Where a drug sits when PEG crystallizes: time-resolved X-ray scattering of six drug/PEG dispersions

Background and motivation

Polyethylene glycol (PEG) is one of the oldest carriers for solid dispersions. A poorly water-soluble drug is melted with PEG, the melt solidifies, and the product often dissolves much faster than the crystalline drug. The introduction recalls the familiar problems of this approach: poor reproducibility, physicochemical properties that change during manufacture, scale-up and storage, and a limited understanding of how dissolution is enhanced (Leuner and Dressman 2000; Chiou and Riegelman 1971; Serajuddin 1999; Craig 2002). C1 The authors place the root of these problems in the microstructure. PEG is semicrystalline, and in a PEG dispersion the drug may crystallize or not, and may sit in different places with different domain sizes; properties and dissolution should depend on which of these occurs (Dordunoo 1991; Ginés 1996; Talibuddin 1996). C1

For PEG blended with other polymers, the picture was already fairly clear. SAXS and DSC studies had shown that a miscible amorphous polymer added to PEG is pushed out of the PEG crystals as PEG crystallizes, either into the amorphous layers between the crystalline lamellae (interlamellar), between bundles of lamellae (interfibrillar) or between spherulites (interspherulitic). How far it is pushed depends on three things: how fast PEG crystallizes, how mobile the additive is (set by the blend glass transition temperature, Tg), and whether the two components interact specifically (Talibuddin 1996; Silvestre 1987; Russell and Ito 1988; Utracki 2002). C2 For small molecules in PEG there were fewer studies, with monoolein, Pluronic-based nanocrystalline dispersions, parabens and lipids (Mahlin 2004; Qian 2007; Unga 2009; Unga 2010), and the role of the same three factors had not been worked out. One of these studies had reported that the solubility of a small-molecule additive in PEG affects the microstructure of the solidified PEG (Unga 2009). C2·C7

The authors' own earlier study (Zhu, Taylor and Harris 2010) had looked at PEG dispersions of several drugs, but only after 24 h of solidification. The present paper adds time: by recording small-angle X-ray scattering (SAXS) and wide-angle X-ray scattering (WAXS) at the same time on a synchrotron, it follows the PEG structure and the crystallinity of drug and polymer from the moment the melt is quenched. It also widens the choice of drugs to cover a range of solubility in, and miscibility with, PEG. Six model compounds were used: haloperidol (HLP), benzocaine (BZC), chlorpropamide (CPM), ibuprofen (IBP), nilutamide (NLU) and ketoconazole (KET), shown in Figure 1, with PEG of molecular weight 3350. M1·M2·M3·M4·M5·M6·M7 The question behind the work is simple to state: when a drug/PEG melt solidifies, does the drug end up between the PEG lamellae, how long does it stay there, and which of its properties decide this?

Background concepts

Semicrystalline PEG and the long period. When PEG crystallizes from the melt, its chains fold back and forth to form thin, flat crystalline lamellae. The lamellae stack on top of each other with layers of amorphous (non-crystalline) PEG in between, and the stacks grow outward as spherulites, as sketched in the paper's graphical abstract. The repeat distance of this stack, one crystalline lamella plus one amorphous layer, is called the long period, $L$. Because the stack is periodic, it scatters X-rays at a small angle, and the position of that SAXS peak gives the long period. The graphical abstract prints the relation used: E4

In plain terms: the long period is inversely proportional to the scattering vector $q$ at which the SAXS peak appears; $n$ is not defined in the paper and is presumably the order of the peak, so for the first-order peak $L=2\pi /q$. A peak at $q$ ≈ 0.047 Å⁻¹ gives about 13.4 nm, which matches the "around 13.3 nm" the paper reports for freshly crystallized PEG, and a peak at about 0.039 Å⁻¹ gives 16.1 nm, the value listed for IBP/PEG at 30 min. E4·R7·R8·R15·R18 The logic of the whole paper rests on this quantity. If a drug is held in the amorphous layers between the lamellae, those layers become thicker and the long period grows; if the drug is pushed out of the stack, or crystallizes separately, the long period should look like that of pure PEG. C33

WAXS: who is crystalline. At wide angles X-rays are scattered by the crystal lattices themselves. Crystalline PEG gives two strong peaks, at $q$ ≈ 1.36 and 1.65 Å⁻¹, and a crystalline drug adds peaks of its own, for example at about 1.40 Å⁻¹ for HLP. R7·R9 Recording SAXS and WAXS on the same sample at the same time therefore tells, frame by frame, both how the PEG stack is spaced and whether the drug has crystallized.

Chain folding in low-molecular-weight PEG. PEG 3350 is short enough that its chains fold only a few times inside a lamella. Following the polymer literature (Cheng and co-workers 1991 and 1992; Arlie 1967; Buckley and Kovacs 1976), the authors assign the first structure formed on cooling to nonintegral folded lamellae, in which the chain folds a non-whole number of times, and the long period of 11.2 nm reached after a day to once-folded lamellae. Over minutes to hours, the nonintegral folded lamellae thin into once-folded ones, and the long period falls. C14·R8

Eutectic and monotectic phase diagrams. A drug/PEG phase diagram plots the temperature at which the last crystal melts (the liquidus) against composition. In a eutectic system, adding drug lowers the melting point of PEG and adding PEG lowers the melting point of the drug, so the liquidus falls from both sides to a minimum, the eutectic. In the monotectic case named in the paper, no such minimum appears: the liquidus rises steeply from the PEG side toward the melting point of the drug. C3·R1·R2·R3·R4 The earlier work by Baird and Taylor (2010), on which this paper draws, related the eutectic composition to the drug's melting point and its ability to interact with PEG. C4

The interaction parameter from melting point depression. How much the drug's melting point falls as PEG is added measures how favourably the two mix. The paper uses the Flory–Huggins form of melting point depression (Flory 1953; Nishi and Wang 1975; Hoei 1992; Marsac 2006): E1

In plain terms: $T_M^{\mathrm{m}\mathrm{i}\mathrm{x}}$ is the melting temperature of the drug in the presence of PEG and $T_M^{\mathrm{p}\mathrm{u}\mathrm{r}\mathrm{e}}$ that of the pure drug; $\Delta H_{\mathrm{f}\mathrm{u}\mathrm{s}}$ is the heat of fusion of the pure drug (taken from Baird, Van Eerdenbrugh and Taylor 2010); $R$ is the gas constant; ${\varphi }_{\mathrm{A}\mathrm{P}\mathrm{I}}$ and ${\varphi }_{\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{e}\mathrm{r}}$ are volume fractions; $m$ is the ratio of the molar volume of PEG to that of a lattice site, here defined by the volume of the drug molecule; and $\chi$ is the interaction parameter. The first two terms in the bracket describe athermal (purely entropic) mixing of a small molecule with a long chain; $\chi$ collects the extra energy of drug–PEG contacts. The authors plot the left side, rearranged, against ${\varphi }_{\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{e}\mathrm{r}}^2$ and take $\chi$ from the slope, using melting points at 50–80% drug, then normalize it by multiplying by the molar volume ratio of PEG to drug and dividing by the number of chain segments in one PEG molecule. A more negative $\chi$ is read as a negative enthalpy of mixing, that is, favourable drug–PEG interaction. E1·E2·P5·C6 The heats of fusion, molar volumes and fitted plots are not printed, so the values in Table 1 cannot be recomputed from the paper. X10

Solubility of the drug in supercooled PEG. The liquidus also gives the solubility of the drug in liquid PEG at each temperature. To estimate it at 25 °C, below the melting point of PEG, the authors use the van't Hoff equation: E3

In plain terms: $x$ is the mole fraction solubility of the drug in liquid PEG, $\Delta H_f$ the heat of solution and $T$ the absolute temperature. If $\Delta H_f$ is constant, $\ln x$ falls on a straight line against $1/T$; the authors fit that line to the liquidus points of Figures 2–5 and extend it to 25 °C, then convert the result to weight percent. Because 25 °C is below the PEG melting point, the number refers to a hypothetical supercooled liquid PEG. E3·P6·C8

Crystallization tendency, supercooling and Tg. Drugs differ in how readily they crystallize from their own undercooled melt; some crystallize as the melt cools, others stay amorphous for hours or days (Baird, Van Eerdenbrugh and Taylor 2010). How fast a melt crystallizes depends on its supercooling, the distance between the temperature at which it is held and its melting point: a smaller supercooling usually means slower nucleation, and for a polymer it also means thicker lamellae. Tg matters because below it molecular mobility is very low. In polymer blends a low blend Tg lets the additive move out of the way of the growing PEG crystals, which is one of the three factors listed above. C2·C15·C20

Methods in detail

The study has two layers. The first characterizes each drug against PEG at equilibrium: its phase diagram, its interaction parameter, its solubility, whether it interacts by FTIR, and how readily it crystallizes on its own. The second follows the structure of 20/80 drug/PEG blends in real time as they solidify. The comparison of the two layers is the argument of the paper.

Samples

CPM, HLP and NLU came from Sigma-Aldrich, BZC from Spectrum Chemical, KET from Hawkins and IBP from Albemarle; PEG 3350 was a gift from Dow. M1·M2·M3·M4·M5·M6·M7 Physical mixtures (1 g of powder in total) were made by geometric mixing, comelted, and solidified at a set temperature; solidified dispersions were stored in desiccators at low humidity. P1 All time-resolved measurements used 20/80 drug/PEG by weight. A1

Phase diagrams, interaction parameters and solubility

For each drug/PEG pair, mixtures across the whole composition range were melted to a homogeneous liquid, solidified, stored for a few days (longer for slowly crystallizing drugs) and lightly ground. The powder was then heated at 1 °C/min on a hot stage under a polarized light microscope, and the temperature at which the last crystal melted was taken as the liquidus. P3 DSC at the same heating rate, in sealed pans after equilibration at 25 °C, was used to confirm the diagrams; the SI shows these thermograms only for IBP/PEG (Figure S1). P4·R19·C5 The BZC, IBP, NLU and KET diagrams were measured here (Figures 2–5); the CPM and HLP diagrams were taken from Baird and Taylor (2010), measured by the same method. P3·R5 From the liquidus the authors derived the interaction parameter by melting point depression and the solubility at 25 °C by van't Hoff extrapolation, as described above. P5·P6

Time-resolved SAXS/WAXS

The scattering experiments were done at beam station 5-ID-D of the Advanced Photon Source, with 17 keV X-rays (wavelength 0.73 Å) and a custom CCD detector. The SAXS detector sat 3032 mm and the WAXS detector 228 mm from the sample, covering $q$ from 0.0085 to 0.23 Å⁻¹ and from 0.6 to 4.6 Å⁻¹; silver behenate, lanthanum hexaboride and glassy carbon served as calibrants. P2·M8·M9·M10 A physical mixture was placed in an aluminum DSC pan on a Linkam hot stage, heated above the melting point of the drug, and quench-cooled to the solidification temperature, 25 °C for all blends. Pure PEG was also crystallized at 40 °C. Frames were recorded for about 30 min (to about 2 h for CPM and IBP) and once more after 24 h; the SAXS profiles are shown as Lorentz-corrected intensity. P2·A2·A3 The paper does not give the heating rate, the time held above the melting point, the frame exposure, the sample mass, or how the samples were kept between the last frame and the 24 h measurement. P2·X9

Crystallization tendency and FTIR

Each pure drug and PEG was melted and cooled to room temperature under the polarized light microscope to see how fast it crystallized on its own. P7·R6 In the SI, ATR-FTIR spectra (500–4000 cm⁻¹, 4 cm⁻¹ resolution, 128 scans) were recorded on the pure molten components and on molten 20/80 mixtures; a "physical mixture" spectrum was calculated by weighting the pure drug and pure PEG spectra, using one reference band for each component. If drug and PEG do not interact, measured and calculated spectra should coincide. P8·C10 The temperature of the molten samples is not stated. X11

Experimental analysis

Six drugs against PEG: interaction, solubility, crystallization and Tg

Figures 2–5 show that BZC, IBP and NLU each form a eutectic with PEG, at 30, 40 and 10 wt% drug and 48.9, 49.6 and 58.4 °C, while KET/PEG is monotectic: its liquidus jumps from about 60.5 °C for PEG to about 101 °C at only 5 wt% KET and reaches about 156 °C for pure KET. R1·R2·R3·R4·R5·C3 From Baird and Taylor's earlier diagrams, CPM/PEG is a eutectic at 35% with a minimum melting point of 55 °C, and HLP/PEG is monotectic. R5.a·R5.c Table A gathers what the paper reports for each drug.

Table A. The six model drugs (from Table 1 and the text; Tg values cited by the paper from Baird, Van Eerdenbrugh and Taylor 2010). R5·R6·M1·M2·M3·M4·M5·M6

The authors read the table as follows. BZC, CPM and IBP have estimated solubilities above 20% in PEG at 25 °C, "consistent with" their negative interaction parameters; all three carry hydrogen bond donors that can bind the ether oxygen of PEG, and their melting points are moderate. C9 NLU also interacts favourably (χ −0.09) but is less soluble (11 wt%), which "can be explained by" its higher melting point, about 160 °C on Figure 4. C12·R3 HLP and KET have positive interaction parameters, which indicates weaker interaction, and both are "almost insoluble" in PEG at room temperature. C13

The FTIR spectra in the SI support the split. For molten BZC/PEG, IBP/PEG and NLU/PEG the measured spectrum departs from the calculated physical mixture, most clearly in the carbonyl band: about 1700 against 1686 cm⁻¹ for BZC, 1735 against 1703 cm⁻¹ for IBP and 1733 against 1721 cm⁻¹ for NLU (read from Figure S2, approximate). For KET/PEG the two spectra almost coincide. R20·C10 The SI attributes the carbonyl shifts to changes in drug–drug hydrogen bonding caused by drug–polymer interaction. C11 Note by the AI writer: the main text says that FTIR confirmed the interactions of BZC, CPM and IBP, but Figure S2 shows BZC, IBP, NLU and KET; no CPM spectrum is given, so for CPM the case rests on the interaction parameter alone. X3

Note by the AI writer: the estimated 25 °C solubility of NLU (11 wt%) is slightly above the NLU content at its eutectic (10% at 58.4 °C). An extrapolation down the drug liquidus to a lower temperature would be expected to give less, not more; the fitted lines are not shown, so the 25 °C values are best treated as rough and used for ranking. X4 Also, DSC places the IBP/PEG eutectic endotherm at about 47 °C (Figure S1), 2–3 °C below the hot-stage value of 49.6 °C, and the order of the traces in Figure S1 appears to be the reverse of its caption: the top trace has the single sharp endotherm of pure IBP at about 75 °C and the bottom trace the single endotherm of PEG at about 57.8 °C. R19·X14·X2

Pure PEG: lamellae that thin within minutes, and a 40 °C control

When molten PEG was quenched to 25 °C, the two WAXS peaks of crystalline PEG, at $q$ ≈ 1.36 and 1.65 Å⁻¹, appeared at once and did not move, while the SAXS peak, first seen at $q$ ≈ 0.047 Å⁻¹, moved steadily to larger $q$ (about 0.054 Å⁻¹ by 28.2 min, read from Figure 6). R7 In long-period terms, $L$ started around 13.3 nm, fell to 11.6 nm within 30 min, and reached 11.2 nm after 24 h, where it stayed (Figure 7). The authors assign the starting structure to nonintegral folded lamellae and the 11.2 nm period to once-folded lamellae. R8·C14

At 40 °C the once-folded lamellae again dominated. $L$ first rose slightly and then fell to 12.6 nm after 40 min, about 1 nm above the 25 °C value at 30 min; after 24 h both temperatures gave the same 11.2 nm. R8 This control matters later. Crystallizing at 40 °C instead of 25 °C lowers the supercooling of PEG by 15 °C, and that changes $L$ by only about 1 nm at 30 min. R8·C21

Drugs that interact with PEG: a larger long period while they stay amorphous

BZC. Pure BZC crystallizes as soon as its melt cools, but in the 20/80 blend no BZC peaks appeared in the WAXS over 30 min (Figure 10). The authors attribute the slower crystallization "most likely" to the lower supercooling of BZC in the blend: the 20/80 dispersion melts at 55 °C, against 92 °C for pure BZC. R6·R11·R1·C20 Meanwhile the SAXS peak sat at lower $q$ (about 0.038 Å⁻¹), and $L$ rose within about 1.5 min to about 16.5 nm and stayed there to 28.2 min, 4–5 nm above pure PEG at the same time (Figure 11). R11·R12

Two explanations were weighed for this increase. BZC lowers the melting point of PEG, so PEG in the blend crystallizes at a lower supercooling, which by itself tends to give thicker lamellae. Alternatively, BZC is incorporated between the PEG lamellae. Because the 25-versus-40 °C control moved $L$ by only about 1 nm, the authors judge the first explanation "unlikely" to account for 4–5 nm and conclude that the increase "can be largely attributed" to BZC in the interlamellar region. C21·R8 After 24 h, BZC had crystallized and $L$ had dropped from 16.5 to 11.4 nm, close to pure PEG; the authors read this as BZC being expelled from the interlamellar region. C22·R18.b Note by the AI writer: Figure 11 plots this 24 h point at about 11.7 nm; the text and Table 2 give 11.4 nm, and the table value is used here. X1

CPM. CPM has the most negative interaction parameter of the six and a high solubility in PEG. In the blend its $L$ fluctuated between about 14.7 and 15.4 nm in the first 15 min and then held near 14.7 nm to about 85 min, falling slightly to about 14.5 nm by about 115 min (Figure 12). At about 28 min this is more than 3 nm above pure PEG, "suggesting interlamellar incorporation". WAXS, described but not shown, found no CPM crystals over this period; after 24 h most of the CPM "appeared" to have crystallized and $L$ had fallen to 11.7 nm. R13·C23·C24·R18.c

IBP. IBP's $L$ rose within about 3 min to about 16.5 nm, 4.9 nm above pure PEG at 20 min, and then held near 16.1 nm for about 100 min (Figure 13). R14 It then dropped suddenly, over a few minutes, to about 13.2 nm, and this was the one system in which the drop was caught in the act. At 100 min there was a single SAXS peak at $q$ ≈ 0.039 Å⁻¹ and IBP was amorphous. At 111 min a new SAXS peak had appeared at $q$ ≈ 0.047 Å⁻¹, and the WAXS showed a crystalline IBP peak (Figure 14). By 116 min the 0.047 Å⁻¹ peak was the dominant one. R15·C25 For a few minutes two long periods coexisted, one falling from about 15.9 to 15.5 nm and one from about 13.7 to 13.5 nm, before $L$ settled near 13.3 nm and finally reached 11.9 nm at 24 h. R14·R18.d

NLU. NLU behaved like the other interacting drugs. Its $L$ passed through about 17.3 nm at 2–3 min, settled near 16.5 and then 16.1 nm, and stayed around 16 nm for the first 30 min (Figure 15). After 24 h the WAXS showed crystalline NLU and $L$ was 11.4 nm, virtually that of pure PEG, "once again indicating" removal of the drug from the interlamellar region. R16·R18.e·C27

Weakly interacting drugs: HLP and KET

HLP crystallized at once: its WAXS peak at $q$ ≈ 1.40 Å⁻¹ was present from the very first frame (Figure 8). R9·C17 The long period of HLP/PEG started slightly above pure PEG (about 13.9 nm) and followed almost the same course, settling near 12.0 nm by about 11 min and reaching 11.2 nm at 24 h (Figure 9). The authors call this "essentially the same" as pure PEG and explain it as HLP being removed from the PEG phase "most likely" by its fast crystallization on cooling. R10·R18.a·C18·C19

KET is the opposite case in one respect: it did not crystallize at all, even after 24 h. R6·R17 Yet its $L$ fell to about 11.8 nm within about 2 min and stayed there to 24 h, close to pure PEG (Figure 16). R17·R18.f·C29 The authors attribute this, as "likely", to KET's positive interaction parameter, which "may indicate" a low tendency to mix with PEG and a large degree of exclusion from the interlamellar region. C30 Note by the AI writer: at 24 h KET's $L$ (11.8 nm) is 0.6 nm above pure PEG (11.2 nm), a residual similar to that left by crystallized CPM (11.7 nm) and IBP (11.9 nm); the paper does not discuss differences of this size. X6

The summary at 30 minutes and 24 hours

Table 2 of the paper condenses the time courses into two snapshots. Table B reproduces it, with the pure PEG values from the text and the difference from pure PEG at 30 min added by the AI writer.

Table B. Long period of PEG and phase state of the drug (from Table 2; pure PEG from the text; differences computed by the AI writer). R18·R8

The authors' summary is that with drugs that can interact with PEG (BZC, CPM, IBP, NLU) the long period is "much larger" than with the non-interacting KET as long as the drug has not crystallized, and "This suggests that interlamellar incorporation is favored" for molecules that interact strongly with PEG; once the drug crystallizes it is removed from the interlamellar region and $L$ falls. C32·C26 In the conclusions they add that the location of the drug "could thus be inferred" from the change in $L$, and that drugs with higher solubility in, and stronger interaction with, PEG gave the larger increase. C33·C32 Note by the AI writer: the data support this as a contrast between two groups, not as a ranking. CPM has the most negative χ and a solubility of 26 wt% but the smallest increase of the four interacting drugs (+3.1 nm), while NLU, with the least negative χ and 11 wt%, gives one of the largest (+4.4 nm). X5·R5·R18

Does the Tg of the drug matter?

Because the polymer-blend literature names additive mobility, through Tg, as one of the factors that decide exclusion, the authors included two drugs with Tg above the 25 °C solidification temperature. C2 NLU (Tg around 33 °C) enlarged $L$ and then lost the effect on crystallizing exactly like the low-Tg drugs, so its higher Tg "did not appear" to change its behaviour. R16·R18.e·C28 KET (Tg 44 °C) stayed amorphous yet left $L$ like pure PEG, and the authors write that "clearly" the higher Tg did not prevent phase separation of the compound. R17·C31 Note by the AI writer: the Tg values are cited rather than measured, only two drugs test this factor, they sit 8 and 19 °C above the solidification temperature, and they also differ in χ and crystallization tendency, so the Tg conclusion is limited to this range. X13

Discussion

The authors' picture

Put together, the results give a simple sequence, which the authors state in their discussion and conclusions. The chain below follows their reading; the location of the drug in step 3 is inferred from the long period, not observed directly.

In this picture, two drug properties act at different stages. The sign of the interaction with PEG decides whether the drug enters the amorphous PEG layers at all; the drug's own crystallization decides how long it stays there. A Tg above the solidification temperature, at least up to 44 °C, changed neither. C32·C26·C28·C31 The authors close by saying that adding a small molecule to PEG "can" change the polymer's structure as seen in the long period, that the location of the drug "could thus be inferred" from it, that combined time-resolved SAXS/WAXS was "extremely useful" for this, and that the results "help explain" the often-noted variability of drug/PEG dispersions with storage time. C34·C33·C35·C36 The abstract adds that these factors are "likely to impact the effectiveness" of drug/PEG systems as drug delivery systems. C37

What this suggests for formulation

Note by the AI writer (not discussed in the paper as such): the most practical message is about time. In these 20/80 blends at 25 °C, the state in which a favourably interacting drug sits amorphous between the PEG lamellae lasted from about 2 h (IBP) to less than a day (BZC, CPM, NLU) before the drug crystallized. R14·R18 A PEG dispersion characterized right after preparation may therefore have a different structure from the same product a day later, which bears on when release or stability tests are run. The paper names this as the likely source of storage-time variability but does not measure dissolution or stability. C36·C37

Note by the AI writer: the data also suggest two separate screens. The sign of χ from a melting point depression study predicted which drugs entered the interlamellar layers here, but neither the size of χ nor the 25 °C solubility predicted how much the long period grew. A drug such as KET, weakly interacting and slow to crystallize, stayed out of the PEG lamellae even while amorphous; in a dispersion it would then be expected to form its own amorphous domains, which this paper did not image. X5·X7

Limitations

Limitations the authors state. The paper does not have a separate limitations section. Its hedges mark the softer steps: the interlamellar location of BZC "can be largely attributed" from the long period, the CPM data are "suggesting" incorporation, HLP's lack of effect is "most likely" due to fast crystallization, KET's is "likely" due to its positive interaction parameter, which "may indicate" exclusion, and NLU's higher Tg "did not appear" to matter. C21·C23·C19·C30·C28 The CPM WAXS data are explicitly "data not shown". R13

The AI writer's own judgement. The time-resolved data are direct and the contrast is large: 14.7–16.5 nm with the four interacting drugs against 11.8–12.0 nm with HLP and KET and about 11.6 nm for pure PEG at 30 min, and for IBP the fall of the long period was recorded in the same frames as the appearance of drug crystals. R18·R8·R14·R15 The weaker points are these:

• The drug's location is inferred. The long period is the sum of the crystalline lamella and the amorphous layer, and the paper does not separate the two; it also assigns every long period to once-folded PEG lamellae. The only alternative explanation it tests is reduced supercooling. A change in PEG chain folding or crystal thickness caused by the drug would also change the long period and was not examined. X8·C14·C21
• KET's exclusion and phase separation are inferred from a long period like that of pure PEG; no measurement shows where the amorphous KET sits. X7
• Each system was run once, as far as the paper reports, and the storage conditions between the last frame and the 24 h measurement are not given. The "30 min" values of NLU and KET in Table 2 have no plotted point at 30 min (the last time-resolved points are at about 26 and 15 min). X9
• The 24 h crystallinity rests on WAXS that is not shown, and its wording varies: Table 2 lists BZC, CPM, IBP and NLU as "partially crystalline", while the text says BZC "crystallized" and "most of" the CPM did. X12
• The interaction parameter comes from melting points at 50–80% drug, on the drug-rich side of the diagrams, while all scattering samples were 20/80; the CPM and HLP values come from earlier diagrams, and the normalization cannot be reproduced from printed inputs. X10
• A few small inconsistencies: the BZC 24 h point plotted at about 11.7 nm in Figure 11 against 11.4 nm in the text and Table 2; the reversed trace order in Figure S1; the NLU 25 °C solubility above its eutectic content; the missing CPM FTIR spectrum. X1·X2·X4·X3
• One drug loading (20/80), one PEG grade (3350) and one solidification temperature for the blends were studied, and the Tg test rests on two drugs. A1·X13

Conclusion

This paper turns a vague statement, that the microstructure of PEG dispersions varies, into a measurable one. Time-resolved SAXS/WAXS of six drugs blended 20/80 with PEG 3350 and solidified at 25 °C shows that drugs with a negative interaction parameter (BZC, CPM, IBP, NLU) hold the PEG long period 3–5 nm above that of pure PEG while they are amorphous, and that the long period returns close to pure PEG as soon as the drug crystallizes, which for IBP was seen happening at about 100–116 min. R18·R8·R14·R15 Weakly interacting HLP, which crystallizes at once, and KET, which stays amorphous, leave the PEG lamellae much as they are. R10·R17·R18 The authors interpret the larger long period as drug held in the interlamellar region, a reading supported by a supercooling control but not by a direct measurement of where the drug sits, and they link the structural changes to the known storage-time variability of drug/PEG dispersions. C21·C32·C36·X8