Characterization methods

A cellulosic hollow-fibre module (100 cm² and 15 nm pores in the tacrolimus study) fed from the dissolution vessel, donor and receiver pumped co-currently at 2 mL/min, receiver read by in-line UV and donor by HPLC; total dissolved drug in donor plus receiver follows crystallization hate2019 P4, P8, C19. Receiver AUC and a test-to-reference ratio with 80–125% limits summarize the result hate2019 P9, E6. The module is described as having 115 cm² of membrane in the nevirapine study [hate2017 M10].

The absorption rate is a property of the apparatus: a closed vessel may overstate crystallization, and an absorptive test may understate it if absorption in vivo is slower hate2019 C12, C35. Two explanations for the missing crystallization of fresh tacrolimus capsules, the polymer and the lower supersaturation from absorption, were not separated, because no run without the sink at the same volume exists hate2019 X15, C22. The area of the same module is 100 cm² in one card and 115 cm² in another [hate2019 P4; hate2017 M10]. The fitted model constants are not reproducible from the printed values (F29) hate2019 X1, X3, X4.

Storage over saturated salts at fixed RH (here 54–94% RH at room temperature) and read-out by IR, DSC, PXRD or microscopy [rumondor2009c P4; rumondor2009b P2, P3; marsac2010 P7].

Evidence table: accelerated stability testing.

In susceptible systems "the route of crystallization may be altered at high RH" and need not represent lower RH [rumondor2009c C52]; the separation threshold depends on time and composition [marsac2010 X9]. Dry and moderate-RH data could not rank PVP against HPMCAS for felodipine within the method error rumondor2009b X4. The 2013 papers state that their results "highlight the difficulty" of choosing accelerated conditions [wegiel2013b C41; kestur2013b C40].

By the loss of the E100 amine bands at 2820 and 2770 cm⁻¹ with rising drug content [wegiel2013a R6]; by a protonated component of the N 1s XPS peak and by carboxylate bands at 1594–1560 and 1410–1343 cm⁻¹ in subtraction IR spectra [hiew2022a R19, R20]; by ATR-FTIR of melts at equimolar functional groups (spectra not shown) [vaneerdenbrugh2011a R8, R15, R17]; by colour change (curcumin with E100) [wegiel2014 R10, C20]: dark red for curcumin–E100 against orange for the others, read as a change of the drug's ionization state; by a dispersion Tg above both components [wegiel2014 C41]. For the weak base lumefantrine with acidic enteric polymers in the physical-stability papers, XPS of the N 1s signal gives the share of protonated drug, and ATR-FTIR shows the polymer carbonyl band moving up and a band near 1560 cm⁻¹ read as carboxylate ("likely") [trasi2020 P15, R9, C11, C12; bhujbal2021 P6, R6]. The pKa of each group is calculated or titrated to judge whether a proton can move [vaneerdenbrugh2011a P7; hiew2022a P17].

Evidence table: acid-base interaction.

The strength of the contact was not measured, and the causal role of the E100 contact in stability was not separated from other properties of E100 [wegiel2013a X10]. The naringenin–E100 spectra are described but not shown, and no polyphenol-side marker is reported [wegiel2013a C17, X10]. The melt spectra behind the PSSA reactions are not shown, and no acid–base reaction was seen for bifonazole [vaneerdenbrugh2011a X13, R16]. The lumefantrine values come from dry compacts and fall after dissolution, except for CAP; an earlier XPS study found no protonation with HPMCAS where this one found 19% [hiew2022a R20, R30, X14]. An ion pair does not guarantee stability: the 50% HPMCAS dispersion, with 1.61 mol drug per mol acid groups, crystallized within a week [hiew2022a R26, C63]. For curcumin the extent of proton transfer was not measured, and no polymer with the same backbone without amine groups was tested [wegiel2014 C37]. For resveratrol the IR-based ranking that put E100 first did not hold in storage, where PVP outperformed it [wegiel2013b X10]. In the lumefantrine powder studies the share of protonated drug ordered the stability of powders made without water but not of powders precipitated into water, where residual water and drying time decided [trasi2020 C36; bhujbal2021 C50, C51]. The polymer that protonated the drug most, CAP, released the least drug at 40% loading [trasi2020 R14; bhujbal2021 R15].

Two Tg events by DSC (one Tg is read as miscible, with the limits in single-Tg miscibility criterion) [vaneerdenbrugh2012a P3, C3]; pyrene and prodan fluorescence and fluorescence imaging, AFM-nanoIR, nanoTA, Lorentz contact resonance and AFM topography purohit2015a P3, P4, P9, P10; li2016c P2, P4, P7, P8; for polymer pairs also the hydrogen-bonded fraction of PVP carbonyls by FTIR [vaneerdenbrugh2012a P5]. In the later process papers: two Tgs by DSC, AFM with LCR or nanoTA, fluorescence microscopy and the drug's autofluorescence, TEM with EDX or HAADF [li2017 P9, P10, P13; mugheirbi2017 P8, P9; li2019 P10]. In the moisture papers: IR against a physical-mixture spectrum built from C–C band ratios; DSC wet and re-dried; sorption against additivity; AFM and TEM [rumondor2009c P7, E7, P9, P12; marsac2010 P14, P15]. In the release-mechanism papers: Pyrene I/III ratio, nile red or prodan confocal fluorescence, AFM and DSC of films and powders after 97–100% RH, or of films dipped in buffer for seconds [purohit2017b R2, R3, R4, R5, R6; indulkar2019a R18; que2019 R14; saboo2019 R19; saboo2020a R23]; pyrene and TEM of spray-dried particles [bhardwaj2018 R15, R16].

Evidence table: amorphous-amorphous phase separation.

A single DSC Tg did not show the separation of itraconazole–HPMC, which the authors attribute to mixing during the scan, a reading supported only by the narrower second-run steps purohit2015a C39, C26, X1. Whether the separation is kinetic (solvent-induced, water-induced) or thermodynamic is not settled, and the phase compositions were not measured: the band-ratio and softening data rank the phases but give no weight fractions purohit2015a C48; li2016c X3, X10. The separation depended on preparation: humidity, solvent water and the alcohol of the solvent changed whether films separated [mugheirbi2017 C39; mugheirbi2018 R1, C50]. For polymer pairs the plot of the miscibility map is not monotonic cell by cell [vaneerdenbrugh2012a X10]. Separation during dissolution or storage was invoked for PVPVA dispersions of lumefantrine but not measured [hiew2022a C43, C61, X11]. DSC and polarised light microscopy may overestimate miscibility [mugheirbi2018 C4]. Whether separation helps or hurts release depends on the resulting microstructure [li2019 C5, C58]. The same separation can lower release (ritonavir–PVPVA at 25% DL) or raise it (lopinavir–HPMC at 33–85% DL) [li2020 C64; li2019 R2]. In the humid felodipine–PVP films the composition of the polymer-rich phase was not determined either; the DSC "second Tg" near 124 °C may be depressed melting [marsac2010 X7, C18]. Faster crystallization after moisture-induced separation was shown across different drugs, not for one drug with and without separation [rumondor2009c X7]. In the 2019 ritonavir–PVPVA study AAPS was seen only in films dipped for seconds or in humidified samples, not in a dissolving tablet, and neither its rate nor that of dissolution was measured [indulkar2019a X6]. A drug-rich porous surface also follows from faster polymer dissolution alone [saboo2020a X6]. In the 2022 ritonavir surfactant study separation was inferred from one lowered Tg of material scraped from the wet tablet, without a second Tg or water content, and was not tested for three of the five surfactants [indulkar2022 X10, X4].

DSC melting data and modulated-DSC reversing heat capacities fitted linearly in T for crystal, glass and supercooled liquid; the Cp method needs an un-ionized, non-solvated crystal that melts without degradation [almeidaesousa2015 P5, P6, C11; mosqueragiraldo2015 P3].

The Cp parameters are not printed, so the Cp values cannot be recomputed [almeidaesousa2015 E9, E10]. The claim that the Cp method is more accurate presupposes that the onset equals the supercooled-liquid solubility [almeidaesousa2015 X4]. The glass carries more free energy than the supercooled liquid below Tg [almeidaesousa2015 C29]. For atazanavir (undercooling 171 K, synthesis: 481 − 310 K) the Hoffman inputs printed give the printed ratio of 62 only with 0.22 read as exp[−I(a)], not I(a) as Table I heads the column (synthesis) [indulkar2017a X1; indulkar2015 M2]; for pazopanib the cycle cannot be run because it degrades on melting [sugihara2018 C17].

Prepared by freeze-drying or melt quenching and checked by DSC [towler2008 P3, P5, R1]; Tg, recrystallization on heating, moisture sorption and storage at high humidity are then measured [towler2008 P6, P8, P9]; later by open-dish storage at 40 °C/75% RH [hiew2022b R7; duong2022a C32].

Stoichiometry and ionization state of the amorphous salts were not reported [towler2008 X8]. Crystallization and moisture uptake did not follow counterion pKa [towler2008 C14, C34]; drug or salt Tg appeared an important predictor for lumefantrine [hiew2022b C37], yet the highest-Tg propranolol salt crystallized [towler2008 R12]. Salt formation did not change delamanid crystallization in dispersion films at pH 3.0 [nguyen2023b C17].

Rotary evaporation from methanol (ritonavir) or methanol/dichloromethane (CRH1) and incremental or single-dose non-sink dissolution [ilevbare2013a P13, P14; almeidaesousa2016 P2, P11]. The dispersions of the miscibility papers were made by dissolving drug and polymer in a common solvent and removing it: rotary evaporation from methanol, ethanol, chloroform or dichloromethane–ethanol, followed by vacuum drying [taylor1997 P3; rumondor2009a P1; vaneerdenbrugh2010b P6]; spin coating on coverslips, ZnSe or KRS5 for films [vaneerdenbrugh2010b P1; marsac2009 P4]; cryomilling of a 10% w/w dispersion with crystalline drug and melt-quenching for the low-polymer growth samples [kestur2010a P1, P2]. A physical mix of amorphous drug and polymer is the control without molecular mixing [taylor1997 C24, C26]. Amorphous state is checked by XRPD or polarized light [taylor1997 R1; vaneerdenbrugh2010b P2]. Further preparations: rotary evaporation at 45–60 °C with 24 h vacuum drying and cryomilling [wegiel2013a P1; hiew2022a P5]; spin coating on KRS-5, ZnSe or quartz purohit2015a P2, P9; freeze-drying of aqueous polymer blends [vaneerdenbrugh2012a P2]; 50 mg compacts for release [hiew2022a P9]. In the process papers the solid form was read by Raman and DSC (single Tg between those of the components; polymer-free marker band) [broman2001 P1, P2, R1, R4]. In the Bioperformance papers the dispersions were made by rotary evaporation from methanol and cryomilling, and checked by PXRD, DSC, PLM and SHG [hate2018 P5, P10; elkhabaz2017 R8]. In the physical-stability papers: spin-coated films (dichloromethane:ethanol, ethanol or chloroform, heated to 90 °C) for nucleation, growth and IR; rotary-evaporated and, for low polymer contents, cryomilled powders for DSC, sorption and PXRD [konno2006 P1, P3; rumondor2009c P1; rumondor2009b P1; kestur2010b P1]. Spin-coated films and rotary-evaporated powders of nine 50:50 drug–polymer pairs [rumondor2010c P1, P5]; rotary evaporation in 5% steps to find the highest amorphous loading [wegiel2013b P1]; rotary evaporation followed by cryomilling [wegiel2014 P2, P3]; a 10% PVP dispersion diluted with crystalline drug by cryomilling, then melted, to reach 3% PVP [kestur2013a P4; kestur2013b P2].

One composition of one polymer each; the medium of the CRH1 ASD runs in Figs. 7 and 8 is unclear [ilevbare2013a X15; almeidaesousa2016 X5]. The preparation route changes the result: films, rotary-evaporated powders and melt-quenched powders of the same pair can differ (PH–PAA) and the three characterization methods of one study used differently prepared samples [vaneerdenbrugh2010b C44; rumondor2009a X4]. Direct rotary evaporation below 5% w/w polymer would let felodipine crystallize, hence the cryomilled dilution, whose uniformity at the molecular level is assumed, not measured [kestur2010a C5]. A single Tg does not prove a single phase (see single-Tg miscibility criterion). Thermodynamically stable dispersions are hard to achieve; most are supersaturated and kinetically stabilized [marsac2009 C34]. The amorphous and single-phase state of the halogen-bond dispersions (up to 50 wt% for XPS, 90 wt% for IR) was not verified [que2020 X10]. Neat naringenin gave an amorphous film only by spin coating, quercetin not at all [wegiel2013a R7]. Amorphous probucol occurred only with PVP present, so carrier and form are confounded [broman2001 X13]; amorphous does not mean fastest release [broman2001 C35]. Two drugs in one dispersion can form a co-amorphous system whose amorphous solubilities are lower than those of the single-drug dispersions [trasi2019 C38]. In the Bioperformance papers making an ASD does not by itself remove dissolution-rate-limited absorption (high loading) [indulkar2016 C41]. The lowest loading did not always give the best dispersion (10% enzalutamide in HPMCAS crystallized), and the best solution inhibitor of 24 polymers made a dispersion no better than the crystals (F8, F24) [wilson2018 R6, R16; wilson2020 R1, R6]. A dispersion that is one phase when dry can separate under humidity and stay separated after drying [rumondor2009c C2; marsac2010 C53]. Nucleation was measured on films and Tg on powders made differently, so "one phase" is shown on one sample form and assumed for the other [konno2006 X12]. In the later felodipine growth papers too, below about 5% polymer rotary evaporation was expected to crystallize felodipine partly, hence the two-step dilution [kestur2013b C5]. Interactions were read on spin-coated films at other compositions than the stored powders [wegiel2014 X8; wegiel2013b X12]. Equal composition does not mean equal structure: six tacrolimus products with about one part HPMC per part drug differed in stability because of the manufacturing route [trasi2017 R1, R6, C47].

By calculation (Hoffman or heat-capacity free energy, sorption isotherm) and by the onset [almeidaesousa2015 P9, P10]; also by dissolving an amorphous film [ilevbare2013a P4]. In the ASD process papers: crystalline solubility by 72 h shake flask; amorphous solubility as the onset of light scattering during solvent-shift addition (liquid–liquid phase separation), in some cases only with a polymer present because the drug crystallised too fast without it [moseson2020a P7, P8; moseson2021 R1, R27]. In the Bioperformance papers it is estimated from melting data or as onset ÷ crystalline solubility raina2015c P3, P6. In the crystallization-from-solution papers: calculated with the Hoffman equation and a water-sorption correction; measured by dissolving a spin-coated film (ritonavir), by pH-induced precipitation, and by the onset of phase separation of a solvent-shifted solution [ilevbare2012b P4–P6; hsieh2012 P11; jackson2014 P6, P7]. In the ASD dissolution papers: calculated from melting data with the Hoffman equation and the measured crystalline solubility [konno2008a P7; alonzo2010 P12; alonzo2011 P8]; tested against flux relative to the crystal alonzo2011 R14.

The calculated advantage fails for high-melting drugs when the Hoffman free energy is used (F7) [mosqueragiraldo2015 C23, C24]. When additives raise crystalline solubility more than the LLPS concentration, the concentration-based ratio falls (felodipine, about 9.2 in buffer to 4.5 in 0.04% vitamin E TPGS and 5.0 in FaSSIF), while the maximum flux stays in the buffer band, so the advantage in activity is retained raina2015c R9, R11, C29. In dispersions, a dose below the amorphous solubility (IDM 50 against about 80 µg/mL) can still lose most of the advantage to crystals; a crystal-free bicalutamide dispersion never reached 40 µg/mL against 150 µg/mL [moseson2020a X17; moseson2021 C41]. The amorphous solubility of ritonavir alone is reduced to about 4 µg/mL in a co-amorphous granule [trasi2019 R4]. In the Bioperformance papers the ratio from concentrations falls with solubilizers although the flux advantage does not raina2015c C43; the atazanavir ratio differs between cards because the crystalline solubility differs (F1). In human fluid the flux of a solution at the amorphous solubility was about 30 times that of a crystalline suspension for posaconazole, so the advantage was not lost although the concentration ratio fell to about 5 (F18) [elkhabaz2019b C30, C45]. The printed atazanavir and darunavir ratios do not follow from the printed solubilities [alkalla2026 X1]. With a polymer in the medium the same ratio applied to the polymer-containing solubility overestimates the ceiling (danazol: 43 against an onset of 8–9 µg/mL) jackson2014 X9. Celecoxib's measured 2.5 µg/mL is far below the calculated value and is not discussed [ilevbare2013b X9]. The efavirenz run started at S about 2.3, the ratio of calculated amorphous to crystalline solubility being 2.4 [ilevbare2013c X4]. The printed estimates for the dihydropyridines differ between media in ways the equation does not allow raina2015b X2. Felodipine dispersion and solution readings of 14.1 and about 11 µg/mL exceed the 9.5 µg/mL estimate of the 2008 paper, measured by UV through a 70-µm filter [konno2008a X5]. The 25 °C indomethacin range (17–25 µg/mL) does not reproduce from its printed inputs (15.8–23.6) [alonzo2010 X2]. That a dispersion reached the same activity as the pure amorphous drug "may be purely coincidental" alonzo2011 C48. The tacrolimus ratio uses an ultracentrifugation value measured with 10 µg/mL HPMC in the buffer [trasi2016 X9]. For praziquantel the ratio is lower for the enantiomer because its crystal is more soluble, not because its amorphous solubility is lower [alvarenga2025 R1].

(1) As the onset of scattering during slow addition of a drug stock, confirmed by a polarity probe [ilevbare2013a P5; almeidaesousa2015 P1, P2]. (2) As the drug concentration in the supernatant after ultracentrifugation of a phase-separated solution [trasi2015a P2; almeidaesousa2016 P7]. (3) By dissolving an amorphous film under non-sink conditions [ilevbare2013a P4; indulkar2015 P13]. (4) As the break point of flux against donor concentration [trasi2015a P6]. (5) By calculation from the crystalline solubility, the free-energy difference and the water activity of the amorphous drug [ilevbare2013a E1]. The papers of 2017–2022 add quantitative ¹H NMR of the dissolved drug beside the drug-rich phase ueda2021 P3, P4, ultracentrifugation and the flux plateau in bile-salt media [lu2017b P3, P4], and an in-situ UV probe after a pH shift [sugihara2018 P6]. In the Bioperformance papers: (1) solvent-shift UV extinction, a methanol stock pumped into stirred buffer and the onset of scattering read at 350–455 nm [raina2014a P7; alhalaweh2016 P9; indulkar2016 P3; hate2018 P2; elkhabaz2017 P4]; (2) ultracentrifugation or centrifugation of a supersaturated solution and assay of the supernatant [mosqueragiraldo2018b P8; alhalaweh2016 P10]; (3) the flux plateau in a diffusion cell [raina2014a R16]; (4) estimate from the crystalline solubility with the Hoffman free-energy difference and a water-sorption correction [raina2014a E3, E5]; (5) the drug's own fluorescence, whose emission intensity rises, and whose peak may shift to shorter wavelengths, when the drug-rich phase forms, at the concentration found by ultracentrifugation in buffers and in pooled human intestinal fluid [elkhabaz2019b R3, C17]; (6) a solvent-shift or antisolvent addition into buffer with the polymer already dissolved, which gives the value in the presence of that polymer [ueda2020a R1; alkalla2026 R3, P7]; (7) for tacrolimus the value of about 50 µg/mL is taken from earlier work and used to define the dose-to-volume conditions of a dissolution test [hate2019 R7, C16; purohit2019 C15].

The onset depends on the method by up to about 25% (ritonavir in water at 37 °C, 34–40 µg/mL) and on the medium (water against 100 mM buffer, twofold), so values are comparable only within one medium [ilevbare2013a X14, X16, C40]. Four of the eight 2013 predictions cannot be recomputed from the printed inputs [ilevbare2013a X1]. The claim that the onset equals the amorphous solubility is partly circular in the free-energy comparison [almeidaesousa2015 X4]. For ionizable drugs the value is for the un-ionized form indulkar2015 C47. For ibuprofen the same medium gave 177.1 µg/mL by extinction and 154 µg/mL by NMR, with crystalline solubilities of 29.8 and 43.1 µg/mL, in two papers of one group ueda2020b R1, X1; ueda2021 R5, R2. In SDS a prediction from the crystalline solubility overshoots about twofold [indulkar2017a C16]. For a drug that degrades on melting (pazopanib) the calculation cannot be made [sugihara2018 C17]. In the Bioperformance papers the onset depends on stirring, addition rate and polymer [raina2014a X15]. The estimate fails in FaSSIF and for high-melting telaprevir [raina2015c C18; mosqueragiraldo2015 C24]; the nifedipine prediction cannot be reproduced [raina2014a X3]. The atazanavir crystalline solubility is 5 µg/mL in one card and about 1.1 µg/mL in two others, so ratios differ [hate2018 R1; alhalaweh2016 R6; indulkar2017a E3]. A second drug or a polymer in the drug-rich phase lowers it (F6, F7). The value depends on the medium and on the polymer in it: the more hydrophobic HPMCAS grade AS-HF lowered it for four drugs by a quarter to well over half, and a co-dissolved crystalline drug lowered it too (F19, F20) [ueda2020a R1, C13; alkalla2026 C1, C39]. In simulated and human intestinal fluids the amorphous solubility rose much less than the crystalline solubility, so the ratio of the two fell from about 36 to about 5 for posaconazole (F18) [elkhabaz2019b R1, R2]. The printed enzalutamide estimate cannot be reproduced from its inputs [wilson2018 R3, X1]. The amorphous solubility of atazanavir at 37 °C is 75 (cited), 79, 81.5 and 93.1 µg/mL in four later papers, with crystalline values of 1.4, 1.14 and 1.6 µg/mL, so the ratio is 56, 71 or 58 depending on the paper [hate2022 M1; elkhabaz2019b R1, R2; ueda2020a R1; alkalla2026 R3] (synthesis: 79/1.4, 81.5/1.14, 93.1/1.6).

Cloud point on cooling at 0.5 °C/min and clear point on heating at a known concentration (90° scattering at 474 nm), taking their mean as TL-L; or titration onset at a fixed temperature, which agrees closely with it [ilevbare2013a P12, C30]. For ECP films the phase boundaries of the ECP–methanol–water diagram were not determined and Fig. 12 is a schematic li2017 C38, X9. For the ritonavir–PVPVA spray solution the region is computed along the trajectory, not mapped: pseudo-ternary sections (total solvent, PVPVA, ritonavir) at 11.5:88.5, 15:85 and 30:70 show the region growing with the water content of the feed [li2020 R18]. In the release-mechanism papers: Position of droplet formation in time-lapse confocal images [yang2022a R10, R11]; polymer content of particles held in buffer with PVPVA [deac2023b R4, R5].

The binodal was measured on the low-temperature branch only, four concentrations, and the scan direction on that branch is unresolved [ilevbare2013a X6, X7, X17]. Raina and co-workers describe the binodal points as the theoretical onset of LLPS in a partially miscible drug–buffer system [raina2014a C27]. In the film papers the quaternary system with polymer is "much more complex" than the ternary one drawn, and Fig. 12B lacks the letters that the text uses li2017 C36, X9. The computed binodal depends on χ values that are partly assumed and inconsistently described, and keeps the 10:90 path outside the region although the powders separate [li2020 X10, X12]. The binode drug loads stated in the 2023 text do not exactly match the nanoparticle data [deac2023b X18]. MePHPH films separated at 80% RH where the computation predicts one phase, and single-phase PHPH mixtures lie inside the computed gap [deac2024b C34].

Not measured; derived from solubility ratios and used to reason about the effect of a second component on the amorphous solubility (F11) [trasi2015a E1, E4]. In the Bioperformance papers it is inferred from flux or from the shift of the LLPS concentration [alhalaweh2016 R12, R13; mosqueragiraldo2018b C37]. Through the membrane flux ratio, which follows the activity [lu2017 E2]. In the ASD dissolution papers it is inferred from flux plateaus and from the fit of ideal mixing [jackson2015 R9; trasi2015b R5].

None printed in the amorphous solubility and LLPS chapter; the Bioperformance, Crystallization from solution and ASD dissolution papers give the following.

A polymer mixing into the drug-rich phase is later reported to lower the drug's chemical potential and hence the amorphous solubility, as does a second drug in a co-amorphous system [ueda2020a C36; trasi2015b C36; alkalla2026 C12]. Ideal mixing was tested for one pair only [trasi2015a X9]. The chemical potential of the drug in the polymer-containing droplets was not measured in any of the polymer papers ueda2020b X4; ueda2021 X5, X6. The activity of telaprevir in the CA Sub-rich phase was not measured, so mixing and complexation are not separated [mosqueragiraldo2018b X12]. For the darunavir case the mixing is inferred; the centrifuged drug-rich phase was not analysed and the light-scattering onset with darunavir present was not measured (F20) [alkalla2026 X6]. A higher concentration is not a higher chemical potential in micellar media [jackson2016 C31]. In the two-drug dispersions the lower chemical potential is inferred, not measured [trasi2015b C36]. CAAdP 0.85 raised the ritonavir LLPS concentration by about 17%, unlike the other polymers [ilevbare2013d X7]. The lower chemical potential in the lopinavir dispersions is inferred from concentrations; χ was not measured and the model is used qualitatively [li2018 C46]. For delamanid in HPMCP-55 the authors invoke stronger or more extensive drug–polymer interactions lowering the chemical potential of the drug, which were not measured [nguyen2023c C41, C42].

Not measured quantitatively in the library: defects are inferred from Fourier-filtered TEM images (inverse FFT of one spot pair), from TEM contrast of grain boundaries, and from SEM and PLM of growth irregularities [moseson2018a P8; moseson2019 R1, R2, R3].

Defect density was not quantified, and bending or beam damage of thin organic crystals is not excluded as a source of bend contours moseson2018a X5, X6. The cited faster dissolution of crystals with higher defect density is not a measurement made in the ASD process papers [moseson2019 R23, X6]. The surface area and defect density of the residual crystals were not measured [moseson2020a X9; moseson2021 X10]. The inference for the agglomerated lot rests on the 5 min point; at 4 min L37 was still higher than L38 and their coarse tails differ [moseson2022c R10, R1, X5].

Isothermal hot-stage microscopy without mixing (time to loss of birefringence), variable-temperature XRD, SEM-EDX chlorine maps of drug in the polymer and TEM of films held at temperature [moseson2018 P10, R13; moseson2019 P3, P4, P5, P6, P7]. Nonisothermal TGA of a dispersion against crystalline physical mixtures with micronized (3.5 µm) or large (93 µm) crystals [moseson2020 P3, R13]; the broad endothermic shift of the non-reversing heat flow over 120–170 °C of a 30% crystalline extrudate moseson2021 R7. Residence-time series in a 5 mL recirculating micro-extruder (50 rpm, 145 ± 1 °C, 2–20 min) read by PXRD, with a population balance model fitted to the crystallinity [moseson2022c P15, P16, P18, P19].

The hot-stage sample was a partly dissolved extrudate, not a physical mixture moseson2018 X8. Between Tc and Tm dissolution was slower than melting because of lower thermal input, higher viscosity and no mixing moseson2018 C33. The nanoscale sequence of fragmentation is assembled from different particles [moseson2019 X11]. In DSC of miscible extrudates the crystals dissolve during the scan, so DSC under-reads them moseson2021 C12, C36. The fitted dissolution rate constant lies 5–40 times below the assumed values (0.5 and 4 µm/s), which the authors relate to melt viscosity; the printed sign conventions disagree and the size-dependence parameters are undetermined [moseson2022c R16, C30, X11, X21]. The supporting images are quiescent PLM at 170 °C, 25 °C above the melt temperature of the extrusions [moseson2022c X18].

Seeded growth from a supersaturated solution at a stated supersaturation and polymer concentration, with the growth rate from the desupersaturation profile; adsorption by AFM in liquid (globule spacing) and crystal habit by SEM [moseson2020a P10, P15, P16; moseson2021 P12]. In the crystallization-from-solution papers: Seeded desupersaturation: a fixed mass of seed crystals (0.010 g in 50 mL for ritonavir, celecoxib and efavirenz; about 2.5 mg in 18 mL for felodipine) is placed in buffer with predissolved polymer, supersaturation is created by adding drug in methanol, and a fibre-optic UV probe on second-derivative spectra gives the slope of the first 2 min (5 min for felodipine) as the growth rate [ilevbare2013b P10; alonzo2012 P4]. Eg is the ratio of slopes, in triplicate [ilevbare2012c P12]. Rates can be normalized by seed mass and BET area [ilevbare2013b E6]. Unseeded controls and turbidity checks test that the slope is growth [ilevbare2012a R9; ilevbare2013b R5]. Rotating-disc seeds (felodipine) and seeds grown in situ in the vessel (telaprevir); the ratio is R_0/R [schram2015 P2; schram2016 P5; lu2017a P1, P2]. In the ASD dissolution papers: initial slopes of desupersaturation curves with equal crystal surface areas assumed (Eq. 5) [konno2008a E5, P6, C28]; particle size against time by light scattering alonzo2011 P5; seeded desupersaturation with 7 mg seeds in 50 mL, initial slope without polymer against the slope over 16 h with polymer [xie2016a P4, P11]. In the supercooled liquid: by comparing seeded growth rates of the neat drug and of drug with a low polymer level at the same temperatures [trasi2014a P3; trasi2014b P2], with FTIR to assess drug–polymer hydrogen bonding [trasi2014b P4]. For the felodipine, bifonazole and nimesulide dispersions: a 10% w/w polymer dispersion made by rotary evaporation is diluted with crystalline drug in a cryogenic mill to 0.5–4.5% w/w, melted between cover glasses, nucleated over P2O5 at 40 °C (or at ambient temperature) and grown on a hot stage at 70–110 °C (or 18–100 °C), the slope of spherulite diameter against time being the rate [kestur2010a P1, P2, P3, P4; kestur2011 P1, P2, P3]. Whether the slope is a diameter or radius rate is not always stated, and the felodipine Fig. 3 slopes are about twice the Fig. 4 rates [kestur2010a X3; kestur2011 X12]. Hydrogen bonding is read separately by IR on films with 20–50 wt% polymer [kestur2010a P6, C6].

E_G falls as supersaturation rises and depends on seed preparation, polymer concentration and polymer chemistry, so values from different systems are not comparable [ilevbare2012b C56; schram2016 C13; moseson2020a X12]. Adsorption was imaged on bulk crystals only, not on residual crystals [moseson2020a X5]. The threshold for "effective" is applied loosely: Eg above 1 is the stated rule, but counts of effective polymers match a cut-off near 1.2, and polymers at 1.0–1.4 are called ineffective [ilevbare2013b X3; ilevbare2012c X7]. Eg depends on supersaturation, polymer concentration, pH and buffer strength, so ratios from different papers are comparable only roughly; absolute rates cannot be compared because seed mass, volume and methanol fraction are often not printed [ilevbare2012c X16; ilevbare2012b X12]. Most 2013 ritonavir values were reported before [ilevbare2013b X15]. A polymer that raises the solubility lowers S and so flatters Eg [ilevbare2013c X7; ilevbare2012a X6]. The ratio depends on the seed route and the dose [schram2016 R2]; rates are not normalised to area in the felodipine and telaprevir papers [schram2016 X12; lu2017a C5]. For a bile salt mixture the "intermediate" ratio rests on an off-scale value for STCDC and a wide spread for the mixture (about 16.7–54.5), and secondary nucleation at δ 7.9 was assumed small [lu2018 X7, C8]. Polymers that form globules on an amorphous drug surface "may" need a higher concentration to block growth, which was not measured [mosqueragiraldo2018a C41]. No precipitate was examined in the 2008 felodipine study and surface areas were assumed equal [konno2008a X4, X7]. Seeded and unseeded tests of felodipine with HPMC later found growth at 0.607 of the polymer-free rate but an estimated nucleation-rate ratio of 7.05 × 10⁻⁴, placing the main effect on nucleation [alonzo2012 R5, R17]. For nifedipine polymorph conversion the polymers are read as inhibiting α nucleation, not growth [raina2014b C32]. The celecoxib slopes are not printed and R_p was taken over 16 h while the plot covers 4 h, so the ratios cannot be reproduced; seeded growth was measured with HPMCAS only, not HPMC [xie2016a X8, X5]. A polymer that blocks nucleation need not stop growth: HPMCAS held unseeded paclitaxel at S = 88 for more than 16 h but only slowed growth on seeds [que2018 R3, C30]. In the supercooled liquid: inhibition of growth and of nucleation need not coincide either: for acetaminophen, HPMCAS inhibited nucleation best but growth least [trasi2012a C46]. Some additives accelerate growth [sato2016 R11]. The molar basis of the PVP level was not defined [trasi2014a X7]. The text of the felodipine paper overstates two points against its own figures: the 105 → 70 °C fall is about 51-fold, not 3 orders of magnitude, and PVP still inhibits about 3-fold at 110 °C where the polymers "appear to have no effect" [kestur2010a X1, X2]. The nimesulide text gives 2–10-fold, the plot up to about 21-fold [kestur2011 X7]. The hydrogen-bond order was measured at 20–50 wt% polymer and one temperature, and bifonazole–polymer bonding was not measured [kestur2010a X5; kestur2011 X9]. The melting-point argument used form I in physical mixtures while form II grows [kestur2010a X7]. Only growth was measured, not nucleation [kestur2010a X10]. The two media are explained differently: in solution the authors propose surface poisoning by adsorbed polymer [moseson2020a C48, C50], in the supercooled liquid drug–polymer hydrogen bonding and mobility, with demixing at the growth front as a "speculation" [trasi2014a C33; kestur2010a C40, C42], so ratios from the two media are not interchangeable (synthesis: different media and proposed mechanisms).

SEM of crystals recovered after a growth run, length and width of 50 crystals; the habit expected for the crystal structure from the BFDH tool in Mercury; PXRD relative peak intensities as a proxy for the exposed faces [ilevbare2012c P4, P7, P13; ilevbare2012a P6]. In the ASD dissolution papers: polarized light and SEM images; length over width [raina2014b P4; ilevbare2013d P8; chen2015a P10].

Means of 50 crystals have overlapping spreads; lengths with polymer are not below the seed length, so the ratio falls through width; three of the five S 7.6 rows are cited from earlier papers [ilevbare2012a X9; ilevbare2012b X4]. The printed 5.2 for the CAP Adp 0.85 sample recomputes to 5.14 [ilevbare2012b E15, X5]. A habit change does not by itself show how much growth is slowed: HPMC gave the smallest crystals and PAA the habit nearest the prediction [trasi2015 R12, C19]. The celecoxib needle direction along the c-axis is stated without an indexing basis [chen2015a X19]; DLS sizes of needle-shaped crystals are spherical equivalents and "likely to be unreliable" [ilevbare2013d C19]. The fall of the long-needle aspect ratio is read as fine crystals from secondary nucleation [que2018 C43].

Seeded desupersaturation: bulk crystals (D50 21.9 µm) added at 10, 25 or 50 µg/mL to 50 µg/mL indomethacin, with the growth rate in µg/s [moseson2020a P10, R4]; external form 1 seeds equal to 10% of a bicalutamide dose added to a crystal-free dispersion and read by Raman as a relative growth rate moseson2021 P13, P17, R22. Potency of residual seeds is read by comparing extrudates with seeded crystal-free dispersions at the same amorphous dose [moseson2020a P13, P14]. Seeded desupersaturation of posaconazole in FaSSIF V2: 20 or 100 µg/mL drug, HPMCAS at 0–9 times the drug, seeds of 0–10 µg/mL, percent left at 30 min shah2024 P16, R22. In the tacrolimus capsules crystals were read by XRPD (limit of detection 10%) and inferred from desupersaturation in non-sink dissolution purohit2018 R2, R7. For self-seeding in pH-shift tests: crystals are detected in the acid stage by SHG, which gives signal only from crystals, beside two-photon excited fluorescence of all drug, and their ratio estimates the share of the undissolved drug that is crystalline elkhabaz2019a P10, P11, E1; on dispersion films and powders by polarized light and SEM after 1 h in acid, with PXRD of the powders [nguyen2023c P12, P13, P14]. The effect is read from the release after the pH shift against a run without an acid stage elkhabaz2019a R6, R7.

Surface area and defects of residual crystals were not measured, so the extra potency is an inference [moseson2020a X9; moseson2021 X10]. Bulk seeds at a larger mass can grow more slowly than a small mass of embedded residual crystals; micron-sized seeds would "still not approximate" crystals below 100 nm [moseson2021 R22; moseson2020a C42]. Seed potency also depends on supersaturation and on the polymer [moseson2020a C54, C56]. The focused beam reflectance signature read as secondary nucleation for the crystal-containing 25% dispersion has the same form as the one read as primary nucleation for the crystal-free one, and the separate rates of growth and nucleation were not measured shah2024 K3, K4, X14, X13. Added seeds may not mimic crystals formed inside the ASD purohit2018 C7. Growth in the tacrolimus tests was inferred from concentration purohit2018 X9. Seeds did not always cause crystallization: at 240 and 100 mL the lower maximum supersaturation and the membrane loss kept growth small, at 40 mL it was large (F28, F29) hate2019 C46, C49. The rate units printed with the seeded runs are inconsistent with the profiles hate2019 X1. No seeding experiment was run in either pH-shift study, and the crystals on the delamanid dispersions were not identified, so the step from gastric crystals to lower release is inferred [elkhabaz2019a X8; nguyen2023c X11]. The 50% posaconazole dispersion also lost its supersaturation without an acid stage, so its crystal burden is not the only cause elkhabaz2019a C44.

Shake flask with excess crystals, about 48 h at the test temperature, separation by ultracentrifugation, HPLC or UV assay [ilevbare2013a P1; almeidaesousa2015 P3; indulkar2015 P3; mosqueragiraldo2015 P1; trasi2015a P1; almeidaesousa2016 P5]. A small amount of the stock solvent was added in one study to mimic the onset composition [almeidaesousa2015 P3].

Crystalline nicardipine turned amorphous in water within 2 h, so no crystal solubility exists for it indulkar2015 R1. The paclitaxel value is that of a hydrate [trasi2015a X12]. Solubility at pH 8 is quoted at pH 9 in the text of one paper indulkar2015 X2. Media differ between drugs in the 2013 table [ilevbare2013a X16]. Pazopanib's value is very low and carries a high experimental uncertainty [sugihara2018 C16]; the same drug and medium gave 29.8 and 43.1 µg/mL for ibuprofen in two papers ueda2020b R1; ueda2021 R2.

Spin-coated films of drug and polymer at several ratios, scored by polarized light microscopy at day 0, 1, 3 and 7 and expressed as an amorphicity index [vaneerdenbrugh2010b P1, P2, E1]; hot-stage growth of spherulites in melt-quenched films with low polymer levels after nucleation at 40 °C [kestur2010a P2, P3, P4]; rotary-evaporated bulk powders by DSC crystallinity [vaneerdenbrugh2010b P8, E5]. Only growth, not nucleation, was measured in the felodipine and bifonazole–nimesulide studies [kestur2010a X10; kestur2011 P3]. Induction time and seeded desupersaturation in solution; DSC of undercooled melts for the dispersion (a melting peak on reheating means nucleation during the hold); matrix crystallisation by microscopy and Raman of particles in buffer [moseson2021 P12, P15, P16; moseson2020 P7]. In the crystallization-from-solution papers: seeded growth rates and unseeded induction times with and without predissolved polymer under identical conditions, combined into a relative nucleation rate (F5, F6) [alonzo2012 P10, E9, E10]; induction times by UV extinction or absorbance [ilevbare2013c P9]; pH-shift precipitation followed by polarized light, WAXS and HPLC [hsieh2014 P2–P7]; fluorescence and SHG for droplet and crystal read-outs jackson2014 P7, P10. In the physical-stability papers: nucleation-site counts in films [konno2006 P2; marsac2006a P4; konno2008 P4], melt or film growth rates [kestur2010b P3; rumondor2010 P5], PXRD crystallinity of stored powders rumondor2009b P5. In the ASD dissolution papers: non-sink dissolution of dispersions or of amorphous drug with polymer in the medium; solvent-shift solutions with predissolved polymer; induction times by extinction or pyrene; Raman of slurries for the solid [konno2008a P3, P5; alonzo2010 P2, P4, P5; jackson2015 P3; ilevbare2013d P6].

Evidence table: crystallization inhibition.

Inhibition of growth, of nucleation and of overall crystallinity need not coincide: the best growth inhibitor of acetaminophen promoted nucleation [trasi2012a C46]. Polymers rank differently for different drugs [vaneerdenbrugh2010b C20; kestur2011 C29]. No polymer property tested at low levels (Tg, melting point, molecular weight) explained the order [kestur2010a C32, C34, C35]. The hydrogen-bond mechanism is a rank correlation over four polymers, and the demixing step is the authors' "speculation" [kestur2010a C38, C42]. Film results carried over to bulk powders for six of seven drugs [vaneerdenbrugh2010b C42, C43, C44]. The IR ranking of the polyphenol polymers separated the best and worst but not the middle ones, and polymer content differed from 30 to 80% in the stability comparison [wegiel2013a X4, X3]. The same ionic contact that protected lumefantrine dispersions slowed the dissolution of the enteric polymers [hiew2022a C53]. A polymer that inhibits in solution does not necessarily inhibit in the hydrating solid (HPMCAS and bicalutamide; polymers predissolved with amorphous nifedipine) [moseson2021 C51; raina2014b C11]. The mixing condition fails when the dispersion has phase separated, so water-induced separation can remove the benefit of the polymer [rumondor2009c C6]. Celecoxib with HPMC or HPMCAS at 22 µg/mL and 5 µg/mL polymer gave induction times of about 10–13 min in one laboratory and more than 8–16 h in another [ilevbare2013c R9; xie2015 R1; xie2016a R3]. Growth and nucleation data for the same polymers are often measured at different supersaturations [ilevbare2013b X15]. A polymer that raises the crystalline solubility or lowers the onset of phase separation changes the supersaturation it is credited with (F19) jackson2014 X9. In the solid state, equal effectiveness holds for felodipine nucleation when dry, not for growth (polymers differed several-fold) or in humid air [konno2006 C14; kestur2010a R6; konno2008 C10]. After moisture-induced phase separation the PVP level stopped mattering for crystallization at 93% RH rumondor2009b C20. The mechanism (barrier at the nucleus, hydrogen-bond coupling, polymer encaged in the crystal) was argued, not measured [konno2006 C40; kestur2010b C35, C42]. A polymer's effects on nucleation and on growth need not agree [trasi2012a C53]. Equal inhibition of two polymorphs is read as action in the liquid, unlike adsorption in solution [kestur2013a C6, C28]. PVP slowed felodipine bulk growth about sevenfold but surface growth only twofold [kestur2013b C32]. A polymer can also speed growth (PHB with high-Tg compounds) [sato2016 C29]; the factor depends on the polymorph and on the side of the growth maximum [tian2017 C37, C35]; and two grades of one polymer can favour opposite enantiomers [sato2015 C28]. PAA, the most effective polymer for bifonazole in melt-quenched films and the best growth inhibitor for acetaminophen melts, was ineffective on acetaminophen in solution (synthesis: contrast between papers) [kestur2011 C9; trasi2012a C46; trasi2015 C26]. In the ASD dissolution papers the ranking depends on the drug (PVP worked for indomethacin, not for felodipine, danazol or ritonavir) [alonzo2010 C22; konno2008a R7; jackson2015 R5; ilevbare2013d R12]. Solution screens predict only the rank order, and HPMC was misranked for felodipine [konno2008a C31; jackson2015 X8]. An additive can stabilize droplet size yet promote crystallization [ilevbare2013d R14].

By the DSC cooling–reheating classification [baird2010 P1]; by spin coating and dry storage [vaneerdenbrugh2010a P9]; by seeded growth rates [trasi2014a P1]; by induction times or precipitate crystallinity in supersaturated solution [ilevbare2013c R8; raina2015b R6]; by storage of neat glasses [baird2010 P7; nguyen2023a R4]. In dispersions: the fraction P(t) of 50 neat melts that had started to crystallize at 25 °C, followed for about 500 h [zhu2012 P3]; onset and t_{50} of the drug in the comelt by XRD and by two Raman bands [zhu2012 P5, P6, R23]; simultaneous SAXS/WAXS of the blend from 0.2 min to 24 h [zhu2011 P2]; crystallization of each neat melt cooled to room temperature under polarized light [zhu2011 P7, R6]. In solution: by the order of extinction onset and visible crystals, with and without a nucleation inhibitor raina2015a R4, R5, and by how long a drug-rich phase persists before the probe signal returns to the aqueous value (lopinavir 94 min; ritonavir not within 213 min) tres2018 R20, R21. For drug–polymer films of the spin-coating screen, by scoring crystalline area over 7 days and expressing it as an amorphicity index, averaged over drug/polymer ratios, polymers or drugs [vaneerdenbrugh2010b P2, P3, E1, E2, E3, E4]. The class of the pure drug is cited from the group's melt and spin-coating classifications [vaneerdenbrugh2010b R11]. In the process papers: isothermal DSC holds of the undercooled melt between Tg and Tm followed by reheating (a melting peak means nucleation during the hold): posaconazole 1 h at 130–160 °C and bicalutamide 1 h at 130–180 °C nucleated, indomethacin did not in 4 h at 130–150 °C [moseson2020 P7, R7]; in solution by induction time and seeded growth moseson2021 P12, R3. In raina2014a the tendency is cited, not measured: felodipine a slow and nifedipine a faster crystallizer [raina2014a R20]. The crystallization-from-solution papers cite the classes from cooling the melt (Baird 2010) and from rapid solvent evaporation (Van Eerdenbrugh 2010), and measure the solution tendency by the pH-shift titration or the unseeded induction time [hsieh2012 P7; ilevbare2013c P9]. In the physical-stability papers, by nucleation in films and crystallization on heating of melt-quenched glasses [marsac2006a P4, P10]. In the ASD dissolution papers: by time to crystallize in aqueous slurry (Raman) and by microscopy of wetted particles [alonzo2010 P5; raina2014b P4, P6].

The tendency depends on route and medium: rapid and slow drying differ strongly [vaneerdenbrugh2010a C26], and the solid-state ranking need not hold in dilute aqueous solution [hsieh2014 C30]; in the dihydropyridine series, by contrast, the solution ranking and the two drugs that recrystallize from the melt coincide raina2015b C28. No single descriptor captures it [vaneerdenbrugh2010a C45]. In dispersions, heterogeneous nucleation by the carrier can break the link to the pure drug [zhu2012 C41]: the link "was found to break down for the slowly crystallizing compounds" ibuprofen and fenofibrate, because crystalline PEG apparently promotes their heterogeneous nucleation [zhu2012 C41]. The neat ibuprofen and fenofibrate statistics stop at 3 weeks and the dispersion runs have unstated replicates [zhu2012 X5, X6]. The neat-melt entries of the 2011 PEG paper (chlorpropamide, nilutamide and ibuprofen "within a few hours") come from single melts and sit beside the 50-sample statistics of the 2012 paper for ibuprofen [zhu2011 R6; zhu2012 R2]. The polymer fraction needed per drug is not tabulated, although the day-by-day scores show the highest drug fraction that stayed amorphous [vaneerdenbrugh2010b X13, R2]. The class of flurbiprofen depends on the solvent (class I from ethanol–dichloromethane, class II from ethanol), and the HPMC and HPMCAS films used the mixed solvent [vaneerdenbrugh2010b X2, X12]. Class III drugs were left out on the expectation of stability over a week [vaneerdenbrugh2010b C13]. Polymer and chemistry change the ranking within a class, so the class is a risk flag, not a prediction (see F15) [vaneerdenbrugh2010b C20]. The link from the neat drug to the dispersion held in the spin-coating screen [vaneerdenbrugh2010b C1] but broke down for ibuprofen and fenofibrate in PEG [zhu2012 C41] (synthesis: the two papers differ in carrier and test). Posaconazole, indomethacin and bicalutamide were chosen as class III glass formers, yet neat posaconazole and bicalutamide nucleated within an hour between Tg and Tm; an isothermal hold near Tm tests nucleation differently from the cooling and reheating test that defines the class (synthesis) [moseson2020 C5, R7]. The granulation conclusion rests on one pair of slow crystallizers [trasi2019 C53]. In the Bioperformance papers the tendency of the neat melt does not carry into solution with polymer (F10). A rapid crystallizer in the neat state and in solution can still be held by one polymer: HPMCAS-HF kept pretomanid in solution for over 1000 min at S = 5 while most enteric polymers failed within minutes (F27) [nguyen2023a R5]. In the pH-shift study class III covers all type 2 bases and also three of the four classifiable type 1 bases, so it does not separate the types [hsieh2012 X1]. Efavirenz is class III yet crystallizes in minutes at S about 2.3 [ilevbare2013c R9.a, X4]. Failing to crystallize in the DSC screen "does not guarantee" slow crystallization in solution raina2015b C29. The carry-over of the nifedipine–felodipine ranking into PVP dispersions rests on one drug pair and one polymer, and no PVP level giving equal stabilization of the two drugs was found [marsac2006a X7]. A favourable class did not protect a marketed product whose process left the drug outside the polymer: tacrolimus forms a stable glass, yet one generic crystallized faster than the pure amorphous drug [trasi2016 C13; trasi2017 C28, C47]. In the ASD dissolution papers nifedipine and felodipine have similar Tg and moisture sorption (cited) yet differ in water only when polymers are present [raina2014b C27, R17]; the felodipine–nifedipine comparison is cross-study [raina2014b X16]. Delamanid behaved like tacrolimus and bicalutamide, whose residual crystals the polymer does not poison, rather than like indomethacin [duong2022a C58].

By visual particles and birefringence under polarized light raina2015a P7; by the return of a probe signal toward its aqueous value (pyrene I1/I3, PRODAN lifetime) raina2015a R16; tres2018 R21; by a long lifetime component of the drug itself tres2017 R1.g; in outside work by PXRD of compacts [bhardwaj2018 R3].