Rr

KD

a

ARRAA

KAECFMT

1

pdAaai[puacm

nwgcSan

h0

Thermochimica Acta 582 (2014) 77–85

Contents lists available at ScienceDirect

Thermochimica Acta

j ourna l h om epage: www.elsev ier .com/ locate / tca

ole of thermodynamic, kinetic and structural factors in theecrystallization behavior of amorphous erythromycin salts

apil Nanakwani, Sameer R. Modi, Lokesh Kumar, Arvind K. Bansal ∗

epartment of Pharmaceutics, National Institute of Pharmaceutical Education and Research (NIPER), SAS Nagar, Mohali, Punjab 160062, India

r t i c l e i n f o

rticle history:eceived 29 November 2013eceived in revised form 27 February 2014ccepted 2 March 2014vailable online 10 March 2014

eywords:

a b s t r a c t

Amorphous form has become an important drug delivery strategy for poorly water soluble drugs. How-ever, amorphous form has inherent physical instability due to its tendency to recrystallize to stablecrystalline form. In the present study, amorphous forms of erythromycin free base (ED) and its saltsnamely, stearate (ES), phosphate (EP) and thiocyanate (ET) were generated by in situ melt quenching andevaluated for their crystallization tendency. Salts were characterized for kinetic, thermodynamic andstructural factors to understand crystallization behavior. Kinetics of crystallization followed the order

morphous saltsrythromycinrystallization kineticsragilityolecular mobility

as ES > EP > ET > ED. Fragility and molecular mobility does not completely explain these findings. How-ever, configurational entropy (Sconf), indicative of entropic barrier to crystallization, followed the orderas ET > EP > ES > ED. Lower crystallization tendency of ED can be explained by its lower thermodynamicdriving force for crystallization (Hconf). This correlated well with different structural parameters for thecounter ions.

hermodynamic factors

. Introduction

Current drug discovery approaches, including high through-ut screening and combinatorial chemistry, have led to newrug candidates having poor biopharmaceutical performance [1].mongst all the strategies, salt formation is one of the primepproach, to optimize the physicochemical properties of ioniz-ble drug candidates [2,3]. Similarly, amorphous form generations another delivery platform for water insoluble drug candidates4–6]. Recently, a combined approach of salt formation and amor-hous form generation, namely, amorphous salt form, has beentilized to harness the advantages of both the salt form as wells of the amorphous form [7,8]. Atorvastatin calcium, rosuvastatinalcium and quinapril hydrochloride are some examples of com-ercially available amorphous salt forms.Physical stability of an amorphous salt form is affected by the

ature and type of the counterion. For example, amorphous saltsith strongly interacting counterions generally have a higher

lass transition temperature (Tg) and improved physical stability,ompared to amorphous form of the unionized drug [9–13].

imilarly, the recrystallization behavior from the amorphous formlso differs in different salt forms [10–12,14]. Kinetic, thermody-amic and structural factors contribute to the physical stability of

∗ Corresponding author. Tel.: +91 172 2214682×2126; fax: +91 172 2214692.E-mail addresses: akbansal@niper.ac.in, bansalarvind@yahoo.com (A.K. Bansal).

ttp://dx.doi.org/10.1016/j.tca.2014.03.001040-6031/© 2014 Elsevier B.V. All rights reserved.

© 2014 Elsevier B.V. All rights reserved.

amorphous form. Effect of counterions on these factors has beenstudied, with emphasis on ionic interactions [15], ionic radius[11], pKa of the counterion [9,12], electrophilicity index [9,12] andhydrogen bonding [14]. Previous work from our lab reported thecrystallization behavior of amorphous atorvastatin salts [10] andamorphous prazosin salts [14]. It was observed that the strengthof ionic interaction affects the Tg and crystallization behavior ofdrug salts. However, these studies primarily evaluated the effectof kinetic factors (i.e. molecular mobility) on the crystallizationbehavior of amorphous salts, without assessing the role of ther-modynamic factors. This was due to concomitant degradation ofdrug salts along with melting, thus precluding the determinationof thermodynamic parameters.

The present work is aimed at evaluating the role of counterion onboth kinetic and thermodynamic factors, and their impact on phys-ical stability of amorphous salts. Erythromycin dihydrate free base(ED) and its salts, namely, erythromycin stearate (ES), erythromycinphosphate (EP) and erythromycin thiocyanate (ET) were selectedfor this study. Erythromycins (Fig. 1) are macrolide antibiotics, andcontains a 14-membered lactone ring with ten asymmetric centersand two sugars (l-cladinose and d-desosamine) [16]. Salt forma-tion in erythromycins involves generation of a positive characteron the tertiary amino group, thus rendering a basic character to the

desosamine ring of erythromycin (pKa 8.8). Different erythromycinsalts have been reported with counterions such as stearate, phos-phate, thiocyanate, gluceptate, sulfamate, lactobionate and ethylphosphate [16].

78 K. Nanakwani et al. / Thermochimica Acta 582 (2014) 77–85

O

CH3

O

CH3

CH3

CH3

O

H3C

HO

HO

CH3

CH3

O

O

O

H3C N

OH

CH3

CH3

O

OH

OH3C

CH3

HO

H3C

o gro

2

2

mPar

2

(5hfriatv

2

dSa2

2

D

Fig. 1. Structure of erythromycin (encircled portion shows the tertiary amin

. Materials and methods

.1. Materials

ED and EP were received as gratis samples from Century Phar-aceuticals Limited, India, and ES and ET were received from Mehta

harmaceutical Industries, India. All samples had purity of >99.5%nd were used ‘as received’. All other chemicals were of analyticaleagent grade.

.2. Differential scanning calorimetry (DSC)

DSC heating curves of samples were obtained using DSC Q2000TA Instruments, USA). DSC cell was purged with dry nitrogen at0 ml min−1. Instrument was calibrated prior to analysis, usingigh purity standard of indium and sapphire (aluminum oxide)

or temperature-heat flow and heat capacity (Cp) measurement,espectively. Accurately weighed sample (5–8 mg) were crimpedn aluminum pans and run in the temperature range of 298–483 K,t 10 K min−1 heating rate. Tg was reported as the onset tempera-ure, while endothermic transitions were reported as the midpointalue.

.3. Thermogravimetric analysis (TGA)

Weight loss of samples as a function of temperature wasetermined using Mettler Toledo 851e TGA (Mettler Toledo,witzerland). Samples were placed in alumina crucibles and heatedt 10 K min−1, under nitrogen purge (50 ml min−1), in the range of98–483 K.

.4. Hot stage microscopy (HSM)

Thermal behavior of samples was observed using LeicaMLP polarized microscope (Leica Microsystems Wetzlar GmbH,

up that provides a basic character to the desosamine ring of erythromycin).

Germany), equipped with Linkam LTS 350 hot stage. Photomicro-graphs were acquired using JVS color video camera and analyzedusing Linksys32 software. Samples were mounted in silicone oiland heated from 298 K to 483 K at 10 K min−1.

2.5. Optical and polarized light microscopy

Particle characteristics were assessed by optical and polarizedmicroscopy using a Leica DMLP polarized light microscope (LeicaMicrosystems, Germany). Photomicrographs were acquired usingJVS color video camera and processed using Linksys® software.

2.6. Karl Fischer (KF) titration

Water content was determined by KF titration, using pyridine-free reagents and the dead stop end point method (Karl Fischertitrator 794 Basic Titrino, automatic burette 794 for presentation ofsolvent, 703 Ti stand; Metrohm AG, Switzerland). Analytical gradedisodium tartrate dihydrate (15.65% water content) was used asthe KF standard.

2.7. Powder X-ray diffraction (PXRD)

PXRD pattern of samples was recorded at room temperatureon Bruker’s D8 advance Diffractometer (Bruker, Germany) with a2� compensating slit and CuK� radiation (1.54 A), at 40 kV, 40 mApassing through nickel filter. Analysis was performed in a continu-ous mode with a step size of 0.01◦ and scan rate of 1 s/step over anangular range of 3–40◦ 2�. Obtained diffractograms were analyzedwith DIFFRACplus EVA® (version 9.0) diffraction software.

2.8. Generation of amorphous forms

Amorphous ED and its salts were freshly prepared using in situmelt quenching in DSC. Crystalline samples were heated to 5 K

K. Nanakwani et al. / Thermochimica Acta 582 (2014) 77–85 79

rve an

ar

2

wcuftas

2

rirfdm

2

fta(clt

2

a[p

Fig. 2. DSC heating cu

bove their respective melting points at 10 K min−1, followed byapid cooling to 253 K.

.9. Crystallization kinetics

Amorphous samples, as prepared by in situ melt quenching,ere kept at Tg − 20 K + 0% RH, to achieve common degree of under-

ooling, for 1, 2, 4 and 8 days. Humidity condition was generatedsing dry phosphorous pentoxide (0% RH). Samples were analyzedor crystallinity using microscopy, DSC and PXRD. Percentage crys-allinity was determined from the heat capacity difference (�Cp)nd/or enthalpy of fusion (�Hf), for amorphous and crystallineamples respectively.

.10. Fragility and molecular mobility

Fragility of amorphous forms was determined from the heatingate dependence of Tg. Samples were heated to 5 K above their melt-ng points, rapidly cooled to 253 K and then reheated at differentates of 5, 10, 15, 20, 25 and 30 K min−1. Experiments were per-ormed in triplicate and the onset of Tg value was determined. Thisata was further used to draw Angell’s plot and calculate molecularobility with respect to temperature.

.11. Heat capacity (Cp) measurements

Modulated temperature DSC was used to measure the Cp at dif-erent temperatures. Samples were heated in the range of Tg − 20 Ko Tg + 20 K with a modulation period of 60 s, amplitude of ±0.16 Knd an underlying heating rate of 1 K min−1. Crystalline samples5–8 mg) were compressed into discs to ensure good sample to panontact. Cp measurement for each sample was carried out in trip-icate and reversing Cp was used to calculate the configurationalhermodynamic parameters.

.12. Molecular descriptors

Density Functional Theory (DFT) parameters for phosphatend thiocyanate counterions were already reported in literature12,17]. Calculation of DFT parameters for stearate counterion waserformed using Gaussian 03 package. The method and basis set

d HSM images of ED.

utilized was B3LYP/6-31G (d). Electronic properties of stearic acidwere calculated using a single energy calculation for the neutraland ionized forms of stearic acid, following full optimization (14).

3. Results

3.1. Preliminary characterization of the crystalline andamorphous form

Fig. 2 shows the DSC heating curve of crystalline ED. It showeda broad endotherm at about 348–356 K, attributed to water loss(corresponding to 4% weight loss in TGA equivalent to dihydratestoichiometry). Water loss was followed by recrystallization ofdehydrated form of ED (EDD) at 383 K and then melting at around408 K. Upon further heating, recrystallization exotherm attributedto conversion to erythromycin anhydrate (EA) was observed ataround 418 K, followed by its melting at 463 K. This behaviormatches with already reported thermal behavior of ED [18–20].

DSC heating scans for ES, EP and ET showed melting endothermsat 376 K, 424 K and 434 K respectively (Fig. 3). All thermal transi-tions could be observed over the same temperature range in HSM.

Amorphous salts were generated by in situ melt quenching. Halopattern in the PXRD and an absence of birefringence in polarizedlight microscopy confirmed formation of amorphous form. Tg forED, ES, EP and ET were found to be 385 K, 306 K, 389 K and 396 Krespectively.

3.2. Crystallization kinetics

Amorphous ED and its salt forms were evaluated for their physi-cal stability at stress condition of Tg − 20 K + 0% RH by storing for 1, 2,4 and 8 days. PXRD patterns at Tg − 20 K + 0% RH at time periods of 1,2, 4 and 8 days are shown in Fig. 4. The percent crystallinity of thesesamples was calculated from heat of fusion (�Hf) values of partiallycrystallized material in DSC heating scans. At Tg − 20 K + 0% RH, ESshowed the least physical stability with the percent crystallinityof 22% after 8 days. ED was the most stable, with no crystallinity

even after 8 days. In contrast, EP and ET showed partial crystallinityof 16% and 14% after 8 days at similar conditions. Thus, kineticsof crystallization followed the order: ES > EP > ET > ED. This differ-ential kinetics of crystallization at Tg − 20 K + 0% RH was further

80 K. Nanakwani et al. / Thermochimica Acta 582 (2014) 77–85

ating

ip

3

epaad

Fig. 3. Overlay of DSC he

nvestigated for possible role of different factors affecting theirhysical stability.

.3. Fictive temperature

Fictive temperature (Tf) represents the temperature at whichquilibrium supercooled liquid possess the same thermodynamic

roperties as of the non-equilibrium glass. Tf is not a single temper-ture, but varies with the corresponding temperature of the glass,nd has the mathematical boundation as Tk < Tf < Tg. Tf is indepen-ent of DSC heating rate, and thus can be used as a reliable quality

Fig. 4. PXRD pattern of amorphous form of model dr

curves of ES, EP and ET.

control tool for amorphous pharmaceuticals [21]. Tf can be calcu-lated by the following equations (Eqs. (1) and (2)).

1Tf

=(

Cp

Tg

)+ (1 − Cp)

T(1)

where,

Clp − Cg

p

�Cp =

Clp − CX

p

(2)

where Clp, Cg

p and CXp denote the heat capacity of equilibrium super-

cooled liquid, glassy and crystalline state respectively. �Cp denotes

ugs during crystallization studies at Tg − 20 K.

K. Nanakwani et al. / Thermochimica Acta 582 (2014) 77–85 81

Table 1Summary of different parameters affecting crystallization kinetics of model drugs.

Drug pKa of counter ion Tm (K) Tg (K) �Cp Tm/Tg �Cp (J mol−1 K−1) Ea (KJ mol−1) Tk (K) Tk/Tg Tg − Tk m D

ED – 408.95 385.99 0.65 1.06 231.29 342.28 252.64 0.65 133.34 46.31 19.44463.85

ES 4.90 376.29 306.49 0.57 1.22 312.03 376.16 229.97 0.75 76.51 64.09 12.25EP 2.16, 7.21 12.32 424.27 389.16 0.56 1.11 246.99 416.94 277.88 0.71 111.27 55.95 14.75

99

T vation

ttAtbrl

3

Aoda

wp�c

daan

�

wiawb

p

ET −1.28 434.71 396.50 0.46 1.10 209.

m: melting point, Tg: glass transition temperature, �Cp: heat capacity jump, Ea: acti

he relative differences in heat capacity of crystal, real glass andhe equilibrium supercooled liquid. Higher value of �Cp indicatesrrhenius behavior of amorphous system and corresponding lower

emperature dependency of molecular mobility at temperatureselow Tg [22]. �Cp values at 298 K for all the four compounds areepresented in Table 1. Highest value of �Cp for ED indicates itsower temperature dependency of molecular mobility below Tg.

.4. Fragility

Angell introduced the concept of fragility. Strong liquids showrrhenius relationship, with activation energy being independentf temperature. On the other hand, fragile liquids show non-linearependence of relaxation time upon temperature, with a temper-ture dependent activation energy [23].

Heat capacity jump (�Cp) at Tg is used to describe fragility,herein, ‘fragile’ liquids show significantly higher jump as com-ared to the ‘strong’ liquids. Amorphous ES, EP, ED and ET exhibitedCp values of 312, 246, 231 and 209 J mol−1 K−1 respectively, indi-

ating ES to be the most fragile (Table 1).Fragility of drug salts was also calculated from the heating rate

ependence of Tg (Fig. 5). Activation energy (Ea) for structural relax-tion at Tg was determined (Table 1) from the Arrhenius equationnd from the slope of graph ln q (K min−1) vs. 1000/Tg (Fig. 5). Arrhe-ius behavior is given by the following relation (Eq. (3)) [24]:

(Tg) = �0 exp

[−Ea(Tg)

RTg

](3)

here �(Tg) is the viscosity at Tg, �0 is the viscosity value at annfinitely high temperature, Ea (Tg) refers to the activation energyt Tg and R is a universal gas constant. Ea for molecular mobilityas the least for ES, when compared to ET and EP, indicating ES to

e more fragile as compared to EP and ET.Fragility parameter (m) is another measure of fragility of a com-

ound, with higher values indicating higher fragility behavior. This

0

0.5

1

1.5

2

2.5

3

3.5

4

2.4 2.6 2.8 3 3.2 3.4

ln q

(K.m

in-1

)

1000/Tg (K-1)

ED EP ET ES

Fig. 5. ln q vs. 1000/Tg curves for model drugs.

393.09 274.20 0.69 122.39 51.83 16.45

energy, Tk: Kauzmann temperature, m: fragility parameter, D: strength parameter.

value may be obtained from the activation energy of model com-pounds using Eq. (4) [24].

m = 12.303

Ea(Tg)RTg

(4)

m value followed the order: ES > EP > ET > ED, thus indicating ES tobe the most fragile and ED to be the least fragile (Table 1).

Strength parameter (D) is also a measure of the fragility of com-pounds and was calculated using Eq. (5):

D = (ln 10)m2

minm − mmin

(5)

where mmin is the minimum value of fragility (∼16).Higher value of D (∼30) indicates lower fragility of a com-

pound and vice versa. D values followed the order: ED (19.44) > ET(16.45) > EP (14.75) > ES (12.25), thus indicating ES to be the mostfragile and ED the least fragile (Table 1).

Kauzmann temperature (Tk) indicates the temperature of zeromobility or the temperature at which all the forces involved in crys-tallization are absent [25]. Tk value was calculated from the m valueas well as from the Tg value, using Eq. (6).

Tk = Tg

(1 − mmin

m

)(6)

Angell’s plots were generated to evaluate the temperature depen-dency of relaxation time constant � (Fig. 6). Straight linecorresponds to a strong behavior (Arrhenius behavior) whilenon-linearity in the curve indicate fragility of a compound. Allamorphous forms of erythromycin showed fragile behavior, asindicated by the non linear behavior of relaxation time with tem-perature (non-Arrhenius behavior). However, non-linearity was

higher in case of ES, thus indicating a rapid change of molecularmobility with temperature.

-6

-5

-4

-3

-2

-1

0

1

2

0.8 0.85 0.9 0.95 1

log

(sec

)

Tg/T

EDESEPET

Fig. 6. Plot of log � vs. Tg /T equivalent to Angell’s plot.

82 K. Nanakwani et al. / Thermochimica Acta 582 (2014) 77–85

3

cA

�

wo�cbe

tv

�

wtD

Hwo

�

Ftmt

3

fe

TR

Table 3Configurational quantities of model drugs at Tg − 20 K.

Drug Hconf (J mol−1 K−1) Sconf (J mol−1 K−1) Gconf (J mol−1 K−1)

ED 1663.4 2.8 560.1ES 7142.4 15.3 2483.9

ES, were successfully generated by in situ melt quenching and

Fig. 7. Relaxation time for model drugs below Tg .

.5. Molecular mobility

Molecular mobility for amorphous forms of all the fourompounds under investigation, was calculated based on thedam–Gibbs equation (Eq. (7)).

= �0exp(c/T�Sconf) (7)

here � is molecular relaxation time constant expressed in sec-nds, �0 is a constant (∼10−14 s), T is the absolute temperature,Sconf is the configurational entropy and C is a material dependent

onstant. Configurational entropy refers to the entropy differenceetween the amorphous and crystalline phases due to the differ-nces in the number of configurations.

VTF equation (Eq. (8)) mentions the relaxation times at variousemperatures above Tg to construct a curve of reduced temperatures. mobility.

= �0 exp(

DTo

T − To

)(8)

here T0 denotes the zero molecular mobility temperature and hashe same physical meaning as the Kauzmann temperature TK and

is the strength parameter.Below Tg, amorphous systems are in non-equilibrium condition.

ence, calculation of molecular motions at temperatures below Tg

as performed by using modified equation (Eq. (9)), wherein valuesf Tf were taken into consideration.

= �0 expDTo

T(1 − (To/Tf ))(9)

ig. 7 shows the temperature dependence of molecular relaxationime below Tg. Table 2 summarizes the values of � at isother-

al crystallization conditions of Tg − 20 K. The molecular relaxationime (�) at Tg − 20 K followed the order ES > ET > EP > ED.

.6. Configurational thermodynamic quantities

Researchers have correlated the crystallization tendency of glassormers with molecular mobility [26–29], thermodynamic prop-rties of amorphous form [28,29], molecular interactions [30],

able 2elaxation time (�) for model drugs at Tg − 20 K.

Sample Tg (K) Tg − 20 (K) Tg/(Tg − 20) Relaxation time (�) (s)

ED 385.99 365.99 1.054 3.2 × 103

ES 306.49 286.49 1.069 52.78 × 103

EP 389.16 369.16 1.054 9.23 × 103

ET 396.50 376.50 1.053 12.43 × 103

EP 8140.0 17.1 1642.3ET 33,452.6 74.1 4848.2

moisture content [31], as well as with the mode of generation ofamorphous form [32]. Bhugra et al. observed that compounds withhigher heat of fusion (�Hf) and thus having greater thermodynamicdriving force, crystallized easily when compared to compoundswith lower �Hf [33,34]. Similarly, Graeser et al. correlated thekinetic and thermodynamic factors with the amorphous form sta-bility of a series of drugs and found a good correlation betweenthermodynamic parameters like configurational entropy (Sconf) andtheir physical stability [35,36].

In the present investigation, configurational terms like Sconf,configurational enthalpy (Hconf) and free energy (Gconf) were cal-culated from the modulated DSC thermal scans of crystalline andamorphous sample, using Eqs. (10)–(13).

�Sm = �Hm

Tm(10)

Hconf = �Hm + Cp confdt (11)

Sconf = �Sm + Cp conf

Tdt (12)

Gconf = Hconf(T) + TSconf(T) (13)

where Tm is melting point, �Hm and �Sm are enthalpy and entropyof fusion, and Cp conf is the difference between the heat capacity ofcrystalline and amorphous form at temperature T.

Configurational quantities were calculated in a temperaturerange of Tg − 20 K and Tg + 20 K (Fig. 8). Table 3 shows the calculatedconfigurational parameters at Tg − 20 K and are useful indicators ofthe stability of amorphous form.

3.7. Structural parameters

DFT (density functional theory) can explain various energy fac-tors responsible for intramolecular and intermolecular interactions[37]. In the present investigation, electrophilicity index (ω) wasfound to be higher in case of thiocyanate, as compared to phosphateand stearate counterion (Table 4). Furthermore, chemical potential(�), which indicates the strength between the acid and base, wasfound in the order: ET > EP > ES (Table 4). These structural factorscan be correlated with the physical stability of amorphous salts.

4. Discussion

Pure amorphous form of ED and its salts, namely ET, EP and

characterized for their glass transition temperature (Tg) and thecrystallization behavior at Tg − 20 K + 0% RH (to achieve commondegree of undercooling). Tg followed the order: ET > EP > ED > ES.

Table 4DFT parameters of counterions studied.

Counter ion A I � S � ω

Stearic acid 0.68 8.39 7.71 0.12 −4.54 1.32Phosphoric acid 0.01 10.50 10.49 0.10 −5.26 1.32Thiocyanic acid 2.94 9.06 6.12 0.16 −6.00 2.94

A: electron affinity; I: ionization potential; �: hardness; S: softness; �: electronicchemical potential; ω: electrophilicity index.

K. Nanakwani et al. / Thermochimica Acta 582 (2014) 77–85 83

odel d

HaTptr

4

aamEawiaHfo

srecssdpos

4

o

Fig. 8. Thermodynamic configurational quantities for m

owever, the crystallization kinetics at Tg − 20 K + 0% RH of thesemorphous forms differed and followed the order: ES > EP > ET > ED.hus, there could be factors other than Tg that have an effect onhysical stability of these amorphous forms. In present investiga-ion, all possible factors, viz. kinetic, thermodynamic and structural,esponsible for crystallization kinetics were systematically studied.

.1. Fragility

The concept of fragility proposed by Angell, classifies liquidss ‘strong’ or ‘fragile’, based on magnitude of change in Cp (�Cp)t Tg or on the differences in temperature dependence of meanolecular relaxation time (�) [23]. Higher value of �Cp at Tg for

S indicates its higher fragility as compared to ET and EP. Further,ctivation energy (Ea) for molecular mobility was the least for ES,hen compared to ET and EP, thus signifying ES to be more frag-

le as compared to ET and EP. Evaluation of fragility parameter (m)nd strength parameter (D) also indicates ES to be the most fragile.owever, �Cp at Tg for ED was higher than that of ET and Ea value

or ED was less than that of ES, thus indicating possible effect ofther factors on the physical stability of ED.

Angell’s plot (Fig. 6) revealed that amorphous ED and its saltshowed fragile behavior, as indicated by the non-linear behavior ofelaxation time with temperature (non-Arrhenius behavior). How-ver, non-linearity was higher in case of ES, thus indicating a rapidhange of molecular mobility with temperature. In the presenttudy, the kinetics of crystallization of amorphous materials wastudied at a temperature below Tg, where a strong temperatureependence on molecular mobility, exhibited by fragile material isreferred [38]. However, fragility data does not support rank orderf crystallization kinetics shown by all the four compounds, thusuggesting contribution of additional parameters.

.2. Molecular mobility

Molecular mobility plays critical role in physical stabilityf amorphous form. Molecular motions, including translational,

rugs in the temperature range of Tg − 20 K to Tg + 20 K.

vibrational and rotational motion, increase with temperature,resulting in decreased stability of amorphous forms [39]. Molecularmotions determine crystallization kinetics below Tg as moleculardiffusion is slow at such temperatures [40]. Graeser et al. reported acorrelation between molecular mobility and crystallization kineticsat above and below Tg [35,36].

The relaxation time (�) followed the order ES > ET > EP > ED atTg − 20 K (Table 2), indicating lowest molecular mobility of ES andhighest molecular mobility of ED below Tg. Thus molecular mobil-ity data below Tg also does not support differential crystallizationkinetics of different erythromycin amorphous forms.

4.3. Thermodynamic factors

Configurational entropy (Sconf), configurational enthalpy (Hconf)and free energy (Gconf) signify the thermodynamic barrier to crys-tallization, thermodynamic driving force for crystallization and theoverall thermodynamic force for crystallization, respectively [41].

Change in Cp conf (difference between the heat capacity of crys-talline and amorphous form) values with temperature is shown inFig. 8a, wherein, the magnitude of Cp conf jump at Tg was higher forED, compared to its salts. Higher value of Cp conf jump may suggestbetter physical stability of ED. Furthermore, a comparison of theHconf curves for the model drugs (Fig. 8b) indicate lowest thermo-dynamic driving force for crystallization for ED, further suggestingits higher stability, compared to the salt forms.

Sconf is considered to be one of the major factors in physicalstability of amorphous form as it is directly related to the numberof accessible conformations during nucleation. Lower Sconf meansfewer conformations thus making it easier to achieve optimumconformation required for nucleation. Compounds with flexiblestructure exist in different molecular conformations in the super-cooled liquid state and are expected to exhibit large Sconf values,

thus showing lower crystallization tendency. Sconf curves signifythe entropic barrier to crystallization (Fig. 8c), which was the high-est for ET, indicative of lower probability of ET molecule to achievethe preferred orientation for crystallization. Furthermore, EP had

8 ochim

hmhwE

4

aSmTialaob((bp

idw�k

5

cootomEcluoatTppnuia

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

4 K. Nanakwani et al. / Therm

igher Sconf and lower Gconf (Fig. 8d) value (that signify overall ther-odynamic force for crystallization) compared to ES, thus showing

igher crystallization tendency of the latter. These results correlateell with the PXRD analysis (Fig. 4) at Tg − 20 K and 0% RH, wherein

S showed least physical stability.

.4. Structural factors

The nature and strength of the molecular interactions playn important role to predict the crystallization tendency of salts.trong intermolecular interactions lead to higher Tg and decreasedobility of the amorphous forms, thereby increasing its stability.

hiocyanate with higher electrophilicity index (ω) showed strongerntermolecular interactions (Table 4) and thus highest Tg amongstll the salts under investigation. On the other hand, stearate withower electrophilicity index showed weaker intermolecular inter-ctions and thus lowest Tg. Higher molecular mobility of ES fromther model drugs can also be explained similarly. The strengthetween the acid and base, indicated by chemical potential value�), was found to be higher in ET when compared to EP and ESTable 4). This also confirms a higher interaction potential andetter stability of erythromycin with thiocyanate counterion, com-ared to its phosphate and stearate counterpart.

In nutshell, a comparison of amorphous erythromycin free base.e. ED with its salts establishes predominance of thermodynamicriving force in determining its crystallization kinetics. Moreover,ithin the salt forms, Sconf coupled with structural parameters,

and ω, played decisive role in determining their crystallizationinetics.

. Conclusion

The present study systematically assessed the impact ofounterion on the crystallization kinetics of amorphous saltsf erythromycin with reference to the effect of counterionn the kinetic and thermodynamic factors affecting crystalliza-ion. Kinetics of crystallization at Tg − 20 K + 0% RH, followed therder as ES > EP > ET > ED. However, the order of fragility andolecular mobility below Tg does not support this observation.

valuation of the configurational thermodynamic quantities likeonfigurational enthalpy, entropy and free energy revealed theowest crystallization tendency of ED amongst all the compoundsnder investigation. Similarly, highest crystallization tendencyf ES can be correlated to the lowest thermodynamic barriernd higher driving force for crystallization. This data was fur-her supported by calculation of different structural parameters.hus in conclusion, counterions govern the stability of amor-hous salts by interplay of kinetic, thermodynamic and structuralarameters with most dominant role being played by thermody-amic and structural parameters. This study provides a holisticnderstanding of role of counter ions on crystallization behav-

or and can help in rationalizing stabilization approach for themorphous form.

eferences

[1] M. Rabinowitz, N. Shankley, The impact of combinatorial chemistry on drugdiscovery, in: O.D.J.C.G. Smith (Ed.), The Process of New Drug Discovery andDevelopment, Informa Healthcare, New York, 2006, pp. 56–77.

[2] S.M. Berge, L.D. Bighley, D.C. Monkhouse, Pharmaceutical salts, J. Pharm. Sci. 66(1977) 1–19.

[3] A.T.M. Serajuddin, Salt formation to improve drug solubility, Adv. Drug Deliv.Rev. 59 (2007) 603–616.

[4] B.C. Hancock, G. Zografi, Characteristics and significance of the amorphous statein pharmaceutical systems, J. Pharm. Sci. 86 (1997) 1–12.

[5] A.M. Kaushal, P. Gupta, A.K. Bansal, Amorphous drug delivery systems: molec-ular aspects, design, and performance, Crit. Rev. Ther. Drug Carrier Syst. 21(2004) 133.

[

ica Acta 582 (2014) 77–85

[6] L. Yu, Amorphous pharmaceutical solids: preparation, characterization and sta-bilization, Adv. Drug Deliv. Rev. 48 (2001) 27–42.

[7] Y. Kumar, T.K. Sharma, P. Bose, Amorphous form of losartan potassium,US20060241305 A1, 2003, pp. 1–3.

[8] B. Vijayaraghavan, T. Sharma, N. Kumar, Novel amorphous form of omeprazolesalts, US 20030212274 A1, 2003, pp. 1–3.

[9] P. Guerrieri, A.C. Rumondor, T. Li, L.S. Taylor, Analysis of relationships betweensolid-state properties, counterion, and developability of pharmaceutical salts,AAPS PharmSciTech 11 (2010) 1212–1222.

10] V.M. Sonje, L. Kumar, V. Puri, G. Kohli, A.M. Kaushal, A.K. Bansal, Effect of coun-terions on the properties of amorphous atorvastatin salts, Eur. J. Pharm. Sci. 44(2011) 462–470.

11] P. Tong, L.S. Taylor, G. Zografi, Influence of alkali metal counterions on theglass transition temperature of amorphous indomethacin salts, Pharm. Res.19 (2002) 649–654.

12] C.S. Towler, T. Li, H. Wikström, D.M. Remick, M.V. Sanchez-Felix, L.S. Taylor,An investigation into the influence of counterion on the properties of someamorphous organic salts, Mol. Pharm. 5 (2008) 946–955.

13] L.H. Nielsen, S. Gordon, R. Holm, A. Selen, T. Rades, A. Müllertz, Preparation ofan amorphous sodium furosemide salt improves solubility and dissolution rateand leads to a faster Tmax after oral dosing to rats, Eur. J. Pharm. Biopharm. 85(2013) 942–951.

14] L. Kumar, D. Popat, A.K. Bansal, Investigation of the atypical glass transitionand recrystallization behavior of amorphous prazosin salts, Pharmaceutics 3(2011) 525–537.

15] P. Tong, G. Zografi, Solid-state characteristics of amorphous sodiumindomethacin relative to its free acid, Pharm. Res. 16 (1999) 1186–1192.

16] P.H. Jones, E.K. Rowley, A.L. Weiss, D.L. Bishop, A.H.C. Chun, Insoluble ery-thromycin salts, J. Pharm. Sci. 58 (1969) 337–339.

17] K. Gupta, D. Roy, V. Subramanian, P. Chattaraj, Are strong Brønsted acidsnecessarily strong Lewis acids? J. Mol. Struct. Theochem. 812 (2007)13–24.

18] P. Allen, P. Rahn, A. Sarapu, A. Vanderwielen, Physical characterization of ery-thromycin: anhydrate, monohydrate, and dihydrate crystalline solids, J. Pharm.Sci. 67 (1978) 1087–1093.

19] E. Laine, P. Kahela, R. Rajala, T. Heikkilä, K. Saarnivaara, I. Piippo, Crystal formsand bioavailability of erythromycin, Int. J. Pharm. 38 (1987) 33–38.

20] I. Miroshnyk, L. Khriachtchev, S. Mirza, J. Rantanen, J. Heinämäki, J. Yliruusi,Insight into thermally induced phase transformations of erythromycin A dihy-drate, Cryst. Growth Des. 6 (2006) 369–374.

21] D.Q.M. Craig, P.G. Royall, V.L. Kett, M.L. Hopton, The relevance of the amorphousstate to pharmaceutical dosage forms: glassy drugs and freeze dried systems,Int. J. Pharm. 179 (1999) 179–207.

22] M.E. Matteucci, M.A. Miller, R.O. Williams III, K.P. Johnston, Highly super-saturated solutions of amorphous drugs approaching predictions fromconfigurational thermodynamic properties, J. Phys. Chem. B 112 (2008)16675–16681.

23] C.A. Angell, Relaxation in liquids, polymers and plastic crystals – strong/fragilepatterns and problems, J. Non-Cryst. Solids 131 (1991) 13–31.

24] J.J.M. Ramos, N.T. Correia, R. Taveira-Marques, G. Collins, The activation energyat Tg and the fragility index of indomethacin, predicted from the influ-ence of the heating rate on the temperature position and on the intensityof thermally stimulated depolarization current peak, Pharm. Res. 19 (2002)1879–1884.

25] A.M. Kaushal, P. Gupta, B. AK, Amorphous drug delivery systems: molecularaspects, design, and performance, Crit. Rev. Ther. Drug Carrier Sys. 21 (2004)133.

26] Y. Aso, S. Yoshioka, S. Kojima, Relationship between water mobility, measuredas nuclear magnetic relaxation time, and the crystallization rate of amorphousnifedipine in the presence of some pharmaceutical excipients, Chem. Pharm.Bull. 44 (1996) 1065–1067.

27] Y. Aso, S. Yoshioka, S. Kojima, Relationship between the crystallization ratesof amorphous nifedipine, phenobarbital, and flopropione, and their molecularmobility as measured by their enthalpy relaxation and (1)H NMR relaxationtimes, J. Pharm. Sci. 89 (2000) 408–416.

28] P. Gupta, G. Chawla, A.K. Bansal, Physical stability and solubility advantage fromamorphous celecoxib: the role of thermodynamic quantities and molecularmobility, Mol. Pharm. 1 (2004) 406–413.

29] D. Zhou, G.G.Z. Zhang, D. Law, D.J.W. Grant, E.A. Schmitt, Physical stabilityof amorphous pharmaceuticals: importance of configurational thermo-dynamic quantities and molecular mobility, J. Pharm. Sci. 91 (2002)1863–1872.

30] P. Gupta, A.K. Bansal, Molecular interactions in celecoxib-PVP-meglumineamorphous system, J. Pharm. Pharmcol. 57 (2005) 303–310.

31] S.L. Shamblin, G. Zografi, The effects of absorbed water on the proper-ties of amorphous mixtures containing sucrose, Pharm. Res. 16 (1999)1119–1124.

32] R. Surana, A. Pyne, R. Suryanarayanan, Effect of preparation methodon physical properties of amorphous trehalose, Pharm. Res. 21 (2004)1167–1176.

33] C. Bhugra, M.J. Pikal, Role of thermodynamic, molecular, and kinetic fac-

tors in crystallization from the amorphous state, J. Pharm. Sci. 97 (2008)1329–1349.

34] M. Siniti, J. Carre, J. Letoffe, J. Bastide, P. Claudy, The thermal behavior of hexitols.Part 1. Vitrification and crystallization of iditol, mannitol, sorbitol, and dulcitol,Thermochim. Acta 224 (1993) 97–104.

ochim

[

[

[

[

[

[molecular mobility and crystallization kinetics of amorphous griseofulvin, Mol.

K. Nanakwani et al. / Therm

35] K.A. Graeser, J.E. Patterson, T. Rades, Applying thermodynamic and kineticparameters to predict the physical stability of two differently prepared amor-phous forms of simvastatin, Curr. Drug Deliv. 6 (2009) 374–382.

36] K.A. Graeser, J.E. Patterson, J.A. Zeitler, K.C. Gordon, T. Rades, Correlating ther-modynamic and kinetic parameters with amorphous stability, Eur. J. Pharm.

Sci. 37 (2009) 492–498.

37] R.G. Parr, W. Yang, Density-functional theory of the electronic structure ofmolecules, Annu. Rev. Phys. Chem. 46 (1995) 701–728.

38] R.H. Hatley, Glass fragility and the stability of pharmaceutical preparations-excipient selection, Pharm. Dev. Technol. 2 (1997) 257–264.

[

ica Acta 582 (2014) 77–85 85

39] S.P. Bhardwaj, K.K. Arora, E. Kwong, A. Templeton, S.-D. Clas, R. Suryanarayanan,Correlation between molecular mobility and physical stability of amorphousitraconazole, Mol. Pharm. 10 (2013) 694–700.

40] D. Zhou, G.G.Z. Zhang, D. Law, D.J.W. Grant, E.A. Schmitt, Thermodynamics,

Pharm. 5 (2008) 927–936.41] A.M. Kaushal, A.K. Bansal, Thermodynamic behavior of glassy state of

structurally related compounds, Eur. J. Pharm. Biopharm. 69 (2008)1067–1076.