How the alcohol co-solvent and stray water decide whether an itraconazole–HPMC film mixes or separates

Background and motivation

An amorphous solid dispersion (ASD) keeps a poorly soluble drug in its amorphous form inside a polymer. The amorphous form has a higher free energy than the crystal, dissolves faster and can supersaturate, but it tends to crystallize; the polymer is there to stop that (Simonelli 1976, Rumondor 2009, Caron 2013, refs 1–3) C1. Whether the mixture is stable depends on how much drug the polymer can hold: above the amorphous miscibility limit the blend splits into a drug-rich and a polymer-rich phase, a process called amorphous–amorphous phase separation (AAPS), and the drug-rich domains can then crystallize and erase the solubility advantage (refs 4–6) C2.

Most ASDs in research, and many in production, are made by evaporating a solvent, for example by spray drying. The authors point out that few studies ask how the solvent itself shapes the result; most of that work concerns blends of two polymers (Bank 1971, Paudel 2012, Ansari 2008, refs 7–9). A drug and a polymer usually dissolve very differently in a given solvent, so the solvent is "expected" to change the phase behavior of the blend, and mixed solvents make this more complicated (Al-Obaidi 2009, Wulsten 2009, refs 10–11) C3. Mixed solvents are hard to avoid with HPMC, because most HPMC grades do not dissolve in a single organic solvent C31. Low-resolution methods such as DSC and polarized light microscopy "may" also overestimate miscibility (refs 8, 12) C4.

The paper continues a series on itraconazole (ITZ) with hydroxypropyl methylcellulose (HPMC). Purohit and Taylor (2015, ref 12) found that ITZ–HPMC made from dichloromethane (DCM) and methanol (MeOH) at uncontrolled humidity was phase separated, with ITZ-rich discrete domains C5·R6. The group then linked this separation to water present during production: ITZ–HPMC films from DCM:MeOH were miscible only when spin coated at 7% relative humidity (RH), and as little as 1% water in a DCM:MeOH spray solution caused phase separation (Mugheirbi 2017, ref 13) R5·R7. Water in the solvent is known to separate other dispersions too (Saboo 2017, Rumondor 2011, refs 33–34) C30.

Against this background, the objective is to study how the starting solvent composition changes the conformation of HPMC and, through it, the phase behavior of ITZ–HPMC ASDs made by spin coating A1. Five alcohols serve as co-solvents with DCM: methanol (MeOH), ethanol (EtOH), isopropanol (IPA), n-propanol (PrOH) and n-butanol (BuOH) M3–M7. Films are examined with high-resolution imaging, and the solvent, the drug and the polymer are each studied separately to explain what the films show.

Background concepts

Azeotropes and what is left when a solvent evaporates. When a mixture of two liquids evaporates, the more volatile one usually leaves first, so the remaining liquid becomes richer in the other. An azeotrope is a composition that evaporates without changing, as if it were a single liquid. The authors cite these values: EtOH and water form an azeotrope at 95.5:4.5 % EtOH:H₂O (boiling at 78.2 °C; later given as 95:5 % by volume), IPA at 87.9:12.1 % (80.4 °C), and PrOH and BuOH at 71.7 and 55.5 % alcohol, while MeOH forms none M4·M5·M6·M7·M3·X6. The consequence they build on is simple: a MeOH–water mixture loses MeOH first and becomes richer in water, whereas an EtOH-rich mixture below its azeotropic water content loses water along with the alcohol and does not become richer in water C13·C34.

Hansen solubility parameters. Hansen splits the cohesive energy of a substance into three parts: dispersion (non-polar) forces, polar forces and hydrogen bonding, each expressed as a partial solubility parameter ${\delta }_d$, ${\delta }_p$, ${\delta }_h$ in MPa^1/2. Equation 1 combines them into the total solubility parameter E1:

In plain terms: a substance is a point in a three-dimensional space whose axes are the three partial parameters, and $\delta$ is its distance from the origin; the subscript $i$ names the species. For a polymer, this point is the centre of a sphere, the Hansen solubility sphere, whose radius $R_0$ is called the interaction radius (Figure 1). For a mixture of solvents, the parameters are averaged by volume fraction (equation 2) E2:

In plain terms: each component $1,2,\dots$ contributes its solubility parameters ${\delta }_1,{\delta }_2$ in proportion to its volume fraction $\Phi$, so adding a second solvent moves the mixture's point through the space. The distance between a polymer $p$ and a solvent $s$ is equation 3 (the paper prints the outer bracket as square brackets) E3:

In plain terms: $R_{ps}$ is the distance between polymer and solvent in Hansen space, with the dispersion difference counted four times. If $R_{ps}$ is smaller than $R_0$, the solvent lies inside the sphere and is expected to dissolve the polymer; if larger, it is not E5·C6. The authors use the HPMC values reported by Archer (1992, ref 18) for the E3 grade: ${\delta }_d$ = 17.4, ${\delta }_p$ = 14.9 and ${\delta }_h$ = 19.3 MPa^1/2, with $R_0$ = 7.6 MPa^1/2 M2. They warn that solubility parameters "cannot be used to quantitatively predict" polymer solubility and that the assumption of purely additive, non-specific interactions "may" cause inaccuracies; they use the approach only to judge potential miscibility C7.

Radius of gyration and radial distribution function. The molecular dynamics (MD) part of the paper describes an HPMC chain by two quantities. The radius of gyration $R_g$ measures the size of the chain (equation 4) E4:

In plain terms: $R_g$ is the mass-weighted root-mean-square distance of the chain's $N$ atoms (mass $m_i$, position ${\overrightarrow{r}}_i$) from its centre of mass ${\overrightarrow{r}}_{COM}$; a larger $R_g$ means a more extended chain, a smaller one a more coiled chain. The radial distribution function (RDF) gives the relative likelihood of finding solvent atoms at a given distance from the polymer atoms; a value of 1 is the bulk average, so a peak above 1 marks a solvent shell that is enriched next to the polymer R20·R21.

Imaging tools for phase separation. ITZ is autofluorescent, so in fluorescence optical microscopy (FLOM) drug-rich regions appear bright; the method cannot resolve domains below the optical diffraction limit, which is why the authors add bright-field transmission electron microscopy (TEM), which shows nanometer-scale domains P6·P7·C52. Lorentz contact resonance (LCR) is an atomic force microscopy (AFM) mode in which the probe's contact-resonance frequency depends on the local stiffness, so driving the probe at one frequency highlights one component. Nanothermal analysis (nanoTA) heats the AFM tip and records the temperature at which it starts to sink into a softening region, which the authors take as an indication of the glass transition P8·P9. Finally, a clear point is the temperature at which a drug slurry fully dissolves on heating (here taken as the crystalline solubility), and a cloud point the temperature at which the cooled solution turns turbid P10·C16.

Methods in detail

The study has one fixed formulation and five solvent systems. Every film is a 30:70 w/w ITZ:HPMC blend (HPMC as METHOCEL E3) dissolved at 10 mg/mL total solids in a 50:50 v/v mixture of one alcohol with DCM P1·M1·M2. Around this core the authors set up five lines of work, two on the films and three to explain them.

Films: spin coating under controlled temperature, speed and humidity

Problem. The group's earlier work blamed water for the phase separation of ITZ–HPMC made from DCM:MeOH. If water acts through the solvent, the alcohol paired with DCM should change how much water matters C5. What they did. A 100 µL aliquot of each stock solution was dispensed onto a spinning substrate (500 rpm for 3 s, then 2000 rpm for 30 s, the protocol of ref 13) at an environmental RH of 50 ± 5%. A heat lamp above the KW-4A spin coater set the sample zone to 21, 30, 35, 40 or 50 °C, checked with an infrared thermometer P2. To separate the effect of temperature from that of evaporation rate, a slower protocol (500 rpm for 15 s, twice) was run at the same temperatures P3. Films were also spin coated at 21 °C and 20% RH for all five alcohols and at 7% RH for IPA, PrOH and BuOH; the paper does not describe how these lower humidities were set P4·X8. All films were kept under vacuum for at least 24 h P2.

How the films were read. Fluorescence microscopy (excitation 330–380 nm, emission from 420 nm, ×100 objective) screened films on quartz, and bright-field TEM (LaB6 source, 200 keV) examined films spun directly onto carbon-coated copper grids P6·P7. One film, made from DCM:IPA at 21 °C and 20% RH, was mapped by AFM topography, Lorentz contact resonance (spectra from 1 to 200 kHz, imaging at the resonance of one component) and nanothermal analysis, with the tip temperature calibrated on polycaprolactone, polyethylene and poly(ethylene terephthalate) (melting at 55, 116 and 235 °C) P8·P9·M10.

The solvent: how water changes evaporation

Problem. If water concentrates as some solvents evaporate, this should show in how quickly a solvent mixture loses weight. What they did. About 50 µL of each DCM:alcohol mixture, dry or with 4% added water (and DCM:EtOH also with 10%), was placed in a platinum TGA pan at 24 ± 0.5 °C under a nitrogen purge of 60 mL/min, and its weight was logged every minute until it stopped changing for at least 10 min; the derivative of the weight curve gives the evaporation rate P12. The methods state that the drug-polymer stock solutions were measured too, but only the pure solvent mixtures are shown, and the basis of the water percentage (volume or weight) is not given P12·X7.

The drug: how soluble is ITZ in each alcohol

Slurries of crystalline ITZ (1 to 5.5 mg/mL) in each pure alcohol were heated at 0.5 °C/min with stirring to 5 °C below the alcohol's boiling point, recording the clear point, and then cooled at the same rate to 10 °C, recording the cloud point. The precipitate was filtered and examined by powder X-ray diffraction and modulated DSC (25 to 200 °C at 5 °C/min) P10·P11·P14. Because the starting material is crystalline, the authors treat the clear point as the crystalline solubility C16.

The polymer: is HPMC dissolved as single chains or as aggregates

Problem. A clear solution can still contain polymer aggregates. If HPMC is aggregated in some solvents, the drug and polymer may never mix intimately. What they did. HPMC E3 at 7 mg/mL in 1:1 DCM:alcohol (the HPMC concentration of the 10 mg/mL, 30:70 stock solutions, but without drug) was filtered and split. One half evaporated under a nitrogen purge ("dry"), the other at 50% RH ("wet"), until about half the volume had gone; dynamic light scattering then measured the hydrodynamic diameter, averaging at least four measurements and excluding peaks below 5% intensity P13. Species below 10 nm were counted as single chains and those above 10 nm as aggregates, acknowledging that only relative sizes are compared C18. The same HPMC solutions were spin coated dry (nitrogen purge) or at 50% RH and their topography imaged by AFM P5. Note that the caption of Figure 8 calls the dry condition 20% RH, which conflicts with the nitrogen purge in the methods X4.

Calculation: Hansen distances

Using equations 1–3 with the Archer values for HPMC and alcohol parameters from the Hansen website, the authors computed the distance from HPMC to each pure alcohol, and, using equation 2 with the solubility parameters reported by Jiang 2014 (ref 21), to IPA, PrOH and BuOH mixed with 0 to 100% water P15·R16·R17. The axis of Figure 9b is labelled % w/w, while equation 2 mixes by volume fraction; the parameters used for the alcohols and water are not printed, so the distances cannot be recomputed from the paper X7.

Model: molecular dynamics of HPMC chains

Problem. DLS gives size but not conformation, and it cannot follow the polymer as the solvent composition shifts. What they did. A 20-monomer HPMC chain (537 atoms) was built so that its 60 substitution sites carry 30% methoxyl, 10% hydroxypropoxyl and 60% hydroxyl groups, the pattern listed in Table 1 P16·X10. It was placed in pre-equilibrated boxes of about 15,000 solvent molecules (cubic boxes with edges up to 12.0 nm) of 100% alcohol, 70:30 or 50:50 alcohol:DCM (molar fractions), for EtOH, PrOH and IPA, using GROMACS 5.0 with the CHARMM force field. After minimisation and 5 ns each of NVT and NPT equilibration at 294 K and 1 bar, a 20 ns production run gave the radius of gyration (equation 4) and the RDF between all polymer atoms and all solvent atoms P16. A second set with four chains compared alcohol–DCM with 55:30:15 alcohol:DCM:water P17. To check that the chain can sample its conformations, the backbone torsion C2-C1-O1-C4 was scanned rigidly on 19 dimers cut from the model P18·R22. MeOH and BuOH were not simulated, and the water-free ratio of the four-chain runs is not stated X9.

Experimental analysis

Result 1: humidity flips the outcome, in opposite directions for lower and higher alcohols

Table A collects the film outcomes. At 21 °C and 50% RH, films from MeOH, EtOH and IPA showed bright, high-intensity regions in fluorescence, which the authors read as a non-uniform drug distribution, while films from PrOH and BuOH were uniformly dark, "inferred" to be miscible (Figure 2) R1·C8. TEM of the samples with the clearest fluorescence evidence showed the same pattern: MeOH films carried a few large ring-shaped domains on an interconnected two-tone background, which the authors call a second form of phase separation in the continuous phase; EtOH films carried many round domains of widely varying size; a BuOH film was featureless, "confirming miscibility" despite the 50% RH (Figure 3) R2.a–e·C10. No TEM of a PrOH film at 50% RH is shown R2.

Table A. Phase state of 30:70 ITZ:HPMC films spin coated at 21 °C, from Figures 2–4 R1·R2·R4·R5. Descriptions of the textures are readings of the micrographs; the authors did not quantify domain size or area.

Lowering the humidity to 20% RH changed the picture. MeOH films stayed phase separated, "in good agreement" with earlier work, but EtOH films became homogeneous in both fluorescence and TEM, which the authors take as a miscible system (Figure 4a, b) R4.a·R4.b·C12. The higher alcohols went the other way: IPA films were heterogeneous without discrete domains, and PrOH and BuOH films were, "somewhat surprisingly", phase separated (Figure 4c–e) R4.c–e·C12. Dropping to 7% RH did not help IPA, PrOH or BuOH; their TEM images showed a mottled or network-like dark and light texture (Figure 4f–h). The authors call this "counterintuitive", because MeOH films had become miscible at 7% RH in their earlier study R4.f–h·R5·C12. The summary is that water had opposite effects: with the lower alcohols dry conditions "were necessary" for a mixed film, while with PrOH and BuOH humid air was needed C32·C31.

Result 2: faster evaporation gives smaller domains

Within the 50% RH series, the domains in MeOH and EtOH films became smaller as the spin-coating temperature rose (Figure 2). In fluorescence, MeOH films went from coarse domains at 21 °C to fine dots at 30–40 °C and an almost uniform film at 50 °C, and EtOH films stayed densely textured up to 40 °C and looked uniform at 50 °C R1.a·R1.b. IPA films lost their discrete domains: in TEM they had dark domains on a fine worm-like background at 21 °C, fewer and smaller domains at 30 °C, few at 35 °C, almost none at 40 °C and none at 50 °C, where the film "appeared more homogeneous" (Figure 3f–j) R2.f–j·C9. When DCM:MeOH films were spun at 500 rpm instead of 2000 rpm, domains remained visible at every temperature up to 50 °C (Figure S1) R3. The authors conclude that the smaller domains at high temperature are "mainly" due to faster evaporation ("suggesting") C11. Note by the AI writer (not discussed in the paper): the domain sizes were judged by eye and the film evaporation rate was not measured, so this is a qualitative trend X11.

Result 3: in the IPA film, HPMC forms the raised network

The one film analysed for composition, made from DCM:IPA at 21 °C and 20% RH, had an uneven surface with protruding features in a semi-continuous arrangement (Figure 5a, 5 × 5 µm, height range ±25 nm) R8.a. Contact-resonance spectra peaked at about 129.3 kHz in one phase and about 130.0 kHz in the other; the authors take ITZ to be the stiffer phase, so at a 130 kHz drive ITZ-rich regions appear bright and HPMC-rich regions dark (Figure 5b, c) R8.b·R8.c. Nanothermal analysis gave softening at about 115 °C for the raised features (the plotted trace turns down at about 111 °C) and about 60 °C for the continuous phase, which the authors call "consistent with" the glass transitions of HPMC and ITZ; they therefore read the raised network as HPMC-rich and the surrounding phase as ITZ-rich (Figure 5e) R8.e·C14. The Figure 5 caption locates this film "in Figure 4a", which is the MeOH film; the IPA film at 20% RH is Figure 4c X1.

Result 4: water slows the evaporation of the methanol blend

Table B. Time to complete evaporation (min) of pure DCM:alcohol mixtures at 24 °C, read from Figure 6 (approximate), with the authors' description R9·C15.

The water effect on evaporation follows the azeotrope pattern of the background section (Table B). With 4% water, DCM:MeOH started at about the same rate as the dry mixture but then slowed sharply and took about 40 min instead of about 12 R9.a·R9.b. DCM:EtOH was not slowed by 4% water but took about 73 min with 10% water R9.c–e. DCM:IPA was 2.5 min slower with 4% water, and DCM:PrOH and DCM:BuOH were unchanged R9.f–k. The two slowed mixtures both showed a small rise in rate just before the end R9.b·R9.e. The authors present this as "support for these suppositions": water concentrates in DCM:MeOH, and in DCM:EtOH only above about 5% water C36·C37. Two cautions apply. The plotted ratios, about 3.3-fold for MeOH and 5.4-fold for EtOH against the dry mixture, are smaller than the 4-fold and 7-fold stated in the text, although the direction agrees X3. And these are solvents without drug or polymer, evaporating at 24 °C from a pan, not a spinning film P12.

Result 5: the drug is least soluble in isopropanol

Clear points rose with ITZ concentration in every alcohol (Figure S2a). At a given temperature crystalline ITZ was most soluble in MeOH and least in IPA: 1 mg/mL cleared at about 37.5 °C in MeOH, 39 °C in PrOH, 46 °C in EtOH, 48 °C in BuOH and 53 °C in IPA, and 3 mg/mL at about 49, 56, 60, 60 and 68 °C respectively R10.a–e·C17. On cooling, turbidity appeared far below the clear points; for the four alcohols other than IPA the cloud points of 4–5.5 mg/mL solutions (4.5–5.5 mg/mL for MeOH) fell between about 10 and 34 °C, while IPA solutions clouded at only 1.5 to 3 mg/mL (Figure S2b) R11·C17. The precipitates were predominantly amorphous: PXRD showed a halo with some Bragg peaks for MeOH, EtOH, IPA and PrOH and a halo only for BuOH (Figure S3) R12. The authors describe the DSC traces as a weak glass transition plus melting for the lower alcohols, and a glass transition followed by two liquid-crystalline transitions (guides at about 60, 73 and 90 °C) for the higher alcohols (Figure S4) R13·C17. Note by the AI writer: in the printed Figure S4 all five traces, including the BuOH one with no Bragg peaks, show a large endotherm near 160–170 °C, so the DSC grouping is less clear-cut than the text says X5.

Result 6: HPMC aggregates in the higher alcohols unless water is present

Table C. Hydrodynamic diameter of drug-free HPMC (7 mg/mL in 1:1 DCM:alcohol) and the share of scattered intensity, read from Figure 7 (approximate; the MeOH and EtOH aggregate percentages are the values given in the text) R14.

Light scattering of drug-free HPMC solutions (Table C) showed single chains of about 4–5 nm in every starting solution, but only the MeOH and EtOH solutions were free of aggregates; IPA, PrOH and BuOH solutions also contained a population above 40 nm, which carried most of the scattered intensity R14.a·R14.d·R14.g·R14.j·R14.m·C19. Evaporating half the solvent under nitrogen made HPMC aggregate in every solution, but MeOH and EtOH kept most of the polymer as single chains (aggregates 17 ± 4% and 11 ± 5% of the intensity), whereas IPA, PrOH and BuOH lost their single-chain population entirely R14.b·R14.e·R14.h·R14.k·R14.n·C19. Evaporating at 50% RH gave the opposite result for PrOH and BuOH: their aggregates disappeared and only single chains remained, while IPA was the only alcohol with substantial aggregation (about 117 nm) R14.i·R14.l·R14.o·C19. Drug-free HPMC films matched this: smooth when the solution held single chains (MeOH and EtOH, dry and wet; PrOH and BuOH, wet), rough and corrugated when it held aggregates (IPA, dry and wet; PrOH and BuOH, dry) (Figure 8) R15·C20. All the drug–polymer stock solutions had looked visually clear R23. The authors conclude that water was "necessary" to solubilize HPMC in the PrOH and BuOH systems C41.

Note by the AI writer (not discussed in the paper): DLS percentages are intensity-weighted, and large particles scatter far more light than small ones, so a 70% intensity share of aggregates may correspond to a small fraction of the polymer by mass X12. The measurements were also made without the drug X15.

Result 7: Hansen distances point the same way

Table D. Hansen distances (MPa^1/2) between HPMC and the alcohols, read from Figure 9 (approximate). Water was added in 10% steps; all three higher alcohols converge at about 4.52 for pure water R16·R17.

The calculated distances place MeOH and EtOH inside the HPMC solubility sphere and the three higher alcohols outside it (Table D, Figure 9a) R16. The authors read this as HPMC solubility falling with alcohol chain length, and also with branching in the case of IPA, "clearly indicating" a decrease in polymer solubility with chain length C21. Adding water pulls all three higher alcohols inside the sphere, so HPMC "is expected to be more soluble in alcohols containing water" (Figure 9b) R17·C22. The note by the AI writer here is that Figure 9a draws the R₀ line at about 7.75 although the text gives 7.6, which does not change any inside or outside call X17.

Result 8: in simulations, water swells HPMC only in n-propanol

Table E. Average radius of gyration (nm) of HPMC in the MD runs, read from Figures 10 and 11 (approximate); "with water" is 55:30:15 alcohol:DCM:water R18·R19.

For a single chain, the smallest average R_g occurred at 50:50 alcohol:DCM for all three alcohols, the largest in pure PrOH and IPA, and for EtOH at 70% alcohol (Table E, Figure 10) R18. The authors read this "roughly" as a more extended conformation that "may be favored" in alcohols C23. The RDFs showed two solvation layers at about 0.55 and 0.95 nm; at 70:30 alcohol and DCM were equally likely near the chain, while at 50:50 alcohol became more likely and DCM less (Figure 12) R20·C26. With water present, two water shells formed at about 0.41 and 0.78 nm, with water the most likely neighbour of the polymer and more so in the PrOH and IPA systems than with EtOH (Figure 13) R21·C26. The four-chain R_g rose with water in PrOH (about 2.17 to 2.33 nm), barely in EtOH and not at all in IPA, which the authors describe as the polymer "clearly" expanding in PrOH (Figure 11) R19·C25. The torsion scans gave an average backbone barrier of about 7 kcal/mol, which the authors judge "moderately weak" at around 300 K, so it "may have limited impact" on sampling R22·C24.

Note by the AI writer (not discussed in the paper): each run spanned about 1.4–2.9 nm in R_g, far wider than the differences between compositions, and the standard-deviation boxes of the four-chain runs overlap for EtOH and PrOH; with one trajectory per condition and no statistical test, these are indications rather than measured differences X13·X9.

Discussion

The authors' explanation. The discussion sorts the alcohols into "lower" (MeOH, EtOH) and "higher" (IPA, PrOH, BuOH) and explains the films through what the evaporating liquid can still dissolve C31. The chain below follows Sections 5 and Figure 14 of the paper; the hedges are the authors'.

Lower alcohols. The authors take it as "well established" that more water is absorbed at higher humidity, and state that for EtOH at 20% RH the equilibrium water content is below 3% by volume, just under the 95:5 v/v azeotrope C33. An EtOH-rich solution below the azeotrope leaves an EtOH-rich residue, so water does not concentrate; above about 5% water, water concentrates. MeOH has no azeotrope and is "anticipated" to evaporate faster than water, so water concentrates, which is why MeOH-based ITZ–HPMC needs very low humidity (7% RH) C34·C37. With ITZ's low aqueous solubility, discrete drug domains are "anticipated" when water concentrates and the evaporation is slow, and the authors say this framework "largely" rationalizes the observations C37. They add that the analysis ignores how DCM and the dissolved drug and polymer shift the azeotropic compositions, which they call unknown factors C35.

Higher alcohols. For IPA, the raised HPMC-rich features in the AFM image together with HPMC aggregation in solution and in drug-free films "suggest" that the separation at 20% RH comes from the polymer's incompatibility with the solvent C38. At higher humidity and lower temperature, IPA films show discrete, spherical domains in an inhomogeneous background; since ITZ is least soluble in IPA and water lowers its solubility further, these domains are "likely" drug-rich, so IPA combines poor drug and poor polymer solubility C39. PrOH and BuOH form azeotropes at high water content, so they concentrate in alcohol rather than water, which is "likely" why water did not change their evaporation time C40. In these alcohols water was "necessary" to dissolve HPMC, judged from the separated films at low RH, the aggregation on dry evaporation and the dry film texture C41. The authors also bring in viscosity: the literature values rise from 0.55 cP (MeOH) to 2.95 cP (BuOH), and higher viscosity is "expected" to slow the diffusion of drug molecules into drug-rich domains C42·M3–M7.

The role of the polymer. Although all starting solutions looked clear, HPMC was aggregated in the higher alcohols, and the authors hold this solvophobic behavior responsible for the irregular HPMC films made under dry conditions C43·R23. Their central statement is that the data "suggests" HPMC must exist as discrete molecules in an extended conformation for an intimately mixed ASD; discrete domains and spinodal-like textures "seem" to come from drug–water incompatibility, and network-like corrugations are "likely" caused by solvent incompatibility with HPMC C44·C45. Water is "anticipated" to keep HPMC as single extended chains in the higher alcohols, and extended conformations have been reported to promote drug–polymer interaction and prevent crystallization (Mosquera-Giraldo 2016, ref 40) C46. The simulations are used as support: the solvation shell around HPMC "may" change continuously as the alcohol:DCM ratio shifts during evaporation, so film miscibility should depend on the polymer's conformation throughout evaporation, and water "will" solvate HPMC in mixtures that take up water at high humidity, its effect on R_g in PrOH being "in agreement with" the change in film texture C27·C28. Figure 14 summarises the reading: a film mixes only where both ITZ and HPMC are compatible with the solvent, that is, MeOH and EtOH under dry conditions and PrOH and BuOH under wet conditions C53.

What it suggests for formulation. The authors state that this solvent-triggered behavior "highlights the importance of conducting a thorough screening of various solvents" before preparing ASDs by solvent evaporation such as spray drying, and that the observations "may help" the rational design of such ASDs C50. Note by the AI writer (not discussed in the paper): the practical message for HPMC made from DCM and an alcohol is that the acceptable water level depends on the alcohol; a humidity or water limit that protects a methanol process could harm an n-propanol one. A clear feed solution does not prove that the polymer is dissolved as single chains; the higher-alcohol solutions in this paper looked clear yet contained HPMC aggregates R23·R14. Spray drying evaporates much faster than spin coating, and faster evaporation gave smaller domains here, so the same chemistry could give finer separation in spray-dried particles that low-resolution methods might miss; this was not tested.

Limitations

Limitations stated by the authors. Solubility parameters "cannot be used to quantitatively predict" polymer solubility, and the assumption of non-specific, additive interactions "may" cause inaccuracies C7. The azeotrope analysis does not consider the effect of DCM or of the dissolved solutes on the azeotropic composition, which are "unknown factors" C35. DLS sizes are compared only relatively, since "the actual size is challenging to measure" C18. Sampling in polymer simulations "can be challenging", which they address with the torsion scans C24.

The AI writer's own judgement (1): the mechanism is inferred, not measured. The argument rests on water concentrating in the evaporating solution, but the water content of the solutions during or after spin coating was never measured, and the paper does not say how 20% and 7% RH were set X8. The evaporation profiles that support it are for solvents without drug or polymer at 24 °C, and the fold changes stated in the text (4-fold, 7-fold) are larger than the plotted ones (about 3.3 and 5.4) X3. The polymer argument rests on drug-free HPMC solutions, intensity-weighted DLS percentages and small MD effects; the state of HPMC in the drug-containing solutions was not measured X15·X12·X13. The composition of the discrete domains in the EtOH and IPA films, and of the network-like phases in the PrOH and BuOH films, was not measured either; the one film mapped by AFM-LCR had no discrete domains, so the drug-rich identity of the domains rests on earlier work with MeOH films and on solubility reasoning X14. The text describes the IPA films at 20% RH as network-like "at all of the investigated temperatures", but only the 21 °C film is shown X16.

The AI writer's own judgement (2): scope and reporting. Only one drug, one HPMC grade, one drug loading (30%) and spin-coated films were studied; domain sizes were judged by eye, and spray drying, stability and dissolution were not tested X11. The MD covers EtOH, PrOH and IPA only, uses molar compositions at 294 K rather than the volume ratio and temperatures of the experiments, and one trajectory per condition X9. Several labels and cross-references are wrong: the Figure 5 caption points to Figure 4a (the MeOH film), the cloud-point data are cited as "Figure 7b" (they are Figure S2b), the evaporation data are once cited as "Figure 5", and the Figure 8 caption defines the dry condition as 20% RH while the methods use a nitrogen purge X1·X2·X4. In Figure S4 every precipitate, including the one from BuOH that shows no Bragg peaks, has a large endotherm near 160–170 °C, which the text assigns to the lower alcohols only X5. None of these errors changes the main film observations, which are consistent across fluorescence and TEM R1·R2·R4.

Conclusion

The paper shows that, for ITZ–HPMC spin coated from DCM with an alcohol, water is a "double edged sword" for drug–polymer miscibility: with MeOH and EtOH, humid air gave phase-separated films and only dry air gave mixed ones, while with PrOH and BuOH mixed films needed humid air, and IPA films were heterogeneous in every condition C29·R1·R4. Solvent evaporation data, drug solubility, polymer aggregation, solubility parameters and simulations all point the same way, and the authors build from them a picture in which a film stays mixed only if the evaporating liquid keeps both the drug and the polymer dissolved: water that concentrates rejects ITZ, and a dry higher alcohol rejects HPMC C53·C44. In their own words, the phase behavior of a specific system "can vary even with slight changes in solvent composition", and the observations "may help in the rational design" of ASDs made by solvent evaporation C49·C50. The direct measurements are the film images, the evaporation profiles and the drug-free polymer data; the water content of the evaporating films and the state of HPMC with drug present remain to be measured X8·X15.