Physical stability

How it was shown

Konno and Taylor (2006) asked whether a polymer's ability to slow nucleation of a molecularly dispersed amorphous drug follows the Tg it gives the dispersion or its hydrogen bonding with the drug konno2006 C6. Felodipine and PVP K29/32, HPMCAS (AQOAT AS-MF) or HPMC (Pharmacoat 606) were dissolved in dichloromethane:ethanol 1:1, spin-coated at about 2500 rpm in a nitrogen glove box below 10% RH, heated briefly to 90 °C and stored over phosphorus pentoxide (0% RH) at 22 °C konno2006 P1, P2, M2, M3, M4. Birefringent nucleation sites were counted under polarized light in 12 areas per film with a 10× objective, converted to a volume density with the depth of field of the objective (0.0144 mm), and the nucleation rate was taken as the slope of site density against time; sites at the film edge were excluded and each composition was run in triplicate at 0, 3, 7, 10, 15, 20 and 25% w/w polymer konno2006 P2, E1, A2, R18. Marsac, Konno and Taylor (2006) used the same film and counting method for nifedipine and felodipine with 0–25 wt% PVP K29/32 marsac2006a P1, P4, A2. Konno and Taylor (2008) reproduced the 2006 dry rates as the 0% RH panel of their moisture study, and Marsac and co-workers (2008) measured both drugs again at 0% RH with 7–25 wt% PVP konno2008 R1, X10; marsac2008a R1, P1, P3.

What the data show

Felodipine alone nucleated at about 1.2–1.35 × 10⁵ m⁻³ s⁻¹; 3% polymer gave about 2.3–3.0 × 10³, 7% about 1.1–1.8 × 10³, 15% about 0.75–1.1 × 10³, 20% about 1.8–5.7 × 10² and 25% PVP about 3 × 10² konno2006 R4. The first 3% therefore removed roughly a 40–60-fold factor (synthesis: 1.3 × 10⁵/3.0 × 10³ = 43 for PVP and 1.2 × 10⁵/2.3 × 10³ = 52 for HPMCAS), and the next 22% about one decade more konno2006 R4.a, R4.b, R4.g. From 3 to 15% the three polymers lay within a factor of about 1.5 of each other; at 20% HPMCAS sat about threefold lower inside error bars spanning about 70 to 1.05 × 10³, and at 25% only the PVP point was plotted konno2006 R4, X9. In the PVP study nifedipine fell from about 1.2 × 10⁵ to 1.4 × 10³ m⁻³ s⁻¹ at 7 wt% PVP and to 4.3 × 10² at 25 wt%, felodipine from about 3 × 10⁴ to 2.7 × 10² and 7 × 10¹ marsac2006a R16. The 2008 dry baseline gave felodipine 1.4 × 10⁵ → 1.1 × 10³ (7% PVP) → 3 × 10² (25% PVP) and nifedipine 4.7 × 10⁵ → 5.6 × 10³ → 1.7 × 10³ m⁻³ s⁻¹ marsac2008a R1.a, R1.b, R1.f, R1.g, R1.h, R1.l.

Figure

On the log axis the drug-alone point sits about 50-fold above every 3% point, and the three symbols (filled circle PVP, triangle HPMCAS, open diamond HPMC) overlap at most levels; at 25% only PVP is visible konno2006 R4, X9.

Both drugs drop about two orders of magnitude at 7 wt% PVP, and nifedipine (circles) stays above felodipine (triangles) at every level marsac2006a R16.

Why

The authors' account, with their hedges: three candidate effects of the polymer were weighed. A lower driving force through mixing "probably cannot account" for the drop, because Flory–Huggins theory with data for comparable systems predicts little change in drug activity below a polymer volume fraction of about 0.5 konno2006 C34, C35, R20. A pure diluent effect would scale with polymer fraction, and a 3% addition giving a 1.5-order drop rules it out ("clearly not the case") konno2006 C37. What remains is a raised kinetic barrier: "it seems reasonable to speculate" that the polymers increase the barrier to moving molecules onto the nucleus, possibly through polymer rejected by the growing nucleus and piling up at its surface, and after the first large drop the effect "appears" set by the amount of polymer rather than its type konno2006 C39, C40, C41. In the PVP study the nucleation activation energy below Tg rose with 3% PVP, significantly for felodipine (145 to 201 kJ/mol), which the authors read as a higher barrier ("it appears") marsac2006a R3, C20, C48. None of these barriers, nor the polymer at the nucleus surface, was measured (see F4) konno2006 X14. (synthesis: the flat part above about 7% is what a barrier saturating with polymer at the interface would give, but the papers do not test this.)

Conditions and limits

Dry storage (0% RH) at 22 °C, thin spin-coated films heated to 90 °C, 3–25% polymer, two dihydropyridine drugs konno2006 P1, P2; marsac2006a P1, P4. Rates are counts per depth-of-field volume and comparable only within a counting method; the 2006 conversion is described as multiplication where division is meant, and absolute rates differ about fourfold between Table II and the figures of the PVP study konno2006 X6; marsac2006a X1, X10. Between papers the felodipine-alone rate is about 3 × 10⁴ in Fig. 9 of marsac2006a but 1.3–1.4 × 10⁵ in the other three papers (synthesis: about fourfold) marsac2006a R16.a; konno2006 R4.a; marsac2008a R1.g. Growth was not measured in these studies konno2006 X14.

Counterexamples and conflicts

Equality holds for nucleation and only when dry: at 33–75% RH the three polymers separated, PVP being weakest at low content (F11) konno2008 C10. For melt crystal growth of felodipine at 3% polymer the polymers differed several-fold (PVP > PVP/VA > HPMCAS > PVAc), so "polymer type does not matter" does not carry over from nucleation to growth kestur2010a R6, R7; konno2006 C14. Across drugs no polymer was best for all (see Miscibility and drug-polymer interactions F15). Nor does the equality carry over to other drugs: in supercooled acetaminophen with 10% w/w polymer the number of nuclei per cooling–reheating cycle ranged tenfold, from 2.5 with HPMCAS to 26 with PAA against 10.8 for the drug alone, and in flutamide two of five polymers lowered nucleation while the others had no significant effect (see F17) trasi2012a R12, C46; trasi2012b R11, C18. (synthesis: the felodipine result of 2006 is therefore a result for one drug in thin films at 22 °C, not a rule for polymers.) A later paper cites the 2006 study for a 150-fold drop of felodipine nucleation with 3% PVP at 30 °C kestur2013b R22, larger than the 40–60-fold drop read from the 2006 rates at 22 °C (synthesis: 1.3 × 10⁵/3.0 × 10³ = 43 konno2006 R4); the 2006 card holds no rates at 30 °C, so the cited value cannot be checked here.

Transfer check

Count nucleation sites in dry spin-coated films of the neat amorphous drug and with 3, 7 and 15% of each candidate polymer at one storage temperature below Tg. A drop of one to two decades at 3–7% followed by a flat curve, with the polymers within about twofold of each other, puts the drug in the felodipine pattern. Repeat at the storage humidity before choosing a polymer, since equal dry behaviour hid a 30–70-fold difference at 75% RH (synthesis: 8.6 × 10⁵/1.3 × 10⁴ = 66 and 8.6 × 10⁵/2.0 × 10⁴ = 43 at 3% konno2008 R3.b).

History

Matsumoto and Zografi (1999, outside the library) had seen PVP inhibit indomethacin crystallization at levels too low to raise Tg, which framed the 2006 question konno2006 C3, R21. Konno and Taylor (2006) first measured the polymer effect on nucleation in the library and found the three polymers equal when dry konno2006 C14; Marsac and co-workers (2006) confirmed the same two-decade drop for nifedipine and felodipine with PVP marsac2006a C18. In 2008 both groups used the dry state as the reference for adding water and found the equality lost in humid air konno2008 C9, C10; marsac2008a C8, C9. Later in the library a polymer was found to cut melt growth rates by a broadly similar factor for fast and slow growers (see Crystallization tendency F19). In 2012 Trasi and Taylor counted nuclei in supercooled acetaminophen and flutamide with several polymers and found the polymers far from equal, and in opposite order to their effect on growth for acetaminophen trasi2012a C46, C53; trasi2012b C45; in 2013 Kestur and co-workers read the 60–77-fold slower crystallization of felodipine powders with 3% PVP as mainly a nucleation effect kestur2013b C36 (see F17).

How it was shown

Marsac, Konno and Taylor (2006) compared the two dihydropyridine calcium channel blockers alone and in PVP K29/32 dispersions under dry conditions: nucleation rates in spin-coated films at 22 °C and, for the Arrhenius activation energy, at several temperatures between about 25 and 75 °C, alone and with 3 wt% PVP marsac2006a P4, P5, A3. They measured Tm, heat of fusion and heat capacities by DSC and modulated DSC to integrate the configurational enthalpy, entropy and free energy of crystallization, checked them against the Hoffman equation, aged glasses at 23.5 °C for enthalpy recovery (KWW τ and β), read hydrogen bonding from the NH stretch, and computed drug activity in PVP with Flory–Huggins theory and cited interaction parameters marsac2006a P7, P8, P9, P6, P11. Marsac and co-workers (2008) stored the same pairs with 0–25 wt% PVP at 0, 11, 33, 58 and 75% RH, measured nucleation in films, water sorption and Tg of bulk powders at 25 °C, re-plotted nucleation against water content and Tg − T, and measured crystal solubility in 1-methyl-2-pyrrolidone (NMP), water and NMP–water as stand-ins for PVP and the PVP–water mixture marsac2008a P1, P2, P3, P6, P7, P8, P9.

What the data show

Dry: Table II gives 0.39 against 0.17 × 10¹⁰ m⁻³ day⁻¹ for nifedipine and felodipine glasses; the Fig. 2 slopes and Fig. 9 give larger absolute values, but every source ranks nifedipine first by a factor of about 2–4 marsac2006a R4.a, R1, R16.a, X1. Only nifedipine crystallized on heating a melt-quenched glass at 2 °C/min (metastable form at 87 °C) marsac2006a R5, C11. The free energy of crystallization at 25 °C was 9.6 against 6.2 kJ/mol (Hoffman 8.8 and 6.2), from enthalpies of 22.1 and 16.1 kJ/mol against entropy penalties TΔS of 12.5 and 10.0 kJ/mol marsac2006a R4.i, R4.j, R4.k, R4.l. The activation energy below Tg was 122 (11) against 145 (3) kJ/mol, above Tg 78 (8) against 95 (12) kJ/mol, not significantly different marsac2006a R3, C9. In PVP, nifedipine stayed above felodipine at 7–25 wt% (for example 4.3 × 10² against 7 × 10¹ m⁻³ s⁻¹ at 25 wt%), while Tg against composition, enthalpy recovery and drug–PVP hydrogen bonding per donor were alike and Flory–Huggins activity at a PVP volume fraction of 0.2 was 0.84 against 0.83 marsac2006a R16, R17, R7, R15, R19, C43.

With water the ranking held in all 30 combinations of PVP content and RH marsac2008a R1, C7. Nifedipine dispersions were the more RH-sensitive: with 7% PVP the rate rose about 400-fold from 0 to 75% RH for nifedipine and about 100-fold for felodipine marsac2008a C10. Nifedipine systems sorbed more water (75% RH: 1.8 against 1.1 g/100 g pure; 5.5 against 4.6 with 25% PVP), yet at a given water content the two drugs' Tgs were reported statistically equivalent marsac2008a R2, C12, C20, X9. At equal Tg − T nifedipine still nucleated faster, for example 2.0 × 10⁴ at 14.5 °C against 1.5 × 10³ m⁻³ s⁻¹ at 15.1 °C with 25% PVP; only the pure drugs converged when Tg − T was below 15 °C marsac2008a R8.e, R8.f, C23, X7. Crystalline nifedipine was half as soluble in NMP (mole fraction 0.05 against 0.10, γ 0.09 and 0.30), so the difference lies in its lattice term; in water felodipine was 20 times less soluble (γ 2.43 × 10⁵ against 1.89 × 10³) marsac2008a R10, R11, C35, C38. Counting only polymer plus water as the drug's co-solvent, water made up 23 wt% of it in the nifedipine 10% PVP dispersion at 75% RH against about 16 wt% for felodipine marsac2008a R13, E6, C45.

Figure

The zero line is the glass; nifedipine (circles) lies below felodipine (triangles) at every temperature under its melting point, and the solid Hoffman lines run close to the points marsac2006a R11, C14.

At the same distance below Tg the open nifedipine curves lie above the filled felodipine curves, most clearly with 25% PVP (panel c) marsac2008a R8.

Why

The authors' account, with their hedges: Tg and DSC relaxation time are the same for both glasses, so either other factors dictate nucleation or this measure of mobility misses the motion involved ("suggests either") marsac2006a C37. Thermodynamically, nifedipine has the larger driving force and the difference is enthalpic; an entropy-dominated view (glasses with larger entropies of crystallization crystallize less readily, Zhou and co-workers) would have ranked felodipine first, and Turnbull's rule linking interfacial energy to heat of fusion would also favour felodipine marsac2006a C15, C35, C36, R21, R22. Kinetically, felodipine's higher barrier below Tg is "evidence that kinetic factors are also important"; because nifedipine gains hydrogen-bond strength on crystallizing (crystal NH band about 3318 cm⁻¹, below the glass) while felodipine loses it (about 3363 cm⁻¹, above), "it would be reasonable to speculate" that felodipine must first weaken its amorphous hydrogen bonds marsac2006a C38, C39, R12, C17. Which factor dominates is left open marsac2006a C41. With water, the 2008 authors treat polymer plus water as a co-solvent: nifedipine's crystal is less soluble below about 25 wt% water in that co-solvent, and its dispersions sorb more water, so "it is reasoned" that nifedipine is more supersaturated in every system marsac2008a C42, C43, C48, C49. No supersaturation was computed for any dispersion, NMP stands in for PVP, and the printed cosolvency fits do not reproduce their own points and cross at 16.5 rather than about 25 wt% water marsac2008a X4, C33, X2, X3. (synthesis: both papers point to the crystal side of the free-energy balance; the Hoffman estimate from one DSC melting run already ranked the pair correctly.)

Conditions and limits

One drug pair, one polymer, films at 22–25 °C (the 2008 paper states 25 °C in Fig. 1 and 22 °C in Fig. 7) marsac2008a X1. Nucleation was measured on spin-coated films, Tg, heat capacity and sorption on differently prepared bulk samples marsac2006a X11; marsac2008a X12. The below-Tg restriction of the activation-energy difference is dropped in the abstract marsac2006a X6. Growth rates were not measured.

Counterexamples and conflicts

The pure drugs converged at Tg − T below 15 °C, and two pure felodipine points sit at or above the nifedipine curve, so "always" holds for the dispersions only marsac2008a X7. In the library the driving force pointed the right way for this pair but not for indomethacin against felodipine: indomethacin has the larger driving force yet crystallized less readily under the conditions compared alonzo2010 C42 (see Crystallization tendency F6). In supersaturated aqueous solution the same order appeared: felodipine, a slow crystallizer, exceeded its amorphous solubility, whereas nifedipine crystallized just below it raina2014a C31 (see Crystallization tendency F18).

Transfer check

For two candidates of similar Tg, compute the Hoffman free energy of crystallization at the storage temperature from Tm and heat of fusion; a 1.5-fold or larger difference, as here (synthesis: 8.8/6.2 = 1.4 by Hoffman marsac2006a R4.l), predicts the faster nucleator. Confirm by counting nucleation in neat glass and at the planned PVP level, dry and at 75% RH. Measure the crystal solubility in a liquid analogue of the polymer (NMP for PVP) and in water: the drug with the lower analogue solubility and the higher water uptake falls on the nifedipine side here marsac2008a R10, R2. The evidence covers dihydropyridines in PVP only.

History

Tang, Pikal and Taylor (2002) had shown that hydrogen-bond strength differs among the crystals of these dihydropyridines but not among their glasses marsac2006a R20. Marsac and co-workers (2006) first compared the two drugs dry, alone and in PVP, and concluded that the crystallization tendency of drugs from dispersions "may be dependent" on that of the pure drugs marsac2006a C52. Marsac and co-workers (2008) extended the comparison to sorbed water and replaced the dry driving-force argument with a co-solvent supersaturation argument marsac2008a C49, C51. The 2009 moisture study recomputed the Hoffman driving forces of both drugs (8.2 and 6.1 kJ/mol) and found that this property did not decide moisture-induced phase separation (F13) rumondor2009c M1, M6, C32. The neat-drug classification later placed nifedipine in class II and felodipine in class III baird2010 R24, R25 (see Crystallization tendency F17).

How it was shown

Kestur and co-workers (2010) asked how PVP chain length, PVP amount and temperature set the growth rate of felodipine crystals from the melt, and whether the slowing is thermodynamic or kinetic kestur2010b A1, A2, A3. A 10% PVP dispersion made by rotary evaporation was cryomilled with crystalline felodipine to 0.5–4.5 wt% PVP, melted between two cover glasses, quenched to −5 °C and stored dry at 40 °C until nuclei formed; the nuclei were grown on a hot stage at 70, 80, 90, 95, 100, 105 and 110 °C and the growth rate was the slope of crystal diameter against time kestur2010b P1, P2, P3. Raman showed that all bulk crystals were the metastable form II; form I crystals at the cover-glass edges of pure felodipine were excluded kestur2010b R2, C7. Tg (onset), one Tm (3% K-12), the NH stretch of 20 wt% PVP films at 25 °C and melt viscosity at 90–110 °C were measured alongside kestur2010b P5, P6, P7, P4. Later melt-growth series used the same hot-stage method: ATP enantiomers with 5% w/w PVP K-12 at 40–90 °C, indomethacin polymorphs with 10% w/w PVP K30 from 80 °C to near their melting points, and celecoxib with 10% PVP K29/32 at 80–100 °C sato2015 P4; tian2017 P4; xie2016b P9.

What the data show

Pure felodipine grew at about 3.5 × 10⁻⁹ m/s at 70 °C, 6.6 × 10⁻⁸ at 90 °C and 1.7 × 10⁻⁷ m/s at 105 °C, the maximum, and 1.5 × 10⁻⁷ m/s at 110 °C kestur2010b R7, C9. With 3 wt% PVP the rate at 70 °C was about 3.4 × 10⁻¹⁰ (K-12), 2.3 × 10⁻¹⁰ (K-29/32) and 1.6 × 10⁻¹⁰ m/s (K-90), and at 110 °C about 5.3 × 10⁻⁸, 5.3 × 10⁻⁸ and 2.7 × 10⁻⁸ m/s kestur2010b R7.a, R7.g. In the K-29/32 series the reductions were 2- and 6-fold at 110 °C for 1.5 and 4.5 wt% but 3- and 35-fold at 70 °C kestur2010b R13, C18. Plotted on a log axis against concentration each temperature gave a straight line, steeper at lower temperature kestur2010b R14, C19; with the printed constants each added wt% multiplies the growth rate by about 0.44 at 70 °C and 0.61 at 105 °C (synthesis: e^{-0.822} and e^{-0.486} kestur2010b E1). The crystal shape stayed spherulitic and the growth stayed linear in time over up to 60 h kestur2010b R5, R12, R19.

Three years later the same PVP grade at 3% was tested on both felodipine polymorphs from 60 to 100 °C. The ratio of the growth rate without polymer to that with polymer fell with temperature for both: about 15.8, 17.5, 13.2, 6.8, 3.9 and 3.5 for form I and 18.7, 15.4, 12.5, 8.2, 7.2 and 6.5 for form II at 60, 70, 80, 90, 95 and 100 °C kestur2013a R15. The trend of the 2010 study therefore holds down to 60 °C and for the stable form I as well (see F19). For three other drugs measured in the same years, the PVP factor did not follow the felodipine trend (synthesis: ratios of plot readings): with 10 mol% PVP K12 flutamide grew about 11-, 10-, 8- and 14-fold slower at 40, 50, 60 and 70 °C (1.4/0.13, 5.7/0.59, 20/2.6, 77/5.6) trasi2012b R3.a, R3.h; with 10% w/w PVP acetaminophen grew about 4.7-fold slower at 70 °C and 9.2-fold slower at 100 °C (0.93/0.20, 45/4.9) trasi2012a R6.a, R6.e; and with 10% PVP celecoxib grew about 20-fold slower at 70 °C, 35-fold at 100 °C and 29-fold at 120 °C (0.0020/0.00010, 0.73/0.021, 12/0.42) mosqueragiraldo2014 R1.a, R1.b.

The papers of 2015–2017 measured the PVP factor for three more compounds. With 5% w/w PVP K-12 both enantiomers of the model compound ATP grew about 10 times slower at 50 °C, 5–6 times at 60 °C, 3–4 times at 70 °C and 2–5 times at 80–90 °C, the felodipine trend sato2015 R5.e, R5.f. With 10% w/w PVP K30, indomethacin δ grew about 25 times and α about 16 times slower than without polymer at 80 °C; with rising temperature the δ ratio fell to about 5.7 at 120 °C, while the α ratio rose to about 34 at 120 °C and 558 at 135 °C, above the temperature of fastest growth with PVP (see F29) tian2017 R12, R15. With 10% PVP K29/32, celecoxib grew about 330, 280 and 160 times slower at 80, 90 and 100 °C (synthesis: 0.010/0.00003, 0.14/0.0005 and 0.97/0.006 µm/s xie2016b R21).

Figure

The top curve (circles, pure felodipine) peaks near 105 °C; the gap to the three PVP curves (triangles K-12, diamonds K-29/32, squares K-90) is the inhibition, widest at 70 °C kestur2010b R7.

One straight line per temperature, 105 °C at the top and 70 °C at the bottom; the lower lines are steeper kestur2010b R14.

Why

The authors' account, with their hedges: over 70–110 °C growth slowed with cooling although the driving force rises, so molecular transport dominates and growth is diffusion-limited ("indicating") kestur2010b C22, C23. Tg (43.4–44.1 °C against 43.3 °C) and the one melting point (140.7 against 141.5 °C with 3% K-12) did not change, which rules out antiplasticization and makes a thermodynamic cause "unlikely"; the effect is "most likely kinetic in origin" kestur2010b R1, C26, C28, C34. "We speculate" that drug–PVP hydrogen bonds lower felodipine mobility, possibly by coupling drug and polymer motions, and the weakening of these bonds on heating "would also provide a potential explanation" for the weaker inhibition at high temperature kestur2010b C35, C37. The hydrogen-bond signal (NH 3341 → 3337 cm⁻¹) was the same for all three grades and was recorded at 25 °C and 20 wt% PVP, so it cannot explain the molecular-weight effect, and mobility at the growth temperatures was not measured kestur2010b R9, C30, X12. A companion study of polymer chemistry offered a related picture in which growth needs drug–polymer demixing at the front, harder when the bonds are strong and easier at high temperature where they weaken kestur2010a C42, C43. (synthesis: both readings predict the observed loss of inhibition on heating; neither was tested against a measured interfacial composition.)

Conditions and limits

One drug and one polymer chemistry, form II only, bulk growth in films sandwiched between cover glasses at 27–67 °C above Tg (synthesis: 70 − 43 and 110 − 43), polymer 0.5–4.5 wt% kestur2010b A1, A2, A3, M1. The authors themselves restrict Eq. 1 to temperatures well above Tg and below the growth maximum; the 110 °C series is not in the fit, and the printed constants give 40-fold where 35-fold was measured at 70 °C and 4.5 wt% kestur2010b C21, X14. The supercooling is computed from the form I melting point although form II grew kestur2010b X15.

Counterexamples and conflicts

The fall of pure felodipine growth from 105 to 70 °C is about 49-fold in this paper (synthesis: 1.7 × 10⁻⁷/3.5 × 10⁻⁹ kestur2010b R7.a, R7.f; text "approximately 2 orders of magnitude" from 110 to 70 °C kestur2010b C9, X9) but "3 orders of magnitude" in the polymer-chemistry study of the same year kestur2010a C10. At 110 °C K-12 and K-29/32 gave the same 3-fold reduction and the larger step lay between the two highest grades, the opposite of the 70 °C trend kestur2010b R8, X13. Some additives accelerate growth instead (see Crystallization tendency F19 and F22). The rise of the PVP factor on cooling is a felodipine result: for acetaminophen the factor fell on cooling, from about 9 at 100 °C to about 5 at 70 °C, and for flutamide and celecoxib it showed no trend over 40–70 and 70–120 °C (synthesis from the ratios above) trasi2012a R6; trasi2012b R3; mosqueragiraldo2014 R1. The temperature windows differ relative to each drug's Tg and growth maximum, so the comparison is not at matched conditions (synthesis: felodipine 70–110 °C against Tg 43 °C; acetaminophen 70–110 °C against Tg 21.2 °C; celecoxib 70–120 °C against Tg 55.3 °C) kestur2010b R1; trasi2012a R9.a; mosqueragiraldo2014 R5.a. The trend also failed above the temperature of fastest growth: for indomethacin α the PVP factor rose with temperature from 105 °C upward, after the polymer had lowered that temperature from 135 to 125 °C (see F29) tian2017 R12, R15. For celecoxib two papers disagree on the size of the PVP factor at 10% polymer, about 20–35-fold at 70–100 °C in 2014 and about 160–330-fold at 80–100 °C in 2016 (synthesis from the ratios above), with different preparation and growth set-ups mosqueragiraldo2014 P1, P3; xie2016b P9.

Transfer check

Measure melt growth of the drug's polymorph of interest at three temperatures between about Tg + 30 °C and the growth maximum, with 0, 1.5 and 3 wt% of the candidate polymer, and fit ln growth rate against concentration. A slope that steepens on cooling, as here (−0.49 to −0.82 per wt% kestur2010b R15), means the polymer protects most where the melt is most viscous; do not extrapolate the fit to storage temperatures near Tg kestur2010b C21. Choose the middle molecular weight first: K-29/32 gave most of the chain-length benefit while K-90 raised melt viscosity 10–20-fold kestur2010b C15, C29.

History

Konno and Taylor (2006) had seen irregular spherulites with dark droplets at 3% polymer and speculated that rejected polymer slows growth, without measuring growth konno2006 R2, C11, X14. In 2010 two papers from the group measured felodipine growth directly: the polymer-chemistry study ranked four polymers at 3% (see Miscibility and drug-polymer interactions F10) kestur2010a R6, and this study varied PVP amount, chain length and temperature kestur2010b A1, A2, A3. In 2012 the growth method was carried to flutamide and acetaminophen and in 2014 to celecoxib, where PVP slowed growth about tenfold or more but without the felodipine temperature trend trasi2012b C7; trasi2012a C15; mosqueragiraldo2014 C8; in 2013 Kestur and Taylor confirmed the felodipine trend from 60 to 100 °C for form I and form II kestur2013a R15. Later work extended the comparison of polymer effects across many drugs (see Crystallization tendency F19) trasi2014a R6, C11. Sato and Taylor (2015) and Tian and co-workers (2017) added enantiomer and polymorph comparisons with PVP, and Xie and Taylor (2016) a celecoxib series sato2015 C10; tian2017 C28; xie2016b C36; the title of this finding is unchanged.

How it was shown

Konno and Taylor (2006) wrote the nucleation rate with classical nucleation theory as I = A\exp\left(-\left(\Delta G^{*} + \Delta G_{a}\right)/kT\right), a thermodynamic barrier set by interfacial energy against driving force plus a kinetic barrier for molecules crossing onto the nucleus, and estimated the change in drug activity from the Flory–Huggins equation with literature parameters for PVP–sucrose and PVP–water konno2006 E8, E9, C33, C35. Marsac and co-workers (2006) used the same activity equation with χ from melting point depression (−3.8 nifedipine, −4.2 felodipine) and measured activation energies from Arrhenius plots of nucleation with and without 3% PVP marsac2006a E10, R18, E2, R3. Kestur and co-workers (2010) tested the driving force through melting point depression and modelled PVP diffusion around a disk-shaped crystal growing at the measured rate (2.37 × 10⁻⁸ m/s, 3% K-29/32, 100 °C) with a Stokes–Einstein tracer diffusivity (4.98 × 10⁻¹⁶ m²/s) and a fraction ε of polymer expelled by the crystal kestur2010b P6, P10, R23, E9, E11. Sato and Taylor (2016) measured melting points of physical mixtures with 1–20% PHB and IR of melt films sato2016 P4, P5; Tian and co-workers (2017) wrote the Flory–Huggins free energy of mixing without inputs tian2017 E2.

What the data show

The predicted activity reductions were small and nearly the same for both drugs (0.96, 0.84 and 0.67 at PVP volume fractions of 0.1, 0.2 and 0.3), so on the driving-force term alone nifedipine "would be expected" to keep nucleating faster at any PVP level marsac2006a R19, C47. With 3% PVP the activation energy rose for both drugs (nifedipine 122 → 127 below Tg and 78 → 103 kJ/mol above; felodipine 145 → 201 and 95 → 111), significantly for felodipine below Tg and nifedipine above Tg marsac2006a R3, C20. The nucleation rate fell 40–60-fold with 3% of any of three polymers that differ in Tg and hydrogen bonding, whereas dilution alone would scale with the polymer fraction konno2006 R4, C37 (see F1). In the growth model, expelled PVP piled up just ahead of the front: with ε = 1 the surface weight fraction reached 0.5 within about 52 s, with ε = 0.1 about 0.46 at 500 s and even with ε = 0.01 about 0.07 and still rising, while the measured growth stayed constant over up to 60 h kestur2010b R21, R19, C41.

The papers of 2012–2014 measured the melting point with each additive. With 10 mol% of six polymers flutamide melted at 108–110.1 °C against 110 °C, so the authors judge the degree of supercooling hardly changed trasi2012b R7, C33; with 10% w/w of seven polymers acetaminophen melted at 166.7–168.6 °C against 169 °C ("similar for all systems") trasi2012a R9, C19. Celecoxib with 10% of PVP or one surfactant melted 1.7–4.5 °C below the pure drug (163.46 °C), and with 10% TPGS plus 10% PVP 12.0 °C below (151.49 °C) mosqueragiraldo2014 R5. For felodipine, 3% PVP slowed form I and form II by similar factors at 60–80 °C (ratios 15.8–13.2 and 18.7–12.5) kestur2013a R15, and in powders it slowed the early crystallization 60–77-fold but growth only about 2-fold (surface) and 7-fold (bulk) kestur2013b R15, R12.

The papers of 2016–2017 again found small melting-point changes and added a driving-force argument. With 1–10% PHB the melting points of six compounds moved by 0–2 °C, and by up to 4 °C at 20%, and IR showed no band shifts, so the authors infer an unchanged driving force and an effect on mobility (see F28) sato2016 R7, R9, C18, X11. For indomethacin with 10% HPMC, HPMCAS or PVP, Tian and co-workers wrote the Flory–Huggins free energy of mixing and argued that stronger drug–polymer interaction "would" lower the driving force, so that PVP acts "most likely" through both lower mobility and lower driving force; above the temperature of fastest growth they name melting point depression and a smaller driving force as "possible reasons" for the steep rise in inhibition tian2017 E2, C30, C31, C36. Neither the interaction parameter nor the driving force was measured tian2017 X13.

Figure

The nifedipine (solid) and felodipine (dashed) curves nearly coincide and stay above about 0.8 up to a volume fraction of 0.2 marsac2006a R19.

If the crystal pushes all polymer out (dotted), the front becomes polymer-rich within a minute; even 1% expulsion (solid) keeps the front composition rising kestur2010b R21.

Why

The three kinetic proposals, with the authors' hedges: (1) a miscible polymer cannot enter the drug crystal, is rejected and diffuses away slowly below Tg, so a polymer-rich layer builds at the nucleus that "would impede" drug transport and "may also" raise the surface cost konno2006 C39; the dark droplets on crystallites grown with 3% polymer are "thought" to be such a phase konno2006 R2, C11, C42. (2) PVP raises the activation energy for nucleation, more for felodipine, whose amorphous hydrogen bonds are stronger than its crystal's marsac2006a C48, C39. (3) drug–PVP hydrogen bonds couple drug and polymer motion and lower drug mobility ("We speculate"), and because a constant growth rate requires a constant front composition, "it seems reasonable to conclude" that the chains are encaged in the crystal rather than expelled kestur2010b C35, C38, C42. (synthesis: proposals 1 and 3 disagree on where the polymer goes; they come from different regimes, nucleation in films at 22 °C, about 20 °C below Tg, and growth in melts at 100 °C, 57 °C above it konno2006 P2; kestur2010b R23, so both may hold in their own regime, but no paper measured the polymer at an interface.) The driving-force argument is robust within its own assumptions; the same conclusion was reached for higher PVP contents by the miscibility papers, which found most dispersions supersaturated and their stability "predominantly" kinetic (see Miscibility and drug-polymer interactions F9) marsac2006b C41. (4) Two polymorphs of felodipine have different crystal surfaces but grow from the same liquid, so a polymer acting at the faces should slow them by different factors; the similar ratios "most likely" mean that PVP acts through the amorphous phase kestur2013a C1, C6. (5) For felodipine powders, the authors argue that growth inhibition alone cannot explain a 60–77-fold slower crystallization, so PVP "must" also slow nucleation (see F17) kestur2013b C34, C35. (synthesis: reading 4 is compatible with the coupling of drug and polymer motion proposed in 2010 (proposal 3) and less so with a polymer layer at the face (proposal 1), but neither it nor reading 5 involved a measurement at an interface.)

Conditions and limits

Activities are Flory–Huggins estimates for non-equilibrium glasses with cited χ values and an unprinted volume ratio marsac2006a C46, E10. The felodipine–polymer χ was not known for HPMCAS or HPMC konno2006 C35. Only one Tm (3% K-12) was measured, and the crystal–melt interfacial energy, the other thermodynamic lever, was not measured kestur2010b X11, C27. The model's tracer diffusivity cannot be reproduced from the stated solvent viscosity, the radial front speed is taken as the diameter slope, and predicted weight fractions exceed 1 kestur2010b X4, X5, X16.

Counterexamples and conflicts

For felodipine nucleation, hydrogen-bond strength (PVP > HPMC > HPMCAS) did not correlate with inhibition konno2006 C31, whereas for felodipine melt growth the hydrogen-bond order of four polymers matched the growth-inhibition order kestur2010a C26 (see Miscibility and drug-polymer interactions F12). (synthesis: hydrogen bonding may matter for growth and not for nucleation, or the 70%-polymer IR used by konno2006 misrepresents 3–25% films konno2006 X11.) Konno's droplets suggest polymer outside the crystal, while the 2010 PXRD and model place PVP inside it without changing the lattice; the PXRD shows only three reflections of one family kestur2010b C43, X10. Where melting-point depression is large, the driving force may matter after all: with 10% TPGS plus 10% PVP celecoxib melted 12 °C lower and grew 0.03 times as fast as with PVP alone at 120 °C, which the authors say "could contribute" through a smaller driving force near Tm; but SLS plus PVP, with a 7.3 °C depression, grew faster than PVP alone, so the depression does not order the mixtures (see F22) mosqueragiraldo2014 R5, R2.b, C33, X13. At 90–100 °C the felodipine polymorph ratios differed 1.2–1.9-fold with non-overlapping error bars, and the form I ratios at 70–80 °C do not follow from the plotted rates, which weakens the equal-factor reading kestur2013a X6, X7. Tian and co-workers set their result against the liquid-phase reading of the felodipine polymorph test: each polymer slowed three indomethacin polymorphs by different factors, which they read as action at each crystal–melt interface (see F29) tian2017 C38, C39.

Transfer check

Measure the melting point depression of the drug with 3–5% of the polymer at 1 °C/min: a shift within about 1 °C, as here, means the polymer cannot act through the driving force at that level kestur2010b R1.e. Estimate the drug activity with Flory–Huggins and a measured χ; above about 0.8 at the planned polymer fraction the polymer must act kinetically marsac2006a R19. To tell the kinetic routes apart, image the crystal–amorphous interface (for example by nanoscale IR) for polymer enrichment, which none of these papers did.

History

Crowley and Zografi (2003, outside the library) had proposed polymer accumulation at the crystal front for indomethacin with low PVP konno2006 C39. Konno and Taylor (2006) and Marsac and co-workers (2006) first argued in the library that a few percent of polymer acts on kinetics, not on the driving force konno2006 C40; marsac2006a C47, C48. Kestur and co-workers (2010) tested the front-accumulation idea quantitatively and replaced it by encaging kestur2010b C42. In the same year the polymer-chemistry study linked growth inhibition to drug–polymer demixing at the front (see Miscibility and drug-polymer interactions F12) kestur2010a C42. In 2012 the unchanged melting points with 10 mol% or 10% w/w polymer were found again for flutamide and acetaminophen trasi2012b C33; trasi2012a C19; in 2013 Kestur and co-workers added the polymorph test, placing the PVP effect in the liquid, and the powder argument that PVP acts mainly on nucleation kestur2013a C6; kestur2013b C36; in 2014 a polymer–surfactant mixture gave the first large melting-point depression in the chapter, offered as a partial cause of slower growth mosqueragiraldo2014 C33. The short part of this finding gained one sentence for these papers; its title is unchanged. In 2016 a low-Tg polymer without interactions sped growth with an unchanged melting point (see F28), and in 2017 the indomethacin polymorph study moved the kinetic reading from the liquid to the crystal–melt interface sato2016 C18; tian2017 C39.

How it was shown

Konno and Taylor (2006) measured onset Tg of rotary-evaporated dispersions at 0–100 wt% polymer by DSC and enthalpy relaxation of 9:1 dispersions aged at 2 °C (KWW fit), and compared both with the nucleation rates of spin-coated films konno2006 P3, P4, P6, R6, R9. Marsac and co-workers (2006) compared Tg and enthalpy recovery at 23.5 °C of the two drugs and their 10 wt% PVP dispersions marsac2006a P7, P9. Konno and Taylor (2008) plotted the nucleation rates of humid films against T/Tg, using Tgs of bulk dispersions equilibrated at each RH konno2008 P9, R13; Marsac and co-workers (2008) plotted nucleation against Tg − T marsac2008a P9, R8. Kestur and co-workers (2010) measured Tg and melt viscosity of felodipine with 3 wt% PVP and scaled growth by viscosity kestur2010b P4, P5, R16. Rumondor and co-workers (2010) compared drug-alone growth from 13 to 75% RH with an Adam–Gibbs–Vogel relaxation time computed from cited Tg values at each RH rumondor2010 P8, E1, R7. Trasi and Taylor (2012) measured onset Tg of flutamide melts with 10 mol% of six additives and of acetaminophen dispersions with 10% w/w of seven polymers by DSC at 10 °C/min after cooling at 15 °C/min, and compared them with growth rates of the same compositions trasi2012b P5, R7; trasi2012a P4, R9. Wegiel and co-workers measured polymer and dispersion Tg (second heating, modulated DSC for PVP) and set them against storage outcomes for resveratrol (2013) and curcumin (2014) wegiel2013b P6, P7, R18; wegiel2014 P7, R13. Mosquera-Giraldo and co-workers (2014) replotted celecoxib growth rates with 10% additive against T − Tg mosqueragiraldo2014 P7, R12. Later papers compared Tg of ATP–polymer glasses with enantiomer growth, Tg of celecoxib dispersions after 3 days at 56–94% RH with storage, Tg of humidified tacrolimus and its HPMC dispersion with storage at 40 °C/75% RH, and Tg of precipitated lumefantrine dispersions with storage sato2015 P6; xie2016b P7; trasi2017 P3; bhujbal2021 P7; the PHB paper replotted growth against T − Tg sato2016 P6.

What the data show

At 7, 15 and 25% polymer the dry Tgs were about 44, 48 and 55 °C with PVP, 42, 44 and 44 °C with HPMCAS and 45, 42 and 43 °C with HPMC, against about 44 °C for felodipine, and the PVP increase starts only near 15% konno2006 R6, X15; the nucleation rates of the three systems were equal konno2006 R4. At 2 °C the 9:1 dispersions relaxed with τ of 208 h (PVP), 319 h (HPMC) and 687 h (HPMCAS), an order that differs from the Tg order and was not related to nucleation by the authors konno2006 R9, X13. Nifedipine and felodipine glasses had Tg onsets of 42.3 and 43.1 °C and τ of 6.3 and 4.2 h; with 10 wt% PVP τ became 42.1 h for nifedipine and 16.7 h for felodipine, so the faster-nucleating dispersion was the slower-relaxing one marsac2006a R4.c, R4.n, R7. Felodipine with 10 wt% PVP lost mobility (65% against 84% of the maximum enthalpy recovery at 16 h) although its Tg fell to 40.5 °C marsac2006a R8, C26. After 75% RH all felodipine films with up to 25% polymer had Tgs of about 30–35 °C, below dry felodipine, yet most nucleated more slowly than dry felodipine konno2008 R7, C27. At T/Tg ≈0.68–0.69 the rates ranged from about 2.9 × 10² (25% HPMC) to 9.4 × 10³ m⁻³ s⁻¹ (7% PVP) konno2008 R13. At a given Tg − T more PVP lowered felodipine nucleation further, about 1.7 orders with 10% and 2.2 orders with 25% PVP at about 15 °C (plot readings) marsac2008a R8, X8. With 3 wt% PVP the zero-shear viscosity at 100 °C was about 58 (K-12), 156 (K-29/32) and 690 Pa·s (K-90) against 44 Pa·s, and viscosity-corrected growth with K-12 lay below that of pure felodipine kestur2010b R10, R16, C24. Between T/Tg ≈0.94 and ≈0.97 log(growth rate) + 1 rose from about −0.59 to −0.23, much less steeply than the AGV mobility line rumondor2010 R7, C26.

In flutamide (Tg −3.1 °C) the six polymers gave onset Tg of −1.07 to 0.6 °C with little difference between them, while their growth effects ranged from none (PAA, Tg not obtained because it crystallized on cooling) to tenfold (PVP); nilutamide raised Tg most (to 2.5 °C) but was not the best inhibitor trasi2012b R7, R3, C33. In acetaminophen (Tg 21.2 °C) the seven polymers gave 22.9–25.4 °C: PAA (+4.2) and PVP (+3.9) were strong growth inhibitors, HPMC (+3.4) and E100 (+3.0) weak ones trasi2012a R9, R6, C18. For resveratrol the polymer Tg order PVP K29/32 (159 °C) > HPMC (145) > CMCAB (137) > PAA (128) > HPMCAS (111) > PVP K-12 (91) > E100 (41 °C) did not follow storage stability, in which PVP K-12 behaved like K29/32 and E100 was second only to PVP; dry dispersion Tg values were 63–85 °C wegiel2013b M2, M3, M4, M5, M6, M7, M8, R12, R18, C33. For curcumin at 75% drug, the dispersions had Tg of 61.7–68.3 °C; the most stable (E100, 62.4 °C) and the least stable (PVP, 62.8 °C) differed by 0.4 °C wegiel2014 R13, C22. With 10% surfactant the Tg of celecoxib fell from 55.34 to 44.81 (TPGS) and 44.46 °C (sucrose palmitate) and growth at 70 °C rose about 85- and 7-fold; replotted at equal T − Tg the binary curves came together to within about 12-fold at about 25 °C above Tg, but TPGS stayed about 8 times above the drug and the PVP-containing ternaries stayed far apart mosqueragiraldo2014 R5, R1, R12, X4.

The papers of 2015–2021 add four cases. HPMCAS-HF slowed one enantiomer of ATP about twice as much as the other while every sample kept a Tg of 14.3–18.4 °C sato2015 R1, R5, C6. Celecoxib dispersions held 3 days at 94% RH had Tgs of 37.4–42.1 °C with four polymers, yet the PVP dispersion crystallized within a month and the PVPVA and HPMCAS ones lasted 150 and 115 days; the authors state that Tg "was not predictive" xie2016b R17, R23, C46. Pure amorphous tacrolimus and its 1:1 HPMC dispersion had similar Tgs after 75% RH (51.4 and 49.0 °C), but only the drug alone crystallized within 4 weeks at 40 °C/75% RH trasi2017 R15, R16. In lumefantrine dispersions the 20:80 compositions had the lowest Tg (45 °C with HPMCAS, 88 °C with HPMCP) and stayed amorphous as long as the others bhujbal2021 R10, R4, C46. The 2014 limit also holds for a polymer: 5–10% PHB lowered the Tg of three high-Tg compounds by 5–12 °C and sped their growth up to tenfold, and against T − Tg the rates fell close to those of the pure compounds (see F28) sato2016 R7, R10, C21.

Figure

At the same abscissa the rates scatter over more than a decade, with no single trend; the plotted values reproduce 25/Tg in °C rather than a kelvin ratio, which keeps the order but not the meaning of a reduced temperature konno2008 R13, X1.

The 13–75% RH growth points follow the dashed line with a coupling exponent of 0.5, not the solid AGV line; the four high-RH points lie far above both rumondor2010 R7, X9.

Why

The authors' account, with their hedges: a higher Tg "would certainly contribute" to inhibition but "is not the only factor" konno2006 C27; either factors other than Tg decide nucleation, or DSC relaxation does not describe the motion involved marsac2006a C37; modes of mobility not reflected by Tg "must also be considered", and binary mixtures can be less mobile than the pure drug at a similar Tg marsac2008a C30; marsac2006a C26. Bulk viscosity "cannot be used as a precise measure" of felodipine mobility in PVP melts kestur2010b C24, C29. Growth faster than AGV mobility predicts may reflect decoupling of diffusion and viscosity or growth at the surface of uncovered films, much faster than in the bulk ("Possible explanations") rumondor2010 C28, C29. (synthesis: Tg-scaling assumes that every system has the same mobility at its Tg and that crystallization follows that mobility one to one; the polymer effects of F1 and F3 act at low polymer fractions that barely move Tg, and between drugs the crystal side of F2 decides, so Tg cannot carry the ranking.) The later authors draw the same conclusion: Tg rises "cannot completely explain" growth trasi2012b C33, "cannot explain" the differences between polymers trasi2012a C18, and "does not appear to be the only factor" in storage wegiel2013b C33; wegiel2014 C22. For miscible, low-Tg surfactants the authors read the faster growth "most likely" as an effect on mobility, since the Tg-corrected curves come closer mosqueragiraldo2014 C30. The Fox-equation deviations printed for the curcumin dispersions, read there as signs of interaction, were computed with temperatures in °C and the wrong composition; with kelvin and 75% curcumin only the E100 deviation remains large (+18.8 °C) wegiel2014 R18, X1.

Conditions and limits

Tg and relaxation came from bulk powders or melt-quenched glasses, crystallization from films; Tg was always measured on samples prepared differently from the films konno2006 X12; marsac2006a X11; konno2008 X6. Within one composition Tg still matters: nucleation fell as Tg − T rose in every series, and water lowering Tg speeds crystallization (F9) marsac2008a C21.

Counterexamples and conflicts

None found in the library that ranks polymers by dispersion Tg at the low polymer levels studied here; the miscibility papers reached the same conclusion for growth (see Miscibility and drug-polymer interactions F11) kestur2011 C12. For coated high-load dispersions, onset of crystallization in storage did follow the sign of Tg − T at the storage condition moseson2022a C24, so Tg − T can sort conditions within one formulation even where it cannot sort polymers (synthesis). A large Tg drop from a miscible additive does change growth: TPGS and sucrose palmitate lowered the Tg of celecoxib by about 10 °C and sped its growth up to about 85-fold at 70 °C, and the T − Tg replot narrows that spread to about 12-fold (synthesis: about half of the spread on a log scale) mosqueragiraldo2014 R1, R5, R12; the Tg rule fails for small Tg changes between polymers, not for large ones caused by a plasticizer. A large Tg drop from a non-interacting polymer acts the same way: PHB (Tg 0 °C) sped growth of the three compounds whose Tg it lowered and not of the three whose Tg it left unchanged (see F28) sato2016 R7, R11.

Transfer check

Do not screen polymers or drug candidates by dispersion Tg or T/Tg at polymer levels below about 25%. Use Tg only to compare the same formulation across storage conditions, and measure nucleation or growth directly at each condition. For the neat drug, check whether growth below Tg follows its relaxation time; a coupling exponent well below 1, as here (0.5 rumondor2010 C27), warns that Tg-based shelf-life estimates will be optimistic.

History

Van den Mooter and co-workers (2001, outside the library) had stressed the Tg increase from high-Tg polymers konno2006 C2. Konno and Taylor (2006) first showed in the library that no Tg increase was required konno2006 C26; Marsac and co-workers (2006) found the same for two drugs and added that DSC relaxation does not rank them marsac2006a C13, C37. In 2008 both moisture studies showed that neither T/Tg nor Tg − T orders humid films konno2008 C26; marsac2008a C23, and in 2010 melt viscosity and AGV mobility were found equally poor guides kestur2010b C29; rumondor2010 C26. The neat-drug classification reached the matching conclusion for reduced Tg (see Crystallization tendency F12). The papers of 2012–2014 extended the result to flutamide and acetaminophen growth and to resveratrol and curcumin storage trasi2012b C33; trasi2012a C18; wegiel2013b C33; wegiel2014 C22, and the 2014 surfactant study supplied the limit that a 10 °C Tg drop does speed growth mosqueragiraldo2014 C30. The short part and the type line of this finding were extended for these papers; the title is unchanged. Papers of 2015–2021 added a chiral test at constant Tg, Tg after humidity for celecoxib dispersions, the tacrolimus drug–dispersion pair and lumefantrine dispersions in which the lowest-Tg composition was among the most stable sato2015 C6; xie2016b C46; trasi2017 R15; bhujbal2021 C46, and the PHB study a second case of a large Tg drop sato2016 C21; the type line was extended for these papers and the title is unchanged.

How it was shown

Trasi and Taylor (2012) melt-quenched acetaminophen with 10% w/w of seven polymers, measured spherulite growth on a hot stage, found the nucleation temperature zone by cooling the melt in the DSC to successively lower end temperatures and reheating, and counted the nuclei that grew after one hot-stage cycle to −30 °C trasi2012a P1, P2, P6, P7. In a companion study they used 10 mol% of six simple polymers and nilutamide in flutamide, located the nucleation zone the same way, and read nucleation from the growth rate at the temperature of the DSC crystallization maximum: fewer nuclei push the exotherm to a higher temperature, where growth is faster trasi2012b P1, P2, P6, P7. Kestur and co-workers (2013) compared the early crystallization slope of felodipine powders with and without 3% PVP against surface and bulk growth rates measured in films at the same temperature, 30 °C kestur2013b R15, R12, E2.

What the data show

For acetaminophen the growth order at 10% w/w was PAA > PVP, PVPh, PVPVA > HPMCAS, E100, HPMC, at 100 °C about 1.6–2.8, 5–7 and 11–15 µm/s against about 45 µm/s trasi2012a R6, C15. The nuclei counts ran in a different order: HPMCAS 2.5, PVP 5.2, PVPh 9.2, HPMC 9.3, PVPVA 12.3, E100 13.1 and PAA 26, against 10.8 for the drug trasi2012a R12. PAA also moved the nucleation zone from 50–60 °C down to −10–30 °C, HPMCAS to 10–45 °C trasi2012a R12.a, R12.c, R12.e. In the drug's own zone growth was only about 0.04–0.23 µm/s (50–60 °C), against about 45 µm/s at 100 °C, which the authors give as the reason acetaminophen forms a glass: its nucleation and growth zones "do not substantially overlap" trasi2012a R8.a, C39. For flutamide, at 40 °C the additives gave growth rates of about 0.13 (PVP), 0.28 (PIPA), 0.43 (nilutamide), 1.0 (PVAc) and 1.6 µm/s (PAA) against about 1.4 µm/s trasi2012b R3. At the exotherm maximum, PVAc and PIPA gave growth rates of 7.7 and 7.0 µm/s against 4.4 for flutamide, read as fewer nuclei, while PVP (5.4), PDMA (4.1) and PVPH (5.9) were classed as no effect trasi2012b R11, C18. For felodipine the early slope fell from 2.21 to 0.037 h⁻¹ (size 1) and from 1.41 to 0.018 h⁻¹ (size 2) with 3% PVP, while PVP cut the surface growth rate from 2.7 × 10⁻⁹ to 1.45 × 10⁻⁹ m/s and the bulk rate from 0.47 × 10⁻⁹ to 6.37 × 10⁻¹¹ m/s kestur2013b R15, R12.

Figure

Read the last column against the growth ranking: HPMCAS (2.5) and PVP (5.2) below the drug (10.8), PAA (26) far above it trasi2012a R12.

On the log axis PAA lies lowest and HPMCAS, E100 and HPMC just below the drug, the reverse of the PAA–HPMCAS order in the nuclei counts trasi2012a R6.

Why

The authors' account, with their hedges: classical nucleation theory captures the thermodynamic and kinetic factors, but the polymer effects are "not particularly straightforward" to explain with it, and it does not encompass molecular recognition trasi2012a C48, C49. PAA "may" exert a local ordering effect on acetaminophen that enhances nucleation, by analogy with polymer-directed polymorphs from solution; HPMCAS "may" act like large additives that lengthen induction times in solution, "we can only speculate" trasi2012a C50, C51. For flutamide, nilutamide "may" enter prenuclei by molecular recognition and, being larger, block their growth ("quite possible"), "most likely" because of its small size relative to a polymer trasi2012b C47, C48. For felodipine, since overall crystallization is nucleation followed by growth, "it appears that PVP must" also inhibit nucleation strongly, so its stabilizing effect is attributed "to a large degree" to nucleation kestur2013b C35, C36. The authors of both 2012 papers conclude that the effect of a polymer on nucleation is not predictable from simple properties, unlike growth, which follows hydrogen bonding (see F18) trasi2012a C52; trasi2012b C45. (synthesis: growth inhibition acts where molecules attach to an existing crystal, nucleation where clusters form in the liquid; an additive that binds the drug strongly can slow attachment and still order the liquid locally, which is one way to read the PAA result.)

Conditions and limits

Growth was measured in melt films far above Tg (synthesis: about 30–110 °C above for acetaminophen, Tg 21.2 °C, and 43–73 °C above for flutamide, Tg −3.1 °C), nucleation in the zones found by DSC at 15 or 5 °C/min cooling, whose limits depend on the scan rates trasi2012a R9.a, C13; trasi2012b R7, P6. The acetaminophen counts add up nucleation over the time spent in each zone, and the PAA zone is four times wider; per second in the zone PAA (about 0.065 nuclei/s) would not exceed the drug (about 0.11/s) (card calculation) trasi2012a X8. The flutamide verdicts rest on two Tmax values per sample and an unnamed test, and the PIPA image does not show fewer nuclei trasi2012b X4, X6. The felodipine nucleation effect was not measured in the powders; it rests on rate ratios and a cited 150-fold value kestur2013b X7, R22.

Counterexamples and conflicts

Some polymers act on both: PVP slowed acetaminophen growth about 5–7-fold at 100 °C and halved the nuclei trasi2012a R6, R12.f, and PVP slowed both felodipine nucleation and growth kestur2013b C36. For flutamide the card reads the data as "no consistent relation" between growth and nucleation effects rather than an absence of nucleation effects among good growth inhibitors: PIPA, a strong growth inhibitor, also lowered nucleation trasi2012b X5. The dry felodipine study of 2006 found PVP, HPMC and HPMCAS equally effective on nucleation at equal weight konno2006 C14, while the same three polymers differ in growth inhibition kestur2010a R6; nucleation and growth were not measured in the same samples there.

Transfer check

After a growth screen, test the two or three best polymers and the weakest one for nucleation as well: a nucleation zone by cooling–reheating DSC and a nuclei count after one cycle take little material trasi2012a P6, P7. Store the candidate at its intended temperature and check whether that temperature lies inside the drug's nucleation zone; HPMCAS lowered the zone and the count for acetaminophen, but the authors note it "could potentially" give more nuclei than PVP over time at room temperature trasi2012a C47. For a drug that grows fast but nucleates poorly, such as celecoxib, the nucleation effect is the one to rank mosqueragiraldo2014 R15, C7.

History

Konno and Taylor (2006) measured dry nucleation rates of felodipine and Kestur and Taylor (2010) growth rates, in different samples konno2006 C13; kestur2010a R6. In 2012 Trasi and Taylor measured both in the same drug, for acetaminophen and for flutamide, and found them decoupled, citing the felodipine nucleation work as an earlier sign trasi2012a C53; trasi2012b C45. In 2013 Kestur and co-workers inferred from rate ratios that PVP protects felodipine mainly through nucleation kestur2013b C36. Later papers found the same decoupling in supersaturated solution: HPMC cut felodipine nucleation more than a thousandfold but growth only about twofold, and PAA, the best growth inhibitor for supercooled acetaminophen, did little in solution alonzo2012 C46; trasi2015 C26, C28 (see D. Crystallization from solution F6 and F27); for nifedipine polymorph conversion the polymers acted most likely on nucleation raina2014b C32.

How it was shown

Trasi and Taylor (2012) chose flutamide because it has both a donor (amide NH) and an acceptor (amide C=O) and paired it at a 9:1 molar ratio with six simple polymers carrying only acceptors (PVP, PDMA, PIPA, PVAc), only donors (PVPH) or both (PAA), plus 5% w/w PVP, PVPVA, HPMCAS and HPMC trasi2012b P1, M3, M4, M5, M6, M7. They measured growth at 40–70 °C, IR of 1:1 molar films, NBO charges on the polymer carbonyl oxygen and the energy of flutamide as its amide rotates trasi2012b P2, P4, P9. The acetaminophen study used seven polymers at 10% w/w and three at a 10% molar ratio on a monomer basis, with IR of 30% w/w films trasi2012a P1, P2, P8. The celecoxib study included 10% PVP as the reference inhibitor mosqueragiraldo2014 P3, R1.b. Sato and Taylor (2015) took transmission IR of ATP films with 20–30% polymer and Xie and Taylor (2016) of celecoxib films from 9:1 to 1:9 drug:polymer; both measured melt growth sato2015 P5; xie2016b P2, P9.

What the data show

At 40 °C flutamide grew at about 1.4 µm/s alone and at about 0.13 (PVP), 0.23 (PDMA), 0.28 (PIPA), 0.73 (PVPH), 1.0 (PVAc) and 1.6 µm/s (PAA), in the same order at 50–70 °C trasi2012b R3. The flutamide NH band fell from 3315 to 3287 cm⁻¹ with PVP, 3289 with PDMA and 3293 with PIPA, rose slightly with PVAc (3322) and stayed with PAA (3315); the PVP carbonyl moved from 1679 to 1658 cm⁻¹ trasi2012b R14, R15. The computed carbonyl oxygen charges were −0.65 (PVP), −0.647 (PIPA), −0.64 (PDMA), −0.606 (PAA) and −0.598 (PVAc) trasi2012b R25. In the 5% w/w series growth at 70 °C was about 4.0 (PVP), 12 (PVPVA), 17.5 (HPMCAS) and 31 µm/s (HPMC) against about 78 µm/s trasi2012b R5. The lowest-energy flutamide conformer is planar with the carbonyl next to the ring C(5)–H, and turning the amide costs up to about 23 kJ/mol trasi2012b R23, R24. For acetaminophen PAA moved the drug NH/OH band up to 3320 cm⁻¹ and its own acid carbonyl from 1705 to 1714 cm⁻¹, PVP moved the NH/OH band to 3283 cm⁻¹ and its carbonyl from 1679 to 1657 cm⁻¹, and HPMC, HPMCAS and E100 gave no noticeable shifts trasi2012a R16, R17, R20. At 10% molar the three polymers grew acetaminophen 5–17 times slower than the drug and the two PAA grades no longer differed trasi2012a R8, C16. With celecoxib, PVP moved the lower NH band from 3269 to 3192 cm⁻¹ and its carbonyl from 1681 to 1668 cm⁻¹ mosqueragiraldo2014 R6, R7, C21.

Later papers tested the reading in three ways. In the chiral compound ATP, HPMCP gave new NH and carbonyl bands identical for both enantiomers and slowed them alike, while HPMCAS removed the free NH2 band more for one enantiomer, which then grew about twice as slowly (see F27) sato2015 R12, R13, R14, R5. In celecoxib with 10% polymer at 80 °C, growth was about 0.00003 µm/s with PVP K29/32, 0.00007 with PVPVA and 0.00031 with HPMCAS against 0.010 alone, in the order of the NH shifts (3196, about 3198 and 3237 cm⁻¹ from 3268) xie2016b R21, R3, R4, R5. In indomethacin at 80–100 °C, PVP, the polymer with the strongest bond to the drug in the 2016 fluorescence study, slowed growth most and HPMCAS least, which the authors attribute "perhaps" to weaker interaction; the interactions were cited, not measured tian2017 R12, C32, R16, X13. PHB, with no band shifts, slowed no compound by more than about 2.4-fold (synthesis: 1/0.42) and sped up three (see F28) sato2016 R9, R11.

Figure

Each decade is a tenfold change; PVP (green) lies lowest, PAA (orange) sits on top of pure flutamide trasi2012b R3.

The three polymers that lower the NH frequency most are the three best growth inhibitors of Fig. 2 trasi2012b R14, C39.

Why

The authors' account, with their hedges: polymer carbonyls act, "it appears", as competitive acceptors for the flutamide NH, giving a drug–polymer bond more favourable than the drug–drug one; the NH shifts and the carbonyl charges "strongly support" the view that specific interactions are "extremely important" for growth inhibition trasi2012b C25, C40. PVAc's carbonyl, bonded to an oxygen, carries less charge "essentially", and PVPVA was "presumably" weaker than PVP because vinyl acetate interacts weakly; the cellulose polymers "presumably" have less basic acceptors trasi2012b C27, C41. Flutamide's carbonyl keeps part of its charge in the intramolecular CH···O contact, "highly likely" also in the amorphous state, which makes it "much less favorable" as an acceptor; PAA self-associates in dimers and "very likely" prefers itself trasi2012b C35, C37. For acetaminophen interactions are expected between the best donor and the best acceptor: the drug phenol to the pyrrolidone carbonyl, PAA's acid OH to the drug's amide carbonyl, and the polymers with weak groups do little trasi2012a C32, C28, C27. Equal inhibition per hydrogen-bonding group "further supported" the role of the interaction trasi2012a C33. The authors rule out Tg as the cause, since the Tg rises were small and similar for strong and weak inhibitors (see F5), and the small melting-point changes left the degree of supercooling nearly the same for all polymers trasi2012a C18, C19; trasi2012b C33.

Conditions and limits

Growth was measured in melt films at 9:1 molar or 10% w/w, IR on spin-coated films at 1:1 molar or 30% w/w, so the IR ranks interaction strength but does not show the state of the growth samples trasi2012b X9; trasi2012a X15. The charge correlation rests on four carbonyl polymers; PIPA and PDMA differ by 0.007 in charge and grow alike trasi2012b X8. The intramolecular bond was computed for one molecule in vacuum and was not observed in the glass trasi2012b C35, R23. Growth was measured far above Tg; storage stability was not tested in these papers trasi2012b A1; trasi2012a A1. In the flutamide series every polymer was used at the same drug:repeat-unit molar ratio, so dilution was the same for all trasi2012b C32; for nilutamide itself the authors write that dilution plus a higher Tg "can be explained" as the cause of its growth effect, on a corrected curve that is not shown trasi2012b C42, X11.

Counterexamples and conflicts

PVPh slowed acetaminophen growth about as much as PVP although its IR spectrum changed little trasi2012a X6. PAA crystallized flutamide partly on cooling and appears immiscible at higher PAA content, so its failure may reflect poor mixing rather than the intramolecular bond trasi2012b X12, R4, C38. For curcumin, whose keto–enol group is held intramolecularly, PVP gave the largest IR shifts of five polymers yet was the poorest inhibitor in storage, so a large IR shift did not predict storage stability there (see F24) wegiel2014 X7. Outside this set, PVP slowed five donor drugs by a similar 5–10-fold factor, so its effect did not follow the drug's own growth rate trasi2014a C10 (see Crystallization tendency F19). In indomethacin above about 100 °C the polymer order changed with polymorph and temperature, HPMC overtaking PVP for α, so one interaction order does not fix the growth order at every temperature (see F29) tian2017 C15, R12.

Transfer check

List the drug's donors and acceptors and check whether any of them is held in an intramolecular bond in the crystal structure or a conformer calculation; pick a polymer whose strongest group complements a free group of the drug (an acceptor polymer for a free NH or OH donor, a donor polymer for a free carbonyl) trasi2012b C51; trasi2012a C32. Confirm by IR of a dispersion at the intended composition that the drug band shifts trasi2012b X9. Then measure melt growth at 5–10% polymer at three temperatures trasi2012b P2. Do not rely on Tg rises, which were similar for the best and worst inhibitors trasi2012a C18.

History

Kestur and Taylor (2010) matched the IR order of felodipine–polymer bonding with the order of growth inhibition, and Kestur and co-workers (2011) found that an intramolecular NH···O bond in nimesulide flattened the differences between polymers (see Miscibility and drug-polymer interactions F12 and F14) kestur2010a C26; kestur2011 C27. In 2012 Trasi and Taylor tested the idea with polymers chosen for donor or acceptor groups only, for flutamide and acetaminophen, and showed that the best polymer changed with the drug's free group trasi2012b C8; trasi2012a C31. In 2014 PVP served as the hydrogen-bonding reference inhibitor for celecoxib mosqueragiraldo2014 C27, and Wegiel and co-workers carried the intramolecular-bond argument to curcumin storage (see F24) wegiel2014 C35. In solution the same donor polymer failed, which the authors relate to its ionization trasi2015 C27. Sato and Taylor (2015) used two enantiomers to hold everything but the interaction constant and found grade-specific inhibition by HPMCAS (see F27) sato2015 C27; Xie and Taylor (2016) and Tian and co-workers (2017) added celecoxib and indomethacin series xie2016b C36; tian2017 C28; the type line was extended and the title is unchanged.

How it was shown

Kestur and Taylor (2013) melt-quenched felodipine films 5–10 µm thick and grew form I either at a free surface (top cover glass removed) or in the bulk from seeds placed against the supercooled liquid; form II, which nucleates in unseeded films between two cover glasses, was grown in the bulk kestur2013a P1, P2, P3. Growth rates were measured by polarized light microscopy on a hot stage or in ovens, form I from 18 to 100 °C and form II from 40 °C, with form II data at 70 °C and above taken from earlier papers kestur2013a P6, X4. A 3% PVP K-29/32 dispersion was made by rotary evaporation and cryomill dilution, and the polymer effect was expressed as the ratio of the bulk rate without polymer to that with polymer (in words, Eq. E2) kestur2013a P4, E2, R15. Raman spectra and melting points identified the polymorphs kestur2013a P7. Tian and co-workers (2017) compared three indomethacin polymorphs with 10% of three polymers tian2017 A1, A2.

What the data show

Bulk form I grew at about 5.0 × 10⁻⁹ m/s at 60 °C, form II at about 2.8 × 10⁻¹⁰ m/s; at 40 °C the values were about 2.6 × 10⁻⁹ and 5.5 × 10⁻¹³ m/s kestur2013a R8.g, R8.c. The authors summarize the gap as "1-2 orders of magnitude"; the plotted means give about 4700-fold at 40 °C, about 87-fold at 50 °C, about 15–18-fold at 60–70 °C and only 7–9-fold at 80–100 °C kestur2013a C4, X5. Form II fell smoothly by about five orders of magnitude from 105 to 40 °C kestur2013a R11, R8.n, R8.c. Form I instead rose on cooling from 50 to 45 °C, in the bulk from about 1.7 × 10⁻⁹ to 4.2 × 10⁻⁹ m/s kestur2013a R8.e, R8.d, R10. With 3% PVP the ratios at 60, 70, 80, 90, 95 and 100 °C were about 15.8, 17.5, 13.2, 6.8, 3.9 and 3.5 for form I and 18.7, 15.4, 12.5, 8.2, 7.2 and 6.5 for form II kestur2013a R15. With PVP, form I still jumped near Tg, now between 45 and 40 °C (bulk about 6.8 × 10⁻¹¹ to 3.2 × 10⁻¹⁰ m/s), about 5 °C lower and larger than without polymer kestur2013a R14, R13, X14. For the jump the authors estimate about 500 molecular layers added per structural relaxation time at 45 °C (relaxation time taken as 100 s, molecular radius 9 Å), against 0.1–1 for diffusion-controlled growth and 1700–2400 for the glass-crystal mode of ROY; the card recomputes about 290 from the printed inputs, still in the hundreds kestur2013a R18, R16, X1, X2.

Figure

Black squares are form I, red circles form II; both fall from about 16–19 at 60 °C to 3.5–6.5 at 100 °C, close at 60–80 °C and apart at 90–100 °C kestur2013a R15, X6.

The form I points rise from 50 to 45 °C next to the Tg arrow while the form II curve (triangles) falls smoothly kestur2013a R8, R10.

Why

The authors' account, with their hedges: polymorphs present different crystal interfaces but grow from the same liquid, so equal inhibition of both points to PVP changing the amorphous phase, the PVP–felodipine environment being similar for both ("suggest"; abstract: "most likely"); in growth from solution, by contrast, adsorption of trace polymer to the crystal faces is the accepted mechanism kestur2013a C1, C6, C28. The form I jump is "most likely" the glass-crystal or diffusionless mode, too fast for diffusion, and about 500 layers per relaxation time is "supporting the supposition"; whether a polymorph shows the mode "is thought to be related" to its density and packing, form I being the denser (1.45 against 1.42) kestur2013a C3, C23, C24, M1, M2. Because PVP lowers both rates by about the same factor, faster-growing crystals "will require more polymer" to bring growth to negligible values ("suggesting") kestur2013a C26. The authors add that extrapolating high-temperature growth to near or below Tg assumes a smooth decrease, which the form I jump breaks kestur2013a C7. (synthesis: this is the polymorph counterpart of the cross-drug result that PVP slows fast and slow growers by a similar factor; in both, the inhibition is multiplicative, so the absolute rate left over follows the uninhibited rate.)

Conditions and limits

One drug, one polymer at about 3% (basis not stated), bulk ratios at 60–100 °C only, rates from plot readings without stated replicates kestur2013a X9, R15, X13. The measured front changes habit with temperature (compact spherulites below 40 °C, fiber layers at 50–70 °C, faceted crystals higher), so rates at different temperatures do not describe the same front kestur2013a X12. The relaxation time for the layer count is assumed, and the quoted 2.6 × 10⁻⁹ m/s matches the 40 °C rather than the 45 °C point kestur2013a X1. Form II with PVP below 60 °C, where the jump question matters, was not measured kestur2013a X11.

Counterexamples and conflicts

At 90–100 °C PVP slowed form II about 1.2–1.9 times more than form I and the error bars separate, so "the same extent" holds within about a factor of two kestur2013a X6. Dividing the plotted rates of Figs. 2 and 3 does not reproduce the form I ratios at 60–80 °C (about 13, 29 and 22 against 15.8, 17.5 and 13.2), so the evidence for equality is weakest exactly where it is claimed best kestur2013a X7. In felodipine powders of the same year PVP slowed surface growth only about twofold but bulk growth sevenfold, so the inhibition factor also depends on where the crystal grows (see F21) kestur2013b R12. For indomethacin each of PVP, HPMC and HPMCAS slowed the δ, α and γ polymorphs by very different factors, up to about 2900 against 2.9 at 120 °C with HPMC, which the authors read against the liquid-phase explanation (see F29) tian2017 R12, C39. At 80 °C, well below the temperatures of fastest growth as in the felodipine comparison, the indomethacin ratios of the three forms were within a factor of about 2.3 of each other (5.9, 3.5 and 4.7 with HPMC; 25, 16 and 11 with PVP) tian2017 R12.a.

Transfer check

Identify which polymorph actually forms in the product (powders of felodipine gave form I, the faster grower) and measure the polymer's effect on that form kestur2013b R10; kestur2013a C27. Measure growth at two temperatures below about Tg + 15 °C as well as above, because a jump near Tg can make high-temperature data underpredict low-temperature growth kestur2013a C7, C18. If the stable form grows ten times faster than the metastable one, assume it needs proportionally more polymer and test a higher polymer level kestur2013a C26.

History

Kestur and Taylor (2010) measured PVP inhibition of form II only (see F3) kestur2010b R2. In 2013 they compared both polymorphs, found the glass-crystal jump for form I, and placed the PVP effect in the liquid kestur2013a C3, C6; the particle-size study of the same year explained why powders crystallize to form I (see F20, F21) kestur2013b C26. A year later the comparison across five drugs gave the same multiplicative picture between drugs (see Crystallization tendency F19) trasi2014a C10, C11. Tian and co-workers (2017) found polymorph-specific factors for indomethacin and argued against the liquid-phase reading (see F29) tian2017 C38; the title is unchanged.

How it was shown

Wegiel and co-workers (2014) stored 75:25 curcumin dispersions with five polymers at six conditions and identified the crystals by PXRD reflections (form I at 8.9, 17.4, 23.5 and 25.1° 2θ; form II at 13.9, 26.0 and 27.1° 2θ) wegiel2014 P5, R15, R17. Trasi and Taylor (2012) identified flutamide crystals grown between quartz cover slips by Raman against the known form, and acetaminophen events on DSC reheating by their melting points trasi2012b P3; trasi2012a P4, R10. Mosquera-Giraldo and co-workers (2014) took XRD of celecoxib crystals grown from the melt with each surfactant mosqueragiraldo2014 P6, R14. Kestur and co-workers (2013) noted which form grew in covered films, at free surfaces and in powders kestur2013a R1, R2; kestur2013b R10. Tian and co-workers (2017) identified the indomethacin forms by morphology, Raman, hot-stage melting and XRD tian2017 R2, R3, R4, R6.

What the data show

Neat amorphous curcumin crystallized as form I at every condition where it crystallized wegiel2014 R15, C23. After 1 day at 98% RH and 60 °C the cellulosic dispersions showed form II reflections and no 17.4° peak, E100 and PVP gave form I; the form II of the dispersions matched a polymorph reported in 2011 as metastable, with a higher dissolution rate and solubility than form I wegiel2014 R17, R20, C28. For flutamide all samples grew the same form (CSD WEZCOT) except 5% w/w HPMC, which grew another, identified by small Raman shifts, whose melting with onset 90 °C appeared in DSC; with PDMA a second small exotherm near 84 °C marked a delayed transition between polymorphs trasi2012b R6, R9, R8.d, C13. Melted acetaminophen first crystallized as form III, which on reheating turned to form II before melting at 156 °C; with polymers the transition was delayed, with E100 and HPMC an exotherm near 133–139 °C trasi2012a R2, R10, C20. Celecoxib grown with SLS showed a peak near 8° 2θ (form I) absent with TPGS and sucrose palmitate, which grew the starting form III mosqueragiraldo2014 R14, C38. In felodipine, confinement between two cover slips gave form II and a free surface always form I, and powders gave form I only kestur2013a R1, R2, R3; kestur2013b R10. A 10% HPMCAS felodipine dispersion stored 491 days at 93% RH gave a mixture with form II rumondor2009b R13.

In indomethacin the polymorph that formed depended on storage temperature, δ at 60 °C and α at 80 °C, γ only from seeds, and the polymers retarded nucleation so that a higher temperature was needed to start crystals tian2017 R1, R5, C7. The ATP enantiomers changed growing form between 90 and 100 °C, so all polymer comparisons were kept at or below 90 °C sato2015 R6, C9, X12. Amorphous tacrolimus in a marketed capsule crystallized as the monohydrate, which the authors relate to the larger effect of humidity than of temperature (see F33) trasi2017 R6, C26.

Figure

The right-hand box marks the form II peaks of the HPMC, HPMCAS and CMCAB traces, the left-hand box the form I peaks of E100 ("EUD"), PVP and crystalline curcumin wegiel2014 R17.

Only the SLS trace (c) has the extra peak near 8° 2θ of form I; sucrose palmitate (b) and TPGS (d) match celecoxib form III (a) mosqueragiraldo2014 R14.

Why

The authors give few mechanisms. For curcumin they state only that the cellulosic polymers "were able to influence" the polymorph wegiel2014 C28. For celecoxib, "one possible explanation" of the surfactant growth effects "could be" that different polymorphs grew, which XRD ruled out for TPGS and sucrose palmitate mosqueragiraldo2014 C37, C38. For felodipine, form I's faster surface growth, especially below Tg, and its tendency to nucleate at the surface "most likely" explain why powders give only form I kestur2013a C25. (synthesis: an additive can change the form either by changing which nucleus appears first or by slowing the growth or conversion of one form more than another; the HPMC–flutamide and cellulosic–curcumin cases do not separate the two.)

Conditions and limits

The curcumin form II reflections are read differently in text and figure, the same CCDC code is printed for both forms, and the CMCAB trace has a small peak near 17.2° 2θ wegiel2014 X3, X11, X12. The flutamide Raman spectra are not shown trasi2012b P3. The celecoxib forms were checked only for the binary surfactant samples, and the SLS sample, which contained some form I, may have given growth rates of two forms mosqueragiraldo2014 X9. All forms come from melt films or stress conditions of hours to days, except the felodipine HPMCAS mixture after 491 days rumondor2009b R13.

Counterexamples and conflicts

Most additive–drug pairs in these papers did not change the form: every flutamide sample but HPMC, the TPGS and sucrose palmitate celecoxib samples, and E100 and PVP with curcumin trasi2012b R6; mosqueragiraldo2014 C38; wegiel2014 R17. The 2014 curcumin authors call a stabilized metastable form a possible benefit for bioavailability but "clearly" not ideal from a regulatory view wegiel2014 C43.

Transfer check

At each stability pull identify the crystal form, not only the crystallinity, and compare it with the form of the neat amorphous drug under the same condition wegiel2014 R15, R17. If a polymer gives a metastable form, check whether it persists and how it dissolves before treating the result as a failure or a benefit wegiel2014 C43, R20. For growth-rate screens with additives, confirm the form of the growing crystal, since a change of form can masquerade as a change of rate mosqueragiraldo2014 C37, X9; kestur2013a C27.

History

The 2009 humidity study noted form II in an old HPMCAS felodipine sample without analysis rumondor2009b R13. In 2012 Trasi and Taylor recorded a polymer-specific polymorph for flutamide and delayed transitions for flutamide and acetaminophen trasi2012b C10, C13; trasi2012a C20. In 2013 the felodipine papers tied the form to the growth site kestur2013a C25; kestur2013b C26. In 2014 surfactants (SLS) and cellulosic polymers changed the form for celecoxib and curcumin mosqueragiraldo2014 C38; wegiel2014 C28. The neat-drug side, a metastable form often growing first from the supercooled liquid, is in Crystallization tendency F7 lu2016 C7. In 2015–2017 the enantiomer and indomethacin growth papers kept their polymer comparisons within one form per temperature range sato2015 X12; tian2017 R1, and the tacrolimus survey identified a hydrate as the form that grew in humid storage trasi2017 C26; the title is unchanged.

How it was shown

Kestur and Taylor (2013) grew form I in melt-quenched films with and without a top cover glass from 18 to 100 °C kestur2013a P1, P2, P3, P6. Kestur and co-workers (2013) ground amorphous felodipine and a 3% PVP dispersion under dry nitrogen, sieved two fractions (150–250 and 500–710 µm), stored them dry at 30 °C, followed crystallinity by Raman and PXRD calibration models, and measured film growth at the free surface and in the bulk at 30 °C kestur2013b P2, P3, P4, P5, P6, P7, P8. For the plateaus they treated each particle as a sphere of the sieve-midpoint radius crystallizing inward from the surface (in words, Eq. E1) kestur2013b E1, C29:

X = 1 - \left(\frac{r - d}{r}\right)^{3}

(synthesis: the shell-volume form of the stated words, with X the crystalline fraction, r the particle radius and d the depth; it reproduces 17.3 µm from 43.5% at r = 100 µm and 16.2% at r = 302.5 µm kestur2013b R16).

What the data show

Without polymer the two form I rates fell by 3–4 orders of magnitude from 100 to 18 °C, to about 5.4 × 10⁻¹⁰ m/s at the surface and 8.7 × 10⁻¹¹ m/s in the bulk; above 70 °C the two coincided kestur2013a R9, R8.a, C2. In the plotted means the surface/bulk factor is about 6 at 18, 30 and 50 °C but about 3.5–3.7 at 40–45 °C and about 1.5 at 70–80 °C kestur2013a X13. With 3% PVP surface crystals grew 1–1.5 orders of magnitude faster than bulk crystals kestur2013a R14. At 30 °C the rates were 2.7 × 10⁻⁹ (surface) and 0.47 × 10⁻⁹ m/s (bulk) without PVP and 1.45 × 10⁻⁹ and 6.37 × 10⁻¹¹ m/s with it kestur2013b R12. Pure amorphous powders began to crystallize within 3–6 h and finished within 72–120 h, the smaller fraction first (Raman about 88 against 53% at 48 h) kestur2013b R5, R3. With PVP both fractions stalled after 50–60 days, at about 45–50% and 15–20% kestur2013b R8, C27. The PVP powder kinetics were biphasic, the early slope 13–14 times the late one, which the authors compare with the surface/bulk ratio of about 23 kestur2013b R14, C33.

Figure

Read across the rows: PVP halves the surface rate but cuts the bulk rate about sevenfold, so the surface/bulk ratio rises from 5.7 to 22.8 kestur2013b R12.

The 200 µm powder (squares) stalls near 42–45%, the 605 µm powder (triangles) near 15–18%, both after about two months kestur2013b R6.

Why

The authors' account, with their hedges: faster surface growth "may be attributed" to mobile molecules at a more liquid-like surface, as surface diffusion in indomethacin glass far exceeds bulk diffusion kestur2013a C16, R17. The particle-size effect of the pure drug "can be attributed in large part" to the surface-to-volume ratio together with the faster surface growth, and it is weaker than for indomethacin (about 100-fold surface enhancement) because felodipine's surface/bulk difference is smaller ("may be due to") kestur2013b C24, C25, R24. The PVP plateaus "may" come from a constant penetration depth of surface crystallization, and the fast first stage is "probably" surface growth kestur2013b C28, C33. Earlier work found that felodipine grows around PVP molecules without enriching PVP at the front, so a polymer-rich layer is not invoked kestur2013b C31, R27. Where surface growth exceeds bulk growth, predicting low-temperature stability from high-temperature data without it "may" overpredict stability kestur2013a C18.

Conditions and limits

One drug, form I, dry conditions, films 5–10 µm thick and two sieve fractions; particle surface area was not measured kestur2013a P1; kestur2013b P3, C24. Raman and PXRD gave different values for the same powders (about 46 against 31% for the small pure fraction at 24 h) and a smaller size gap by PXRD, which the authors put down to Raman's surface sensitivity kestur2013b X13, C10. The bulk rate with PVP corresponds to about 5.5 µm per day, about 300 µm over 56 days, far beyond the estimated 17.3 µm depth, while the crystallinity barely rose between 56 and 161 days, so the plateau is not explained by the film growth rates (card calculation) kestur2013b X8. The early/late slope ratio (powder % per day) and the surface/bulk ratio (film m/s) are different quantities kestur2013b X6.

Counterexamples and conflicts

Pure felodipine powders showed no biphasic kinetics, unlike indomethacin and griseofulvin, although the surface effect was present kestur2013b C21, R26. At 50 °C, above Tg, the PVP powder crystallized to about 85% rather than 45%, which the authors read as a smaller role of the surface there ("likely") (see F25) kestur2013b R9, C39. Across 14 compounds the surface order of growth rates did not always match the bulk order, so a bulk screen does not rank surface risk trasi2014a C8.

Transfer check

Measure growth with the top cover glass removed as well as sandwiched, at one temperature below Tg and one above kestur2013a P2, P3. Store at least two particle-size fractions of the dispersion, since milling and sieving change the free surface kestur2013b C2, C12. If the surface rate is much higher than the bulk rate with the polymer, consider protecting the surface: a polymer or gold coating suppressed surface growth in cited work, and atomic layer coatings slowed surface crystallization of ASD particles in later work kestur2013b C3; duong2022c C35.

History

Surface-enhanced growth was known for indomethacin, nifedipine and griseofulvin kestur2013a R17; kestur2013b C23. Rumondor and co-workers (2010) suggested surface growth of uncovered films as one reason felodipine grew faster than mobility predicts rumondor2010 C28. In 2013 the two felodipine papers measured the surface effect, its persistence with PVP and its consequence for powder particle size kestur2013a C15; kestur2013b C13, C20. Later work compared surface and bulk growth across 14 compounds (see Crystallization tendency F21) and coated ASD particles against surface crystallization (see ASD process F43) trasi2014a R4, R5; duong2022c C33.

How it was shown

Mosquera-Giraldo, Trasi and Taylor (2014) cryomilled celecoxib with 10 wt% surfactant, 10 wt% PVP K-12 or both, melted the powder between cover slips, quenched it, seeded the film at the edge and followed the growth front on a hot stage at 70–120 °C mosqueragiraldo2014 P1, P3, X1. A second series used 3, 5 and 7% PVP with and without 10% TPGS mosqueragiraldo2014 P2. Tg and the drug melting point were measured by DSC, interactions by FTIR of spin-coated films with 50% surfactant, growth was replotted against T - T_g, and the grown crystals were checked by XRD mosqueragiraldo2014 P4, P5, P7, P6. Sato and Taylor (2016) measured melt growth with PHB, and Yang and co-workers (2022) imaged surfactant-containing films at high humidity sato2016 P3; yang2022b P4.

What the data show

Pure celecoxib grew at about 0.002 µm/s at 70 °C and 12 µm/s at 120 °C mosqueragiraldo2014 R1.a. At 70 °C the binary systems grew at about 0.17 (TPGS), 0.014 (sucrose palmitate) and 0.0031 µm/s (SLS); at 120 °C at about 17, 7.4 and 13 µm/s mosqueragiraldo2014 R1.c, R1.d, R1.e. The ranking TPGS > sucrose palmitate > SLS holds at 70 and 80 °C only; from 100 °C SLS grew faster than sucrose palmitate, which fell to or below the drug at 110–120 °C mosqueragiraldo2014 X2. Tg was 55.34 °C for celecoxib, 44.81 with TPGS, 44.46 with sucrose palmitate, 55.53 with SLS and 62.40 °C with PVP mosqueragiraldo2014 R5.a, R5.b, R5.c, R5.e, R5.g. In the ternaries, SLS plus PVP grew about 2–20 times faster than PVP alone, sucrose palmitate plus PVP faster below 100 °C and slower above, and TPGS plus PVP slower at every temperature, about 0.00006 µm/s at 70 °C and 0.012 µm/s at 120 °C mosqueragiraldo2014 R2, C11. The TPGS–PVP ternary had the lowest Tg (44.17 °C) of the PVP systems and the lowest drug melting point (151.49 °C, 12 °C below the drug), and at 120 °C grew a large faceted plate instead of a fibrous front mosqueragiraldo2014 R5.f, C18, R13, C34. In the 3–7% PVP series TPGS lowered the 120 °C rate (3% PVP: about 1.9 against 3.0 µm/s) but raised the 80 °C rate (about 0.0089 against 0.0010 µm/s) mosqueragiraldo2014 R3, R4.

A low-Tg polymer acts like the miscible surfactants: 5% PHB (Tg 0 °C) lowered the Tg of griseofulvin, DPB and celecoxib by 5–12 °C and sped their growth up to tenfold, and at equal T − Tg most of the speed-up disappeared (see F28) sato2016 R7, R10, R11. In humid ritonavir–PVPVA films the lipophilic Spans entered the drug-rich phase and sped separation and coarsening, while SDS and Tween 80 delayed visible separation (see F32) yang2022b C7, C58.

Figure

TPGS (purple crosses) lies far above the drug (blue diamonds) at 70 °C and joins it near 120 °C; PVP (red squares) lies far below at every temperature mosqueragiraldo2014 R1.

Each TPGS curve is flatter than the matching PVP-only curve and crosses it on cooling; pairs are told apart by colour mosqueragiraldo2014 R3, R4, X11.

Why

The authors' account, with their hedges: the miscible surfactants "appear to behave" like small-molecule additives that depress Tg and so raise mobility, as water does, and after correction for Tg their main effect is "most likely" on celecoxib mobility (see F5 for how much of the spread T - T_g removes) mosqueragiraldo2014 C29, C30. SLS, immiscible, has "very little impact" alone; adding PVP "appears to increase" its miscibility, after which it speeds growth mosqueragiraldo2014 C31, C32. Near the melting point growth is under thermodynamic control, so the large melting-point depression with TPGS and PVP, meaning a smaller driving force, "could contribute" to its slow growth at high temperature mosqueragiraldo2014 C25, C33. In general a surfactant that mixes with the drug and plasticizes it "would be expected" to raise growth, but the outcome with polymer and surfactant together is "currently difficult" to predict mosqueragiraldo2014 C40, C41. (synthesis: the crossing curves fit two opposing contributions, a lower driving force that matters near Tm and a higher mobility that matters at low temperature; neither was measured.)

Conditions and limits

One drug, one polymer grade, 10% surfactant, growth at 70–120 °C, above the Tg of every system; storage at room temperature was not tested mosqueragiraldo2014 P3, R5.a, A1. The FTIR evidence for miscibility comes from 50% surfactant films made from solvent mosqueragiraldo2014 X8. Tg differences below about 1 °C are unresolved (for example TPGS–PVP 44.17 against TPGS 44.81 °C) mosqueragiraldo2014 X7. The faceted TPGS–PVP crystal may not be the form III whose growth the other systems report mosqueragiraldo2014 X9.

Counterexamples and conflicts

Tm depression does not order the ternaries: SLS–PVP is depressed more than PVP alone yet grows faster at every temperature mosqueragiraldo2014 X13. With 10% PVP no crossing appeared down to 70 °C, so "at least at higher temperatures" understates the TPGS–PVP advantage at that level mosqueragiraldo2014 X5, C13. Sucrose palmitate, though it lowers Tg like TPGS, grew 1.2–3.8 times slower than the drug at equal distance above Tg mosqueragiraldo2014 X4.

Transfer check

Treat any miscible low-Tg excipient (surfactant or plasticizer) as a possible growth accelerator: measure the dispersion Tg with and without it and, if Tg falls, measure melt growth at a temperature 15–25 °C above the new Tg mosqueragiraldo2014 C40, R12. Test the full formulation, not the binary pairs, since a polymer can cancel or reverse the surfactant effect mosqueragiraldo2014 C44. Do not extrapolate a high-temperature advantage of a surfactant–polymer combination to storage temperatures; find the crossing temperature first mosqueragiraldo2014 C15.

History

Before 2014 the effect of surfactants on solid-state crystallization of dispersions had, in the authors' words, "few, if any" studies mosqueragiraldo2014 C5. Later papers of the same group found TPGS and other surfactants plasticizing PVPVA and HPMCAS dispersions nguyen2023a C22; correasoto2022b C39; correasoto2023 C30, surfactants doubling ritonavir growth in solution (see D. Crystallization from solution F18), and a TPGS dispersion losing release after milling (see ASD process F40). In 2016 a low-Tg polymer was found to accelerate growth in the same way (see F28) sato2016 C29, and in 2022 surfactants were found to change water-induced separation according to the phase they enter (see F32) yang2022b C35; the title is unchanged.

How it was shown

Sato and Taylor (2015) asked whether a chiral polymer inhibits crystallization of two enantiomers to different extents, which would isolate the role of compound–polymer interactions, since enantiomers differ only in handedness sato2015 C5, C27. ATP(R) and ATP(S), a compound with one NH2 and one OH group and a Tg of 16–17 °C, were mixed with 5% w/w of achiral PVP K-12 or of the chiral polymers PHB, HPMCP (HP-55), HPMCAS-LF or HPMCAS-HF by rotary evaporation, melted about 5 °C above the melting point between coverslips and quenched to clear films sato2015 M1–M7, P1, P3, R1.a, R1.b. Spherulite size was followed on a hot stage at 40–120 °C and the growth rate taken as the slope of size against time, in triplicate; the racemate, made by cryomilling equal amounts of both enantiomers, was measured at 70–90 °C alone and with PVP or HPMCAS-HF sato2015 P2, P4, E1, A3. DSC gave Tg, the onset of recrystallization on heating (Tc) and melting; transmission IR was taken on films with 30% HPMCP, 30% HPMCAS-LF and 20% HPMCAS-HF, levels chosen to make the spectral changes readable; and 5% HPMCAS-HF glasses were held at 40 °C and followed by IR for 30 h sato2015 P6, P5, C15, P7.

What the data show

The pure enantiomers grew from about 0.022–0.032 µm/s at 40 °C to 21–34 µm/s at 90 °C, ATP(S) about 1.2–1.6 times faster than ATP(R) at every temperature, although the text calls the rates "almost exactly the same" sato2015 R5.a, R5.b, X2, C8. PHB changed growth little (about 4.8 against 5.4–6.7 µm/s at 70 °C, R and S together) sato2015 R5.c, R5.d. PVP slowed both enantiomers alike, about 10-fold at 50 °C and 3–4-fold at 70 °C, with overlapping R and S markers at 60, 70 and 90 °C sato2015 R5.e, R5.f. HPMCP slowed growth more, about 50 times at 60 °C and 20–25 times at 70 °C, and kept ATP(S) about 1.2–1.4 times faster, the pure-compound ratio sato2015 R5.g, R5.h, X2. HPMCAS-HF gave about 0.011, 0.094, 0.47 and 0.24 µm/s for ATP(S) and 0.0055, 0.049, 0.19 and 0.12 µm/s for ATP(R) at 50, 60, 70 and 80 °C; HPMCAS-LF gave about 0.015 and 0.058 µm/s for ATP(S) and 0.029 and 0.089 µm/s for ATP(R) at 60 and 70 °C sato2015 R5.i–l. At 60 °C the size–time lines show the split directly: with HPMCAS-LF ATP(R) reached about 33 units in 480 s against about 18 for ATP(S), with HPMCAS-HF ATP(S) about 45 against 23 for ATP(R) sato2015 R7, R8. Tg stayed at 14.3–18.4 °C in every sample, while Tc rose from 58–59 °C to 61.5–63.6 °C with PVP and to 77.2 (S) and 81.6 °C (R) with HPMCAS-HF sato2015 R1, C6, C7. At 40 °C the ATP(S)–HPMCAS-HF glass was fully crystalline by 15 h while ATP(R) was still partly amorphous at 30 h sato2015 R15, C20. In the IR, HPMCP gave a new NH band at 3253 cm⁻¹ and a new carbonyl band at 1699 cm⁻¹, identical for R and S; HPMCAS gave a new band at 3290 cm⁻¹, and the free amorphous NH2 band at 3390 cm⁻¹ nearly vanished with ATP(S) in LF but stayed stronger with ATP(S) in HF sato2015 R12, R13, R14. The racemate grew at about 0.64, 1.9 and 4.7 µm/s at 70, 80 and 90 °C, and at 0.28–2.65 µm/s with PVP or HPMCAS-HF sato2015 R9.c, R9.d, R9.e.

Figure

Filled markers are ATP(S), markers with a light centre ATP(R); the PVP and HPMCP pairs overlap, while the yellow (HPMCAS-HF) and dark-green (HPMCAS-LF) pairs split apart, in opposite directions sato2015 R5.

Both lines are straight through the origin; ATP(S) (squares) grows about twice as fast as ATP(R) (triangles), the reverse of the LF grade in Fig. 7 sato2015 R8, R7.

Why

The authors' account, with their hedges: the ATP amino group donates hydrogen bonds to polymer carbonyls, as the new NH bands show ("consistent with"; "strongly suggests") sato2015 C16, C24. In HPMCAS some ester carbonyls sit close to the chiral cellulose backbone, so R and S bind to different extents ("it is clear"), and the differential inhibition is "most likely due to" differences in the ability of each enantiomer to form hydrogen bonds with the polymer sato2015 C26, C29. In HPMCP the binding carbonyls lie next to the achiral phthalate group, which "may minimize" contact with chiral centres; PVP, being achiral, binds both alike ("it is suggested"); PHB "presumably" interacts little with either sato2015 C25, C30, C23. Why LF and HF favour opposite enantiomers is "not readily explainable", "most likely linked" to their acetate and succinoyl substitution, which was not measured sato2015 C28. Because everything except chirality is the same, the authors read the result as "strongly" supporting compound–polymer interactions as determinants of growth inhibition sato2015 C27. Not measured: the interactions at the 5% level used for growth, the IR of PVP and PHB films, and nucleation sato2015 X5, X6, C22. (synthesis: this is the cleanest test in the chapter of the hydrogen-bond reading of F18, since Tg, Tm and dilution are the same for both enantiomers; it supports that reading for HPMCAS, while the identical HPMCP spectra show that a chiral polymer need not discriminate.)

Conditions and limits

One non-drug model compound, 5% w/w polymer, dry melt films at 40–90 °C, all below the polymorph change of the pure enantiomers at 90–100 °C sato2015 A1, A3, R6, X12. The racemate was measured only at 70–90 °C and only with two polymers sato2015 A3. The factor of 50 quoted in the text for ATP(S) with HPMCAS-HF has no stated temperature; Fig. 6 gives 14–87 for ATP(S) and 28–105 for ATP(R) sato2015 X4. The absolute LF rates in Fig. 7 are about 2.5 times those of Fig. 6, although the ratio agrees sato2015 X3. DSC values are single runs without replicates sato2015 X8.

Counterexamples and conflicts

Chirality alone did not decide: HPMCP and PHB are chiral and did not discriminate sato2015 C10. In the library's only other chiral case, a supersaturated aqueous solution, HPMCAS-MF and HPMC delayed crystallization of racemic praziquantel at least 83 min but of R-praziquantel only 15.5–37 min, while without polymer both crystallized in about 5 min; the authors there name stereoselective binding as one possibility ("it is not unreasonable to speculate"), next to differences in crystal form and conformation polyzois2024 R4, R3.a, R3.g, R3.h, C28. The cited nitrendipine precedent found the enantiomer effect in nucleation and not in growth sato2015 C22.

Transfer check

For a single-enantiomer drug, screen polymers on the enantiomer itself, not on the racemate: here the racemate responded about twofold where the enantiomers responded 14–105-fold, so a racemate screen misranks the polymers sato2015 R9, X4. Test at least two grades of a cellulose ester such as HPMCAS (LF and HF reversed the preferred enantiomer here) and compare the growth ratio against the pure enantiomers' own ratio, not against 1 sato2015 R5, X2. Use IR of the drug's free donor band at the intended polymer level as a check on which enantiomer binds more sato2015 R13, R14, X5.

History

Miyazaki and co-workers (2011, outside the library) had reported chiral polymers changing the crystallization of nitrendipine enantiomers through nucleation sato2015 C22. Kestur and Taylor (2010) and Van Eerdenbrugh and Taylor (2011) had tied growth inhibition to compound–polymer interactions sato2015 C27 (see Miscibility and drug-polymer interactions F12 and F17). Sato and Taylor (2015) first showed in the library a grade-dependent, enantiomer-specific growth inhibition by HPMCAS sato2015 C11. Polyzois and co-workers (2024) cited it when praziquantel ASD polymers were found weaker for the single enantiomer than for the racemate polyzois2024 C28.

How it was shown

Sato and Taylor (2016) asked how a small amount of PHB, a biodegradable semicrystalline polyester considered for implants, changes crystal growth from the supercooled liquid of compounds with pure-compound Tg from −1 to 90 °C sato2016 C4, C5, M1–M7. Blends with 1–20% PHB (and 5% PVAc, Tg 40 °C, with griseofulvin) were made by cryomilling or rotary evaporation, melted between coverslips about 5 °C above the compound's melting point and quenched sato2016 P1, P2, P3, M8. Spherulite diameter was followed on a hot stage in 10 °C steps, n = 3, below the expected growth maximum of each compound; the rates are diameter slopes, twice a radial rate sato2016 P3, E2, E3, X8. DSC gave Tg of the mixtures and Tm of physical mixtures, IR checked for drug band shifts, and the rates were replotted against T − Tg and as a ratio with to without polymer sato2016 P5, P4, P6, P7, E4.

What the data show

For the high-Tg compounds: griseofulvin (Tg 90 °C) grew at about 0.015 µm/s alone and 0.15 µm/s with 5% PHB at 120 °C, 0.022 µm/s with 1% PHB, and with 20% PHB about 0.23 µm/s at 110 °C where the 5% blend grew at about 0.024 µm/s sato2016 R4.a, R4.b, R4.c, R4.d. Celecoxib (57 °C) went from about 0.025 to 0.16 (5%) and 0.19 µm/s (20%) at 90 °C, but only from 3.4 to 5.0 and 3.7 µm/s at 120 °C sato2016 R4.f, R4.g, R4.h. DPB (53 °C) went from about 0.135 to 0.72 (5%) and 1.4 µm/s (20%) at 80 °C and from about 27 to 41 µm/s (5%) at 120 °C sato2016 R4.i, R4.j, R4.l. The gain shrank to about 1.5-fold at the highest temperatures sato2016 X5. For the low-Tg compounds: flutamide (−1 °C) grew at about 0.031, 0.019 and 0.013 µm/s at 30 °C with 0, 5 and 10% PHB, ATP (17 °C) at about 0.27 and 0.20 µm/s at 50 °C with 0 and 5% PHB, and acetaminophen (21 °C) at about 0.36–13.8 against 0.34–12.8 µm/s at 70–100 °C sato2016 R3.a, R3.b, R3.c, R3.d, R3.e, R3.f, R3.g. The Fig. 12 ratios were about 0.42, 0.75, 0.96, 5.4, 6.5 and 10 for compounds of Tg −1, 17, 21, 53, 57 and 90 °C sato2016 R11. Mixture Tg fell from 90 to 85, 78 and 58 °C for griseofulvin with 1, 5 and 20% PHB, from 53 to 46, 43 and 35 °C for DPB with 5, 10 and 20%, and from 57 to 52 and 41 °C for celecoxib with 5 and 20%, while acetaminophen (21 → 22 °C), ATP (17 → 16 °C) and flutamide (−1 → −2 and 0 °C) did not change; 5% PVAc lowered griseofulvin only to 84 °C sato2016 R7, M1–M6, X4. Near equal T − Tg the rates converged: DPB about 4.2 µm/s (pure, 47 °C) against 3.8 (5% PHB, 44 °C) and 3.7 (10%, 47 °C), griseofulvin 0.74 (pure, 50 °C) against 0.95 (5%, 52 °C); with 20% PHB DPB grew at about 3.7 µm/s at 55 °C against about 16 µm/s for the pure compound at 57 °C, griseofulvin 0.52 at 62 °C against 3.4 at 60 °C, and griseofulvin with PVAc at 0.49 µm/s at 56 °C sato2016 R10. PVAc, with a Tg gap to griseofulvin (50 °C) close to that between PHB and celecoxib (57 °C), left griseofulvin growth at about the pure rate (0.13 against 0.13 µm/s at 130 °C) sato2016 R4.a, R4.e, M1, M2, M7, M8, C26.

Figure

From left to right the points are flutamide, ATP, acetaminophen, DPB, celecoxib and griseofulvin (matched by Tg); the three left points lie below the line at 1 (retardation), the three right points at 5–10 (acceleration); the axis is logarithmic sato2016 R11.

The 5–10% PHB series fall onto the pure-compound lines, while the 20% PHB squares and griseofulvin with PVAc lie below them sato2016 R10.

Why

The authors' account, with their hedges: all growth temperatures lie below the expected maximum near 0.94 Tm, so growth "is expected to be" limited by transport (see Crystallization tendency F2) sato2016 C15, E2. PHB left melting points and IR unchanged, so "it can be inferred" that the driving force is unchanged, and the polymers "most likely" act through the mobility of the crystallizing compound sato2016 C18, C19. In "broad agreement" with Yu and co-workers, who found only the lowest-Tg polymer, PEG, to speed nifedipine growth, the authors relate acceleration to the mobility difference between polymer and host, greatest for the high-Tg compounds ("suggesting that the polymer increased the molecular mobility") sato2016 C23, C24, C28. When the difference is small, the PHB chains "most likely" impede transport to the crystal and slow growth sato2016 C25. PVAc shows that Tg is "only a crude indicator" of mobility sato2016 C26. No viscosity, relaxation or diffusion was measured sato2016 X7. (synthesis: the compounds that PHB accelerated are exactly those whose Tg it lowered, and at 5–10% PHB the speed-up largely disappears against T − Tg, so plasticization of the drug-rich melt by a low-Tg, non-interacting polymer accounts for most of it sato2016 X6; this is the same pattern as miscible surfactants and water in F22 and F9, and the limit of F5 that large Tg drops do change growth even though small ones do not rank polymers.)

Conditions and limits

Drug-rich mixtures (1–20% polymer for growth) above Tg, in dry melt films; the polymer-rich compositions of a PHB implant and storage below Tg were not studied sato2016 A2, A4. Most films were melted below the PHB melting point (176 °C), so mixing of PHB in the acetaminophen and ATP films rests on one Tg each sato2016 X9. The ratio plot gives neither PHB level nor temperature per point; the flutamide point matches 10% rather than 5% PHB sato2016 X3. The authors' "unchanged Tg at low amounts" fits only the three low-Tg compounds sato2016 X4.

Counterexamples and conflicts

PVAc lowered griseofulvin's Tg only 6 °C and did not speed its growth, although its Tg gap to the drug is similar to that of PHB to celecoxib sato2016 R7.d, C26. 20% PHB grew more slowly than its Tg predicts, so the polymer chain also slows growth once enough of it is present sato2016 C21. Usually polymers at a few percent slow growth, often without changing Tg (see F3, F5 and F18) sato2016 C16; the polymers that did so in this chapter have Tg above the drug's or bind it, which PHB does not (synthesis: from the Tg data and IR in F5 and F18).

Transfer check

Before using a low-Tg polymer, surfactant or plasticizer with a high-Tg drug, measure the mixture Tg; a fall of 5 °C or more with 5% additive, as for the three accelerated compounds here, warns of faster growth sato2016 R7, R11. Then measure melt growth at two temperatures and compare at equal T − Tg: if the curves collapse, the effect is plasticization and will also apply at storage temperature; if growth stays below the Tg prediction, as with 20% PHB, the chain also impedes transport sato2016 R10, C21. An additive whose Tg is close to the drug's (here acetaminophen, ATP, flutamide) is not expected to accelerate sato2016 R11.

History

Powell and co-workers (2013, outside the library) had found PEG, the lowest-Tg polymer tested, to accelerate nifedipine growth sato2016 C23. In 2014 miscible low-Tg surfactants were found to speed celecoxib growth (see F22) mosqueragiraldo2014 C9, C30. Sato and Taylor (2016) showed the same with a polymer and tied the direction to the compound's Tg sato2016 C27, C29; the neat-drug chapter lists the result among additives that accelerate growth (see Crystallization tendency F19).