Do chiral polymers tell enantiomers apart? Crystal growth of R and S 2-amino-1,1,3-triphenyl-1-propanol in polymer-containing melts

Background and motivation

Many new drug candidates are poorly soluble in water, and amorphous solid dispersions (ASDs) are a common way to raise their dissolution rate and apparent solubility. The gain comes from the high energy of the amorphous form, and the same high energy drives it back to the crystal during storage and during dissolution, at which point the advantage is lost. Preventing crystallization over the shelf life and during delivery is therefore a central design goal (Introduction; refs 1–4) C1.

Adding a polymer is the usual way to do this, and the literature holds many examples of better physical stability with polymers (refs 5–9). Why polymers work is, in the authors' words, "currently not fully understood". Three mechanisms are commonly proposed: the polymer raises the glass transition temperature ($T_g$) of the mixture and so lowers molecular mobility at the storage temperature (Van den Mooter and co-workers, ref 10); it acts as a diluent and lowers the chemical potential of the drug; and it forms specific drug–polymer interactions, such as hydrogen bonds, that restrict mobility on a local scale (refs 5, 7, 8, 11) C2. The third mechanism is hard to isolate in ordinary experiments, because changing the polymer or the drug changes many properties at once: $T_g$, solubility, miscibility, the driving force for crystallization and the interactions themselves.

The authors found a way around this in chiral chemistry. Many drugs have one or more chiral centres and are increasingly developed as single enantiomers. Two enantiomers have the same chemical and physical properties apart from being mirror images, which is why ordinary chromatography cannot separate them C3. Chiral high-performance liquid chromatography (HPLC) can, and its stationary phases are typically made from derivatives of amylose or cellulose, natural polymers that are themselves chiral (Yashima 2001, ref 15; Tachibana and Ohnishi 2001, ref 16) C3. Under normal-phase conditions these phases engage the analyte through π–π, dipole–dipole and hydrogen-bonding interactions, and when those interactions differ between the two enantiomers, so do their retention times (refs 17–20) C21. The cellulose derivatives used to make ASDs, hydroxypropyl cellulose, hypromellose, hypromellose acetate succinate (HPMCAS) and hypromellose phthalate (HPMCP), are chiral in exactly the same way C4.

One earlier study had looked in this direction: chiral polymers were reported to affect the crystallization kinetics of the R and S enantiomers of nitrendipine differently, an effect attributed to different nucleation rates in the presence of HPMCP, with no difference in crystal growth rates (Miyazaki and co-workers 2011, ref 21) C22. Whether a chiral polymer can slow the growth of two enantiomers differently, and whether any such difference follows their interactions with the polymer, had not been shown.

The paper sets out to test one hypothesis: differences in how the two enantiomers interact with a chiral polymer in the amorphous state will give that polymer a different crystallization-inhibiting effect on each enantiomer, whereas an achiral polymer should affect both alike C5. If the hypothesis holds, the experiment isolates the interaction mechanism, because everything other than chirality is identical between the two enantiomers.

Background concepts

Enantiomers and the racemate. A molecule with one chiral centre exists as two mirror-image forms, labelled R and S by the arrangement of groups around that centre (and + or − by the sign of optical rotation). The model compound here is 2-amino-1,1,3-triphenyl-1-propanol (ATP), with one chiral carbon carrying the amino (NH₂) group next to a carbon carrying a tertiary hydroxyl (OH) group and two phenyl rings (Figure 1a, b) M1·M2. ATP(S) is the (S)-(−) form and ATP(R) the (R)-(+) form. The racemate, ATP(RS), is the 1:1 mixture; it can crystallize differently from either pure enantiomer, which is why it is studied separately P2.

Chiral and achiral polymers. A polymer is chiral when its repeat unit contains chiral centres in a fixed configuration. The glucose rings of the cellulose backbone in HPMCAS and HPMCP carry several chiral centres, marked with asterisks in Figure 1e–f, and hydroxypropyl substituents add more M3·M4·M5. Polyhydroxybutyrate (PHB) is chiral as well, with one stereocentre per repeat unit, but it is not a cellulose derivative M7. Polyvinylpyrrolidone (PVP) is treated as achiral M6. HPMCAS comes in grades (here LF and HF) that differ in their content of acetyl and succinoyl groups; the paper does not give the substitution levels M3·M4.

Crystal growth from an undercooled melt. A melt cooled below its melting point without crystallizing is undercooled; cooled further it becomes a glass. When such a sample is held at a temperature between $T_g$ and the melting point, crystals that form grow outward as spherulites, radial aggregates of crystal fibres that look like the patterned discs in Figure 2 R3. The radius (or diameter) of a spherulite increases linearly with time at fixed temperature, and the slope is the crystal growth rate. Growth rate first rises with temperature, as molecules become more mobile, and then falls again near the melting point, where the driving force for crystallization vanishes. Measuring growth separately from nucleation lets one ask how a polymer acts on molecules attaching to an existing crystal face.

Thermal events in DSC. In differential scanning calorimetry, a glass shows a step at its glass transition ($T_g$). On further heating, an amorphous sample that crystallizes gives an exothermic peak, whose onset is the recrystallization temperature ($T_c$); the crystals then melt at $T_m$ P6. A higher $T_c$ at the same heating rate means that crystallization was held back to a higher temperature, a common screening read-out for crystallization inhibition by polymers.

NH and OH stretching bands in infrared spectra. An NH₂ group shows an asymmetric and a symmetric N–H stretching band. When the NH₂ donates a hydrogen bond, for example to a carbonyl oxygen, its N–H bonds weaken and the bands move to lower wavenumbers; when the hydrogen bonds become weaker or more disordered, the bands broaden and move up. The C=O stretching band of an acceptor carbonyl likewise moves down when it accepts a hydrogen bond. These shifts are the evidence the paper uses to compare how much each enantiomer interacts with each polymer P5.

Methods in detail

The study combines four kinds of measurement on the same materials: DSC of melt-quenched samples, crystal growth rates on a hot-stage microscope, infrared spectra of dispersions, and an isothermal crystallization test followed by infrared. The flow below shows how they connect.

Materials and sample preparation

Both enantiomers of ATP were bought from Sigma-Aldrich and used as received M1·M2. The polymers were HPMCAS grades LF and HF and HPMCP grade HP-55 from Shin-Etsu, PVP K-12 (weight-average molecular weight 2000–3000 g/mol) from BASF, and PHB (molecular weight 50 000) from Advanced Polymer Materials M3–M7. Dispersions with 5% w/w polymer were made by dissolving compound and polymer together, in acetone for HPMCAS and PVP, in dichloromethane for PHB and in a 1:1 mixture of dichloromethane and ethanol for HPMCP, removing the solvent by rotary evaporation and lightly grinding the powder in a mortar P1. The racemate was prepared by mixing equal amounts of ATP(R) and ATP(S) in a mortar and cryomilling for 10 min; racemate dispersions were made only with HPMCAS-HF and PVP P2. The 5% level was used for all growth and DSC work A2.

Crystal growth rates

About 3 mg of sample was melted between two coverslips at about 5 °C above the melting point and quenched immediately to room temperature, giving a clear film that showed no birefringence under crossed polarizers P3·R4. The film was then heated on a hot stage to a set temperature and held there while spherulite growth was recorded by time-lapse photography. Spherulite size rose linearly with time, and the growth rate was taken as the slope from linear regression; each temperature and polymer was measured in triplicate P4·R4·E1. The pure enantiomers were measured from 40 to 120 °C in 10 °C steps; with polymers the range was narrower (50–90 °C for PVP and HPMCP, 40–90 °C for PHB, 50–80 °C for HPMCAS-HF, 50–70 °C for HPMCAS-LF), and the racemate was measured at 70–90 °C A3. The paper does not say how the spherulites were nucleated, nor whether "size" means diameter or radius; the axis of Figures 7 and 8 reads "Size (D x 10-6)", while Figures 6 and 9 give growth rates in µm/s X11.

Thermal analysis

DSC used a TA Q2000 with a refrigerated cooling system, dry nitrogen at 50 mL/min and pinhole-lidded aluminium pans, calibrated with indium and tin. $T_g$ was measured by cooling the melt at 10 °C/min and reheating at 10 °C/min; $T_g$, $T_c$ and $T_m$ were reported as onsets P6. For the pure materials the first heating, the cooling and the reheating were all recorded (Figure 3); for the 5% PVP and 5% HPMCAS-HF samples a single heating trace is shown (Figures 4, 5). The methods also say that melting was measured for physical mixtures, so it is not certain which scan gives the $T_m$ of the polymer-containing samples in Table 1 X9. No sample mass or number of runs is given P6.

Infrared spectroscopy

Transmission FTIR spectra (Bio-Rad FTS 6000, 128 scans, 4000–400 cm⁻¹, dry air purge at about 0% RH) were taken on films on KRS-5 windows, made either by melting compound and polymer between the window and a coverslip and quenching, or by spin-coating an acetone solution at about 2000 rpm P5. Which method was used for which spectrum is not stated X5. To get changes large enough to read, the authors used 30% polymer for HPMCP and HPMCAS-LF and 20% for HPMCAS-HF, rather than the 5% of the growth experiments C15·A4. Spectra of the R and S dispersions were compared in the NH/OH stretching region (3200–3650 cm⁻¹) and, for HPMCP, in the carbonyl region (Figure S1) R12·R13·R14. The crystal structure of the high-melting form of ATP(S) was deposited as CCDC 1401488 to identify the hydrogen bonds in the crystal; the diffraction experiment itself is not described P8·R10.

Isothermal crystallization by infrared

Glasses of ATP(S) and ATP(R) with 5% HPMCAS-HF were held at 40 °C, and NH/OH spectra (3000–3700 cm⁻¹) were taken after melting, after 15 h and after 30 h, against spectra of crystallized pure ATP P7·R15. This test reports overall crystallization, nucleation and growth together, at a temperature below the lowest growth measurement with HPMCAS-HF X14.

What makes the design work. The enantiomer pair is the control. Any property that does not depend on chirality, such as $T_g$, molecular mass, the driving force for crystallization or the interaction with an achiral polymer, is the same for R and S. A chiral polymer that slows the two at different rates therefore reveals an effect that can only come from chirality-sensitive contacts between compound and polymer. PVP serves as the achiral reference, and PHB and HPMCP test whether chirality alone is enough C5·A2. Compared with the nitrendipine study (ref 21), which found differences in nucleation, this paper isolates growth C22.

Experimental analysis

The pure compounds: two nearly identical enantiomers and a slower racemate

As received, ATP(S) and ATP(R) melted with onsets of 144.0 and 142.7 °C and partly crystallized when the melt was cooled (Figure 3, Table 1) R1.a–b·R2. On reheating, both showed a $T_g$ near room temperature (17.0 and 16.3 °C), recrystallized from 58.1 and 59.0 °C into at least one different polymorph and then melted mainly at 119.2 and 116.0 °C R1.a–b·R2. The racemate behaved differently: it melted at 144.6 °C, did not crystallize on cooling, had a $T_g$ of 16.0 °C and on reheating recrystallized from 61.3 °C into a form that melted at 145.8 °C, like the original material R1.c·R2. The text says the racemate "melted at a higher temperature"; in Table 1 this holds against the recrystallized enantiomers (145.8 against 119.2 and 116.0 °C) but not against the as-received enantiomers, whose first-heating onsets are within 2 °C of the racemate's X10.

Table A. DSC onsets at 10 °C/min (from Table 1; melting during the first heating in brackets). No replicate number or uncertainty is given R1·X8.

Crystal growth of the pure enantiomers (Figure 6) rose from about 0.02–0.03 µm/s at 40 °C to a maximum of about 21–34 µm/s at 90 °C, dropped abruptly between 90 and 100 °C and then stayed at about 8–16 µm/s up to 120 °C R5.a–b. The crystals grown at 40–90 °C looked clearly different from those grown above 100 °C, and the authors, from the DSC data and IR of the crystallized films (not shown), assign the low-melting polymorph to growth below 100 °C and the high-melting polymorph to growth at 100 °C and above; the low-melting form cannot grow at 120 °C because it has started to melt R6·C9·X12. All growth comparisons with polymers lie at or below 90 °C, so they concern the low-melting form.

The authors write that ATP(R) and ATP(S) "showed almost exactly the same growth rate for the temperature range studied, as expected" C8. Note by the AI writer (not discussed in the paper): in Figures 6 and 9, ATP(S) is faster than ATP(R) at every temperature, by about 1.2 times at 60–70 °C and about 1.6 times at 80–90 °C R5.a–b·R9.a–b·X2. For two enantiomers in the pure state this difference is not expected thermodynamically and may reflect sample-to-sample variation; it matters because it sets the baseline against which the polymer effects below must be judged.

The racemate grew much more slowly than either enantiomer at the same temperature: about 0.64, 1.9 and 4.7 µm/s at 70, 80 and 90 °C, against about 5.3–6.7, 13–21 and 21–34 µm/s for the enantiomers (Figure 9) R9·C12.

Polymers that slow both enantiomers alike

At 5% w/w, the polymers differed widely in how much they slowed growth (Figure 6) R5. PHB had "minimal impact": with it both enantiomers grew at about 0.95 µm/s at 60 °C and 12.4 µm/s at 80 °C, slightly below pure ATP(S) and close to pure ATP(R) R5.c–d·C10. PVP was effective but "equally effective for both of the enantiomers": at 70 °C both grew at about 1.6 µm/s, a quarter to a third of the pure rates, and the R and S points overlap at 60, 70 and 90 °C R5.e–f·C10. HPMCP was "even more effective", with no differential effect: at 60 °C ATP(S) grew at about 0.035 and ATP(R) at about 0.025 µm/s, about 50 times slower than without polymer R5.g–h·C10.

Table B. Crystal growth rates (µm/s) of each enantiomer with 5% polymer, read from Figure 6 (approximate; log axis). R and S are told apart by the marker style in the legend (filled for S, with a light centre for R) R5.

Note by the AI writer (not discussed in the paper): with HPMCP the ratio of ATP(S) to ATP(R) growth is about 1.2–1.4 at 50–90 °C, the same as without polymer, so "no differential effect" here means no change from the pure-compound baseline rather than identical rates R5.g–h·X2.

HPMCAS: the two grades favour opposite enantiomers

HPMCAS inhibited growth about as much as HPMCP near 60–70 °C, but the two enantiomers no longer grew at the same rate C10. Figures 7 and 8 show spherulite size against time at 60 °C. With HPMCAS-LF (Figure 7), the ATP(R) spherulite reached about 33 size units after 480 s and the ATP(S) spherulite about 18, a ratio of about 1.8 R7. With HPMCAS-HF (Figure 8), ATP(S) reached about 45 units and ATP(R) about 23, a ratio of about 1.9 R8. Figure 6 shows the same split over temperature: with HPMCAS-HF, ATP(S) grew about 2.1, 1.9, 2.5 and 2.0 times faster than ATP(R) at 50, 60, 70 and 80 °C, and with HPMCAS-LF, ATP(R) grew about 1.9 and 1.5 times faster than ATP(S) at 60 and 70 °C R5.i–l.

The paper's text describes this result the other way round: "In the presence of HPMCAS-HF, ATP(R) grew twice as fast as ATP(S). In contrast, in the case of HPMCAS LF grade, ATP(S) had a growth rate twice as fast as the R enantiomer" C11. The figure legends and markers, the DSC onsets, the 40 °C infrared test and the infrared interaction data (all described below) agree with each other and with the reverse assignment; in the judgement of the AI extractor, the sentence in the text is most likely a misprint X1. This article follows the figures: HPMCAS-LF slows ATP(S) more, and HPMCAS-HF slows ATP(R) more. Whatever the assignment, the authors' conclusion, that "HPMCAS clearly has a differential impact on each enantiomer", stands on the figures C11.

Two further points need care. The absolute growth rates in Figure 7 are about 2.5 times those in Figure 6 for HPMCAS-LF at 60 °C (Figure 7 slopes about 0.069 and 0.038 µm/s if size is in µm, against about 0.029 and 0.015 µm/s in Figure 6), whereas Figure 8 matches Figure 6 for HPMCAS-HF (about 0.093 and 0.048 against about 0.094 and 0.049 µm/s); the ratio between enantiomers agrees in both cases X3·E1. And against the pure compounds, HPMCAS-HF slowed ATP(S) about 14–87 times and ATP(R) about 28–105 times depending on temperature; the text's "factor of 50" for ATP(S) carries no temperature and does not match any single temperature in Figure 6 R16·X4. HPMCAS-HF was also not always "similar" to HPMCP: at 80 °C it slowed growth about 5–9 times more X7.

DSC and isothermal crystallization agree with the figures

With 5% HPMCAS-HF, recrystallization on heating began at 77.2 °C for ATP(S) and 81.6 °C for ATP(R), against 58.1 and 59.0 °C without polymer; with 5% PVP it began at 63.6 and 61.5 °C R1.d–e·R1.g–h. $T_g$ did not change (15.6–18.4 °C), "presumably because of the relatively low polymer concentration" R1·C6. The authors read the HPMCAS-HF values as "suggesting that HPMCAS-HF is more effective at inhibiting crystallization of the latter enantiomer", ATP(R), and the PVP values as no difference C7. The gap is 4.4 °C for HPMCAS-HF and 2.1 °C for PVP, without replicates, so the DSC result supports rather than proves the direction X8. The racemate's $T_c$ rose most of all, to 83.3 °C with HPMCAS-HF and 72.2 °C with PVP R1.f·R1.i.

In the 40 °C test with 5% HPMCAS-HF (Figure 15), the ATP(S) spectra at 15 and 30 h match the crystallized pure compound, with a sharp band near 3365 cm⁻¹, whereas ATP(R) shows only a weak sharp band on a broad amorphous background after 30 h R15. The authors write that ATP(S) "appears to have completely crystallized within 15 h" while ATP(R) "is clearly a mixture of crystalline and amorphous material", "in agreement with the growth rate studies" C20. This agreement holds with the figure assignment of the growth data, not with the text's sentence X1.

Infrared: how ATP binds each polymer

In the crystal of the high-melting form, ATP(S) molecules form hydrogen-bonded dimers in which the OH of one molecule donates to the nitrogen of another; the NH₂ hydrogens take no part (Figure 11) R10·C13. The crystal spectrum accordingly shows sharp asymmetric and symmetric NH₂ stretches at 3365 and 3307 cm⁻¹ on a broad OH band (Figure 10) R11. Amorphous ATP(S) shows the same pattern, broadened and shifted up, with the asymmetric NH₂ stretch at 3390 cm⁻¹; the authors read this as the crystal motif persisting in the glass with weaker hydrogen bonds ("most likely") R11·C14.

With 30% HPMCP (Figure 12), the OH band moved slightly up and the NH bands changed more, with a new band at 3253 cm⁻¹; in the carbonyl region (Figure S1) the polymer's band at 1727 cm⁻¹ was joined by a new band at 1699 cm⁻¹ R12. The authors read both changes as the ATP NH₂ group donating hydrogen bonds to polymer carbonyls ("consistent with") C16. The spectra with ATP(R) and ATP(S) were identical, "suggesting that each enantiomer interacts to a similar extent with HPMCP" R12·C17.

With 30% HPMCAS-LF (Figure 13), a new NH band appeared at 3290 cm⁻¹, and the changes were larger for ATP(S): the 3390 cm⁻¹ band of the glass was "drastically reduced" with ATP(S) but stayed with ATP(R), which the authors read as more ATP–polymer interaction for ATP(S) and more remaining ATP–ATP interaction for ATP(R) R13·C18. With 20% HPMCAS-HF (Figure 14) the trend reversed: the 3390 cm⁻¹ band stayed more intense with ATP(S), "suggesting that this enantiomer forms less interactions with HPMCAS-HF" than ATP(R) R14·C19.

Table C. Infrared evidence of ATP–polymer hydrogen bonding against the growth result. The last column follows the figure legends; the text's growth sentence states the reverse R12·R13·R14·R5·X1.

Read together, the enantiomer that the infrared data show to be more bound to the polymer is, for each HPMCAS grade, the one that grows more slowly, and HPMCP, which binds both alike, slows both alike R12·R13·R14·R5·X1. The authors sum up that the IR data "strongly suggests" that both HPMCAS and HPMCP interact with ATP through hydrogen bonds between the polymer carbonyls and the ATP amino group C24.

The racemate: much less inhibition

For the racemate (Figure 9), 5% PVP and 5% HPMCAS-HF were about equally effective: growth fell from about 0.64, 1.9 and 4.7 µm/s to about 0.30, 1.0 and 2.65 µm/s with PVP and about 0.28, 0.91 and 2.2 µm/s with HPMCAS-HF at 70, 80 and 90 °C R9.c–e. That is about a two-fold reduction, against 14–105-fold for the pure enantiomers with HPMCAS-HF R9·R5·X4. The authors note ("Interestingly") that for the enantiomers HPMCAS-HF was "considerably more effective" than PVP, whereas for the racemate the two were similar, and that each polymer inhibited the racemate less C12.

Discussion

How the authors connect interaction and inhibition

The authors' argument runs from the infrared evidence to the growth data through the chemistry of each polymer. The chain below follows their discussion and conclusions; the HPMCAS step is the one their data address directly, the others are their proposals.

For HPMCAS the authors write that "some of the carbonyl groups (the ester linkages) are in closer proximity to the cellulose backbone and it is clear that this polymer interacts differently with the R and S ATP enantiomers", and that "this differential interaction leads to a factor of two variation in the crystal growth rate" C26. In the abstract and conclusions the cause is put as "most likely due to differences in the ability of each enantiomer to form hydrogen bonding interactions with the polymer" C29. For HPMCP the functional groups that interact with ATP "are adjacent to the rigid, achiral phthalate function", which "may minimize the interaction of ATP with the chiral cellulose backbone and other chiral groups"; the conclusions phrase this as interacting groups that are "distant from chiral centers" C25. For PVP "it is suggested" that both enantiomers interact equally because the polymer lacks chirality, and PHB "presumably" had minimal interactions with either enantiomer, which would explain both its weak inhibition and the absence of a difference C30·C23.

The switch between the HPMCAS grades is the open end of the argument. LF and HF are the same polymer with different acetate and succinoyl substitution; the authors write that the switch "is not readily explainable without more detailed studies of the key intermolecular interactions" but "is most likely linked to the different degree of substitution of each polymer grade with respect to the acetate and succinoyl groups" C28. Neither the substitution levels nor which substituent binds ATP are measured.

The broader conclusion is stated firmly: since "all other properties of the system apart from the chirality are identical", the result "strongly supports previous reports" that intermolecular interactions between a low molecular weight compound and a polymer are "critical determinants of the polymer effectiveness as a crystal growth rate inhibitor" (Kestur and Taylor 2010, ref 11; Van Eerdenbrugh and Taylor 2011, ref 22; Kothari and co-workers 2015, ref 23) C27. The authors close by calling the insights "important for understanding amorphous solid dispersions and inhibition of drug crystallization" C31.

Where the result fits, and what it suggests for formulation

Note by the AI writer (not discussed in the paper): the experiment is a clean demonstration that, at a polymer level too low to move $T_g$, a polymer can change crystal growth rates by one to two orders of magnitude, and that the size of the change depends on molecular contacts that are sensitive to chirality R1·R5. That places specific interactions, rather than mobility or dilution, at the centre of growth inhibition for this system.

Note by the AI writer (not discussed in the paper): three practical points follow for chiral drugs formulated with cellulose esters. First, HPMCAS grades are not interchangeable as growth inhibitors: here LF and HF protected opposite enantiomers, each by about two-fold at 60 °C R7·R8. Second, results on a racemate do not transfer to a single enantiomer: the racemate grew several times more slowly than either enantiomer and was inhibited only about two-fold by PVP or HPMCAS-HF, against 14–105-fold for the enantiomers R9·R5·X4. Third, a simple infrared marker, the loss of the 3390 cm⁻¹ NH₂ band of the glass, tracked which enantiomer a grade protected, which suggests a quick way to rank grades by interaction before running growth or stability studies, at least for compounds with a comparable donor group R13·R14.

Limitations

Limits the authors state. The authors say that the opposite behaviour of HPMCAS-LF and HPMCAS-HF "is not readily explainable without more detailed studies of the key intermolecular interactions in the systems" C28. They attribute the unchanged $T_g$ "presumably" to the low polymer level and the lack of a PHB effect "presumably" to minimal interaction C6·C23. They chose 30% polymer for the infrared spectra to obtain readable changes C15.

The AI writer's own judgement (1): the swapped sentence and the baseline. The text assigns the faster enantiomer opposite to Figures 6–8 for both HPMCAS grades, while the figures, the DSC onsets, the 40 °C infrared test and the infrared interaction trends all agree with one another X1. A reader who takes the sentence at face value would conclude that the enantiomer which binds the polymer more grows faster, the opposite of the authors' mechanism. In addition, the pure enantiomers do not grow at "almost exactly the same" rate: ATP(S) is about 1.2–1.6 times faster at every temperature X2. The HPMCAS effect (about 1.5–2.5 times) is larger and reverses between grades, so the conclusion survives, but the margin over the baseline is smaller than "twice as fast" implies.

The AI writer's own judgement (2): how strong the interaction evidence is. The infrared spectra were taken at 20–30% polymer, four to six times the 5% used for growth, and by one of two preparation methods not specified per spectrum; the band changes are described in words, without band areas or bonded fractions X5. PVP and PHB were not examined by infrared at all, so the explanations offered for them are untested X6. The HPMCP argument is qualitative: Figure 1f marks a chiral centre on the hydroxypropyl link of the hydroxypropyl phthalate substituent, next to a phthalate ester carbonyl, so "distant from chiral centers" is not borne out by the drawn structure X13.

The AI writer's own judgement (3): numbers to treat with care. The absolute growth rates for HPMCAS-LF differ about 2.5-fold between Figures 6 and 7 X3; the "factor of 50" for ATP(S) with HPMCAS-HF cannot be placed at a temperature X4; the DSC differences are single values without replicates X8; it is unclear which DSC scan gives the $T_m$ of the polymer-containing samples X9; the polymorph identity of the growing crystals rests on spectra that are not shown X12; and the 40 °C test mixes nucleation with growth at a temperature below the growth measurements X14. Minor typographical slips (Figure S1 caption "HPMC", "HMPCAS-LF" in the text) do not affect the conclusions X15. Finally, all of this rests on one model compound, which is not a drug, one polymer level for growth, and melt-quenched films rather than solvent-made dispersions under storage.

Conclusion

This paper uses the cleanest control available for a drug–polymer interaction: two molecules identical in every respect except handedness. With 5% polymer in an undercooled melt of 2-amino-1,1,3-triphenyl-1-propanol, the achiral PVP and the chiral HPMCP slowed crystal growth of both enantiomers alike and PHB barely slowed it, while HPMCAS slowed one enantiomer about twice as much as the other, with the LF grade favouring ATP(S) and the HF grade ATP(R) according to the figures R5·R7·R8·X1. DSC and an isothermal infrared test agree with this direction for HPMCAS-HF R1.d–e·R15. Infrared spectra show that the ATP amino group hydrogen-bonds to polymer carbonyls, equally for both enantiomers with HPMCP and preferentially for the more inhibited enantiomer with each HPMCAS grade R12·R13·R14. The authors take this as strong support for compound–polymer interactions as critical determinants of crystal growth inhibition, attribute the grade switch "most likely" to acetate and succinoyl substitution, and leave that explanation for future study C27·C28·C29. For formulation, the result is a reminder that the polymer grade, and the stereochemistry of the drug, can change crystal growth inhibition even when nothing else, including $T_g$, changes R1·R5.