protons at the surface, and the protons have to leave by diffusion, by meeting hydroxyl ions or on buffer species that take them up at the surface and release them in the bulk qi2022 C11, C24; bapat2023 C33, C36. The surface pH therefore lies below the bulk pH (about 0.4 units for HP-50 films, cited), and with a polymer pKa near 4–5 and HPMCAS solubility rising steeply between pH 5.0 and 6.5 (cited) a small drop cuts the polymer's solubility at the surface qi2022 R18, C23; bapat2023 C38, M3. In the diffusion-controlled regime the flux is

J_p = \frac{D_p}{\delta}\, C_p

with C_p the polymer solubility at the interface and \delta the boundary layer, which at a rotating disk follows

\delta = 1.612\, D_p^{1/3}\, \nu^{1/6}\, \omega^{-1/2}

so that diffusion-controlled release rises with the square root of rotation speed bapat2023 E2, E3. A buffer with more species, or a pKa that lets it take up a proton at the surface pH, lifts the interfacial pH: BIS-TRIS (6.5) carries protons better than phosphate (7.2), while Tris and TEA, mostly protonated at both surface and bulk pH, "cannot act as proton carriers" qi2022 C26, C27; bapat2023 C39. The authors read the gel pH series as "strong" support, and the HPMCAS salt, which releases no protons, as further support (F24) bapat2023 C39, C41. Because release rose with √rpm without the plateau that marked disentanglement control for Eudragit (cited), and because HPMCAS-MF lies below the entanglement molecular weights estimated for other polysaccharides (about 22–34 kDa), they consider disentanglement control "unlikely" bapat2023 R25, R27, C45. For HP-50 the absence of a cation-size effect makes a percolation model "likely not appropriate" qi2022 C29. Surface pH is also said to set the hydration of the gel layer and hence disentanglement qi2022 C30. (synthesis: in the polymer-controlled regime of F2 the polymer's rate caps drug release; for an enteric polymer that cap is set by the medium as well as by the polymer, so the same dispersion can release several times faster in 50 mM phosphate than in a weak buffer; in the crystallization chapter the least succinoylated HPMCAS grades released posaconazole poorly at low pH for the same reason, release being polymer-controlled (Crystallization from solution F44) shah2025 C17.) Bapat and co-workers put the slow neat polymer in pH 9.0 glycine down to the low buffer capacity of that buffer at the acidic surface, and place three loratadine results on a gel-pH axis (6.0, 6.2, 6.3), only the first value with a stated source bapat2024 C13, R15, X6.

Conditions and limits

HP-50 and HPMCAS-MF compacts at 37 °C and 100 rpm (100–700 rpm in one series) in phosphate, amine, glycine and maleate buffers; no bicarbonate buffer was used, and the FaSSIF the authors call physiological has 12 mM/ΔpH against the cited 2–8 mM/ΔpH of fasted intestinal fluid qi2022 P10; bapat2023 P7, X8, R24. The surface pH, the proton flux and the polymer solubility at the interfacial pH were not measured; the gel pH is an average over scraped gel, near the upper limit of the dye for the salt, and the cited HPMCAS solubility data stop at pH 6.5 qi2022 X16; bapat2023 X5, X6. The buffer capacities of the amine buffers and the composition of the 77 mM/ΔpH medium are not printed, and the medium of the HPMCAS salt is labelled 26 or 51 mM/ΔpH in different figures qi2022 X12; bapat2023 X2, X1. Drug release was measured at 10% (HPMCAS) and 20% (HP-50) loading only bapat2023 X4; qi2022 A6.

Counterexamples and conflicts

Cation size changed HP-50 dissolution little, but sodium (about 28.4% at 60 min) lay below potassium, ammonium and cesium (33.5–35.7%), and the cation concentrations of that series differed qi2022 R1, X11. At pH 9.0, phosphate of only 2 mM/ΔpH still gave HPMCAS about 0.67 mg/min/cm², faster than pH 6.8 phosphate of 3–5 mM/ΔpH, so bulk pH acts beside buffer capacity bapat2023 R5, C15. The rotation data are linear in √rpm but rise above the Levich prediction at 700 rpm (about 2.9–3.3 against 2.65) bapat2023 X10. The stop of HPMCAS release near a gel pH of 5 is an extrapolation of four points bapat2023 R17, X13. In the ASD dissolution chapter the gastric pH and the enteric polymer grade decided the release of delamanid and praziquantel dispersions after a pH shift (Supersaturation from ASD dissolution F25), and SDS dissolved far more HPMCAS from suspended particles at pH 3 (Supersaturation from ASD dissolution F15) chen2015a R18. For a drug that complexes HPMCAS the micelles of FaSSIF-V1 and FeSSIF-V2 changed release several-fold, so the absence of a FaSSIF effect holds for the polymer and non-complexing drugs only (F34) kaur2025 R13, R15, C49. In 100 mM phosphate loratadine complexation also fell from about 70 to 48% at equal pH, a second route for the buffer effect that the authors do not discuss bapat2024 R16, X11.

Transfer check

For a dispersion in HPMCAS, HPMCP or another enteric polymer, run the constant-area release test in pH 6.8 phosphate at 3–5 mM/ΔpH as well as at 26 mM/ΔpH, with the polymer assayed, since the usual 50 mM buffer ran HPMCAS about 5.6 times faster than a 3 mM/ΔpH buffer (synthesis above) bapat2023 R5, R24. A run at several rotation speeds that rises with √rpm without a plateau places the limit in diffusion of the polymer, not in disentanglement bapat2023 C30. Compare FaSSIF with its blank buffer before attributing a medium effect to micelles bapat2023 R7, C20.

History

Lower surface pH at dissolving enteric coatings was known from coating work (Ozturk 1988, Harianawala 2002, cited), and the Nguyen and Fogler model (2005, cited) tied carboxylic polymer dissolution to interfacial solubility and proton carriers qi2022 C7, R18, C11. In the library the critical ionization of HP-50 (about 83%) and the HPMCAS-MF pKa came from a lumefantrine study of the Miscibility chapter qi2022 R19; bapat2023 M3. Qi and Taylor (2022) were the first in this subcategory to vary buffer species and capacity for an enteric dispersion polymer qi2022 C14, C4. Bapat and co-workers (2023) added the gel pH, the rotation test and drug release that follows the polymer across media bapat2023 C51, C30, C49. Bapat and co-workers (2024), Bapat and Taylor (2025) and Kaur and co-workers (2025) took the gel pH of 6.0 in 50 mM pH 6.8 phosphate as the pH at which to measure complexation of HPMCAS by weak bases (F32, F33) bapat2024 C9; bapat2025a C28; kaur2025 C44, and Kaur and co-workers found that micelle-containing fluids change release where the drug complexes the polymer (F34) kaur2025 C49. The title's statement on FaSSIF was narrowed after these papers to the neat polymer and drugs that do not complex it (synthesis).

How it was shown

Purohit and Taylor (2017) took the solutions of fresh films after 4 h of dissolution and measured free ritonavir in the supernatant after ultracentrifugation (35000 rpm, 30 min), added pyrene (0.4 ppm) to 1 µm-filtered solutions, imaged and sized particles by NTA and checked for crystals by polarized light purohit2017b P16, P17, P12, P18. Indulkar and co-workers (2019) compared the released amount with the amorphous solubility line (≈3.3 mg in 100 mL), noted turbidity and sized the 10% medium at 30 min by DLS indulkar2019a R1, R9, P6. Saboo and co-workers (2019) sampled the medium of each compact at about 1 h for 4-Di-2-Asp fluorescence (a blue shift marks a drug-rich environment), DLS and NTA saboo2019 P12, C9. Que and co-workers (2019) recorded cloudiness and sized 3 mL aliquots by DLS from 10 to 60 min (PDI below 0.3) que2019 R5, P9. Indulkar and co-workers (2017) sized the nanodroplets of four nifedipine dispersions by NTA at 5 °C and followed extinction at a non-absorbing wavelength indulkar2017b P6, R10. Saboo and co-workers (2020) compared felodipine concentrations with its amorphous solubility of 8 µg/mL, on unfiltered samples saboo2020a R6, R8, X12.

What the data show

Ritonavir films: free drug after 4 h was 27–30 µg/mL for every HPMCAS film and 28 µg/mL for 10:90 and 20:80 PVP and 10:90 PVPVA, all at the amorphous solubility of 26.8–28.5 µg/mL, against 10–25 µg/mL for the other PVP and PVPVA films; pyrene gave 1.50–1.61 (HPMCAS), 1.63 and 1.66 (PVP 10:90, 20:80) and 1.53 (PVPVA 10:90) against 1.88–1.90 for the rest purohit2017b R10, R1, R11. The LLPS films spiked to apparent concentrations of ≈130 (10:90 PVP), ≈116 (10:90 PVPVA) and ≈119 µg/mL (10:90 HPMCAS); the spikes decayed (for example to ≈17 and ≈21 µg/mL at 240 min) while the free drug stayed near 28 µg/mL and no crystals formed purohit2017b R12, R13, R14, R10, R18. Fresh 10:90 films gave mean droplet sizes of 200 (PVPVA), 250 (HPMCAS) and 315 nm (PVP) purohit2017b R8. Ritonavir tablets: 10% reached ≈3.5 mg in 100 mL by 10 min against the ≈3.3 mg line, the 30% tablet crossed between 30 and 40 min, the 35% tablet reached only ≈1.5 mg by 120 min, and the 10–25% media were visibly turbid indulkar2019a R1, R20, R9. Nilvadipine and cilnidipine: emission maxima of about 560–565 nm for 5 and 10% of both drugs and ≈570 nm for 15% cilnidipine against ≈593–595 nm in buffer and at higher loadings; DLS 237 ± 8 and 246 ± 3 nm (nilvadipine 5, 10%) and 255 ± 1, 300 ± 2 and 366 ± 7 nm (cilnidipine 5, 10, 15%); NTA scattering at these loadings and none above saboo2019 R10, R11, R12. Powders gave turbid media at 5–10% nilvadipine and clear ones at 15–20% saboo2019 R20. Ledipasvir at 2 and 5% turned the medium cloudy above an amorphous solubility below 1 µg/mL, with droplets of ≈225–257 nm over 10–60 min que2019 R5, R12. Nifedipine dispersions at 5 °C gave ≈190–222 nm droplets and extinction that rose once the amorphous solubility was passed indulkar2017b R7, R10. Felodipine from PVP at 5 and 10% reached ≈35 and ≈67 µg/mL at 60 min against 8 µg/mL, and every HPMCAS loading exceeded 8 µg/mL saboo2020a R6, R8. IND at 5 and 10% passed its amorphous solubility of 22.1 µg/mL between 25 and 30 and between 15 and 20 min, the media held 250–350 nm particles and were turbid without birefringence, while 15–25% stayed at 0.1–0.6% release without particles; every INDester medium exceeded 3.9 µg/mL, also after crystallization saboo2020 R6, R7, R12, R8. Felodipine with 5% TPGS reached about 68–165 µg/mL at 10–35% against 9.5 µg/mL, with particles to the limit and none at 40% saboo2021 R11, R16. The analogue media were turbid at or below the limits que2021 R18. Loratadine–PVPVA compacts released about 23% at 35% (line about 3.2%) with droplets and about 0.2% at 40% without, and loratadine–HPMCAS compacts at 30–40% stayed above the line while the polymer ran ahead yang2021 R19, R16, C14, C15. Clopidogrel dispersions at 40 and 50% passed the LLPS concentration within 10 min, the 50% binary although PVPVA ran ahead correasoto2022a R16, R18. With ATZ only the SDS and CTAB dispersions released enough to form colloids correasoto2022b R8, C8. Clotrimazole–PVP dispersions passed the amorphous solubility of 10.3 µg/mL in every release, with a turbid medium, and loratadine from HPMCAS LF reached 17.6 µg/mL at pH 6.8 and 59.1 µg/mL at pH 7.5 after 1 h against amorphous solubilities of 9.0 and 7.6 µg/mL, which the authors did not discuss deac2024a R4, M1; bapat2025a R5, X9.

Figure

Spectra of the congruent loadings peak near 560–565 nm, blue-shifted from the buffer curve near 595 nm; the incongruent loadings lie on the buffer curve saboo2019 R10.

The 10–25% curves pass the dashed line within minutes, the 30% curve after 30–40 min; 40%, 50% and neat amorphous ritonavir stay at the bottom indulkar2019a R1, R20.

Why

The authors' reading is that a polymer-driven release continues past the amorphous solubility because the polymer, not the drug, sets the rate, after which the excess drug separates as drug-rich droplets, as they had shown for nifedipine (F5) indulkar2019a C49, R24. A drug-controlled release slows toward zero as the drug approaches its amorphous solubility and "never" produces LLPS indulkar2019a C48; purohit2017b C50. Purohit and Taylor generalize that colloids form where release is "congruent or close to congruent" purohit2017b C52. The decay of the apparent concentration after the spike is put down to nanodroplets that absorb and scatter UV light and coalesce, since no crystals were seen and the free drug stayed at the amorphous solubility purohit2017b C39. The larger droplets at higher cilnidipine loading "could indicate" lower stability against agglomeration or an effect of loading on the formation route saboo2019 C19. An alternative reading of the dihydropyridine result is that the droplets were absent above the switch simply because the concentration at 1 h was far below the amorphous solubility saboo2019 X14; the two readings agree on the outcome and differ only on whether congruency carries information beyond the release extent. (synthesis: in every system of these papers the colloid appeared when, and only when, the released drug passed the amorphous solubility; congruent release is the way a dispersion of a slowly dissolving amorphous drug gets there within the test time.)

Conditions and limits

Ritonavir films at 25 °C and compacts at 37 °C in pH 6.8 phosphate; nifedipine at 5 °C; DLS at one loading for ritonavir tablets; colloid not measured for felodipine purohit2017b X18; indulkar2019a X5; saboo2020a X12. The NTA images of the ritonavir films do not state the loading, and the pyrene table does not state whether the films were fresh or humidity-exposed purohit2017b X8, X6.

Counterexamples and conflicts

The 20:80 cilnidipine powder rose to ≈0.70–1.33 µg/mL, above its amorphous solubility of 0.6 µg/mL, for most of 4 h without reported LLPS saboo2019 X4, R21. In the ASD dissolution chapter the colloid shed by 90:10 felodipine–HPMC dispersions was read as crystalline, not as drug-rich droplets (Supersaturation from ASD dissolution F9) alonzo2011 R8, C33. Loratadine–PVPVA at 35% formed droplets with incongruent release (rates about 0.36 against 0.12 mg/min/cm²) yang2021 R21, C15, and clopidogrel at 50% passed its LLPS concentration with PVPVA running ahead correasoto2022a R18, R16; both are low-Tg drugs (F18) (synthesis: in both the drug still left at a substantial fraction of the polymer rate, 0.12/0.36 = 0.33 and 3.6/6.7 = 0.54).

Transfer check

Measure the free drug (ultracentrifugation or flux) and a colloid signal (DLS, NTA or an environment-sensitive probe) in the same release run, and compare the released amount with the amorphous solubility measured in that medium purohit2017b P16, P17, P12; saboo2019 P12. Expect droplets only at loadings that release congruently and only once the released amount passes the amorphous solubility; a polymer- or a surfactant-specific size (about 200–370 nm here, about 60–130 nm with surfactant, F13) should be checked by sizing at more than one loading que2019 R12.

History

Nanodroplets from dissolving dispersions at low loading were known from the ASD dissolution chapter (Alonzo 2011, Jackson 2015) indulkar2017b C6. Purohit and Taylor (2017) first tied their formation to congruent or near-congruent release and to a matrix that stays mixed, and Indulkar and co-workers (2019) to at least partly polymer-controlled release purohit2017b C52, C6; indulkar2019a C49. Saboo and co-workers (2019) confirmed the cut-off with three read-outs for two drugs saboo2019 C30, and Que and co-workers (2019) stated that colloids form only below the limit of congruency que2019 C10, C55. Saboo and co-workers (2020, 2021) and Que and co-workers (2021) confirmed the link for further drugs saboo2020 C16; saboo2021 C19; que2021 C21; Yang and co-workers (2021) found droplets from incongruent release of loratadine, which revises the "only" of the title for low-Tg drugs yang2021 C15, C2. In 2024–2025 clotrimazole–PVP and loratadine–HPMCAS dispersions passed the amorphous solubility as well, the latter without comment by the authors deac2024a R4, C18; bapat2025a X9.

How it was shown

Indulkar and co-workers (2017) asked whether the droplets form by dissolution of the drug followed by LLPS in solution (mechanism I) or by water-induced AAPS in the wet solid, whose drug-rich domains are released intact (mechanism II, proposed for anacetrapib–PVPVA by Harmon and co-workers, cited) indulkar2017b C7, C8. They added dispersions of ordinary (proteated) nifedipine to D2O saturated at 110 µg/mL with nifedipine deuterated at its four aromatic positions, in four amounts (0.36–7.70 mg/mL of 10% dispersion), and measured the dissolved H- and D-forms by ¹H NMR right after complete dissolution: the methyl peak (2.29 ppm, 6 H) is common to both, the aromatic peaks (7–8 ppm, 4 H) belong to the H-form only, and the droplets give no NMR signal indulkar2017b P8, P9, E1, C11. For PVPVA a maleic acid external standard replaced the methyl reference indulkar2017b E2. Because exchange between droplets and solution would scramble the label under either mechanism, they first mixed a D-form solution (1× amorphous solubility) with H-form droplets (3×) made by solvent shift in D2O with 1 mg/mL HPMC-E5 and followed the D-share at 5–25 °C indulkar2017b P5, C13. The amorphous solubility was measured by UV extinction at 450 nm with 1 mg/mL HPMC-E5 in D2O, and the wet Tg of amorphous nifedipine was estimated from its dry Tg and an extrapolated water uptake indulkar2017b P2, P3, P4. Purohit and Taylor (2017) dipped ritonavir films (10:90 and 50:50) in buffer for 10 s, dried them and imaged them by confocal microscopy (nile red, brighter in drug-rich regions) and SEM, and sized the droplets of dissolving films by NTA purohit2017b P9, P13, P12. Yang and co-workers (2022) fused 100 mg ritonavir–PVPVA compacts with Nile red and Alexa Fluor 488 to a glass dish, sealed the top so that buffer entered from the edge only, and took confocal images every 30 s at 37 °C for 15–30 min, with manual agitation that the authors state makes rates incomparable between samples yang2022a P7, C31.

What the data show

At 36 µg/mL the four dispersions gave ≈78.1, ≈69.4, ≈77.8 and ≈78.6% D (10% HPMC-E5, 20% HPMC-E5, 10% PVPVA, 20% PVPVA), at 110 µg/mL ≈48.1, ≈44.5, ≈49.4 and ≈55.8%, at 330 µg/mL ≈30.4, ≈25.6, ≈24.2 and ≈27.3%, and at 770 µg/mL ≈11.4 and ≈17.6% for the HPMC-E5 dispersions, the PVPVA signal overlapping the aromatic peaks at that dose indulkar2017b R11, R12. The expected values of Table 1 equal 100 \times 110/(110 + c_H) with c_H the H-form added (extractor arithmetic: 75.3, 50.0, 25.0, 12.5), while the red lines of the figure sit at ≈57.9, ≈29.1 and ≈14.4% for the three higher doses without explanation indulkar2017b X1. In the exchange control the D-share fell from the starting 50% to ≈41.1, ≈39.9, ≈39.3, ≈30.2 and ≈26.7% at 5, 10, 15, 20 and 25 °C, full exchange giving 25%, and at 5 °C 29 points from ≈5 to ≈61 min stayed between ≈39.5 and ≈47% indulkar2017b R3, R4. About 2.5% of the H-form was estimated to sit on droplet surfaces, which explains only a small part of the early ≈9-point drop indulkar2017b R5, X3. The amorphous solubility rose from 110 µg/mL at 5 °C to 156 µg/mL at 25 °C, and the estimated wet Tg was about 17 °C (dry 42 °C, 2.5% water) indulkar2017b R1, M1, E6. Ritonavir: after 10 s in buffer the 10:90 PVP film was uniform while the 50:50 PVP film showed dark domains of about 1–3 µm in a bright drug-rich phase; SEM showed smooth 10:90 surfaces purohit2017b R4, R9. After humidity the domains were about 1–3 µm by confocal microscopy and 5–10 µm by AFM, against droplets of 200–315 nm purohit2017b R3, R5, R8. Ritonavir–PVPVA, real time: at 10% small red droplets formed near the gel–solution interface from the first minute and entered the solution, not throughout the gel; at 20% the same, with droplets starting closer to the water front; cross-sections at 10% showed droplets of 500 nm to 2 µm only at the compact surface yang2022a R10, R11, R6. Dissolving MePHPH–PVPVA compacts formed droplets below 1 µm starting 155 ± 43, 57 ± 14 and 45 ± 20 µm from the glass–gel boundary at 5, 20 and 30%, nearer the core as loading rose and less than the gel thickness of the release set-up, and some were seen leaving into the solution; at 5% PHPH the onset lay at 601 ± 68 µm against a 356 ± 18 µm gel deac2023a R6, R8, R4, C27.

Figure

In a medium saturated with the deuterated drug, mechanism I gives mixed droplets and the ordinary drug in solution; mechanism II gives droplets of the ordinary drug and a solution that stays all deuterated indulkar2017b C11.

Bars near the solid red lines and far below the dashed line at 100% support mechanism I; compare the bars with the Table 1 values (75, 50, 25, 12.5%), since the drawn lines differ from them at three doses indulkar2017b R11, X1.

Why

The authors' reasoning: if the drug dissolved first it mixes with the deuterated drug and lowers the D-share to a dose-dependent value; if droplets left the solid intact, a saturated medium gives them no reason to dissolve and the solution stays all deuterated indulkar2017b C11. Because exchange is slow at 5 °C, they read the ordinary drug in solution as dissolved dispersion drug that replaced the deuterated drug, not as drug released from droplets indulkar2017b C27. They attribute the slow exchange to droplets below their wet Tg ("presumably" faster exchange above it; glassy droplets "appear" to slow it), with the wet Tg taken from the rule

T_{g,wet} \approx T_{g,dry} - 10\ ^{\circ}\mathrm{C} \times w

(w in % water), 42 - 10 \times 2.5 = 17 °C indulkar2017b C18, E6. Their conclusion is that the droplets form by dissolution followed by LLPS ("strongly suggest"), that mechanism II cannot be taken as general for PVPVA dispersions ("we cannot concur"), and that drug–PVPVA dispersions that stay miscible in water "are expected" to follow mechanism I indulkar2017b C26, C37, C38. Purohit and Taylor "find no evidence" of a matrix origin: low-loading films stay uniform after 10 s, and matrix domains are an order of magnitude larger than the droplets, so their release "cannot account for" the nanosized species purohit2017b C56, C57. (synthesis: both tests concern dispersions that release congruently; the regime in which mechanism II was proposed, a matrix that separates on wetting, is the incongruent regime of F3 and F8, where no droplets formed at all in these papers.) Yang and co-workers (2022) propose that drug-rich and polymer-rich domains form together by water-induced separation where the gel's water content brings the drug–polymer–water mixture past the binodal, which is at the gel–solution interface, and that this separation, at the surface in contact with the bulk solution, "would presumably" meet the conditions of the isotope test; for them separation in the gel and LLPS in the solution are parts of one process on one three-component phase diagram ("In reality") yang2022a C34, C35, C48, C50, C51. They set this against drug-rich domains formed throughout the matrix and stabilized by surfactant (Harmon and co-workers, cited) and against the "particle and glue" picture of phase-separated lopinavir films (Li and Taylor 2019, ASD process), since their droplets formed only at the surface yang2022a C48, C49. (synthesis: the 2017 test showed that the droplet drug had mixed with the solution, the 2022 images that droplets form within the outermost wet gel; both concern low loadings where release is congruent, and neither measured the composition of the forming droplets.) Deac and co-workers (2023) infer that during congruent release the drug is "carried" into the diffusion layer by the swelling and dissolving polymer, separating either inside the gel (MePHPH) or after leaving it (PHPH at 5%), and they cite the isotope result as placing droplet formation in a high-mobility environment, in solution or in a low-viscosity region near the interface deac2023a C47, C3.

Conditions and limits

Nifedipine at 10 and 20% with HPMC-E5 or PVPVA at 5 °C only; ritonavir films at 25 °C indulkar2017b A2, A4; purohit2017b P14. The amorphous solubility was measured only for the H-form with 1 mg/mL HPMC-E5, while the NMR samples held up to about 7 mg/mL polymer and the D-form indulkar2017b X6, X7. The 17 °C wet Tg is an estimate from a rule taken from indomethacin indulkar2017b X8. Diffusion of the D-form into the solid dispersion is excluded by inference indulkar2017b X9. The real-time ritonavir set-up wetted the compact from its edge under manual agitation, and the binodal composition and water profile of the gel were not measured yang2022a C31, X10. The onset distances of 2023 come from gently mixed in situ imaging and are compared with gel thicknesses measured at 100 rpm deac2023a X4.

Counterexamples and conflicts

Harmon and co-workers (2016, cited, not a card of the library) found mechanism II for anacetrapib–PVPVA and suggested it may be general for PVPVA indulkar2017b C8; purohit2017b C51. The ritonavir images resolve about 300 nm, comparable with the droplets, so nanoscale domains in low-loading films are not excluded purohit2017b X11, C25. Saboo and co-workers (2021) still describe the origin as unresolved, matrix domains released intact (cited) against phase separation of dissolved drug, "not necessarily" exclusive, and their humid-film domains were larger than the 200–300 nm particles in solution saboo2021 C46, C47.

Transfer check

The isotope test needs droplets that exchange slowly with the solution, so it suits a drug whose water-saturated drug-rich phase is glassy at the test temperature (nifedipine about 17 °C at 5 °C); for a drug with a low wet Tg such as ritonavir (about 0 °C) the premise would have to be checked first with droplets made from the dispersion itself indulkar2017b C13, X5; indulkar2019a M1. For any drug, imaging of the wetted surface at better than 200 nm resolution would be needed to exclude matrix-born droplets purohit2017b X11.

History

Mechanism II was proposed by Harmon and co-workers (2016, cited) indulkar2017b C7, C8. Indulkar and co-workers (2017) and Purohit and Taylor (2017) answered it in the same year for nifedipine and ritonavir indulkar2017b C26; purohit2017b C57, and the 2019 copovidone study builds on the nifedipine result for polymer-assisted dissolution followed by LLPS indulkar2019a R24. Saboo and co-workers (2019) note that the effect of loading on the formation route "has not been widely explored and is under debate" saboo2019 C19. Saboo and co-workers (2021) left open whether TPGS lets molecular drug release first or keeps matrix-born droplets from ripening saboo2021 C52. The temperature dependence of the nifedipine amorphous solubility adds a steady rise from 5 to 25 °C to the ritonavir and efavirenz profiles of the LLPS chapter indulkar2017b R1, C15. Yang and co-workers (2022) imaged the forming droplets in real time and placed them at the gel–solution interface, reconciling the isotope result with water-induced separation at that interface yang2022a C50, C51; this adds the last clause to the title. Deac and co-workers (2023) found both places of formation in two analogues: inside the gel for the methyl ether, beyond it for phenolphthalein deac2023a C27, C31.

How it was shown

Yang and co-workers (2021) followed the z-average and PDI by DLS over 60 min for 2.5% elbasvir and danazol and 10% miconazole and loratadine dispersions dissolving in buffer, the PVPVA dispersions also in buffer containing 400 µg/mL HPMCAS yang2021 P16, A6. Correa Soto and co-workers (2022) made CPD droplets by solvent shift into pH 6.8 phosphate with or without surfactant, held them at 37 °C and 200 rpm and sized them hourly for 4 h by DLS and NTA in triplicate; they measured the zeta potential at 70 µg/mL CPD (about 50 µg/mL as droplets), estimated the surfactant needed for a monolayer from droplet size and literature molecular areas, and sized the media of dissolving dispersions at 90 min correasoto2022a P6, P7, E2, P16. Que and co-workers (2021) sized the 30-min media of the analogue dispersions by DLS and imaged them by NTA, and Saboo and co-workers (2021) the 60-min media of felodipine dispersions que2021 P11, P12; saboo2021 P9, P10. Indulkar and co-workers (2022) sized the droplets in the medium of dissolving 30:70:5 ritonavir–PVPVA–surfactant tablets by DLS from the time the ritonavir concentration passed its amorphous solubility to 90 min, single values without error bars, and viewed the precipitates of the Tween 80 and Span 20 tablets under crossed polarizers indulkar2022 P7, P8, X12.

What the data show

Elbasvir: z-average about 670, 900, 940, 1110 and 1190 nm at 5–60 min from the PVPVA dispersion in buffer, PDI about 0.32 → 0.57, crystals after 30 min; about 280 → 390 nm (HPMCAS dispersion) and 250 → 300 nm (PVPVA dispersion with dissolved HPMCAS), PDI below about 0.21 yang2021 R3. Loratadine: about 820 → 2120 nm (PDI 0.07 → 1.0) from PVPVA against about 420 → 570 nm from HPMCAS yang2021 R6. Danazol: the PVPVA dispersion with dissolved HPMCAS gave about 200 nm at 10 min and about 2780 nm at 60 min, the HPMCAS dispersion about 360–390 nm; miconazole: all three series about 330–370 → 520–560 nm yang2021 R4, R5. CPD in buffer: size about 1.5 times the start after 1 h and about 12 times after 2 h, PDI about 4.2–4.6 times; with SDS (50 µg/mL), TPGS (4) and Tween 80 (5) size ratios about 1.01 → 0.92, 0.99 → 0.96 and 0.89 → 0.81 over 4 h; CTAB (50) about 12 after 1 h with floating agglomerates correasoto2022a R6. More surfactant gave smaller starting droplets, about 520 nm in buffer against 280 nm with 4 µg/mL TPGS correasoto2022a R5. Zeta potential: about −16 mV in buffer, −69 with SDS, +29 with CTAB and −1 to −5 mV with the non-ionics correasoto2022a R7. The calculated monolayer coverage was 0.3–17.8 (SDS), 2.4 (CTAB), 10.4 (TPGS) and 3.1 µg/mL (Tween 80) against effective levels of 50, none, 4 and 5 µg/mL correasoto2022a R24. In dissolution media at 90 min the 40% dispersions gave about 790 nm (binary), 510 (Tween 80), 580 (TPGS), 690 (SDS) and over 2000 nm (CTAB), the 50% dispersions about 1960, 780, 890, 1260 and over 2000 nm correasoto2022a R19. Analogues: 179 ± 8 (TIPP), 193 ± 5 (Me-TIPP), 348 ± 16 (PHPH) and 364 ± 5 nm (Me-PHPH), PDI below 0.3 que2021 R22. Felodipine: about 276–279 nm at 10–20% and 237–270 nm at 30–35% with 5% TPGS, about 278 nm at 10% without saboo2021 R17. Ritonavir: without surfactant the droplets grew from about 940 nm at 20 min to 2200 nm at 90 min; with SDS about 1180 nm when release levelled off at 20 min and about 2400–2470 nm later; with TPGS about 1320 nm at 10 min and 3840 nm at 90 min; with Span 20 about 550 → 1410 nm and with Span 85 about 330 → 960 nm; the Tween 80 precipitate was spherical droplets of about 10 µm, too large for DLS, and the Span 20 precipitate held needle crystals indulkar2022 R11, R13, R14. Clotrimazole droplets made by solvent injection grew from about 540 nm to 1 µm without PVP, from about 340 to 610 nm with K12 and from about 290 to 630 nm with K29/32, at the same normalized rate deac2024a R5. Ritonavir droplets agglomerated in the gel of 30% Tween 80 compacts and entered the medium as chunks at 2 and 5 min yang2023 R10. Bedaquiline precipitated at 100 µg/mL formed particles above 6000 nm without polymer and 413–967 nm with 400 µg/mL HPMCAS, about 98–107 nm in FeSSIF-V2 and 241–1002 nm in FaSSIF-V1 kaur2025 R18.

Figure

Black (PVPVA dispersion in buffer) rises and broadens; red (HPMCAS dispersion) and blue (PVPVA dispersion with dissolved HPMCAS) stay small and narrow yang2021 R3.

Buffer and CTAB bars break the axis; missing bars mean growth that could not be measured, not stable size correasoto2022a R6, X10.

Why

The droplets are metastable and coarsen by ripening, flocculation and coalescence, and the fast coarsening of CPD droplets is "consistent with" the high interfacial energy of a lipophilic drug-rich phase correasoto2022a C2, C37. The low effective surfactant levels are "consistent with" stabilization by surfactant adsorbed at the droplet surface, sterically and/or electrostatically, and the zeta changes "support" adsorption correasoto2022a C39, C13; SDS needs more "most likely" because of its high HLB, and CTAB coarsens the droplets "by an unknown mechanism" correasoto2022a C42, C38. The non-ionic surfactants also partition into the drug-rich phase (aqueous TPGS 50 → 33 ± 5.4 µg/mL, a droplet composition of about 50:17 CPD:TPGS), which the authors judge unrelated to stabilization ("does not appear") correasoto2022a R12, C19, C40. For the polymers, adsorption of HPMCAS on droplet surfaces "has been hypothesized"; charged at the study pH, HPMCAS can add electrostatic to steric hindrance, which neutral PVPVA cannot, and the exact mechanism "clearly requires more investigation" yang2021 C39, C40. Que and co-workers expect colloid size to depend on the compound and excipients ("is likely") and smaller colloids to "potentially" improve absorption through faster reservoir replenishment que2021 C38, C37. (synthesis: because every PVPVA powder test of the 2021 drug-property study was run with 400 µg/mL HPMCAS in the medium to keep grown droplets from being lost on centrifugation, its PVPVA release data already include stabilization by a second polymer yang2021 P12, X18.) For ritonavir Indulkar and co-workers (2022) offer two explanations: the poorly water-soluble Spans "probably" mix into the droplets and lower the chemical potential of ritonavir, which "may" inhibit Ostwald ripening, as a poorly soluble oil did for amorphous felodipine particles (cited), or "may" prevent coalescence by steric repulsion; Tween 80 and TPGS "could possibly" partition into the droplets, lower their Tg and make them coalesce indulkar2022 C49, R25, C50, C48. Droplet composition and Tg were not measured indulkar2022 C48, C49. Deac and co-workers set aside steric stabilization by adsorbed PVP, since neither grade slowed droplet growth in dilute suspension, and propose entanglement in the gel instead ("may be contributing") (F29) deac2024a C44, C45. Yang and co-workers attribute the Tween 80 agglomeration "likely" to Tween 80 accumulating on the droplet surfaces yang2023 C22.

Conditions and limits

Buffers of pH 6.8, 7.5 and 9.0 at 37 °C; droplets made by solvent shift (CPD) or by dissolving dispersions (the others); one level per surfactant in the 4-h series, lower levels in an SI that was not supplied correasoto2022a P6, X10. The DLS series of the drug-property study are single drugs at one loading each, and the third series of two panels is unclear yang2021 A6, X7. Colloid sizes of the analogues were compared at different loadings, Me-PHPH below its limit que2021 X3.

Counterexamples and conflicts

Dissolved HPMCAS did not hold danazol droplets, and miconazole droplets grew alike with either polymer yang2021 X9. CTAB failed although its level was 20 times its calculated monolayer coverage, and TPGS worked below its calculated coverage correasoto2022a R24, X15. TPGS did not change felodipine droplet size, unlike a cited anacetrapib study in which size fell with TPGS saboo2021 C21, R28. TPGS and Tween 80, which held clopidogrel droplets at constant size, let ritonavir droplets grow to about 4 and 10 µm when built into the dispersion, and the Spans, which partitioned into atazanavir droplets, kept ritonavir droplets smallest correasoto2022a R6; indulkar2022 R11, R13; correasoto2022b R17 (synthesis: the surfactants were dissolved with preformed clopidogrel droplets but released from the dispersion together with ritonavir, so the comparison is between set-ups as well as drugs). The abstract of the ritonavir paper names TPGS instead of Span 85 among the smallest droplets, which contradicts its figure and table indulkar2022 X1.

Transfer check

Size the droplets of the target dispersion by DLS every 15–60 min for 1–4 h in the release medium, with the dissolved polymer at the concentration release produces, and check for crystals by polarized light yang2021 P16; correasoto2022a P6. If the droplets grow, test an ionizing polymer (HPMCAS) in the medium or in the dispersion, or 5–50 µg/mL of a non-ionic or anionic surfactant, and avoid a cationic surfactant until it has been tested yang2021 R3; correasoto2022a R6, C43. For a surfactant built into the dispersion, size the droplets in the release medium over the whole test, since the ritonavir ranking of surfactants differed from the clopidogrel ranking obtained with preformed droplets indulkar2022 R11; correasoto2022a R6.

History

Que and co-workers (2019) found smaller colloids with SDS and poloxamer 407 (F13) que2019 R12. Que and co-workers (2021) found the iodinated analogues giving colloids about half the size que2021 C24, and Saboo and co-workers (2021) no size effect of TPGS saboo2021 C20. Yang and co-workers (2021) compared the polymers and ran PVPVA powders with dissolved HPMCAS yang2021 C4, P12. Correa Soto and co-workers (2022) first screened surfactants for 4-h droplet stability and compared the effective levels with monolayer coverage, and propose this screen for selecting surfactants correasoto2022a C12, C41, C43. In the earlier chapters of this category additives decided the size and life of ritonavir and delamanid droplets (Supersaturation from ASD dissolution F12), the polymer decided whether ibuprofen droplets stayed dispersed (Amorphous solubility and LLPS F22), and HPMC-AS kept danazol and enzalutamide droplets small (Crystallization from solution F15); the papers here extend the polymer comparison to droplets released by dispersions. Indulkar and co-workers (2022) sized droplets released by surfactant-containing ritonavir dispersions and found the Spans best indulkar2022 C47; this adds the ritonavir clause to the title. Deac and co-workers (2024) separated droplet growth in solution, unchanged by the PVP grade, from agglomeration in the gel, which the grade decided deac2024a C20, C38.

How it was shown

Each paper dissolved a loading series under the conditions of F1: ritonavir–PVPVA tablets at 10, 20, 25, 30, 35, 40 and 50% indulkar2019a A1; ledipasvir–PVPVA at 2, 5, 7.5, 10 and 20% que2019 A1; nilvadipine at 5–20% and cilnidipine at 5–25% in 5% steps saboo2019 A2; felodipine with PVP, PVPVA and HPMC at 5, 10, 15, 20 and 30% saboo2020a A2; spray-dried and extruded ITZ–PVPVA at 10, 20, 25 and 50% bhardwaj2018 A1. Purohit and Taylor (2017) tested ritonavir films at 10–50% in 10% steps with LLPS, not congruency, as read-out purohit2017b A2, R11.

What the data show

Ritonavir: percent released at 60 min (drug/polymer) was ≈100/≈96.5 (10%), ≈100/≈94 (20%), ≈97/≈100 (25%), ≈29/≈62 (30%), ≈1.7/≈8 (35%), ≈0.8/≈5.5 (40%) and ≈0/≈3.1 (50%) indulkar2019a R11. The 30% tablet lagged for about 15 min, released both components together to about 40 min and then let PVPVA run ahead (≈96 against ≈48.5% at 120 min); the 35% tablet released ≈49% polymer and ≈11% drug by 300 min indulkar2019a R5, R6, R3. The authors sort the tablets into three categories, fast, complete and congruent (10–25%), initially congruent and then incongruent (30%), and slow and incomplete (35–50%, with 35% mixed) indulkar2019a C32. The 30% extent differed about 1.6-fold between two series indulkar2019a X3. Ledipasvir: about 100% of both components by 60 min at 2 and 5%, no drug up to 60 min at 7.5% que2019 R3, R4. Nilvadipine and cilnidipine: polymer/drug ratios about 1 up to 10% and 15%, then about 20, 40, 40 and 105 saboo2019 R5, R6. Felodipine: PVP 3.8 → 0.020, PVPVA ≈1.9 → ≈0.009, HPMC ≈0.14 → ≈0.04 → ≈0.005 mg/min/cm² across the switch, the HPMC drop coming in two steps saboo2020a R14, X11. ITZ: spray-dried 64 (25%) → 113 (20%) → 199 µg/min/cm² (10%), extruded 55 → 48.8 → 199 bhardwaj2018 R4. Ritonavir films formed LLPS up to 20% with PVP and only at 10% with PVPVA purohit2017b R10, R11. (synthesis: the drops across the switch are 2.13/0.035 = 61 and 2.18/0.006 = 363 for nilvadipine and cilnidipine, 3.8/0.020 = 190 and 1.9/0.009 = 211 for felodipine with PVP and PVPVA, but only 113/64 = 1.8 (spray-dried) and 199/48.8 = 4.1 (extruded) for ITZ in 0.1 N HCl.) Limits in PVPVA from the papers of 2020–2022: IND 10% and INDester at least 25% (5-min rates) saboo2020 C14; PHPH 5, TIPP 10, Me-TIPP 30 and Me-PHPH 40% que2021 R19; felodipine 15% without TPGS saboo2021 R12; loratadine 30% in compacts, and drops at 25% (loratadine), 40% (miconazole), 5% (danazol) and 10% (elbasvir) in powders, with paclitaxel and ivacaftor dropping between 5 and 10% yang2021 R22, R7, R8, R9, R10, R11, R12, X26; clopidogrel about 40% correasoto2022a C46; ATZ 5% correasoto2022b C12; lumefantrine free base at least 35% and its eight salts 2–25% hiew2022b R20, X7. The one-step drops were large again: IND 3.14 → 0.04 mg/min/cm² (10 → 15%), ATZ from about 100 to below 3% released (5 → 10%) and lumefantrine sulfate 4.5 → 0.10 mg/min/cm² (2 → 5%) saboo2020 R3; correasoto2022b R1; hiew2022b R12. (synthesis: 3.14/0.04 ≈ 80-fold and 4.5/0.10 = 45-fold; 22 forms = 14 drugs and 8 salts in the rows below.) A second ritonavir–PVPVA series of 2022 released about 97–100% of drug and polymer by 60–90 min at 20% and about 35% of the drug at 30%, placing the switch between 20 and 30% in that paper's own data, next to the 25% of 2019 yang2022a R2, X7, R15. In 2023 MePHPH–PVPVA released at least as fast as neat PVPVA to 40% and about 15% of its drug at 50% after a lag, with nothing at 70%, and PHPH–PVPVA about 75% at 5% and nothing at 20%, in line with the cited limits of 40 and 5% deac2023a R1, R2, R14. The drug content of the gel rose at 40 and 50% MePHPH and not at 20%, so by that criterion the MePHPH limit lies between 20 and 40% deac2023a R3, C14, X18. Clotrimazole–PVP dispersions passed the 80% criterion at 20% but not 23% (K25), at 33% but not 35% (K29/32), at 35% but not 38% (K60) and at 38% but not 40% (K90), and failed at 5% with K12 and K17 deac2024a R1, R12, X5. With 10 parts of a Span per 100 parts of drug and polymer, ritonavir–PVPVA released about 87–97% at 30% but 1–8% at 40% (F31) yang2023 R2.

Figure

Both bars are near 100% at 10–25%, both fall at 30% (the drug more than the polymer), and both are near zero from 35% on indulkar2019a R11.

Why

The authors stress that the range over which release switches from polymer- to drug-controlled is "very narrow" and that the switch changes drug release by "orders of magnitude" indulkar2019a C52. They attribute the drop to faster water-induced phase separation relative to dissolution at high loading ("can be attributed"), see F8, and propose that key ritonavir–PVPVA interactions in water are "likely" saturated around 25–30% (see F10) indulkar2019a C33, C34. The size of the drop follows from the two regimes of F2 and F3: the amorphous nilvadipine compact dissolves more than three orders of magnitude more slowly than PVPVA, "explaining" the precipitous fall saboo2019 C43. (synthesis: the drop cannot exceed the ratio of the polymer-controlled to the neat-amorphous rate, about 1700 for nilvadipine–PVPVA and about 4600 for felodipine–PVP at pH 6.8 but about 10.4/0.051 ≈ 200 for ITZ in 0.1 N HCl saboo2019 R7; saboo2020a R13; bhardwaj2018 R9, R4; the observed drops are 60–360-fold for the first two and 2–4-fold for ITZ, so the cliff is steepest where the amorphous drug alone dissolves most slowly.)

Conditions and limits

Compacts at 37 °C in pH 6.8 phosphate (ITZ in 0.1 N HCl), ritonavir films at 25 °C; loading steps of 2.5–10% que2019 A1; purohit2017b A2. Congruency is defined only as simultaneous release indulkar2019a X7. The limit depends on how it is defined and measured: last congruent loading, release of at least 80%, or LLPS hiew2022a X17; purohit2017b R11. Gel drug content gave a lower MePHPH limit (between 20 and 40%) than release did (40%), so the criterion matters deac2023a C14, X18.

Counterexamples and conflicts

Felodipine dispersions in HPMCAS and EUDS showed no limit up to 50%, and 50:50 ritonavir–HPMCAS films still formed LLPS (see F11) saboo2020a R15; purohit2017b R11. The same pair gave different limits by different tests: ritonavir–PVPVA films formed LLPS only at 10% (25 °C, LLPS read-out) while tablets released congruently to 25% (37 °C, congruency read-out), and a cited room-temperature value puts 20% beyond the limit (synthesis: a lower limit at lower temperature is consistent with both) purohit2017b R11; indulkar2019a C2; benson2025 R21. In the ASD dissolution chapter delamanid–HPMCP fell gradually from 25 to 40% and praziquantel–HPMCAS showed no cliff (Supersaturation from ASD dissolution F10) duong2022a R24; polyzois2024 C16. HPMCAS dispersions of three more drugs declined gradually with loading, and SDS dispersions of ATZ declined gradually above 10% (about 60.5 and 39% at 15 and 20%) instead of collapsing yang2021 C2; correasoto2022b R4, X11. The release data shown in the 2024 phase-diagram paper place the MePHPH limit only between 30 and 50%, against the cited 40% deac2024b X9.

Transfer check

Dissolve the candidate drug–polymer pair at 37 °C in 2.5–5% loading steps around the expected switch, with both components assayed que2019 A1; saboo2019 A2. Treat the last congruent loading as an upper bound, choose the product loading with a margin below it, and repeat the first failing loading, whose extent varied 1.6-fold between runs for ritonavir indulkar2019a X3. Check the switch at the temperature of use, since the cited room-temperature limit of ritonavir–PVPVA lay below the 37 °C one benson2025 R21.

History

Simonelli and co-workers (1969, cited) described a loading-dependent change of regime for sulfathiazole–PVP bhardwaj2018 R22. Bhardwaj and co-workers (2018) proposed a threshold drug-to-polymer ratio above which the amorphous drug controls release ("suggests") bhardwaj2018 C66. Indulkar and co-workers (2019) named the limit of congruency and the cliff for ritonavir–PVPVA indulkar2019a C2, C31. Saboo and co-workers (2019) located it for two dihydropyridines and recommended loadings below the "drop-off" concentration when rapid release is needed saboo2019 C33, Que and co-workers (2019) found the lowest value of the period for ledipasvir que2019 C10, and Saboo and co-workers (2020) showed that two polymers have no limit up to 50% saboo2020a R15. Purohit and Taylor's films of 2017 show the same loading switch through LLPS before the term was introduced purohit2017b C52. The papers of 2020–2022 added limits for 22 further drug forms and expressed them per mole when molecular weights differ que2021 E3, R20; this extends the range given in the short part from 5–30% to 2–40%. Deac and co-workers (2024, polymer molecular weight) showed the polymer grade moving the limit at least sevenfold, and Deac and co-workers (2024, ternary phase diagrams) located the limit near the critical point of the drug–polymer–water diagram (F29, F30) deac2024a R12; deac2024b R37.

How it was shown

Purohit and Taylor (2017) imaged 50:50 and 10:90 ritonavir films by SEM after 10 s in buffer purohit2017b P13. Bhardwaj and co-workers (2018) inferred a drug-rich layer from the curving polymer profiles of 50:50 ITZ compacts, without surface analysis bhardwaj2018 C52. Indulkar and co-workers (2019) ejected ritonavir tablets at 0–15 min, blotted them and recorded ATR-IR against amorphous ritonavir and PVPVA, and imaged vacuum-dried tablets by SEM indulkar2019a P7, P8. Que and co-workers (2019) analysed compact surfaces before and after 10 min by XPS (N 1s fitting with pure-component spectra, F 1s components) and imaged dried compacts by micro-CT at 8 µm/pixel que2019 P7, P5. Saboo and co-workers (2019) took compacts either side of the switch after 10 min for micro-CT (3 µm pixels), SEM, SEM-EDX of the N/O ratio (0.5 for nilvadipine, ~0.3 for PVPVA) and XPS of the top ≈10 nm saboo2019 P14, P15, P16, P17, E2. Saboo and co-workers (2020) examined 30% felodipine tablets after 10 min by micro-CT, SEM and EDX of chlorine, which only felodipine contains saboo2020a P8, P9. Yang and co-workers (2022) dissolved ritonavir–PVPVA compacts at 10–50% containing trace Nile red (drug-rich phase) and Alexa Fluor 488 (polymer- and water-rich phase), removed them after 5–30 min, dried the surface with air and imaged surfaces and cut cross-sections by confocal microscopy, measured how far the gel–glass front had advanced into the compact, and analysed the 40% surface by XPS (C 1s fits, five spots on one compact) before and after 30 min yang2022a P1, P5, P6, P10, P9. Deac and co-workers (2023) blotted and dried compacts after 10–60 min of dissolution and read their surface drug content by ATR-FTIR against standard curves, with the PHPH medium spiked with 40 µg/mL PHPH, and imaged the surfaces of dye-labelled compacts by confocal microscopy after 2–20 min deac2023b P12; deac2023a P8.

What the data show

Ritonavir: the 50% surface still resembled the unexposed dispersion at 2 min, resembled amorphous ritonavir from 5 min with the PVPVA peaks diminished, and was closest to it at 15 min; the 20% surface was unchanged at 5 min, and SEM showed a rough, flaky 50% surface at 15 min against smooth 20% surfaces indulkar2019a R16, R15, R17. Ledipasvir: surface drug matched the nominal loading before dissolution (17 against 20, 40 against 50 wt%) and reached ≈54 ± 5 and ≈61 ± 2 wt% after 10 min; the share of the lower-binding-energy fluorine fell from ≈73, ≈42 and ≈18% to ≈9, ≈3 and ≈2% at 10, 20 and 50% que2019 R15, R17. The 10% compact kept its thickness with a distinct top layer at 10 min and bulged with pores and cracks by 30 min que2019 R13. Nilvadipine and cilnidipine: compacts below the switch were homogeneous and smooth, those above it had a porous layer at the dissolving face within 10 min, pits with raised rims up to about 20 µm (nilvadipine) or about 0.1–1 µm (cilnidipine), and porosity that grew at 20 and 30 min saboo2019 R13, R14. EDX on a nilvadipine 15:85 compact gave N/O ≈0.50 ± 0.03 on a pit rim, ≈0.35 ± 0.04 inside and ≈0.31 ± 0.01 before dissolution saboo2019 R16. XPS gave 2 ± 0 → 32 ± 1% (nilvadipine 15:85) and 20 ± 1 → 60 ± 4% (cilnidipine 20:80), while the congruent compacts did not change saboo2019 R18. Felodipine 30%: porous layers for PVP, PVPVA and HPMC, none for HPMCAS and EUDS; pits on PVP and PVPVA, voids on HPMC, angular particles on HPMCAS and EUDS; chlorine as in the short part, with HPMCAS ≈2.9 → ≈3.1 and EUDS ≈3.0 → ≈2.7 (not significant) saboo2020a R19, R21, R22. Ritonavir films after 10 s in buffer: a rough porous network on 50:50 PVP, a granular surface on 50:50 PVPVA, smooth 10:90 films and a smooth 50:50 HPMCAS film purohit2017b R9. Polymer release under the layer: ≈2% of the PVPVA of the 20% ledipasvir compact by 10 min, and PVPVA from felodipine compacts fell from ≈1.9 to ≈0.14–0.19 mg/min/cm² across the switch, whereas PVP only about halved and HPMC did not fall que2019 R11; saboo2020a R14, X15. IND: the 10% tablet surface was unchanged by IR after 5 min while the 15% surface gained the 1634 and 1591 cm⁻¹ bands; the 25% surface reached about 40–50, 50 and 60% drug by matching IR spectra at 5, 10 and 20 min, 41 ± 3, 55 ± 11 and 72 ± 9 wt% by EDX (from 25 ± 2) and 44 ± 5, 52 ± 7 and 60 ± 5 wt% by nanoTA on immersed films; SEM showed smooth 10% and rough, porous 15 and 25% surfaces, and AFM height features grew from about 15 nm at 2 s to about 150 nm at 20 min saboo2020 R15, R16, R17, R21, R28, R19, R24. Felodipine with TPGS: after 10 min the 30% tablets with 0 and 2% TPGS (incongruent) had a porous surface layer, those with 5 and 10% (congruent) a homogeneous section saboo2021 R19. Lumefantrine benzoate at 30%: large pits after 10 min and a rough, irregular surface, read as a PVPVA-rich phase dissolving first ("postulated") hiew2022b R17, C22. Ritonavir–PVPVA in 2022: surfaces carried small discrete drug-rich domains at 10–25%, a mixture of continuous drug-rich patches and discrete dots at 30%, and an interconnected, porous drug-rich network at 40–50% after 5 min that covered the whole surface by 30 min; neat amorphous ritonavir and neat PVPVA gave uniform images yang2022a R4. Cross-sections showed a glassy core, a gel layer with a sharp boundary to the core and, from 30%, a drug-rich surface layer, 5–10 µm at 30% after 5 min and about 12 µm at 40% after 30 min, with polymer- and water-rich regions under it yang2022a R5, R6, R7. The gel–glass front advanced alike for neat PVPVA and 10–20% compacts (about 44–54% of the thickness by 10 min, all of it by 20 min) but reached only about 29 and 17% by 20 min at 30 and 40% and then stopped yang2022a R8. XPS: 35 ± 0.5 wt% ritonavir before and 72 ± 10 wt% after 30 min at 40% yang2022a R14, X13. PHPH and MePHPH in 2023: surface drug of 5% PHPH about 5.6 → 6.2% over 60 min, 15% PHPH about 13.5 → 21.3% by 30 min, 50% PHPH about 49 → 53% in buffer and about 49 → 26% with 1 mg/mL PVPVA in the medium; 30% MePHPH about 33–43%, 50% MePHPH about 44 → 101 → 91% deac2023b R6, R7. Confocal surfaces were uniform at 30% MePHPH, patchy at 40%, a continuous porous drug-rich phase with water-rich pockets at 50%, and porous at 20% PHPH; neat drugs and polymer did not separate deac2023a R9, R10.

Figure

Smooth surfaces below the switch (a, c) against pits with raised rims above it (b, d); note the different scale bars (10, 20, 1 and 1 µm) saboo2019 R14.

Both compacts start near their nominal loading and end far above it que2019 R15.

Why

In incongruent release the polymer leaves faster, so the interface "must" become enriched in drug; the porous structure is "consistent with" polymer loss and the pits "most likely" arise from faster polymer than drug release saboo2019 C38, C21, C22. The authors trace the pits to rapid phase separation at the interface followed by dissolution of the polymer-rich regions ("can likely be attributed"), and compare the drug-rich rims with the drug-rich rings of humidified films ("remarkably similar"; see F8) saboo2019 C39, C40. For ritonavir they add that drug molecules "can" move laterally and that drug-rich nanodomains "can" coalesce into contiguous regions, ritonavir being a supercooled liquid at 37 °C (wet Tg ~0 °C) indulkar2019a C28, C40, M1. For ledipasvir the drug-rich layer makes dissolution drug-controlled, "too slow to detect" for this drug, and "it appears" to stop release of drug and even of soluble PVPVA, leading to swelling que2019 C31, C32, C36. For felodipine the surface enrichment is "a physical barrier to polymer release from subsurface layers" saboo2020a C42. An alternative reading of the felodipine data is that preferential polymer dissolution alone gives a porous, drug-rich surface whether or not phase separation precedes it saboo2020a X6. In the ASD dissolution chapter such a layer had been proposed for 50:50 felodipine–HPMC but not observed alonzo2011 C46. For IND the authors find the agreement of three methods with different probe depths "compelling evidence" of enrichment, take the absence of the acid-dimer band to mean that some hydrogen-bonded polymer remains, and favour phase separation over plain polymer loss because the surface roughened over about 500 nm, a length not measured saboo2020 C44, C22, C42, X14. Yang and co-workers (2022) read the stalled gel–glass front as water ingress "significantly retarded" by the hydrophobic drug-rich layer, which "effectively acts as a barrier" to water, and "may serve" as a physical barrier to polymer release even where water gets in, which they say explains why poor drug release often comes with poor polymer release yang2022a C30, C55. They attribute the surface enrichment to preferential polymer release and the persistence of the layer to a medium saturated at the amorphous solubility, which was not measured yang2022a C40, C53, X11. For the phenolphthalein analogues the surface layer in contact with nearly pure water takes the composition of the dilute drug-rich phase, a "pseudo-equilibrium", while deeper layers do not; the layer of the polymer-binding PHPH therefore holds much polymer (F27) deac2023b C61, C59, C4.

Conditions and limits

Surfaces were examined after 2–15 min (micro-CT of nilvadipine to 30 min) and at one loading above the switch for most systems; SEM and micro-CT samples were dried before imaging saboo2019 A5; indulkar2019a X11; que2019 X22. The 15:85 nilvadipine surface started polymer-rich (2% drug by XPS against 15% nominal) saboo2019 X13. Below 20% ledipasvir the N 1s signal of the drug could not be separated from the polymer's, and there is no XPS at 7.5%, the first failing loading que2019 R15, X21. The low-loading ritonavir surface was examined at 5 min only, while its release continued to ≈40 min indulkar2019a X9. The 2022 ritonavir images are of compacts removed from the medium and air-dried, and the front position is a proxy for water entry; water content was not measured yang2022a P5, C28. The 2023 surface values carry calibration offsets, come from dried compacts, and the MePHPH values near 90–101% lie beyond the highest standard deac2023b X8, X17.

Counterexamples and conflicts

The barrier did not suppress polymer release in every polymer: from felodipine compacts PVP release about halved and HPMC release did not fall across the switch saboo2020a X15. The 20% ritonavir surface showed round features of ≈0.5–3 µm at 5 min that the text does not discuss indulkar2019a X10. Polymer still left the 40% ritonavir compact in the release test (about 14% in 90 min) although no polymer dye was seen beyond its drug-rich layer in real-time images yang2022a R2, R13, X8.

Transfer check

For a candidate loading, eject compacts after 5 and 10 min and compare the surface with the dry compact by ATR-IR or XPS, using an element that only the drug carries (Cl, F, or the N/O ratio) for EDX indulkar2019a P7; que2019 P7; saboo2019 E2; saboo2020a P9. A surface that gains drug within minutes and turns porous marks the drug-controlled side of the switch; an unchanged surface the congruent side saboo2019 R18; saboo2020a R22.

History

A drug-rich layer at the dissolving surface was proposed for sulfathiazole–PVP (Simonelli 1969, cited) and for 50:50 felodipine dispersions in the ASD dissolution chapter purohit2017b R23; alonzo2011 C46. Purohit and Taylor (2017) first imaged a porous surface on high-loading films after 10 s in buffer purohit2017b R9; Indulkar and co-workers (2019) recorded the drug-rich tablet surface by IR indulkar2019a C41; Saboo and co-workers (2019) and Que and co-workers (2019) quantified it by XPS and EDX saboo2019 C25; que2019 C17; and Saboo and co-workers (2020) showed its absence with HPMCAS and EUDS saboo2020a C22. Later studies outside the subcategory measured the same enrichment for lumefantrine and bicalutamide hiew2022a C67; moseson2022b C39. Saboo and co-workers (2020) followed the time course of enrichment by three methods, and Saboo and co-workers (2021) showed the porous layer forming only where release was incongruent saboo2020 C44; saboo2021 C22. Yang and co-workers (2022) imaged the layer in cross-sections of dissolving compacts and showed that it slows the advance of water yang2022a C21, C30. Deac and co-workers (2023) found the layer's composition to approach that of the drug-rich phase the drug forms in water, polymer-rich for a hydrogen-bonding drug deac2023b C59, C4.

How it was shown

Purohit and Taylor (2017) followed ritonavir films at 97% RH by the pyrene I/III ratio (0–96 h), confocal microscopy with nile red, AFM after 4 days and DSC of powders after 24 h and drying, and dipped films in buffer for 10 s before confocal imaging and SEM purohit2017b P7, P8, P9, P10, P11, P13. Indulkar and co-workers (2019) dipped nile red films at 10–50% (20, 30 and 40% shown) in water for a few seconds and checked them under polarized light indulkar2019a P9, P10. Que and co-workers (2019) held nile red films at 5, 10 and 50% at 100% RH and 37 °C for 3 days que2019 P6. Saboo and co-workers (2019) held a 10:90 nilvadipine compact in its die and a matching film with prodan and rhodamine 6G at 97% RH for 12 h, then dissolved the compact for 15 min and imaged the film saboo2019 P13, P18. Saboo and co-workers (2020) held 30% felodipine films with prodan at 97% RH for 24 h saboo2020a P10.

What the data show

Ritonavir: at 97% RH the pyrene ratio of 50:50 PVP fell from ≈1.50 to ≈1.30 within 1 h and every 20–50% PVP and PVPVA film fell to ≈1.25–1.35 by 72–96 h, near pure amorphous ritonavir (≈1.27–1.30), while 10:90 films fell slowly (to ≈1.35 PVP and ≈1.45 PVPVA at 96 h) and HPMCAS films stayed at ≈1.39–1.47 purohit2017b R2, X9. Confocal images showed drug-rich domains in PVP and PVPVA films and uniform HPMCAS films; AFM showed 5–10 µm domains on 10:90 PVP and PVPVA; DSC showed a ritonavir-rich step at ≈47–50 °C in every PVP and PVPVA powder and one step (≈63–97 °C) for HPMCAS purohit2017b R3, R5, R6. After 10 s in buffer, 50:50 PVP showed 1–3 µm dark domains in a bright drug-rich continuous phase and SEM showed porous and granular surfaces on 50:50 PVP and PVPVA, while 10:90 films and both HPMCAS films stayed uniform; 50% ritonavir–PVP and –PVPVA dispersions had separated within 10 s of water contact in earlier work purohit2017b R4, R9, R20. Ritonavir–PVPVA films dipped in water: 20% uniform, 30% sparse bright spots of a few µm, 40% dense spots, none birefringent indulkar2019a R18, R19. Ledipasvir films after 3 days: as in the short part que2019 R14. Felodipine films: discrete prodan-rich domains for PVP, PVPVA and HPMC, uniform HPMCAS and EUDS at two magnifications saboo2020a R23. Humidity also changed release: the conditioned 10:90 nilvadipine compact released ≈0.34% of its drug and ≈2.6% of its PVPVA in 15 min against ≈15% of both for the control, and its film showed drug-rich circles and rings of roughly 5–15 µm in a continuous polymer-rich phase (see F12) saboo2019 R9, R19. Clopidogrel films after 5 min at 100% RH: uniform at 40%, bicontinuous at 50%, drug-continuous at 60% correasoto2022a R20. Felodipine films at 30% with 0–10% TPGS all separated after 12 h, with different morphology (F20) saboo2021 R21. The binary 10% atazanavir film showed no separation for 3 h at 37 °C and separated within 10 min at 50 °C correasoto2022b R9, R12. Lumefantrine free-base and benzoate films separated, the sulfate film did not hiew2022b R19. In dissolving ritonavir–PVPVA compacts imaged in 2022, phase separation appeared at every loading from 10 to 50% at the gel–solution interface; its shape changed with loading, discrete droplets at 10–25% and a continuous drug-rich layer from 30% (F25) yang2022a R4, R6, R10, R11, R12, R13. In 2023 dissolving MePHPH–PVPVA gels separated at every loading imaged (5–50%), segregatively, and PHPH gels associatively, with both dyes in the drug-rich phase (F25, F27) deac2023a R6, R7, C25, C28. MePHPH–PVPVA films at 10–50% stayed clear at 62 and 75% RH and separated from 80% RH, where matching powders held 9.6–14.6 wt% water deac2024b R1, R2. Ritonavir–PVPVA 30% films separated at 84% RH and 14.0% water, with SDS at 92% and 24.5%, and with Span 20 or Tween 80 already at 75% and 10.5–12.5% yang2023 R21. Clotrimazole films with PVP K12 or K29/32 at 10% separated into droplets within 3 h at 100% RH deac2024a R7.

Figure

All films are uniform before water; afterwards the 20% film stays uniform while the 30% and 40% films show bright drug-rich spots, more at 40% indulkar2019a R18.

The 10:90 film (a) is uniform; the 50:50 film (b) has split into dark domains in a bright drug-rich phase purohit2017b R4.

Why

The thermodynamic argument is the Flory–Huggins free energy of mixing of water, drug and polymer,

\frac{\Delta G_{mix}}{RT} = n_w \ln\phi_w + n_D \ln\phi_D + n_p \ln\phi_p + n_w \phi_D \chi_{WD} + n_w \phi_p \chi_{WP} + n_D \phi_p \chi_{DP}

purohit2017b E2, in which a hydrophobic drug interacts unfavourably with water (large \chi_{WD}) and the polymer favourably (small \chi_{WP}); a large \Delta\chi = |\chi_{WD} - \chi_{WP}| drives separation, which the authors call "a classic case of solvent-induced immiscibility" purohit2017b E4, C43, C44. They expect PVP and PVPVA, more hygroscopic and with acceptors only, to separate more readily than HPMCAS ("can be expected") purohit2017b C45, C46. All five papers then propose a race: if the hydrating surface dissolves faster than it separates, drug and polymer leave together; if it separates first, the polymer-rich domains dissolve and leave a drug-rich surface purohit2017b C5; indulkar2019a C10, C30; que2019 C35; saboo2019 C42; saboo2020a C41. Higher loadings "can be expected" to separate more readily and faster, being closer to the miscibility limit, and at low loading there may be too little drug for a barrier indulkar2019a C30; saboo2020a C42. Humidity that converts a congruent compact into an incongruent one is "further evidence" for this role of matrix phase separation saboo2019 C41. Saboo and co-workers (2020) add that the water absorbed matters, so the low uptake of HPMCAS and EUDS "conceivably" keeps them below the separation threshold (F11) saboo2020a C44. Outside the subcategory moisture-induced separation followed polymer hygroscopicity for felodipine and quinidine, and felodipine–PVP has a large Δχ (χ about 3.3 against 0.5) rumondor2010c C38, C31; marsac2010 C43. (synthesis: in four drug–polymer pairs the loading at which films separate on wetting lies within or just above the release bracket, which is the main support for the race; neither the time to separate nor the time to dissolve was measured on the same compact, so the race remains an inference.) Saboo and co-workers (2020) and Yang and co-workers (2021) keep the race for IND and for the cliff in general ("would be expected"; "It has been suggested") saboo2020 C40, C41; yang2021 C17, Yang and co-workers explain the tendency to separate by the Δχ effect rather than log P yang2021 C21, E2, and Correa Soto and co-workers write that at high loading "AAPS is proposed to occur rapidly" correasoto2022a C53. For high-Tg forms the atazanavir and lumefantrine papers place the failure in low mobility instead: a glassy matrix hydrates and disentangles slowly, and a high-Tg salt cannot separate from the hydrated polymer (F18, F22) correasoto2022b C37; hiew2022b C47. Indulkar and co-workers (2022) restate the race for ritonavir with a third process: dissolution, water-induced separation and crystallization are concurrent kinetic processes, and separation is "inevitable" for hydrophobic drugs in PVPVA at high loading, so a dispersion has to release faster than it separates and crystallizes, or resist both indulkar2022 C23, C24, C25. Yang and co-workers (2022) place the separation in the gel layer, whose water content is highest next to the solution, so that separation is "favored" there yang2022a C44. (synthesis: the measured diagrams of 2024 put most of the MePHPH gel-layer compositions inside the miscibility gap, so separation occurs below the limit as well, as the compact images of 2022–2023 showed (F25, F30) deac2024b R28, C37.)

Conditions and limits

Phase separation was tested on spin-coated or cast films and on compacts held at 97–100% RH for 12 h to 4 days, or dipped in water for seconds and then dried, never on a compact during dissolution saboo2019 X10; indulkar2019a X6; que2019 X17. Films were imaged at 20, 30 and 40% (ritonavir), 5, 10 and 50% (ledipasvir) and 30% (felodipine), so the separation was not followed across each bracket indulkar2019a X6; que2019 K3; saboo2020a X5. Confocal resolution was about 300 nm purohit2017b P9. From 2022 separation was also imaged in dissolving ritonavir compacts, after removal from the medium and in real time with edge wetting, but still without a rate of separation yang2022a P5, P7, C31.

Counterexamples and conflicts

A cited study found that water-induced phase separation did not change the dissolution of a 40% dispersion, which the authors explain by that dispersion being drug-controlled already or by different test conditions saboo2019 R22, C44. Outside the subcategory phase separation raised release of lopinavir–HPMC films at 33–85% loading, where polymer-rich phases dissolve first (F12) li2019 C38. A drug-rich porous surface also follows from faster polymer dissolution alone, so the surface data do not prove the separation saboo2020a X6. Water uptake of the neat polymers does not rank the limits within the hydrophilic group: HPMC takes up less water than PVPVA yet has the lower limit saboo2020a X7. The binary 10% atazanavir film and the lumefantrine sulfate film failed without visible separation, while atazanavir films with surfactants separated within 10 min and released correasoto2022b R9, R10, R11, R2; hiew2022b R19. Since separation appeared at the gel–solution interface of ritonavir compacts below the limit as well (droplets at 10–25%), it is the shape of the separated phase, not its presence, that changed at the limit in that system (F20, F25) yang2022a R4, R6, C58. MePHPH gels separated at 5–30%, below the limit, into discrete droplets, so separation alone does not mark the limit there either deac2023a R6, C46.

Transfer check

Cast films of the drug–polymer pair at the target loading and one step above and below, dip them in the dissolution medium for seconds and expose them to 97–100% RH, then image with a drug-rich probe (nile red or prodan) indulkar2019a P9; saboo2020a P10. Separation at the target loading, especially within seconds, predicts incongruent release; uniform films are consistent with congruent release but do not prove it, because the race has not been measured indulkar2019a X6.

History

Moisture-induced AAPS was established for PVP dispersions in storage work of the library (Rumondor 2009, 2010, 2011, cited), and 50% ritonavir dispersions in PVP or PVPVA were known to separate within 10 s of water contact (Purohit 2015) purohit2017b C9, R20; que2019 C34; saboo2020a C39. Purohit and Taylor (2017) first linked such separation to the loss of LLPS on dissolution purohit2017b C53. Indulkar and co-workers (2019), Que and co-workers (2019) and Saboo and co-workers (2019) placed the film separation against the limit of congruency and proposed the race indulkar2019a C9; que2019 C35; saboo2019 C42; Saboo and co-workers (2020) added the polymer contrast and named water-induced AAPS "likely" the cause of the limit saboo2020a C43. Correa Soto and co-workers (2022) matched film separation with the clopidogrel limit, and the atazanavir and lumefantrine studies of 2022 found failure without visible separation for high-Tg forms, which adds the exception to the title (F20) correasoto2022a C52; correasoto2022b C18; hiew2022b C48. Yang and co-workers (2022) were the first to image the separation in dissolving compacts, and found it below the limit too, which qualifies "above the limit and not below it" of the title for that system yang2022a C5, C45; the last clause of the title was added for this reason. Deac and co-workers (2024) measured the water at which separation starts and the compositions of the separated phases, and Yang and co-workers (2023) the shift of that water by surfactants (F30, F31) deac2024b C13; yang2023 C9.