PEG dispersions

How it was shown

Baird and Taylor (2011) made fused dispersions of chlorpropamide, aceclofenac, loratadine or haloperidol with PEG 3350: 500 mg physical mixtures were melted on a hot plate, cooled slowly, ground and stored over P2O5 for 24 h, in 10 wt% steps from 0 to 100 wt% drug, with extra points at 5 wt% (loratadine) and 15 and 25 wt% (haloperidol) baird2011 P1, A2. They read the offset of melting in hermetic DSC pans at 1 °C/min and, separately, the temperature at which the last solid particle melted by hot-stage microscopy at 1 °C/min, 2–3 mg per sample baird2011 P2, P3. Zhu and co-workers (2011) used the hot-stage method at 1 °C/min on benzocaine, ibuprofen, nilutamide and ketoconazole with PEG 3350, mixtures stored a few days and lightly ground, with DSC only as a check for ibuprofen; their chlorpropamide and haloperidol diagrams were taken from Baird and Taylor zhu2011 P3, P4, R19, R5. Zhu and co-workers (2012) added fenofibrate, by the same two methods, in the supporting information zhu2012 P9, P10, R24.

What the data show

The measured eutectic or monotectic composition is given for each system in the table below. Near the eutectic a single endotherm appeared below both pure melting points, and away from it two endotherms, a low-temperature one from the eutectic or monotectic composition and a high-temperature one from excess PEG or drug baird2011 R4, C13. In haloperidol–PEG the PEG melting stayed near 59–61 °C and the liquidus rose steeply from 10 wt%; in chlorpropamide–PEG the liquidus stayed at about 55–60 °C from 10 to 40 wt% drug and in aceclofenac–PEG from 10 to 30 wt% baird2011 R5, R6, R8. The chlorpropamide value of 35 wt% was not a tested composition, since the lowest liquidus lay at 40 wt% baird2011 X9, R5.e. For benzocaine, ibuprofen and nilutamide the eutectic temperatures were 48.9, 49.6 and 58.4 °C, and chlorpropamide's was 55.0 °C zhu2011 R5.b, R5.d, R5.e, R5.c. Ketoconazole was monotectic, with its liquidus at about 101 °C already at 5 wt% drug zhu2011 R4.b, R5.f. Fenofibrate's liquidus fell from about 82.4 °C for the pure drug to about 58.0 °C at 15 and 20 wt% and rose again toward about 60.5 °C for PEG, so the 20/80 dispersion lay at the eutectic composition zhu2012 R24.o, R24.c, R24.d, R24.a, C40. The low-temperature endotherm of the chlorpropamide and aceclofenac dispersions drifted to about 39–49 °C in the drug-rich traces, which a simple eutectic would not show, so their eutectic temperatures are not established by those traces baird2011 R4, X8.

Figure

Panels (a) chlorpropamide and (b) aceclofenac lie far below the predicted line between 20 and 60 wt% drug, while (c) loratadine and (d) haloperidol lie on or above it; squares are DSC offsets and triangles hot-stage values baird2011 R5, R6, R7, R8.

Why

Baird and Taylor define a eutectic dispersion as one in which the drug dissolves appreciably in the molten polymer and the eutectic point is the maximum drug solubility in the molten polymer at the eutectic temperature, against a monotectic system in which the drug barely dissolves and hardly lowers the PEG melting point baird2011 C1. They write that a drug's solubility in molten polymer depends mainly on the crystal lattice energy (melting point and heat of fusion) and on the strength of specific drug–polymer interactions relative to the pure liquids, so that weak lattices or strong interactions "would be expected" to give high eutectic compositions baird2011 C10. Zhu and co-workers repeat this, as described "previously", for their drugs zhu2011 C4. For the proposed interaction they point to hydrogen-bond donors: benzocaine, chlorpropamide and ibuprofen carry donors that can interact with the PEG ether oxygen and have moderate melting points, whereas nilutamide's lower solubility "can be explained by" its higher melting point zhu2011 C9, C12. Baird and Taylor consider it "likely" that the drug donates a hydrogen bond to the abundant ether oxygens of PEG (about 75 per chain against 2 hydroxyl end groups) baird2011 C30, R12. (synthesis: nilutamide and haloperidol show that a donor group is not sufficient: nilutamide, which the authors count among the interacting drugs, reaches only 10 wt%, and haloperidol, which has an OH group, is monotectic with no carbonyl shift zhu2011 R5.e, C12; baird2011 M4, R10, C32; the lattice term in F2 is the other half of the explanation.) A lower-melting drug without donors reached an intermediate eutectic: fenofibrate, with no sign of hydrogen bonding to PEG, has a eutectic at 15–20 wt% zhu2012 R24, R25, C49.

Conditions and limits

PEG 3350 only; fused and slowly cooled dispersions; eutectic compositions on a 5–10 wt% grid baird2011 X9. DSC and hot-stage liquidus values differ by up to about 16 °C at some compositions (haloperidol at 20 wt%: about 128 against 144 °C by DSC and hot stage) baird2011 X7, R8.d. The crystal forms of the drugs as received and after fusion are not stated, and pure chlorpropamide shows an extra event at about 121–122 °C before melting baird2011 X14, R4. Different PEG grades, cooling histories and drug loadings above the eutectic were not studied. The neat PEG 3350 melts at about 60.5 °C in the phase diagrams, against 55 °C used as a model input baird2011 R4, M5, X10.

Counterexamples and conflicts

None found in the library for the existence of the spread. Donor groups do not predict the eutectic alone (see the synthesis note above). The text of Zhu and co-workers says DSC confirmed the benzocaine, nilutamide and ketoconazole diagrams, but only the ibuprofen thermograms are shown zhu2011 C5, R19.

Transfer check

For a new drug in PEG 3350, record the drug's melting point and heat of fusion and list its hydrogen-bond donor groups. Then measure the liquidus from 0 to 100 wt% by hot-stage microscopy at 1 °C/min on ground, fused samples, with steps finer than 10 wt% near the expected eutectic and with DSC as a second method baird2011 P1, P2, P3. A drug that melts above about 100 °C and has no donors should be expected to give only a few wt% (F2). Treat the eutectic composition as an upper limit at the eutectic temperature of roughly 50–58 °C, not as a storage-temperature solubility (F4).

History

Baird and Taylor (2011) measured the first four phase diagrams and tied the eutectic to lattice energy and interaction baird2011 C10, C36. Zhu and co-workers (2011) added benzocaine, ibuprofen, nilutamide and ketoconazole and took the other two from Baird and Taylor zhu2011 R5, C3. Zhu and co-workers (2012) added fenofibrate and concluded that haloperidol–PEG is the only monotectic system among their four zhu2012 C40.

How it was shown

The model of Lacoulonche and co-workers applies Flory–Huggins lattice theory to the liquidus curves of drug and polymer: one branch for crystalline drug and one for crystalline PEG, each depending on the heat of fusion, melting point and molar volume of that component and on an interaction energy \Delta W_{d,p} baird2011 E1, E2, C5. Baird and Taylor set \Delta W_{d,p} = 0 (athermal mixing), took heats of fusion, melting points and molar volumes from their Table 1, and called the drug volume fraction at which the two branches give the same temperature the eutectic, converted to weight percent baird2011 P6, E4, C6. To map the dependence on melting point they ran the model for a hypothetical drug (molar volume 250 cm³/mol, heat of fusion 35 kJ/mol, molecular weight 325 g/mol) with melting points from 80 to 350 °C and \Delta W_{d,p} of 0, −2.5, −5.0 and −10.0 kJ/mol baird2011 P7, R2. They then back-calculated the interaction energy that reproduces each observed eutectic and converted it to a Flory–Huggins parameter with \chi_{dp} = \Delta W_{d,p}/(R\,T_{eut,th}) baird2011 P8, P10, E3, E8.

What the data show

The predicted and observed eutectics, the back-calculated energies and the parameters are in the first table. The predicted lines start near 55 °C for PEG, dip to minima of about 52.5–54.5 °C and rise to each drug's melting point baird2011 R5, R6, R7, R8. At 40 wt% aceclofenac the measured liquidus was about 86–88 °C against about 135.5 °C predicted, and at 40 wt% chlorpropamide about 55–56 °C against about 98 °C baird2011 R6.e, R5.e. For loratadine and haloperidol the authors call the eutectic "comparable" and the abstract "accurately predicted", but the measured liquidus at intermediate compositions lay above the model line (loratadine at 20 wt%: about 103 °C by hot stage against about 90.5 °C predicted), so the agreement is one of eutectic composition only baird2011 C14, X6, R7.d. The model map gives, at a drug melting point of 100 °C, about 10.0, 18.9, 27.9 and 41.2 wt% for 0, −2.5, −5.0 and −10.0 kJ/mol, and at 150 °C about 2.7, 6.1, 11.7 and 24.9 wt% baird2011 R2.c, R2.h. The authors divide the melting-point axis into four regions at the 10 wt% crossings: region I below 100 °C (high eutectics even without interaction), region II 100–160 °C (appreciable loading only with significant interaction), region III 160–240 °C (very strong interaction needed) and region IV above 240 °C (even very strong interactions "are not enough") baird2011 R3, C12. All four real drugs melt in region II, between 126.7 and 151.7 °C baird2011 C19, M1, M2, M3, M4.

Eutectic composition in wt% drug, read from the model map for an average drug in PEG 3350 baird2011 R2.

In the recomputation of the card, the chlorpropamide prediction comes out at 9.45 wt% instead of 12.8, while the other three predictions and the map reproduce baird2011 E1, X2. The printed expression for the interaction energy is not the rearrangement of the two liquidus equations, and solving them gives −7.35, −14.5, +0.71 and −1.60 kJ/mol for the four drugs instead of −6.85, −18.2, 0.34 and −0.11 baird2011 E3, X3. The ranking aceclofenac, chlorpropamide, haloperidol, loratadine holds in every variant, so only signs and ranking are robust baird2011 X3.

Figure

Diamonds are 0, squares −2.5, triangles −5.0 and circles −10.0 kJ/mol; the horizontal line marks 10 wt% and the vertical lines the region borders at about 100, 160 and 240 °C. The higher the drug melting point, the stronger the interaction required for a eutectic composition above 10 wt% baird2011 R2, R3.

Why

The authors' account: with \Delta W_{d,p} = 0 only the lattice term is left, so the melting point and heat of fusion alone set the prediction: haloperidol and aceclofenac melt at nearly the same temperature (150.9 and 151.7 °C) but haloperidol's heat of fusion is higher (53.3 against 42.2 kJ/mol), so its predicted eutectic is lower, and both have higher lattice energies than loratadine and chlorpropamide ("it is clear") baird2011 C9, M1, M2, M3, M4. Molecular weight and molar volume are said to play an insignificant role, taken from earlier work and not tested here baird2011 C8. A negative back-calculated energy is read as an extra driving force to mix with molten PEG from a favourable enthalpy of mixing ("imply") baird2011 C18. In the conclusions the accuracy of the model is "delineated" by the presence or absence of specific interactions, "for the systems investigated", and the eutectic prediction is "interdependent" on the lattice energy and the nature of the interactions baird2011 C36, C37. The evidence for the interaction itself is in F3. The assumptions behind \chi_{dp} < 0 meaning exothermic mixing are on the Flory-Huggins interaction parameter page.

(synthesis: an independent check from another paper supports the athermal half of the model. Fenofibrate, whose pure-drug liquidus point is about 82.4 °C and whose molten mixture with PEG shows no spectral sign of interaction, has an observed eutectic of 15–20 wt%; the map gives about 18.8 wt% at 80 °C with no interaction zhu2012 R24.o, R24.c, R24.d, R25, C49; baird2011 R2.a. The map was drawn for a heat of fusion of 35 kJ/mol, and fenofibrate's is 33.03 kJ/mol baird2010 M38, so the agreement is approximate. In the same way, ibuprofen, melting at about 77.5 °C, reaches 40 wt%, about twice the athermal map value at 80 °C (synthesis: 40/18.8 = 2.1), and the authors list ibuprofen among the interacting drugs zhu2011 R2.t, R5.d, C9.)

Conditions and limits

Four real drugs, one polymer grade, all melting in region II; regions I, III and IV were not tested with real drugs, and the 200 °C condition for use of the model is stated differently in the abstract (compounds melting below 200 °C) and in the conclusions (above 200 °C), while the map has no boundary at 200 °C baird2011 R3, C12, C38, X5. Predictions and measurements are for the eutectic composition on a coarse grid baird2011 X9. Experimental eutectic temperatures were not reported baird2011 X8.

Counterexamples and conflicts

None found in the library for the existence of the upper limit of the eutectic composition under athermal mixing. The chlorpropamide value of 12.8 wt% does not follow from the printed inputs, and the gap to the observed 35 wt% remains under either value baird2011 X2. Loratadine is called close to ideal in PEG 400 although its measured solubility is about 0.42 of the ideal value (see F3) baird2011 C24, X13.

Transfer check

Compute the athermal eutectic from the drug's melting point, heat of fusion and molar volume with the two liquidus equations, or read it from the map by melting point. Measure the phase diagram. If the measured eutectic is more than about 2 times the prediction, treat a specific interaction as present and test it with the readings in F3. Do not apply the map to drugs melting below 100 °C or above 160 °C without checking, since no real drug outside 126.7–151.7 °C was measured in this paper (fenofibrate and ibuprofen in F1 add two low-melting cases from other papers) baird2011 R3, C12.

History

The model comes from Lacoulonche and co-workers (1997) and was tested for flurbiprofen and other organic compounds, with the cause of its errors "still debatable" baird2011 C5. Baird and Taylor (2011) tied the misses to specific interactions for four drugs baird2011 C36. Zhu and co-workers (2011) took over the view that the eutectic composition is set by the drug melting point and its ability to interact with PEG zhu2011 C4.

How it was shown

Four kinds of reading were used. (1) Interaction energy or parameter from the phase diagram. Baird and Taylor back-calculated an interaction energy \Delta W_{d,p} that closes the gap between the athermal prediction and the observed eutectic (F2) and divided it by R\,T_{eut,th} to give \chi_{dp} baird2011 P8, P10, E3, E8. Zhu and co-workers fitted the Flory–Huggins melting point depression equation to the drug liquidus at 50–80% drug, with the drug volume as lattice site, and normalized the result per PEG segment zhu2011 P5, E1, E2. (2) Activity coefficient in PEG 400. Excess drug was stirred for 24 h at 25 °C in liquid PEG 400 (Baird and Taylor; Marsac and co-workers, with 0.45 µm and 0.2 µm filters respectively), assayed by UV, and the activity coefficient was taken as the ideal solubility divided by the measured mole-fraction solubility, the ideal solubility coming from the melting point and heat of fusion baird2011 P5, P9, E7; marsac2008b P5, P9, E2. PEG 400 was chosen because activity coefficients in viscous, semicrystalline PEG 3350 are hard to measure baird2011 C21. (3) Infrared spectra of the melt. ATR-FTIR spectra of 25:75 drug–PEG 3350 melts, taken at 130, 160, 140 and 155 °C for chlorpropamide, aceclofenac, loratadine and haloperidol, were compared with a spectrum calculated as the weighted sum of the pure molten components; Zhu and co-workers did the same for 20/80 melts of benzocaine, ibuprofen, nilutamide and ketoconazole (temperature not stated) and, in 2012, fenofibrate baird2011 P4; zhu2011 P8; zhu2012 P11. (4) Solubility in supercooled PEG at 25 °C, extrapolated from the liquidus (F4) zhu2011 R5.

What the data show

The table gives the readings per drug. For chlorpropamide the carbonyl band sat at 1717 cm⁻¹ in the melt against 1663 cm⁻¹ in the additive spectrum, and for aceclofenac at 1744 against 1728 cm⁻¹; loratadine and haloperidol showed no or only nominal shifts baird2011 R10.a, R10.c, R10.e, R10.g, R10.h. In PEG 400 at 25 °C, chlorpropamide (194 mg/mL) and aceclofenac (233 mg/mL) dissolved above their ideal solubility, with activity coefficients 0.614 and 0.121, while loratadine (49 mg/mL) and haloperidol (14 mg/mL) gave 2.369 and 0.842 baird2011 R9. Benzocaine, ibuprofen and nilutamide showed carbonyl bands at about 1700, 1735 and 1733 cm⁻¹ against 1686, 1703 and 1721 cm⁻¹ in the calculated spectra, read from the figure, whereas ketoconazole's measured and calculated spectra were superimposed zhu2011 R20. The fenofibrate spectrum had the same peak positions as its calculated spectrum zhu2012 R25, C49. Marsac and co-workers measured ibuprofen at 0.28 and ketoconazole at 0.02 mole fraction in PEG 400, with activity coefficients 0.64 and 0.09 marsac2008b R14.d, R14.f. The Flory–Huggins parameters and the 25 °C solubilities are in the table.

Carbonyl bands marked ≈ were read from the figures of the papers. The two entries in the χ column for chlorpropamide and haloperidol are zhu2011 first, baird2011 second, from the same published phase diagrams.

Figure

Solid lines are the measured melts and dashed lines the additive spectra; the carbonyl band (left arrow) moves for (a) chlorpropamide and (b) aceclofenac but not for (c) loratadine or (d) haloperidol baird2011 R10. Table 3 of the same paper (figs/table03.png) lists the solubilities, activity coefficients and \chi_{dp} baird2011 R9.

Why

The authors' reading of the infrared result: the large carbonyl shifts of chlorpropamide and aceclofenac to higher wavenumber are "indicative" of fewer drug–drug hydrogen bonds because the drug now bonds to PEG; with about 75 ether oxygens against 2 hydroxyl end groups per chain, it is "likely" that the drug donates the hydrogen bond to an ether oxygen baird2011 C29, C30, R12. Haloperidol's absence of a carbonyl shift "may" be due to its tertiary nitrogen being the better acceptor baird2011 C32. For the activity coefficients, the reading holds only if the entropy of mixing is close to ideal: then a value below 1 "suggests" favourable drug–PEG interactions and a value near 1 supports their absence baird2011 C23, C24; Marsac and co-workers read their PEG 400 values below 1 the same way, as exothermic mixing, and the large values in water as endothermic mixing marsac2008b C14. A more negative \chi is "consistent with" a negative enthalpy of mixing zhu2011 C6. In the supporting information of Zhu and co-workers the carbonyl shifts are put down to changes in drug–drug hydrogen bonding caused by drug–polymer interactions, without a separate test zhu2011 C11. What was not measured: no PEG ether or hydroxyl band was analysed, so the bond to PEG is inferred baird2011 X15; the entropy of mixing was not measured baird2011 C23. The abstract of Baird and Taylor says the interactions were "confirmed" and "demonstrated" by infrared spectroscopy, which is stronger than the shifts support baird2011 C36, X15.

Conditions and limits

Nine drugs, PEG 3350 for the melts and PEG 400 for the solubility. The melts of Baird and Taylor were spectra at 130–160 °C, the PEG 400 solubilities at 25 °C, far below the eutectic temperatures of about 52–61 °C, so activity coefficients are comparable between drugs but are not values for molten PEG 3350 baird2011 P4, X16, R5, R6, R7, R8. PEG 400 has end groups that may make it a different solvent from PEG 3350 baird2011 C22, and Marsac and co-workers note that end-group effects would favour a higher solubility in PEO and entropy effects a lower one marsac2008b C19. The temperature of the infrared spectra of Zhu and co-workers is not stated, while the 20/80 liquidus ranges from about 54.5 to 133 °C zhu2011 X11. Each paper defines its own lattice site and normalization for \chi (drug volume and per PEG segment in zhu2011; interaction energy over R\,T_{eut,th} in baird2011; the volume of an ethyl-pyrrolidone analogue in marsac2009), so values should be compared within a paper only zhu2011 X10; baird2011 E8; marsac2009 E2.

Counterexamples and conflicts

(a) Ketoconazole: an activity coefficient of 0.09 in PEG 400, so a solubility above the ideal value, against a positive \chi of 0.02, a solubility of 0 wt% in PEG at 25 °C and an unshifted infrared spectrum in PEG 3350 marsac2008b R14.f; zhu2011 R5.f, R20. (synthesis: both can be true at once, since ketoconazole's ideal solubility is itself tiny (2.0 × 10⁻³ mole fraction) so a mole-fraction solubility of 0.02 is about ten times ideal while still only about 2.6 wt% of the PEG 400 solution, 0.02 × 531.43 / (0.02 × 531.43 + 0.98 × 400); but the two readings measure different quantities and do not agree on "interacts".) marsac2008b R14.f, M6, M8. (b) Haloperidol: \chi_{dp} −0.04 from one phase diagram and +0.08 from the other, the same published diagram; the sign is not robust near zero baird2011 R9.d; zhu2011 R5.a. (c) Chlorpropamide: −2.53 against −0.50 for the same system, a factor of about 5 (synthesis: 2.53/0.50) baird2011 R9.a; zhu2011 R5.c. (d) Loratadine is grouped with haloperidol as close to ideal in PEG 400 although its measured solubility is about 0.42 of the ideal value baird2011 C24, X13. (e) Haloperidol has an OH group but no carbonyl shift baird2011 M4, R10.g. (f) In the 2011 paper of Zhu and co-workers the text says the chlorpropamide interaction was confirmed by infrared spectroscopy but no chlorpropamide spectrum is shown zhu2011 X3. (g) All six drugs of Marsac and co-workers had activity coefficients below 1 in PEG 400 (0.02–0.64) although felodipine and ketoconazole kept their crystallinity in PEO at 94% RH, so this reading did not discriminate there (see F11) marsac2008b R14, R16.

Transfer check

Use at least two readings, not one, and compare a new drug only with drugs measured in the same way: (i) the liquidus from 0 to 100 wt% and the eutectic against the athermal prediction (F2); (ii) activity coefficient in PEG 400 at 25 °C, with the ideal solubility computed from the same melting point and heat of fusion; (iii) the infrared carbonyl and amine or hydroxyl bands of a 25:75 drug–PEG melt recorded 10–20 °C above the drug melting point against the additive spectrum baird2011 P4, P5, P6. Do not mix \chi values across papers. If two readings disagree, as for ketoconazole, record the drug as unresolved and test the real dispersion (F8).

History

Baird and Taylor build on earlier infrared work on drug–polymer hydrogen bonding, including Taylor and Zografi (1997) in the library, and take the additive-spectrum procedure from a 2009 library paper on moisture-induced immiscibility baird2011 C28, P4. Marsac and co-workers (2008) used PEG 400 solubility and activity coefficients first, to rank the driving force for drug dissolution into PEO marsac2008b C18. Baird and Taylor (2011) combined the melt phase diagram, PEG 400 solubility and infrared spectra for four drugs and proposed PEG 400 solubility as a screen baird2011 C25. Zhu and co-workers (2011) added \chi from melting point depression and the 25 °C solubility for four more drugs zhu2011 R5. Zhu and co-workers (2012) added a non-interacting drug, fenofibrate zhu2012 C49.

How it was shown

Zhu and co-workers took the mole-fraction solubility x of each drug in liquid PEG at each temperature from its liquidus, fitted a straight line to \ln x against 1/T (the van't Hoff relation, d\ln x/dT = \Delta H_f/(R T^2)), extended it to 25 °C, below the PEG melting point, and converted the result to weight percent; the estimate therefore refers to supercooled liquid PEG zhu2011 P6, E3, C8. Zhu and co-workers used the same route for fenofibrate from the supporting-information diagram zhu2012 M2, E3. Marsac and co-workers could not measure the solubility in amorphous PEO because the transformation is too slow when dry, so they took the solubility in liquid PEG 400 at 25 °C as a stand-in, assumed that only the amorphous PEO dissolves drug, and calculated how much of the 50 mg of drug in a 500 mg compact the amorphous 9% of the 450 mg of PEO could dissolve marsac2008b C18, C20, E4, P10.

What the data show

The estimated solubilities at 25 °C were 28 wt% (benzocaine), 26 (chlorpropamide), 25 (ibuprofen), 11 (nilutamide) and 0 (haloperidol, ketoconazole) zhu2011 R5, and 6 for fenofibrate and under 1 for haloperidol zhu2012 M2, M1. For chlorpropamide, benzocaine and ibuprofen they lie below the eutectic content (35, 30 and 40 wt%), as a solubility that falls on cooling should; for nilutamide, 11 wt% at 25 °C is above the eutectic content of 10 wt% at 58.4 °C, which the extractor of the card notes as unexpected zhu2011 R5.c, R5.b, R5.d, R5.e, X4. (synthesis: at the 20/80 loading used in the dispersions, benzocaine and ibuprofen are below their estimated solubility and fenofibrate and haloperidol above it zhu2012 M3, M4, M2, M1.) Yet benzocaine started to crystallize after 120–180 min and ibuprofen after about 100 min, and haloperidol, which is insoluble, within seconds zhu2012 R23.a. The extractor suggests a reason that the authors do not discuss: PEG crystallizes first, within about 2 min, so the drug is concentrated in the remaining non-crystalline fraction zhu2012 X13, R23.c, R23.d. For Marsac and co-workers the PEG 400 mole fractions were 0.32 (acetaminophen and ketoprofen), 0.22 (naproxen), 0.28 (ibuprofen), 0.16 (felodipine) and 0.02 (ketoconazole), which translate to 14, 24, 13, 17, 15 and 2% of the drug dissolvable by amorphous PEO; for ketoprofen the card recomputes 0.09 × 450 mg × 0.299 = 12.1 mg of 50 mg, 24.2% marsac2008b R14, R16, E4. The authors find no clear correlation with the observed loss: felodipine (15%) hardly changed while acetaminophen (14%) lost all its crystallinity marsac2008b C22, R16.

Rows of the Marsac and co-workers estimate are in the order acetaminophen, ketoprofen, naproxen, ibuprofen, felodipine, ketoconazole marsac2008b R16.

The two estimates are not on the same scale: Zhu and co-workers give weight percent of drug in the saturated PEG solution, Marsac and co-workers the percent of the drug in a compact that a small amorphous polymer fraction could take up (synthesis: for ibuprofen, 25 wt% in supercooled PEG 3350 against 17% of the drug in a PEO compact zhu2011 R5.d; marsac2008b R16.d).

Figure

Negative \chi and solubilities of 11–28 wt% for benzocaine, chlorpropamide, ibuprofen and nilutamide; positive \chi and zero solubility for haloperidol and ketoconazole zhu2011 R5.

Why

The authors link a solubility above 20% to a negative \chi, hydrogen-bond donors that can bind the PEG ether oxygen and moderate melting points (benzocaine, chlorpropamide, ibuprofen), and nilutamide's lower value to its higher melting point; positive \chi leaves haloperidol and ketoconazole "almost insoluble" zhu2011 C9, C12, C13. Marsac and co-workers find no clear correlation between the PEG 400-based estimate and the loss, relate the outcome instead to the drug's ability to dissolve in the water–amorphous PEO cosolvent (F11), and write that the true solubility in PEO "remains elusive" because end-group effects and mixing entropy act in opposite directions marsac2008b C22, C38, C19. For the dispersions, Zhu and co-workers do not use the solubilities to explain the crystallization times; they attribute benzocaine's delay to melting point depression and the early onset of ibuprofen and fenofibrate to nucleation on crystalline PEG (F8, F9) zhu2012 C32, C34.

Conditions and limits

The 25 °C values refer to supercooled liquid PEG, not to the largely crystalline polymer in a stored dispersion zhu2011 C8. The regression lines, heats of fusion and molar masses behind them are not printed zhu2011 E3. Nothing in these papers measured the crystallinity of PEG 3350 in the dispersions or the drug concentration in its amorphous fraction; Marsac and co-workers estimated PEO 100000 as about 91% crystalline from 197 against 216 J/g marsac2008b R15, E3. Only a 20/80 loading was studied for the dispersions.

Counterexamples and conflicts

The nilutamide value above its eutectic content zhu2011 X4. The dispersions of benzocaine and ibuprofen crystallized although nominally below saturation zhu2012 X13. In the PEO compacts, felodipine and acetaminophen had similar PEG 400 estimates and opposite outcomes marsac2008b R16.

Transfer check

Treat an extrapolated or PEG 400-based solubility as a ranking between drugs measured the same way, not as a loading limit. To decide a loading, compute the drug concentration in the amorphous fraction of the polymer (needs the polymer's crystallinity, for example from the melting enthalpy as in marsac2008b E3) and compare it with the estimate, then store the real dispersion at the intended loading and follow the drug by XRD or Raman over hours to days (F8) zhu2012 P5, P6.

History

Marsac and co-workers (2008) used the PEG 400 stand-in first and found no correlation with outcome marsac2008b C22. Zhu and co-workers (2011) applied the van't Hoff extrapolation, a method the card traces to the group's 2010 study, to six drugs zhu2011 E3, P6. Zhu and co-workers (2012) used the estimates to describe the four drugs of their kinetic study zhu2012 M1, M2, M3, M4.

How it was shown

Zhu and co-workers mixed each drug with PEG 3350 at 20/80 w/w, put the physical mixture in an aluminium DSC pan on a Linkam stage in the beam of the Advanced Photon Source (17 keV), heated it above the drug melting point and quench-cooled it to 25 °C; pure PEG was quenched to 25 and 40 °C zhu2011 P2, A1, A2. Small-angle (q 0.0085–0.23 Å⁻¹) and wide-angle (q 0.6–4.6 Å⁻¹) scattering were recorded together for about 30 min (about 116 min for chlorpropamide, about 131 min for ibuprofen) and once more at 24 h zhu2011 P2, A3. The long period was taken from the SAXS peak position as L = 2n\pi/q zhu2011 E4. The drug phase state came from WAXS: PEG shows crystal peaks at q about 1.36 and 1.65 Å⁻¹ zhu2011 R7.

What the data show

On quenching pure PEG to 25 °C the WAXS crystal peaks appeared at once and did not move, while the SAXS peak, first seen at q about 0.047 Å⁻¹, shifted to larger q; L fell from around 13.3 nm to 11.6 nm within 30 min and to 11.2 nm at 24 h zhu2011 R7, R8. At 40 °C L rose slightly, then fell to 12.6 nm after 40 min, about 1 nm above the 25 °C value at 30 min, and reached 11.2 nm at 24 h zhu2011 R8. The authors assign the first structure to nonintegral folded lamellae and the 11.2 nm period to once-folded lamellae zhu2011 C14.

In the blends, benzocaine raised L to about 16.5 nm within about 1.5 min, and the value remained there to 28.2 min, with no benzocaine crystals by WAXS although pure benzocaine crystallizes at once zhu2011 R12, R11, R6. Ibuprofen raised L to about 16.5 nm within about 3 min, and the value remained at about 16.1 nm until about 110 min, 4.9 nm above pure PEG at 20 min zhu2011 R14. With nilutamide the long period overshot to about 17.3 nm at 2–3 min and then remained at about 16.5, later 16.1 nm zhu2011 R16. With chlorpropamide the long period remained at about 14.7 nm from about 18 to 85 min, more than 3 nm above pure PEG at about 28 min, with no crystallization by WAXS (data not shown) zhu2011 R13. The ibuprofen change was followed through: at 100 min one SAXS peak at q about 0.039 Å⁻¹ and an amorphous drug; at 111 min a second SAXS peak at about 0.047 Å⁻¹ and a crystalline ibuprofen WAXS peak; at 116 min the 0.047 Å⁻¹ peak dominated; two long periods coexisted for a few minutes (about 15.9–15.5 and 13.7–13.5 nm) and L settled near 13.3 nm before reaching 11.9 nm at 24 h zhu2011 R14, R15. At 24 h L was 11.4, 11.7, 11.9 and 11.4 nm with benzocaine, chlorpropamide, ibuprofen and nilutamide, which Table 2 lists as partially crystalline, while the text says benzocaine "crystallized" and "most" of the chlorpropamide did zhu2011 R18, C22, C24, X12.

Figure

L sits at about 16.5 nm from 1.5 to 28 min against about 11.6 nm for pure PEG, and drops at 24 h; the 24 h point is plotted at about 11.7 nm while text and Table 2 give 11.4 nm zhu2011 R12, X1.

About 16.5 then 16.1 nm to about 110 min, then a second, smaller L and a fall to 11.9 nm at 24 h zhu2011 R14.

Why

The authors' explanation, step by step: the long period is the repeat of the crystalline lamella and the amorphous layer between lamellae; a drug that mixes with the amorphous PEG and is located in that layer thickens it and enlarges L zhu2011 X8, E4, C32. For benzocaine, the alternative that the drug lowers the PEG melting point and so the supercooling of PEG, which could also enlarge L, was weighed: pure PEG crystallized at 40 instead of 25 °C changed L by only about 1 nm at 30 min, so the 4–5 nm increase "can be largely attributed" to benzocaine in the interlamellar region zhu2011 C21, R8. For chlorpropamide, "suggesting interlamellar incorporation" zhu2011 C23. When the drug crystallizes it is removed from the interlamellar region and L falls: "crystallization of the drug results in its exclusion from the interlamellar regions" zhu2011 C26. In the wording of the general conclusion, "This suggests that interlamellar incorporation is favored" for molecules that interact strongly with PEG, and "the location of the API could thus be inferred" from L zhu2011 C32, C33. From the polymer-blend literature the authors expect exclusion from PEG crystals to depend on the PEG crystallization rate, the additive's mobility through the blend Tg and specific interactions zhu2011 C2. What was not measured: the thicknesses of crystalline and amorphous layers were not separated, a change of chain folding in the blends was not examined, and the drug was not located by any other method zhu2011 X8. (synthesis: the 24 h values with a crystallized drug, 11.4–11.9 nm, are still 0.2–0.7 nm above the 11.2 nm of pure PEG at 24 h, the same kind of residual that the extractor notes for ketoconazole zhu2011 R18, R8, X6.)

Conditions and limits

One drug/PEG ratio (20/80), PEG 3350, quench to 25 °C; one run per system and no replicates reported; the storage between the last time-resolved frame and the 24 h measurement is not stated zhu2011 X9. The interaction parameters were fitted at 50–80% drug, not at 20/80 zhu2011 X10. The 24 h crystallinity rests on WAXS patterns that are not shown, and chlorpropamide WAXS is not shown at any time zhu2011 X12. The drug content in the amorphous layer and its time dependence are not known.

Counterexamples and conflicts

The size of the increase does not follow the size of the interaction or the solubility: chlorpropamide has the most negative \chi (−0.50) and 26 wt% solubility but the smallest L at 30 min (14.7 nm), while nilutamide (\chi −0.09, 11 wt%) gives 16.0 nm, so only the sign of \chi separates interacting from non-interacting drugs zhu2011 R5.c, R5.e, R18, X5. The BZC 24 h value differs between figure and table zhu2011 X1. None found in the cards of this chapter, since no other card measured the long period.

Transfer check

Prepare the dispersion at the intended ratio and quench under the production conditions; record SAXS and WAXS together from seconds to 24 h, with a pure-PEG control made the same way and a second control at another crystallization temperature to bound the effect of supercooling. An increase of L by 3 nm or more with no drug WAXS peaks is consistent with an amorphous drug between lamellae; confirm with a second method before relying on it, since L alone does not locate the drug zhu2011 X8. A fall of L toward the pure-PEG value signals drug crystallization, as in F8.

History

The group had probed drug/PEG microstructure after 24 h only in an earlier study that is not part of the PEG dispersions set, and the polymer-blend literature on additive exclusion is outside the library zhu2011 C2, C7. Zhu and co-workers (2011) introduced the time-resolved SAXS/WAXS view in this library. Zhu and co-workers (2012) later followed the crystallization of the same ibuprofen and benzocaine dispersions by XRD and Raman and found onsets consistent with the SAXS/WAXS result (F8) zhu2012 R14, R11.

How it was shown

The same SAXS/WAXS protocol as in F5 was used for haloperidol and ketoconazole at 20/80 and 25 °C, with the pure-PEG control zhu2011 P2, A1. The drugs were chosen for differing crystallization tendency, interaction with PEG and Tg: haloperidol rapidly crystallizing and weakly interacting, ketoconazole slowly crystallizing, weakly interacting, with a Tg of 44 °C; the Tg values are cited from the 2010 classification paper zhu2011 C16, M5, M6. Zhu and co-workers (2012) later followed haloperidol/PEG by synchrotron XRD at 20/80 and 25 °C, recording the order of crystallization zhu2012 P4, R7.

What the data show

In haloperidol/PEG the haloperidol WAXS peak at q about 1.40 Å⁻¹ was present from the 0 min frame and persisted as a shoulder of the 1.36 Å⁻¹ PEG peak zhu2011 R9. L started at about 13.9 nm, settled near 11.95 nm from about 11.5 to 35 min, about 0.4 nm above pure PEG, and was 11.2 nm at 24 h; the authors call this "essentially the same" as pure PEG zhu2011 R10, C17, C18. In ketoconazole/PEG L dropped to about 11.8 nm within about 2 min and stayed there to 24 h, with no ketoconazole crystals by WAXS after 24 h zhu2011 R17. At 24 h ketoconazole's L is 0.6 nm above pure PEG (11.8 against 11.2 nm), a residual similar to that of crystallized chlorpropamide (11.7) and ibuprofen (11.9) and not discussed by the authors zhu2011 X6, R18. For nilutamide, L rose to about 16–17 nm and fell to 11.4 nm once the drug had crystallized by 24 h, like the low-Tg drugs zhu2011 R16, R18.e, C28. In the 2012 paper haloperidol began to crystallize before PEG and was completely crystalline within a few seconds, and PEG crystallization began immediately afterwards zhu2012 R7, C2.

Figure

L is about 11.8 nm from 2 min to 24 h, close to the PEG control; the drug stays amorphous zhu2011 R17.

Why

The authors' explanation: haloperidol's lack of influence "most likely results from its rapid removal from the PEG phase" by crystallization on cooling zhu2011 C19. Ketoconazole's lack of effect is "likely due to the positive interaction parameter", which "may indicate a low tendency to mix with PEG and a large extent of exclusion from the interlamellar region"; "clearly" its higher Tg did not prevent phase separation zhu2011 C30, C31. In the polymer-blend literature the additive's mobility, through the blend Tg, is one of three factors that decide exclusion, which is why Tg was examined zhu2011 C2. (synthesis: with the neat-drug classes of the chapter Crystallization tendency, the two drugs belong to opposite ends of the classification, haloperidol in class I-A (crystallizes on cooling, even in liquid N2) and ketoconazole in class III, yet both leave the long period unchanged, so the drug's own crystallization tendency does not decide incorporation; the interaction does baird2010 R1, M17, M42.)

Conditions and limits

Two weakly interacting drugs, one ratio, one PEG grade and one solidification temperature. Ketoconazole's exclusion and phase separation are inferred from L; the amorphous ketoconazole was not imaged and no separate domain was observed zhu2011 X7. The Tg conclusion rests on nilutamide and ketoconazole, whose Tg lie 8 and 19 °C above the 25 °C solidification temperature, which also differ in \chi and crystallization; the Tg values of the other four drugs are given only as below 25 °C zhu2011 X13. (The card gives haloperidol's Tg as below 25 °C, while the classification paper lists 33 °C calculated from the molecular weight; see card-issues.md zhu2011 M1; baird2010 M17.)

Counterexamples and conflicts

None found in the library.

Transfer check

For a drug with a positive \chi or no sign of interaction, do not expect the long period to show its amorphous state: ketoconazole was amorphous but did not change the long period. Check the drug's state directly by WAXS or XRD over 24 h and look for separate amorphous drug domains by another method zhu2011 X7. Do not extend the Tg result beyond about 44 °C at a 25 °C solidification zhu2011 X13.

History

Zhu and co-workers (2011) showed the contrast between F5 and F6 in one study. Zhu and co-workers (2012) confirmed that haloperidol crystallizes ahead of PEG and is unaffected by it zhu2012 R7, C2.

How it was shown

Zhu and co-workers melted pure PEG 3350 or 20/80 physical mixtures in aluminium DSC pans in a Linkam stage in the synchrotron beam (17 keV), held them above the drug melting point until all diffraction peaks had disappeared and quench-cooled them to 25 or 40 °C zhu2012 P4. The two PEG crystal peaks (q 1.35 and 1.65 Å⁻¹) were integrated against the total area by curve fitting to give an apparent crystallinity, normalized to its final value to give the relative crystallinity X(t) = \Phi_{peg}(t)/\Phi_{peg}(\infty), with time zero at the start of PEG crystallization zhu2012 P4, E1. The induction time is the time to the first detected crystals and is negative when crystallization began before the set temperature was reached zhu2012 R5. The Avrami equation X(t) = 1 - \exp(-z\,t^{n}) was fitted to the pure-PEG curves zhu2012 E2, R6. In the SAXS/WAXS study the same beamline was used at 25 °C with frames from 0.2 min zhu2011 P2, R7.

What the data show

Pure PEG at 25 °C reached about 13% at 10 s, about 47% at 22 s, about 88% at 30 s, about 96% at 40 s and about 99% at 80–120 s; at 40 °C it stayed near zero for about 12 s, reached about 51% at 64 s and about 89% at 90 s, and about 95% by 120–150 s zhu2012 R4. The induction time was −10 s at 25 °C, meaning that PEG began to crystallize before 25 °C was reached, and 16 s at 40 °C zhu2012 R5. The fitted Avrami exponent was below 1 at both temperatures (values not printed), read by the authors as one-dimensional growth ("This suggested"), as reported for very fast growth zhu2012 R6, C19. In the dispersions at 25 °C PEG started within seconds (induction times −12, 0, 8 and 1 s for haloperidol, benzocaine, ibuprofen and fenofibrate) and reached 50% in 15, 22, 26 and 21 s zhu2012 R23.c, R23.d. The curves still differ somewhat: PEG in haloperidol/PEG reached 50% sooner than pure PEG, PEG in benzocaine/PEG followed pure PEG to about 58% and levelled off lower (about 94% against about 99% at 120 s), and PEG in ibuprofen/PEG lagged slightly zhu2012 R8, R12, R16, R20, X3, X4. In the SAXS/WAXS study the PEG crystal peaks appeared immediately after the quench and did not move, and the crystalline haloperidol peak was already present in the first frame zhu2011 R7, R9.

The order of crystallization differs by drug. Haloperidol began to crystallize before PEG and was completely crystalline within a few seconds, with PEG following immediately zhu2012 R7, C2. In benzocaine/PEG the drug "did not begin to crystallize until PEG had completely crystallized", in ibuprofen/PEG PEG crystallization was "much faster" than the drug's, and fenofibrate/PEG solidified like ibuprofen/PEG zhu2012 R9, C26, C28.

Figure

Diamonds are 25 °C and squares 40 °C; both curves complete in under 2 min, the 40 °C curve about 41 s later at the midpoint (synthesis: 64 − 23 s) zhu2012 R4.

Why

The authors take the similarity of PEG kinetics at 25 °C in different dispersions to arise from the "extremely fast crystallization of PEG" at that temperature, with larger differences at 40 °C (F10) zhu2012 C27, C43. They conclude that in practice PEG crystallizes within a few minutes in all systems, and that how a changed PEG crystallization rate influences the dispersion microstructure is not known zhu2012 C46. (synthesis: since PEG crystallization is nearly complete within one to two minutes, a drug that is still amorphous at that point remains in the non-crystalline volume of PEG, so its later crystallization at 25 °C takes place in a mostly crystalline matrix, which is the setting of F8 and F9. The card's extractor notes that the drug was concentrated in the remaining non-crystalline fraction zhu2012 X13.)

Conditions and limits

PEG 3350, 20/80 w/w, quench to 25 or 40 °C; the cooling rate, hold time above the melting point, frame exposure and sample mass are not printed zhu2012 P4; zhu2011 P2. The fitted Avrami exponent and constant are not printed, and the reading of n < 1 as one-dimensional growth is qualitative, since on the authors' own scheme n = 1 is already the lowest value zhu2012 R6, X9. In Fig. 2 pure PEG at 40 °C reaches 51% at about 64 s, while in Figs. 11 and 15, with the same legend, it reaches 51% at about 48 s, so the 40 °C time scale of pure PEG is uncertain by about 15 s zhu2012 X1. The number of replicate runs is not stated zhu2012 X6. Slower cooling, other PEG grades and bulk samples were not studied.

Counterexamples and conflicts

None found in the library for the speed of PEG solidification. The text calls PEG kinetics in haloperidol/PEG "essentially the same" as pure PEG although PEG in haloperidol/PEG reaches 50% at about 15 s against about 23 s zhu2012 C21, X3.

Transfer check

Measure the relative crystallinity of the carrier alone and in the dispersion after the cooling used for manufacture; if carrier crystallization is complete within minutes, ask whether the drug crystallizes first (a fast crystallizer in the sense of the neat-drug classes) or remains amorphous, and then follow the drug for hours to days (F8). For other cooling rates or PEG grades, the time scale is untested.

History

Zhu and co-workers (2011) saw the PEG crystal peaks in the first seconds and the haloperidol peak in the first frame zhu2011 R7, R9. Zhu and co-workers (2012) quantified the PEG kinetics at 25 and 40 °C, alone and with four drugs, and found the same order for haloperidol zhu2012 R4, R7, R23.

How it was shown

Zhu and co-workers (2012) melted each pure drug on a hot plate and held it at 25 °C under a polarized light microscope; for ibuprofen and fenofibrate 50 samples each were followed for about 500 h, recording P(t), the fraction that had started to crystallize by time t, because their induction times varied widely zhu2012 P2, P3. The 20/80 comelts were stored over phosphorus pentoxide and measured at intervals by laboratory XRD (Cu Kα, 2θ 4–30°) and by Raman spectroscopy, with marker bands at 1573 and 1680 cm⁻¹ (benzocaine), 746 and 784 cm⁻¹ (ibuprofen) and 1419 and 1648 cm⁻¹ (fenofibrate); haloperidol was too fast for XRD zhu2012 P5, P6, R7. The induction time and t_{50} (time from onset to 50% relative crystallinity, induction not included) of each drug are in Table 1 zhu2012 R23. Zhu and co-workers (2011) followed six drugs at 20/80 and 25 °C by SAXS/WAXS for about 30 min (about 116 and 131 min for chlorpropamide and ibuprofen) and once at 24 h, and observed the pure melts under the microscope after cooling to room temperature zhu2011 P2, P7, R6.

What the data show

Neat haloperidol and benzocaine crystallized before the samples reached 25 °C and were fully crystalline within a few seconds zhu2012 R1. Neat ibuprofen stepped from 2/50 within 1 h to 10/50 (0.20) after 3 weeks, neat fenofibrate from 4/50 within 1 h to 17/50 (0.34), then stayed at 0.34 to 504 h zhu2012 R2, R3. In the dispersions, benzocaine showed no XRD peak at 30 min and a peak at 180 min; the Raman band at 1680 cm⁻¹ appeared at 180 min; crystallinity was about 20% at 200 min, about 50% at 290 min and about 92% at 530 min (text: about 10 h to 90%; plot: about 8 h) zhu2012 R9, R10, R11, X12. Ibuprofen's XRD peak was absent at 20 min and present from 120 min; Raman gave about 100 min induction, about 57–61% at 180 min and about 91–94% at 240 min zhu2012 R13, R14. Fenofibrate's XRD peaks appeared from 240 min and Raman crystallinity rose from about 8% at 180 min to about 54% at 330 min and about 92% at 510 min zhu2012 R17, R18. In the SAXS/WAXS study the neat melts of chlorpropamide, nilutamide and ibuprofen crystallized within a few hours and ketoconazole not within 24 h, and in the blends benzocaine and chlorpropamide were amorphous at 30 min (chlorpropamide WAXS not shown, its long period unchanged to about 115 min), ibuprofen for about 100 min, nilutamide at 30 min and ketoconazole at 24 h zhu2011 R6, R11, R13, R14, R16, R17. Haloperidol was crystalline from the first frame zhu2011 R9.

(synthesis: ibuprofen was followed in both papers, at the same 20/80 ratio and 25 °C, by Raman (onset about 100 min) and by SAXS/WAXS (amorphous at 100 min, crystalline at 111 min), so two methods give the same time scale zhu2012 R14; zhu2011 R15.) (synthesis: relative to the neat drug, PEG therefore delayed benzocaine from seconds to hours, did not change haloperidol and advanced ibuprofen and fenofibrate from weeks, for most samples, to hours zhu2012 R1, R2, R3, R23.a.)

(synthesis: the neat-drug classes of the chapter Crystallization tendency sort the first two outcomes but not the third. Haloperidol and benzocaine are class I-A and chlorpropamide class I-B; ibuprofen, fenofibrate, ketoconazole and nilutamide are class III. The class III drugs behave differently in PEG: ibuprofen and fenofibrate began after about 100 and 120–180 min, nilutamide was partially crystalline and ketoconazole not crystalline at 24 h baird2010 R1, R3, M9, M17, M12, M39, M38, M42, M46; zhu2012 R23.a; zhu2011 R18.e, R18.f. See F17 of that chapter.)

Figure

Onset after about 150–200 min, 50% at about 290 min and near-complete crystallinity after 600 min; both Raman bands give the same curve zhu2012 R11, C23.

Onset at about 100 min and more than 90% at 240 min zhu2012 R14.

Why

The authors set out to test whether the crystallization tendency of the pure drug, combined with its ability to interact with the polymer, decides the outcome zhu2012 C13. Haloperidol crystallizes faster than PEG and is practically insoluble in it, so PEG has no effect on its crystallization, in the authors' account zhu2012 C2, C39, M1. For benzocaine the delay "is thought to be caused by the entropic mixing effects" and favourable benzocaine–PEG interactions that depress the benzocaine melting point: at 20:80, below the eutectic composition, it melts at the eutectic, around 48 °C, against around 89 °C for the pure drug, so the supercooling at 25 °C and the driving force are much smaller; with a Tg of −31 °C the authors find the supposition that the cause is thermodynamic rather than kinetic "reasonable" zhu2012 C32, C33, M3. (synthesis: with the cited values the supercooling at 25 °C falls from 89 − 25 = 64 K to 48 − 25 = 23 K, to about 36% of the neat value zhu2012 M3.) Both values are cited from earlier work and the supercooling and mobility in the dispersion were not measured zhu2012 C32, C33. For ibuprofen and fenofibrate the earlier onset despite melting point depressions of about 30 and 20 °C is attributed to nucleation on crystalline PEG (F9) zhu2012 C34, C4. In their summary the link between the crystallization rate of the pure drug and in the dispersion "was found to break down for the slowly crystallizing compounds" zhu2012 C41. Benzocaine in the SAXS/WAXS study stayed amorphous for 30 min in the blend although the neat drug crystallizes at once; the authors relate this "most likely" to the lower supercooling of benzocaine in the blend, 55 against 92 °C zhu2011 C20, R1.

Conditions and limits

One composition (20/80 w/w), PEG 3350, solidification at 25 °C, comelts; the number of replicate kinetic runs is not stated and Table 1 gives ranges zhu2012 X6. Neat-drug statistics stop at 504 h, when 40 of 50 ibuprofen and 33 of 50 fenofibrate samples had not crystallized, so the abstract's "hours to months" is not measured zhu2012 X5, C3. The haloperidol values of Table 1 have no stated method although the text says the kinetics were too fast for XRD zhu2012 X7. The instrument of Figs. 4, 8 and 12 is not stated zhu2012 X8. The 24 h states in the SAXS/WAXS study rest on WAXS patterns that are not shown, and chlorpropamide WAXS is not shown at all zhu2011 X12. Lower loadings, other PEG grades and storage temperatures were not tested zhu2012 A1, A3.

Counterexamples and conflicts

The 2011 paper says that the neat melts of chlorpropamide, nilutamide and ibuprofen crystallized within a few hours, whereas the 2012 statistics of 50 samples show neat ibuprofen starting in only 2 of 50 samples within 1 h and 10 of 50 within 3 weeks zhu2011 R6; zhu2012 R2. (synthesis: the 2012 authors read the spread as the stochastic nature of nucleation, with the chance of starting low over hours to days zhu2012 C16, C17; a single melt observed in 2011 is then not decisive, so whether PEG advances ibuprofen rests on the 50-sample statistics.) (synthesis: without a statistic of the same kind for the neat drugs, nilutamide and ketoconazole cannot be said to be advanced or delayed zhu2011 R6, R16, R17.) The text times to 90% (about 10 h for benzocaine, about 9 h for fenofibrate) are longer than the plots (about 8.3 h) zhu2012 X12.

Transfer check

For a new drug in a PEG comelt: (1) time the neat drug at the storage temperature on many samples (the 2012 study used 50) because the spread is large; (2) quench the dispersion at the intended ratio and follow the drug by XRD and by two Raman bands, or by SAXS/WAXS, from minutes to at least 24 h, noting whether PEG crystallization is complete before the drug begins to crystallize; (3) compare the drug melting point in the dispersion with that of the neat drug to see how much PEG lowers the supercooling; (4) test for nucleation on crystalline PEG as in F9. Until then do not assume that a drug which is slow as a neat melt will stay amorphous in PEG; the authors state that the drug should be fully amorphous or fully crystalline at administration, not a mixture zhu2012 C37.

History

Zhu and co-workers (2011) gave the first time courses, over 30 min and 24 h, for six drugs and showed which stayed amorphous zhu2011 R18. Zhu and co-workers (2012) followed four of them over hours by two methods and compared them with neat-drug statistics, concluding that PEG can delay, promote or have no influence zhu2012 C6. The finding that the neat-drug class does not predict the behaviour of slow crystallizers in PEG dispersions is taken up in the chapter Crystallization tendency.

How it was shown

Zhu and co-workers placed a small piece of crystalline PEG into supercooled liquid ibuprofen or fenofibrate at 25 °C and watched it under a polarized light microscope zhu2012 P7. For the comparison with the neat drugs they recorded P(t), the fraction of 50 melt samples that had started to crystallize by time t at 25 °C, for about 500 h zhu2012 P3. The onset in the dispersions came from laboratory XRD and Raman on 20/80 comelts stored at 25 °C (F8) zhu2012 P5, P6.

What the data show

In the seeding test, an ibuprofen crystal formed at the interface between molten ibuprofen and crystalline PEG after a few minutes, with no nuclei visible in the bulk supercooled liquid; fenofibrate crystallized "immediately adjacent to the PEG particle" within a few minutes with no crystallites elsewhere (scale bars 200 µm) zhu2012 R15, R19. For the neat drugs, P(t) of ibuprofen was about 0.04 (2 of 50) from about 1 to about 62 h, about 0.12 from about 72 to about 134 h, about 0.14 from about 168 to about 315 h and about 0.20 from about 336 to about 504 h; fenofibrate's was about 0.08 at about 1 h, about 0.20 at about 8–14 h and about 0.34 from about 96 to about 504 h zhu2012 R2, R3. In the 20/80 dispersions the onset was consistently earlier: about 100 min for ibuprofen and 120–180 min for fenofibrate, with more than 90% crystallinity at 4 h and 9 h zhu2012 R23.a, R14, R18, C3. The two drugs form eutectics with PEG, with melting point depressions of about 30 and 20 °C (the plot of fenofibrate shows about 24 °C), which would lower the driving force zhu2012 C34, X11, R24. In the SAXS/WAXS study the ibuprofen dispersion crystallized at about 111 min zhu2011 R15.

Figure

The dashed circle marks the ibuprofen crystal at the PEG edge; the surrounding supercooled ibuprofen contains no visible crystals zhu2012 R15.

Ibuprofen steps to 0.2 and fenofibrate levels at 0.34 over about 500 h zhu2012 R2, R3.

Why

The authors' account: "it appears that crystalline PEG acted as a favorable heterogeneous nucleation site", thought to explain the shorter induction times, and the seeding tests were made "to further confirm this hypothesis" zhu2012 C4. Polymers have been shown to act as heteronuclei for organic molecules and even to select the polymorph, though in solution rather than melt crystallization (a study outside the library), and most studies of drugs in polymers find that the polymer retards crystallization, which makes the promotion by PEG of particular interest zhu2012 C35, C36. Because the induction times of the neat drugs vary widely between samples, the authors read the spread as the stochastic nature of nucleation zhu2012 C16, C17. The extractor's reservation: PEG crystallized first in every system, within about 2 min, so the drug was concentrated in the remaining non-crystalline fraction, and by the table of estimated solubilities benzocaine (28%) and ibuprofen (25%) exceed the 20% loading in supercooled PEG yet crystallized; local enrichment of the residual melt could shorten the induction as well, and the seeding tests, run in pure supercooled drug, cannot separate the two routes zhu2012 X13, M3, M4. The authors add that faster nucleation could change crystal size distribution and so the dissolution of the dispersion; nothing was measured on this zhu2012 C42.

Conditions and limits

Two drugs, one PEG grade, 25 °C, one composition. The size of the PEG pieces and the number of seeding tests are not stated, and the number of replicate dispersion runs is not stated zhu2012 P7, X6. Pure-drug induction was followed only to 504 h, so "hours to months" for the neat drugs is not measured zhu2012 X5. Nucleation inside the dispersions was not observed. Whether other class III drugs are nucleated by crystalline PEG was not tested; in the same PEG at the same ratio nilutamide, also class III, was partially crystalline and ketoconazole, class III, amorphous at 24 h zhu2011 R18.e, R18.f; baird2010 R1, M46, M42.

Counterexamples and conflicts

A single melt of ibuprofen observed in 2011 crystallized within a few hours zhu2011 R6, whereas 50 samples in 2012 gave 2 starting within 1 h and 10 after 3 weeks zhu2012 R2. (synthesis: the spread is read by the 2012 authors as stochastic nucleation zhu2012 C16, C17, so one melt is not decisive; see F8.) None found in the library for the seeding result itself.

Transfer check

Test a crystalline carrier by placing a piece of it in the supercooled neat drug at the storage temperature and looking for crystals at the interface zhu2012 P7. Run enough dispersion samples to see the spread (the neat-drug tests used 50). If crystals form at the interface, a slow crystallizer cannot be assumed to remain amorphous in the carrier; read the neat-drug class (the chapter Crystallization tendency) as a risk flag, not as a prediction, and prefer to measure the actual dispersion.

History

The idea that the neat-drug tendency, combined with the interaction with the polymer, decides the outcome in a dispersion was the hypothesis of Zhu and co-workers (2012); they found that it holds for the fast crystallizers and breaks down for the slow ones, with nucleation by crystalline PEG as the added factor zhu2012 C13, C38, C41. The same group's 2011 SAXS/WAXS study had shown ibuprofen crystallizing in PEG at about 111 min zhu2011 R15.

How it was shown

PEG relative crystallinity versus time was measured by synchrotron XRD as in F7 for pure PEG, haloperidol/PEG, benzocaine/PEG, ibuprofen/PEG and fenofibrate/PEG at 25 °C and for ibuprofen/PEG and fenofibrate/PEG at 40 °C zhu2012 P4, R8, R12, R16, R20. The spherulite growth rate was measured by following the front of a PEG spherulite under the polarized light microscope at 40 °C, the temperature at which growth is slow enough, in pure PEG and the four 20/80 dispersions zhu2012 P8, C30. Hydrogen bonding with PEG was taken from earlier infrared work for benzocaine and ibuprofen and measured in the supporting information for fenofibrate zhu2012 C20, R25, M1, M2, M3, M4.

What the data show

At 25 °C PEG's time to 50% was 15 s with haloperidol, 22 s with benzocaine, 26 s with ibuprofen and 21 s with fenofibrate, and the induction times were −12, 0, 8 and 1 s zhu2012 R23.c, R23.d. At 40 °C PEG in ibuprofen/PEG began after 40 s and reached about 62% at 132 s, the last point, with t_{50} 114 s; PEG in fenofibrate/PEG began after 7 s, reached 50% at about 52 s and 96% at about 135 s, close to pure PEG zhu2012 R23.c, R23.d, R16, R20. In the same figures pure PEG at 40 °C reached 51% at about 48 s and about 89% at about 67 s, while in the pure-PEG curve of Fig. 2 these points come at about 64 s and about 90 s zhu2012 R16, R20, R4, X1. The spherulite front advanced about 205 µm in 10 s in pure PEG and the text gives 10 µm/s; the plotted pure-PEG length reaches about 470 µm at about 24 s, a slope near 20 µm/s by the extractor's calculation zhu2012 R21, R22.a, X2.

The pure-PEG value at 25 °C is the interpolated half point of the curve (about 47% at 22 s and about 55% at 24 s) zhu2012 R4.

Figure

The slopes fall from pure PEG to haloperidol and fenofibrate, then ibuprofen, then benzocaine zhu2012 R22.

At 40 °C PEG crystallization with ibuprofen (triangles) is much slower than in pure PEG (squares); at 25 °C the curves are close zhu2012 R16.

Why

The authors write that the different behaviour of PEG with ibuprofen and fenofibrate "is probably due to the differing ability of the two APIs to interact with PEG": ibuprofen can hydrogen-bond with PEG, whereas fenofibrate lacks donor groups and PEG has mainly acceptor groups; the spherulite growth rates, slowest with benzocaine and ibuprofen, are taken as "further confirmed" zhu2012 C44, C45. The expectation comes from a polymer-blend study outside the library, in which a second component that interacts favourably with PEG reduces its spherulite growth rate zhu2012 C14. They explain the similar PEG rates at 25 °C by the extremely fast crystallization of PEG there zhu2012 C27. (synthesis: an alternative the authors do not weigh is the supercooling of PEG. The 15 K step from 40 to 25 °C shortens pure PEG's t_{50} from about 48–64 s to about 23 s zhu2012 R4, R16. A dissolved drug lowers the melting point of PEG: the temperature at which the last crystal melts at 20 wt% drug is about 54.5 °C with benzocaine, about 57 °C with ibuprofen and about 58.0 °C with fenofibrate against about 60.5 °C for PEG, a depression of 6.0, 3.5 and 2.5 K (synthesis: 60.5 − 54.5, 60.5 − 57, 60.5 − 58.0), while PEG melting stays near 59–61 °C with haloperidol zhu2011 R1.g, R2.g, R1.a; zhu2012 R24.d; baird2011 R4. That order, benzocaine, ibuprofen, fenofibrate, haloperidol, is the order of the spherulite rates (1, 3, about 8, about 8 µm/s), so hydrogen bonding and reduced supercooling are not separated by these data. The numbers are read from plots.)

Conditions and limits

Bulk PEG kinetics at 40 °C only for ibuprofen/PEG and fenofibrate/PEG; the spherulite ranking covers four drugs at one temperature zhu2012 X14. The hydrogen bonding of ibuprofen and benzocaine with PEG is cited from earlier work and was not measured here zhu2012 X14, C20. The number of replicate runs is not stated zhu2012 X6. "Slightly slower" spherulite growth with haloperidol and fenofibrate holds only with the text value of 10 µm/s for pure PEG zhu2012 X2. The authors note that the practical effect is small because PEG still crystallizes within a few minutes in all systems and that the influence of a changed PEG rate on the dispersion microstructure is not known zhu2012 C46.

Counterexamples and conflicts

The text says PEG kinetics at 25 °C show negligible differences, but benzocaine/PEG levels off lower (about 94% against about 99% at about 120 s), haloperidol/PEG starts sooner and ibuprofen/PEG lags zhu2012 C43, X3, X4. Pure PEG's reference curve and spherulite rate are inconsistent between figures and text zhu2012 X1, X2.

Transfer check

To see whether a drug slows PEG crystallization, measure the PEG relative crystallinity with and without the drug in the same run at 40 °C, where PEG is slow enough to resolve, and measure the PEG melting point in the blend to separate a supercooling effect from an interaction effect. Treat the result as relevant only if the microstructure of the dispersion matters for release, which this chapter does not show.

History

A polymer-blend study outside the library had found that an interacting second component slows PEG spherulites zhu2012 C14. Zhu and co-workers (2011) showed in the library that the drug changes the PEG lamellar structure (F5) zhu2011 R18. Zhu and co-workers (2012) measured the effect on PEG crystallization kinetics and spherulite growth for four drugs zhu2012 C5.

How it was shown

Marsac and co-workers sieved drugs and PEO (Polyox WSR N-10, average molecular weight about 100000, dried over phosphorus pentoxide for at least a week) to 45–75 µm, made 500 mg physical mixtures with 10 wt% drug by geometric mixing and pressed them in a 16 mm die at 2040 psi for 1 s marsac2008b P1, M7. The compacts were stored at 0% and 94% RH (saturated potassium nitrate, 22 °C as stated for the moisture work; acetaminophen also at 75% RH) for up to 28 days marsac2008b P2, X6. Drug crystallinity was followed by FT-Raman (1064 nm, 4 cm⁻¹ resolution, at least 256 scans, compact rotating) against two references, the crystal and the drug at its solubility limit in PEG 400 (amorphous for ketoconazole), and by XRPD (Cu Kα, 5–35° 2θ) at 0, 6 and 17 days (acetaminophen) or 0 and 28 days marsac2008b P3, P4, P2. Pure amorphous acetaminophen, naproxen and ibuprofen could not be made as references because they crystallized on cooling marsac2008b R1, C9. Solubility in PEG 400 was measured by UV after 24 h at 25 °C, and activity coefficients were calculated from the ideal solubility (from the DSC melting data) and from the measured or cited solubility in PEG 400 and in water marsac2008b P5, P8, P9, E2.

What the data show

For acetaminophen the crystalline Raman bands at 1610 and 1648 cm⁻¹ shrank and the dissolved-state bands at 1619 and 1661 cm⁻¹ grew, with no crystalline drug detected after 14 days; the XRPD drug peaks (15.5, 18.2, 20.4, 24.3 and 26.6° 2θ) showed little sign after 6 days and a completely X-ray amorphous halo after 17 days marsac2008b R2, R10. Ketoprofen's Raman band moved from 1655 to 1657 cm⁻¹ (1658 in PEG 400) and stayed narrower than in PEG 400, naproxen's from 1630 to 1631 cm⁻¹ (1633 in PEG 400) with significant broadening, and ibuprofen's from 1608 to 1609 cm⁻¹ (1613 in PEG 400); their XRPD peaks became weak, though some ibuprofen peaks may remain; the shifts are below the 4 cm⁻¹ resolution, so the grade rests on the broadening and the XRPD marsac2008b R4, R5, R6, R10, X11. Felodipine and ketoconazole compacts after 28 days were very similar to the crystalline spectra and several drug XRPD peaks persisted; for felodipine, Raman barely distinguishes the crystal from the PEG 400 solution, so its grade rests mainly on XRPD marsac2008b R7, R10, X21. Compacts of acetaminophen became gelatinous within a few days, those of ketoprofen, naproxen and ibuprofen noticeably softer, and those of felodipine and ketoconazole lost a little integrity; the acetaminophen compact took up about 0.46 g water per g dry solid after 28 days against about 0.16–0.215 for the others marsac2008b R17, R13. Acetaminophen compacts also lost crystallinity "significantly", to a lesser extent, at 75% RH (data not shown) marsac2008b R3.

The aqueous solubilities behind the water coefficients range from 1.7 × 10⁻³ (acetaminophen) to 3.0 × 10⁻⁹ mole fraction (ketoconazole), cited from the literature marsac2008b R14. Several Table 2 rows are not internally consistent if \gamma = S^{ideal}/S, so the printed coefficients serve as order-of-magnitude values marsac2008b X5, X18.

Figure

The bottom trace in each panel is the fresh compact with arrowed drug peaks; the large off-scale peaks are PEO. Acetaminophen (a) is gone at 6 days and a halo at 17 days, ketoprofen, naproxen and ibuprofen (b–d) are weak at 28 days, felodipine and ketoconazole (e, f) persist marsac2008b R10.

Similar solubility in PEG 400 for all drugs; activity coefficients in water from 10¹ to 10⁵ in the same order as the crystallinity loss marsac2008b R14.

Why

The authors' account, with their hedges: compaction brings crystalline drug into contact with amorphous PEO regions, which "can be dissolved by PEO" (a hypothesis); the process is too slow to see when dry at 25 °C marsac2008b C16, C17, R8. Water, which is most likely located in the amorphous PEO, lowers the driving force for dissolution because every drug is far less soluble in water than in PEG 400 marsac2008b C24. Water nevertheless speeds the loss by "two potential mechanisms": it plasticizes the amorphous PEO (water Tg −136 °C against about −67 °C for PEO), lowering its viscosity, and it slowly deliquesces PEO, enlarging the disordered volume (F12) marsac2008b C25, C26, C30. The larger volume of water–PEO cosolvent dissolves drugs with a favourable interaction with water, and those with an activity coefficient of about 10⁵ stay crystalline marsac2008b C33, C35, C38. Dissolved drug "will most likely anti-plasticize" the amorphous PEO, since the drug Tg values (−43.6 to 44.4 °C) are above that of PEO marsac2008b C27, M1, M2, M3, M4, M5, M6. The authors conclude that the degree of transformation "was related to the ability of the drug to dissolve in the water-amorphous PEO" cosolvent and that "the thermodynamics of the binary systems were useful in describing the dynamic nature of the ternary system" marsac2008b C38, C39. What was not measured: where the drug is located (in amorphous PEO, in sorbed water or as amorphous drug), the mobility of the wet PEO, a PEO solution phase and the crystalline fraction lost marsac2008b X17, C26, X9, X8. The extractor adds that acetaminophen lost crystallinity at 75% RH, below the cited deliquescence point of PEO, and that its compacts took up about twice the water, which could reflect the drug dissolving in the sorbed water; neither point is discussed marsac2008b X10, R3, R13.

Conditions and limits

One loading (10 wt%), one PEO grade (about 100000, not PEG 3350), one high humidity (94% RH; 75% for acetaminophen only) and up to 28 days; the storage temperature of the spectroscopy compacts is not printed marsac2008b X7, X6. The controls that make water and PEO necessary (0% RH compacts, pure drugs at 94% RH, pure-drug water uptake) and the 75% RH result are "data not shown" marsac2008b X7. The aqueous solubilities come from three cited sources at unstated temperatures marsac2008b X18. Whether the drugs recrystallize on drying, and the effect on release, were not examined marsac2008b C8.

Counterexamples and conflicts

The liquid analogue PEG 400 sorted the drugs in the melt study of F3 but not here: all six drugs had \gamma between 0.02 and 0.64 in PEG 400, including the two that kept their crystallinity baird2011 C25; marsac2008b R14, R16. (synthesis: the two studies ask different questions, a molten dispersion at 50–160 °C and a humid compact at about 22 °C with water as a third component, so a PEG 400 stand-in is not a general proxy for the polymer.) The authors' wording differs between sections: the discussion calls the trend with aqueous solubility "a clear trend", the abstract and conclusions "qualitatively correlate" marsac2008b C35. None found in the cards read for this chapter; no other card of the chapter exposed a PEG dispersion to humidity.

Transfer check

For a crystalline drug in a PEG or PEO matrix, compute the activity coefficient in water from the melting point, heat of fusion and aqueous solubility at the storage temperature (ideal solubility from the melting data, then the ratio to the measured solubility). A drug in the range of acetaminophen (about 10¹) is the case to test first; drugs near 10⁵ stayed crystalline here marsac2008b R14, R16. Store compacts at ≥75% RH and 94% RH for weeks and follow the drug by Raman and XRPD against references, quantify the crystalline fraction, and weigh the compacts for water: a jump in water uptake coincided with the largest loss in this study (synthesis: 0.46 against 0.16–0.215 g/g marsac2008b R13). Do not extrapolate from 10 wt% and PEO 100000 to other loadings or PEG 3350.

History

A study outside the library (Schachter and co-workers, 2004) had shown that ketoprofen in PEO lost crystallinity on heating below the PEO melting point, with the drug thought to reside in the amorphous regions marsac2008b R19, C4. Marsac and co-workers (2008) added water as the stress and the activity coefficient in water as the discriminating property, in the earliest paper of this chapter. The later PEG papers of the chapter studied melts and dispersions at 25 °C without humidity (F1–F10).

How it was shown

Pure PEO 100000 was equilibrated at each RH for several hours and for 5 months, and the weight change recorded (instrument, temperature and sample form not printed) marsac2008b P7, X6. Whole compacts were weighed after 28 days at 94% RH over saturated potassium nitrate at 22 °C, and the water was given per dry mass (a dry-basis fraction, which the text reports as wt%) marsac2008b P6, X15. The PEO peaks were followed by XRPD in the compacts before and after storage marsac2008b P4. The crystallinity of the PEO used was estimated from its enthalpy of fusion, 197 J/g against 216 J/g for perfectly crystalline PEO marsac2008b M7, E3.

What the data show

The isotherm points are in the table below. The authors write that PEO takes up significantly more water after 5 months at 75% RH and above; in the plot the large gap is at 94% RH marsac2008b C12, X20. After 28 days at 94% RH pure PEO and the compacts held about 0.16–0.215 g water per g dry solid, acetaminophen compacts about 0.46, which the text gives as roughly 20 and 45 wt% and, normalized to PEO, about 22 and 50 wt% (synthesis: 20/0.9 = 22.2 and 45/0.9 = 50) marsac2008b R13, E6. In XRPD, after 28 days at 94% RH the compacts with ketoprofen, naproxen, ibuprofen, felodipine and ketoconazole showed a decrease in the intensity of the two main PEO peaks (called "dramatic"), a shift to slightly higher angles and clear broadening; with acetaminophen the peaks were weak at 6 days and absent at 17 days, and the compact was a gelatinous mass marsac2008b R9, R17. The PEO used was about 91% crystalline (197/216 = 0.912; the text calls 91% the "amorphous fraction", a misprint, since Table 3 uses 9% amorphous) marsac2008b E3, R15, X1.

Values are read from Fig. 5; the two series were measured at slightly different RH values marsac2008b R11.

Figure

Diamonds are after several hours and circles after 5 months; the curves diverge at 94% RH, about 10 against 37% marsac2008b R11.

Why

The authors' account: water sorbs mostly into the amorphous regions of PEO and increases the disordered fraction, "as evidenced by" the decrease of the PEO peaks, and the very slow equilibration shows that equilibrium states are hard to reach experimentally marsac2008b C28, C29, C7. In the discussion "it is suggested that PEO may slowly deliquesce": the crystalline polymer dissolves in the sorbed water, which keeps lowering water activity and drawing in more water, and the slowness "most likely" reflects the large polymer molecules; support is the peak loss and a cited deliquescence point of water activity 0.93 at 20 °C from a study outside the library marsac2008b C30, R22. The reduced and broadened peaks are "consistent with" deliquescence even if the drug may alter its thermodynamics marsac2008b C31. The softening of the compacts is "qualitatively consistent" with the loss of PEO crystallinity marsac2008b C32. The abstract states that PEO "undergoes a very slow deliquescence" and concludes that it did so, which the discussion's wording does not support; no PEO solution phase was observed marsac2008b X9. The statement that partially crystalline polymers may not allow Arrhenius extrapolation in accelerated stress testing is offered with "may" and without temperature-dependent data marsac2008b C40.

Conditions and limits

PEO 100000 only, not PEG 3350; the isotherm method, temperature and sample form are not printed, nor is the DSC method for the PEO enthalpy of fusion marsac2008b X6, X19. Water contents are given as wt%, weight change and dry-basis fractions in different places, and the type of the error bars and the number of replicates are not stated marsac2008b X15, X16. The traces for day 0 and day 28 are not labelled in panels b–f of the XRPD figures, so the peak changes cannot be checked quantitatively marsac2008b X13. The PEO crystallinity after storage was not measured marsac2008b C28.

Counterexamples and conflicts

The cited deliquescence point (water activity 0.93 at 20 °C) lies above 75% RH, yet acetaminophen lost crystallinity at 75% RH, so deliquescence of PEO is not a precondition of the drug loss marsac2008b R22, R3, X10.

Transfer check

For a PEG or PEO product, follow water uptake and polymer diffraction over weeks to months at the storage humidities of interest, not for hours; a short equilibration at 75–85% RH understates uptake only slightly here, but at 94% RH it understated it almost fourfold (synthesis: 37.3/9.8 = 3.8) marsac2008b R11. (synthesis: expect states at 94% RH that do not exist at moderate humidity, and read high-RH stress data with the polymer's own change in mind marsac2008b R9, R11.)

History

A deliquescence point of PEO and the destruction of polymer crystals by water in poly(vinyl alcohol) were known from studies outside the library marsac2008b R22, C6. Marsac and co-workers (2008) are the only paper of this chapter to have followed PEO under humidity.