Application of Low-Frequency Raman Scattering Spectroscopy toProbe in Situ Drug Solubilization in Milk during DigestionMalinda Salim,† Sara J. Fraser-Miller,‡ Joshua J. Sutton,‡ Karlis Berzins,‡ Adrian Hawley,§

Andrew J. Clulow,† Stephane Beilles,∥ Keith C. Gordon,‡ and Ben J. Boyd*,†

†Drug Delivery, Disposition and Dynamics, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus),381 Royal Parade, Parkville, VIC 3052, Australia‡Dodd-Walls Centre for Photonic and Quantum Technologies, Department of Chemistry, University of Otago, Dunedin 9054, NewZealand§SAXS/WAXS Beamline, Australian Synchrotron, ANSTO, 800 Blackburn Road, Clayton, VIC 3169, Australia∥Sanofi, 371, Rue du Pr. Blayac, 34181 Montpellier cedex04, France

*S Supporting Information

ABSTRACT: We have recently shown that real-timemonitoring of drug solubilization and changes to solid stateof the drug during digestion of milk can be achieved usingsynchrotron small-angle X-ray scattering. A complementarylaboratory-based method to explore such changes is low-frequency Raman spectroscopy, which has been increasinglyused to characterize crystalline drugs and their polymorphs inpowders and suspensions. This study investigates the use ofthis technique to monitor in situ drug solubilization in milkduring the process of digestion, using a lipolysis model/flow-through configuration identical to that used previously for insitu synchrotron small-angle X-ray scattering studies. Anantimalarial drug, ferroquine (SSR97193), was used as themodel drug for this study. The Raman spectra were processed using multivariate analysis to extract the drug signals from themilk digestion background. The results showed disappearance of the ferroquine peaks in the low-frequency Raman region(<200 cm−1) after approximately 15−20 min of digestion when milk fat was present in the system, which indicated drugsolubilization and was in good agreement with the in situ small-angle X-ray scattering measurements. This proof-of-conceptstudy therefore suggests that low-frequency Raman spectroscopy can be used to monitor drug solubilization in a complexdigesting milk medium because of the unique vibrational modes of the drug crystal lattices.

Milk is a natural lipid-based formulation that has beenshown to facilitate solubilization of a range of poorly

water-soluble drugs during digestion1,2 and improve the oralbioavailability.3−6 Digestion of the triglycerides in lipid-basedformulations (including milk) results in the release ofdiglycerides, monoglycerides, and fatty acids (the relativeratios of which depend on the extent of digestion) that form aprogression of colloidal structures7 into which poorly water-soluble compounds in suspension can dissolve.8 This cancircumvent the considerable risk of the drug precipitating outduring digestion due to loss of solubilization capacity when thedrug is predissolved in the lipid formulation.9,10 Techniquesthat enable in situ monitoring of drug solubilization and solid-state transformations (if any) during intestinal digestion aretherefore valuable as these processes typically occur over arelatively short time scale.2,11 Separation of the digested lipid/drug samples by ultracentrifugation and analysis of the amountof drug precipitated in the denser pellet phase using high-performance liquid chromatography (HPLC) have tradition-

ally been used to estimate the distribution of solubilized andnonsolubilized drug, based on the assumption that excessnonsolubilized drug was partitioned to the pellet phase aftercentrifugation.12 This approach is time-consuming and lesswell suited to providing in situ information on the solid-stateform of drugs during digestion. Consequently, approacheswere developed to enable in situ determination of drugprecipitation and solubilization during digestion of lipid-basedformulations using synchrotron small-angle X-ray scattering(SAXS).2,13 By tracking the increase or reduction of intensityor shifts in positions of characteristic diffraction peaks from thedrugs to different scattering angles, the precipitation11 andsolubilization or polymorphic transformations,2,11,14 respec-tively, were able to be quantified. However, this techniquerequires a high-intensity synchrotron source to provide the

Received: March 7, 2019Accepted: April 18, 2019Published: April 23, 2019

Letter

pubs.acs.org/JPCLCite This: J. Phys. Chem. Lett. 2019, 10, 2258−2263

© 2019 American Chemical Society 2258 DOI: 10.1021/acs.jpclett.9b00654J. Phys. Chem. Lett. 2019, 10, 2258−2263

This is an open access article published under a Creative Commons Attribution (CC-BY)License, which permits unrestricted use, distribution and reproduction in any medium,provided the author and source are cited.

Dow

nloa

ded

by 1

30.1

94.1

44.1

36 a

t 00:

44:5

8:33

1 on

May

23,

201

9fr

om h

ttps:

//pub

s.ac

s.or

g/do

i/10.

1021

/acs

.jpcl

ett.9

b006

54.

necessary sensitivity and time resolution to conduct suchstudies and thus is not amenable to routine use in formulationdevelopment and screening. Alternative approaches that aresensitive to the presence of crystalline material and havesufficient time resolution are therefore required.Raman spectroscopy is a light-scattering technique that has

been widely used to characterize the solid-state forms of activepharmaceutical ingredients (APIs) in the pharmaceuticalindustry because of its rapid analysis and nondestructivenature.15 It has also been used to monitor drug precipitationduring in vitro lipolysis of a self-microemulsifying drug deliverysystem.16 Measurements typically acquire spectral informationin the 200−1800 cm−1

fingerprint region to probe changes inmolecular structures by characterizing the intramolecularvibrations that are sensitive to local functional groupenvironments.17,18 Recent advancements in filter technologiessuch as volume Bragg gratings have allowed spectralacquisitions close to the laser frequency (within 10 cm−1) indispersive Raman systems.19 This low-frequency region (<200cm−1) can provide direct information on the intermolecularinteractions within different solid-state forms of com-pounds.19,20 Intense Raman peaks related to vibrationalmodes of the crystal lattices (phonons) were observed incrystalline samples, while amorphous compounds that lackorder typically give a broad peak.20 As a result, this techniquehas been increasingly used in recent years to detect solid-statetransformation of drugs as powdered and suspension samplesfollowing heat treatment,21 quench cooling,19 or grinding;22 tostudy the kinetics of amorphization of drugs during milling;23

and to identify different crystals in powder mixtures.24

However, to our knowledge, low-frequency Raman spectros-copy has not been used to track crystallinity and solubilizationof poorly water-soluble drugs during digestion of lipid-basedformulations. Herein, we explore the use of low-frequencyRaman spectroscopy to probe the solubilization of a lipophilic

drug in real time during digestion of milk under simulatedsmall intestinal conditions. Ferroquine, a poorly water-solubleweakly basic antimalarial drug with pKa’s of ∼7.0 and 8.5 wasused as the model drug in this study.25

In vitro digestion of full cream bovine milk and buffercontaining ferroquine was performed using the apparatusdepicted in Figure 1 with experimental details being providedin the Supporting Information. Briefly, ferroquine (granulescontaining 50 wt % API provided by Sanofi) was dispersed inmilk or buffer as a suspension (186 mg granules in 2.75 mL ofwater and 17.5 mL of milk, which is approximately equivalentto a clinical dose in a glass of milk).14 The nutritionalinformation on the bovine milk used in this study issummarized in Table S1 (Supporting Information). In anadditional experiment, pretreatment of ferroquine with hydro-chloric acid solution prior to the addition of milk was alsocarried out to mimic transiting through the gastric environ-ment as was performed in the previous X-ray scatteringstudies.2 The samples were circulated through a standingquartz capillary (1.5 mm outer diameter) for 2 min prior toinitiation of digestion. Digestion was initiated by injection of2.25 mL of lipase suspension in tris buffer as describedpreviously.2 Raman spectra were recorded over a −360 to 2030cm−1 spectral window, with 5−7 cm−1 resolution, using a 785nm excitation laser and a 135° backscattering geometry relativeto the collective lens.19,23 Spectra were collected every 0.5 minfor a total of 45 min (60 accumulations of 0.5 s exposure time).Further details of the experimental Raman setup andmeasurements are outlined in the Supporting Information.The Raman spectrum of ferroquine exhibited strong phonon

peaks at 34 and 46 cm−1(Figure 2a) that are considered to beassociated with lattice vibrations of highly ordered ferroquinecrystals.20 These low-frequency peaks were still visuallyobservable after dispersion in tris buffer and milk (Figure2c−e) at a ferroquine concentration of about 4.6 mg mL−1

Figure 1. Schematic of the experimental configuration for in situ measurements of low-frequency Raman spectra during the digestion of milkcontaining ferroquine in suspension. It is noteworthy that this configuration has an identical lipolysis apparatus, pump, tubing, and capillary used inpreviously reported diffraction studies,2 and as such the findings are directly comparable.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.9b00654J. Phys. Chem. Lett. 2019, 10, 2258−2263

2259

with no formation of new peaks or peak shifts observed during

stirring. This indicated that no solid-state transformations had

occurred during dispersion of the solid drug in the buffer or

milk media. The ability to detect low-frequency Raman spectra

of drugs in suspension has also been demonstrated by Larkin et

al. with indomethacin, albeit not to monitor a process or asuspension in a complex medium like milk.26

Addition of lipase to start lipolysis of the fat component inmilk caused a reduction in the intensities of the distinctiveRaman scattering peaks of ferroquine, and the peakssubsequently disappeared during the course of digestion

Figure 2. Low-frequency Raman spectra of (a) ferroquine (FQ) reference powder and the (b) milk background before and after digestion at pH6.5 in the absence of ferroquine. Panels c−e show the Raman spectra of suspensions of ferroquine powder before and after digestion at pH 6.5 in(c) tris buffer, (d) milk, and (e) milk with ferroquine that has been pretreated with hydrochloric acid solution. The arrows point to the positions ofthe Raman peaks associated with crystalline ferroquine.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.9b00654J. Phys. Chem. Lett. 2019, 10, 2258−2263

2260

(Figure 2d,e). However, in tris buffer that contained nodigestible lipids, there was no change in the intensity of theseRaman bands on the addition of lipase (Figure 2c).Disappearance of the peaks suggested loss of long-rangeorder of the crystalline drug that could be attributed to drugsolubilization and/or formation of an amorphous complex withthe digested lipids. Inclusion of a gastric step where theferroquine was exposed to hydrochloric acid prior to intestinaldigestion also showed complete disappearance of the low-frequency Raman peaks after digestion. Less drug was initiallypresent in crystalline form during dispersion after gastricpretreatment because of the greater solubility of basicferroquine in the acidic environment (see Figure S1 in theSupporting Information). The presence of lipids and theprocess of digestion were therefore critical to the solubilizationof ferroquine during digestion.

A closer examination of the Raman spectra was subsequentlyperformed using principal component analysis (PCA) withOrange software version 3.16 to qualitatively interpret spectralvariations between the samples.27 Linear baseline correctionwas applied to the Raman plots between −250 and 250 cm−1,and the spectral region between 8 and 200 cm−1 was subjectedto vector normalization prior to PCA. The plot of the first twoprincipal components (which covered 95% of the samplevariance with 56% in PC1) in Figure 3a showed separateclusters for milk and milk with ferroquine as well as ferroquinein tris buffer due to spectral differences between the samples.Inspection of the loadings plot (shown in Figure 3b) revealedthat ferroquine peaks were correlated with positive PC1, whileno distinctive separation between ferroquine and the milksignals could be obtained from PC2 (Figure S2 in theSupporting Information). It was clear that all the samplescontaining crystalline ferroquine were clustered in positive

Figure 3. (a) Two-dimensional PCA scores plot for milk and milk/tris-containing ferroquine (FQ) based on analysis of the low-frequency Ramanshift region from 8 to 200 cm−1. Circled regions belong to clusters during dispersion, i.e., before lipase injection. (b) Corresponding loadings plotfor PC1 (56% of sample variance) that described the ferroquine signals. (c) Plot of PC1 against dispersion (<0 min) and digestion (>0 min) timefor all of the samples. (d) Percentage crystalline ferroquine in milk with gastric pretreatment during dispersion (<0 min) and digestion (>0 min)analyzed using SAXS. Values were determined from area under the diffraction peak associated with ferroquine at q = 1.30 Å−1, which wasnormalized to 100% of crystalline drug being present during dispersion. Lipase was injected at time = 0 min to initiate digestion. The experimentalconfiguration for the SAXS runs was as previously described and used apparatus identical to the Raman scattering measurements (Figure 1).2

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.9b00654J. Phys. Chem. Lett. 2019, 10, 2258−2263

2261

PC1 values prior to digestion, and no changes were seen intris/ferroquine samples with the progress of digestion(signifying no changes to the drug signal). Shifts in PC1toward negative values were observed for the digesting milk/ferroquine samples. A plot of PC1 against dispersion (time < 0min, suspension stirring before lipase injection) and digestion(time > 0 min, after addition of lipase) time in Figure 3cshowed a gradual decrease in PC1 for the milk/ferroquinesamples with and without gastric steps, which then plateauedaround 15−20 min after lipase addition. Solubilization offerroquine may therefore have reached completion within thistime.Although the fat content of the full-cream milk used for the

Raman studies was slightly lower than that for the equivalentX-ray scattering studies (3.3% vs 3.8%, due to differences inthe fat content of commercially available milk in New Zealandand Australia, respectively), the results obtained from theanalysis of the low-frequency Raman spectra were in goodagreement with the SAXS data where diffraction peaksassociated with crystalline ferroquine also disappeared afterabout 15−20 min of digestion (Figure 3d). This signifies that0.58 g of milk fat (17.5 mL of 3.3% milk fat) is sufficient toprovide complete solubilization of 93 mg of ferroquine API(equivalent to 186 mg of ferroquine granules), as verified bythe SAXS measurements. Another interesting observation wasthat analogous to differences between the ferroquine peakintensities between panels c and d of Figure 2, reduced PC1values were also observed during dispersion in the milk/ferroquine sample that has been pretreated with hydrochloricacid. It is also worth noting that although the PC1 values in thetris/ferroquine sample stayed relatively constant throughoutthe digestion, a slight decrease in the PC1 value was observedafter lipase injection, which could arise because of the slightdilution of the ferroquine as lipase suspension contributed∼10% of the total digestion volume.Conclusions. We have demonstrated that low-frequency

Raman spectroscopy can be used to monitor in situsolubilization of suspended ferroquine in milk during digestion.The presence of digestible fat in milk was necessary to achievedrug solubilization. Complete disappearance of the phononpeaks in the low-frequency Raman spectra was observed afterabout 15−20 min of digestion in milk when analyzed usingmultivariate analysis. This indicates complete drug solubiliza-tion, which is in agreement with previously obtainedsynchrotron small-angle X-ray scattering data. Further studiesare now underway to investigate the wider utilization of thistechnique to monitor solubilization of other types of drugs incomplex formulations (such as those containing excipients andbile salts) with simultaneous detection of solid-state trans-formations.

■ ASSOCIATED CONTENT

*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jp-clett.9b00654.

Experimental methods for flow-through in vitro lipolysisand low-frequency Raman spectroscopy, nutritionalinformation for the milk used in the Raman scatteringand SAXS measurements, loadings plot for PC2, andpH-dependent solubility of ferroquine (PDF)

■ AUTHOR INFORMATIONCorresponding Author*Monash Institute of Pharmaceutical Sciences, MonashUniversity (Parkville Campus), 381 Royal Parade, Parkville,VIC 3052, Australia. Tel.: +61 3 99039112. Fax: +61 399039583. E-mail: ben.boyd@monash.edu.ORCIDKarlis Berzins: 0000-0001-6545-5522Andrew J. Clulow: 0000-0003-2037-853XKeith C. Gordon: 0000-0003-2833-6166Ben J. Boyd: 0000-0001-5434-590XNotesThe authors declare the following competing financialinterest(s): Stephane Beilles is an employee of Sanofi, whichowns the model drug ferroquine used in these studies.

■ ACKNOWLEDGMENTSThis work was funded by the Bill and Melinda GatesFoundation Grant Number [OPP1160404]. Funding is alsoacknowledged from the Australian Research Council under theDiscovery Projects scheme (DP160102906), and A.J.C. is therecipient of a Discovery Early Career Research Award(DE190100531). The SAXS experiments for this work wereconducted on the SAXS/WAXS beamline of the AustralianSynchrotron, part of ANSTO. K.B. and J.J.S. thank theUniversity of Otago for Ph.D. scholarships. The support of theDodd-Walls Centre for Photonic and Quantum Technologiesis gratefully acknowledged.

■ REFERENCES(1) Boyd, B. J.; Salim, M.; Clulow, A. J.; Ramirez, G.; Pham, A. C.;Hawley, A. The impact of digestion is essential to the understandingof milk as a drug delivery system for poorly water soluble drugs. J.Controlled Release 2018, 292, 13−17.(2) Salim, M.; Khan, J.; Ramirez, G.; Clulow, A. J.; Hawley, A.;Ramachandruni, H.; Boyd, B. J. Interactions of Artefenomel (OZ439)with Milk during Digestion: Insights into Digestion-DrivenSolubilization and Polymorphic Transformations. Mol. Pharmaceutics2018, 15 (8), 3535−3544.(3) Batty, K. T.; Salman, S.; Moore, B. R.; Benjamin, J.; Lee, S. T.;Page-Sharp, M.; Pitus, N.; Ilett, K. F.; Mueller, I.; Hombhanje, F. W.;Siba, P.; Davis, T. M. E. Artemisinin-naphthoquine combinationtherapy for uncomplicated pediatric malaria: a pharmacokinetic study.Antimicrob. Agents Chemother. 2012, 56 (5), 2472−2484.(4) Darpo, B.; Ferber, G.; Siegl, P.; Laurijssens, B.; Macintyre, F.;Toovey, S.; Duparc, S. Evaluation of the QT effect of a combinationof piperaquine and a novel anti-malarial drug candidate OZ439, forthe treatment of uncomplicated malaria. Br. J. Clin. Pharmacol. 2015,80 (4), 706−715.(5) Macheras, P. E.; Reppas, C. I. Studies on Drug-Milk Freeze-Dried Formulations I: Bioavailability of Sulfamethizole and DicumarolFormulations. J. Pharm. Sci. 1986, 75 (7), 692−696.(6) Charkoftaki, G.; Kytariolos, J.; Macheras, P. Novel milk-basedoral formulations: Proof of concept. Int. J. Pharm. 2010, 390 (2),150−159.(7) Carey, M. C.; Small, D. M.; Bliss, C. M. Lipid Digestion andAbsorption. Annu. Rev. Physiol. 1983, 45 (1), 651−677.(8) Porter, C. J.; Trevaskis, N. L.; Charman, W. N. Lipids and lipid-based formulations: optimizing the oral delivery of lipophilic drugs.Nat. Rev. Drug Discovery 2007, 6 (3), 231.(9) Khan, J.; Rades, T.; Boyd, B. The Precipitation Behavior ofPoorly Water-Soluble Drugs with an Emphasis on the Digestion ofLipid Based Formulations. Pharm. Res. 2016, 33 (3), 548−562.(10) Sassene, P. J.; Michaelsen, M. H.; Mosgaard, M. D.; Jensen, M.K.; Van Den Broek, E.; Wasan, K. M.; Mu, H.; Rades, T.; Mullertz, A.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.9b00654J. Phys. Chem. Lett. 2019, 10, 2258−2263

2262

In Vivo Precipitation of Poorly Soluble Drugs from Lipid-Based DrugDelivery Systems. Mol. Pharmaceutics 2016, 13 (10), 3417−3426.(11) Boetker, J.; Rades, T.; Rantanen, J.; Hawley, A.; Boyd, B. J.Structural Elucidation of Rapid Solution-Mediated Phase Transitionsin Pharmaceutical Solids Using in Situ Synchrotron SAXS/WAXS.Mol. Pharmaceutics 2012, 9 (9), 2787−2791.(12) Dressman, J. B.; Vertzoni, M.; Goumas, K.; Reppas, C.Estimating drug solubility in the gastrointestinal tract. Adv. DrugDelivery Rev. 2007, 59 (7), 591−602.(13) Khan, J.; Hawley, A.; Rades, T.; Boyd, B. J. In Situ Lipolysis andSynchrotron Small-Angle X-ray Scattering for the Direct Determi-nation of the Precipitation and Solid-State Form of a Poorly Water-Soluble Drug During Digestion of a Lipid-Based Formulation. J.Pharm. Sci. 2016, 105 (9), 2631−2639.(14) Salim, M.; Khan, J.; Ramirez, G.; Murshed, M.; Clulow, A. J.;Hawley, A.; Ramachandruni, H.; Beilles, S.; Boyd, B. J. Impact ofFerroquine on the Solubilization of Artefenomel (OZ439) during inVitro Lipolysis in Milk and Implications for Oral CombinationTherapy for Malaria. Mol. Pharmaceutics 2019, 16, 1658−1668.(15) Strachan, C. J.; Rades, T.; Gordon, K. C.; Rantanen, J. Ramanspectroscopy for quantitative analysis of pharmaceutical solids. J.Pharm. Pharmacol. 2007, 59 (2), 179−192.(16) Stillhart, C.; Imanidis, G.; Kuentz, M. Insights into DrugPrecipitation Kinetics during In Vitro Digestion of a Lipid-Based DrugDelivery System Using In-Line Raman Spectroscopy and Mathemat-ical Modeling. Pharm. Res. 2013, 30 (12), 3114−3130.(17) Savolainen, M.; Kogermann, K.; Heinz, A.; Aaltonen, J.;Peltonen, L.; Strachan, C.; Yliruusi, J. Better understanding ofdissolution behaviour of amorphous drugs by in situ solid-stateanalysis using Raman spectroscopy. Eur. J. Pharm. Biopharm. 2009, 71(1), 71−79.(18) Lipiainen, T.; Fraser-Miller, S. J.; Gordon, K. C.; Strachan, C. J.Direct comparison of low- and mid-frequency Raman spectroscopyfor quantitative solid-state pharmaceutical analysis. J. Pharm. Biomed.Anal. 2018, 149, 343−350.(19) Mah, P. T.; Fraser, S. J.; Reish, M. E.; Rades, T.; Gordon, K. C.;Strachan, C. J. Use of low-frequency Raman spectroscopy andchemometrics for the quantification of crystallinity in amorphousgriseofulvin tablets. Vib. Spectrosc. 2015, 77, 10−16.(20) Larkin, P. J.; Dabros, M.; Sarsfield, B.; Chan, E.; Carriere, J. T.;Smith, B. C. Polymorph Characterization of Active PharmaceuticalIngredients (APIs) Using Low-Frequency Raman Spectroscopy. Appl.Spectrosc. 2014, 68 (7), 758−776.(21) Inoue, M.; Hisada, H.; Koide, T.; Carriere, J.; Heyler, R.;Fukami, T. In Situ Monitoring of Crystalline Transformation ofCarbamazepine Using Probe-Type Low-Frequency Raman Spectros-copy. Org. Process Res. Dev. 2017, 21 (2), 262−265.(22) Hedoux, A.; Paccou, L.; Guinet, Y.; Willart, J.-F.; Descamps, M.Using the low-frequency Raman spectroscopy to analyze thecrystallization of amorphous indomethacin. Eur. J. Pharm. Sci. 2009,38 (2), 156−164.(23) Walker, G.; Romann, P.; Poller, B.; Lobmann, K.; Grohganz,H.; Rooney, J. S.; Huff, G. S.; Smith, G. P. S.; Rades, T.; Gordon, K.C.; Strachan, C. J.; Fraser-Miller, S. J. Probing PharmaceuticalMixtures during Milling: The Potency of Low-Frequency RamanSpectroscopy in Identifying Disorder. Mol. Pharmaceutics 2017, 14(12), 4675−4684.(24) Hisada, H.; Inoue, M.; Koide, T.; Carriere, J.; Heyler, R.;Fukami, T. Direct High-Resolution Imaging of Crystalline Compo-nents in Pharmaceutical Dosage Forms Using Low-Frequency RamanSpectroscopy. Org. Process Res. Dev. 2015, 19 (11), 1796−1798.(25) Kondratskyi, A.; Kondratska, K.; Vanden Abeele, F.;Gordienko, D.; Dubois, C.; Toillon, R.-A.; Slomianny, C.; Lemiere,S.; Delcourt, P.; Dewailly, E.; Skryma, R.; Biot, C.; Prevarskaya, N.Ferroquine, the next generation antimalarial drug, has antitumoractivity. Sci. Rep. 2017, 7 (1), 15896.(26) Larkin, P. J.; Wasylyk, J.; Raglione, M. Application of Low- andMid-Frequency Raman Spectroscopy to Characterize the Amorphous-

Crystalline Transformation of Indomethacin. Appl. Spectrosc. 2015, 69(11), 1217−1228.(27) Demsar, J.; Curk, T.; Erjavec, A.; Gorup, C.; Hocevar, T.;Milutinovic, M.; Mozina, M.; Polajnar, M.; Toplak, M.; Staric, A.Orange: data mining toolbox in Python. J. Mach. Learn Res. 2013, 14(1), 2349−2353.

The Journal of Physical Chemistry Letters Letter

DOI: 10.1021/acs.jpclett.9b00654J. Phys. Chem. Lett. 2019, 10, 2258−2263

2263