Characterization of Azithromycin hydrates

European Journal of Pharmaceutical Sciences 16 (2002) 175–184www.elsevier.nl / locate/ejps

qC haracterization of Azithromycin hydratesa a b bRajesh Gandhi , Omathanu Pillai , Ramasamy Thilagavathi , Bulusu Gopalakrishnan ,

a a ,*Chaman Lal Kaul , Ramesh PanchagnulaaDepartment of Pharmaceutics, National Institute of Pharmaceutical Education and Research (NIPER), Sector-67, SAS Nagar—160 062Punjab,

IndiabDepartment of Medicinal Chemistry, National Institute of Pharmaceutical Education and Research (NIPER), Sector-67,

SAS Nagar—160 062Punjab, India

Received 15 November 2001; received in revised form 3 April 2002; accepted 21 May 2002

Abstract

Azithromycin (AZI) is a macrolide antibiotic with an expanded spectrum of activity that is commercially available as a dihydrate. Thisstudy was carried out to characterize hydrates of azithromycin. A commercial dihydrate sample was used to prepare monohydrate fromwater /ethanol (1:1) mixture. Hydrates were characterized using DSC, TGA, KFT, XRD, HSM, SEM and FT-IR. TGA showed that thecommercial samples are dihydrate and the sample prepared from water /ethanol (1:1) was a monohydrate. Solubility studies revealed thatmonohydrate converted to dihydrate during solubility studies and as a result there was no significant difference in the equilibriumsolubility of MH and DH. Thermal analysis under various conditions revealed that dehydration and melting took place simultaneously.Anhydrous AZI was found to be hygroscopic and converted to DH on storing at room temperature. Molecular modeling studies revealedthe probable sites of attachment of water molecules to AZI. 2002 Elsevier Science B.V. All rights reserved.

Keywords: Azithromycin; Pseudopolymorphism; Thermal analysis; Powder X-ray diffraction; Molecular modeling

1 . Introduction hydrates can also influence the intermolecular interactions,crystalline disorder, changes in free energy, thermody-

Polymorphism and pseudopolymorphism are important namic activity, solubility, dissolution rate, stability andsolid state properties that influence the performance and bioavailability (Khankari and Grant, 1995). Therefore,processing of solid dosage forms (Morris et al., 2001). characterization of solid state properties at an early stagePolymorphism deals with difference in the internal struc- using appropriate analytical methodology is an essentialture of crystals, and pseudopolymorphism is existence of prerequisite in the development of solid dosage forms bothdifferent solvates of the same chemical compound (Vip- from scientific and regulatory points of view (Byrn et al.,pagunta et al., 2001). The most common solvate en- 1995). Azithromycin (AZI) is a semisynthetic acid-stablecountered in pharmaceutical compounds is hydrate. Water erythromycin derivative (Fig. 1) with an expanded spec-in hydrates can be present either in stoichiometric or in trum of activity and improved pharmacokinetic characteris-nonstoichiometric ratio depending on the crystal lattice tics (Dunn and Barradell, 1996). It is reported to exist as aarrangement as well as the nature of binding of water dihydrate (USP, 1995). Azithromycin samples from differ-molecules (Jeffery, 1969). Hydrates of drug can influence ent manufacturers were found to exhibit variable thermalphysicochemical properties, processing, mechanical and behaviour with either single or two endotherms as showncompaction behavior of a drug. In addition, water in in Table 1. Most of them exhibited two endotherms, while

the USP reference standard showed a single endotherm.Therefore, this study was carried out to investigate theqNiper communication no. 73.presence of different solid state forms of azithromycin and*Corresponding author. Tel.:191-172-214682x87; fax:191-172-further there is no literature report on the solid state214692.

E-mail address: [email protected](R. Panchagnula). behaviour of azithromycin. In this preliminary study,

0928-0987/02/$ – see front matter 2002 Elsevier Science B.V. All rights reserved.PI I : S0928-0987( 02 )00087-8

mailto:[email protected]

176 R. Gandhi et al. / European Journal of Pharmaceutical Sciences 16 (2002) 175–184

therms in DSC was used in this study. CS when kept under85% relative humidity (saturated KCl) at room temperaturefor 2 weeks in a dessicator and then subjected to DSCanalysis showed a single endotherm and the sample was adihydrate (DH). Monohydrate (MH) was prepared bydissolving an excess amount of CS with continuous stirringin a water /ethanol mixture (50:50) and warming thesolution in a water bath at 608C. The undissolved drugfrom the saturated solution was filtered and then the filtratewas cooled at room temperature. AZI crystals obtainedwere stored in a vacuum dessicator till further analysis.Anhydrous AZI was prepared by heating the CS, DH andMH in a hot-air oven at 1008C for 30 min and also bystoring under phosphorus pentoxide.

Fig. 1. Chemical structure of azithromycin.2 .3. Differential scanning calorimetry (DSC)

thermal methods, FT-IR, X-ray diffraction, microscopy, The thermograms of hydrates were recorded on a DSCesolubility studies and molecular modeling were used for (Mettler, Toledo DSC 821 Switzerland) using the Mettler

esolid state characterization of AZI. Star system. The temperature axis and cell constant ofDSC cell were calibrated with indium. A heating rate of10 8C/min was employed over a temperature range of

2 . Materials and methods 25–2508C with nitrogen purging (80 ml /min). The sample(1–5 mg) was weighed into an aluminum pan and analysed

2 .1. Materials either in open condition or sealed with and without pinholes and an empty aluminum pan was used as the

Azithromycin was procured as gratis sample from seven reference. To further characterize the melting endotherm ofdifferent manufacturers in India (JK Pharmaceuticals, New AZI, the samples were held in the DSC pan for 30 min atDelhi; Sarabhai Chemicals Ltd., Baroda; Pfizer India Ltd., 1028C and then heated at 108C/min up to 2008C.Mumbai; Alembic Chemicals Ltd., Baroda; Kopran IndiaLtd., Mumbai; Pradeep Drugs Company Ltd., Chennai; Bal

2 .4. Thermogravimetric analysis (TGA)Pharma, Mumbai). The solvents and reagents used wereeither HPLC grade or analytical reagent-grade.

Thermogravimetric analysis was carried out using TGAe(Mettler, Toledo TGA/SDTA 851 , Switzerland) with the2 .2. Preparation of AZI hydrates eMettler Star system. The water loss was determined by

placing the sample (1–5 mg) in alumina crucibles andA commercial sample (CS) which exhibited two endo-heating up to 2508C at a rate of 108C/min under nitrogen

Table 1 purge (20 ml /min). The pan was either open or closed withThermal behavior of azithromycin samples supplied by different manu- a lid in order to compare the results from DSC studies. Tofacturers

further characterize the melting endotherm of AZI, thebAzithromycin DSC analysis samples were held in the TGA pan for 30 min at 1028C.

asamplesTemperature range (8C) Heat of fusion (J /g) Samples were then cooled in the pan to room temperature

and again heated at 108C/min from 25 to 2008C withAZI 1 133.78–143.08 63.54simultaneous recording of TGA and DTA.149.58–156.97 30.71

AZI 2 137.88–146.75 95.83AZI 3 133.27–144.49 61.13

2 .5. Hot-stage microscopy (HSM)145.24–150.60 31.44AZI 4 129.60–142.41 65.10

148.15–154.40 29.99 Thermal events were observed on a hot stage (Mettler,AZI 5 132.91–145.38 61.46 FP 80) under a polarized light microscope (Leitz, Ger-

149.58–156.97 31.44 many) equipped with a 35-mm camera (Leica, MPS 52).AZI 6 132.90–145.39 56.53

The sample was placed over a drop of silicone oil, covered149.39–157.69 30.09with a cover slip and heated at a rate of 108C/min. TheAZI 7 137.66–147.89 91.63

(USP reference) temperature at which liberation of bubbles correspondinga to escape of water vapor and melting events occurred wereAzithromycin obtained from different sources.b DSC thermograms were recorded in sealed pans. observed to characterize the hydrates.

R. Gandhi et al. / European Journal of Pharmaceutical Sciences 16 (2002) 175–184 177

Fig. 2. DSC thermograms of different forms of AZI. Upper curve depicts the commercial sample (CS) of AZI; middle curve shows the endotherm ofdihydrate (DH) and lower curve shows the endotherm of monohydrate (MH). The thermograms were generated using a sealed pan.

2 .6. Karl Fischer titration (KFT) 2 .8. Powder X-ray diffraction (XRD)

Water content in different forms of AZI was determined Powder X-ray diffraction patterns for different forms ofusing a Karl Fischer titrimeter (Metrohm, 716 DMS, AZI were acquired at room temperature on an X-raySwitzerland). Samples (20–25 mg) were accurately diffractometer (Siemens, D-5000, Germany) using Cu Ka

weighed and quickly transferred to the titration vessel radiation (tube operated at 40 kV, 30 mA). Data werecontaining anhydrous methanol. collected over an angular range from 2 to 508 2u in

continuous scan mode using a step size of 0.038 2u and a2 .7. Fourier transform infrared spectroscopy (FT-IR) step time of 0.5 s.

Spectra were recorded in a FT-IR spectrophotometer 2 .9. Scanning electron microscopy (SEM)(Nicolet, Impact 410, USA) using the KBr pellet method in

21the region of 4000–400 cm . AZI samples were viewed by scanning electron micro-

Table 2Thermal analysis and KFT of different forms of AZI

aSample DSC analysis Water contentname d eTemperature range (8C) Heat of fusion (J /g) TGA (%) KFT (%)

b bCS 132.29–143.09 62.7367.13 4.45260.18 4.5760.03c c149.83–155.48 30.4162.93

DH 134.65–141.35 92.9968.58 4.15560.41 4.3560.28MH 139.88–156.31 92.8362.41 2.47260.41 2.3960.79

a DSC thermograms were recorded in sealed pans.b First endotherm.c Second endotherm.d Thermogravimetric analysis; mean6S.D.; n53.e Karl Fischer analysis; mean6S.D.; n53.


Fig. 3. DSC thermograms of DH with different pan types. From top to bottom sealed pan, sealed pan with pin hole and open pan.

Fig. 4. TGA thermograms of different forms of AZI. Upper curve indicates stoichiometric weight loss of two water molecules in CS; middle curveindicates weight loss of two water molecules in DH and lower curve indicates weight loss of one water molecule in MH.


scopy (Jeol electron microscope, D-6000, Japan). The3 . Results and discussionsamples were sputter coated with gold before examination.

DSC thermograms of CS, DH and MH are shown in Fig.2 and the thermal characteristics are shown in Table 2. The

2 .10. Solubility studiesCS sample showed two endothermic peaks, while DH andMH showed single endotherms. DSC analysis using differ-

An excess amount of CS, DH and MH were placed inent pan types showed a shift in the melting endotherm of

glass vials with 5 ml of water in a shaker water bath at 100AZI (Fig. 3). Since melting is preceded by release of

rpm and 378C. Samples were withdrawn at specified timewater, the dehydration temperature shifts with increase in

intervals till equilibrium was reached and analyzed usingpressure from open pan to sealed pan. As can be seen from

the HPLC method which is described elsewhere (Gandhi etthe thermograms (Fig. 3), the melting peak is sharp in the

al., 2000). Excess solid remaining after the last sample wassealed pan due to the rapid attainment of equilibrium under

subjected to DSC and TGA analysis.the ‘pressurized’ pan conditions and the same was alsoobserved in the case of MH (data not shown). When all

2 .11. Molecular modeling three samples were observed in HSM, it was found that inthe case of DH and MH, release of bubbles started at

Molecular modeling studies were carried out using around 1008C and continued up to 120–1308C, whereSYBYL (version 6.6) molecular modeling software (TRI- crystals were found to melt simultaneously with generation

POS Associates, USA) installed on a Silicon Graphics of bubbles. But, in the case of CS, water release waspower ONYX workstation using IRIX 6.5. The AZI accompanied by partial melting of some of the crystals atstructure was sketched and energy minimization was around 1208C and the complete melting of all the crystalscarried out using Tripos force field for 1000 cycles. The took place at 1308C.charges were calculated using the Gasteiger–Huckel meth- A stoichiometric weight loss of two water moleculesod. Docking of water molecules to AZI structure was done (theoretical weight loss 4.58%) was found for CS and DH

using the ‘Dock’ option in SYBYL and the energy of the by TGA, while MH showed a weight loss corresponding todocked complex was minimized using conjugate gradients one molecule of water (theoretical weight loss 2.29%) as(Powell, 1977). shown in Fig. 4. The results were in good agreement with

Fig. 5. Simultaneous TGA and DTA scans of different hydrates of AZI (see details for text). Upper two thermograms depict TGA and DTA curves forDH; Lower two thermograms depict TGA and DTA curves for MH. The straight line indicates that there is negligible water loss in TGA.


those found by KFT as shown in Table 2. Since release of tion and melting of anhydrous form, respectively, whereaswater and melting took place simultaneously, it was in DH and MH there were single peaks. The first endo-difficult to interpret the thermal features, therefore, TGA therm in CS had an enthalpy of 62.73 J/g (Table 2), whichand DTA were recorded simultaneously (Fig. 5) after was in good agreement with the reported enthalpy valuesholding the sample in the TGA pan for 30 min at 1028C to for dehydration which generally lie in the range 60–80release the water, cooling followed by reheating at 108C/ J/g, depending upon the nature of binding of water (Hanmin from 25 to 2008C. As can be seen from Fig. 5, TGA and Suryanarayanan, 1997; Khankari et al., 1992). Theshowed negligible water loss (,0.3%) while DTA showed second endotherm was indicative of melting as was foundan endotherm corresponding to the melting temperature of from HSM and the enthalpy of fusion was 30 J/g. On theAZI. There was a difference in the melting temperature other hand, in the case of DH and MH, dehydration andbetween DSC scans and simultaneous TGA/DTA scans melting took place simultaneously as is evident from thedue to the difference in pan type and the ‘pressure’ inside enthalpy values which is additive of dehydration andthe pan (Byrn et al., 1995). In CS, there were two melting enthalpy values. In SEM, it was found thatendothermic peaks (Fig. 2), which is indicative of dehydra- although the crystal habits of all three samples of AZI

Fig. 6. SEM of different forms of azithromycin. Microphotographs are arranged from top to bottom in the following order: CS, DH and MH.


Fig. 7. DSC thermograms of anhydrous azithromycin prepared by keeping the samples under phosphorus pentoxide overnight. From top to bottom: CS,DH and MH.

were similar, MH was found to contain more irregular in a dessicator overnight followed by TGA and KFT,particles which can probably explain the slight shift in the analysis showed a complete loss of water and melting wasmelting temperature of MH compared to DH and CS (Fig. seen at 125–1328C in DSC analysis (Fig. 7) with an6). enthalpy value of 32–36 J/g implying the formation of the

When the CS sample was gently ground before DSC completely anhydrous form. The same was also confirmedanalysis, there was only one single broad endotherm at by TGA and KFT analysis. But the anhydrous form was1208C (enthalpy value additive of dehydration and melt- found to be hygroscopic and convert to dihydrate oning) similar to DH and MH. This is not unusual since the storage at room temperature.dehydration temperature can be influenced by imperfec- Equilibrium solubility of the samples in water did nottions of a common lattice structure (Bettinetti et al., 1999)or occluded impurities in the crystal defects (Holgado etal., 1995) which may result in artifact peaks in DSC.Moreover, when the samples were held at 1028C for 30min followed by heating at 108C/min up to 2008C, therewas only one endotherm with enthalpy values of 30–45J/g, which matched with the enthalpy values shown inTable 2 corresponding to melting of anhydrous AZI.Variation in enthalpy values may probably be due to a verysmall amount of water still remaining in the sample as wasshown in a similar experiment with TGA as discussedearlier. When the three samples (CS, DH, MH) wereheated in an hot-air oven at 1008C for 30 min followed byDSC and TGA, analysis showed similar results as de-scribed above. Further, KFT showed comparable (0.1–0.5%) water loss as shown by TGA analysis. However, Fig. 8. Solubility of azithromycin samples as a function of time. Eachwhen the samples were stored under phosphorus pentoxidedata point represents mean6S.D. (n53).


differ significantly (P.0.05), although DH (1.9860.11 AZI in MH and DH. An attempt was made using molecu-mg/ml) had slightly higher solubility than MH lar modeling as a tool to study the interactions of water(1.8060.081 mg/ml). However, the solubility results at with AZI. It was found that in the case of DH, one of theearlier time points showed that MH was more soluble and water molecules is involved in hydrogen bonding with theby 48 h converted to DH resulting in similar solubility hydroxyl group at the 6th position (Fig. 10). This watervalues to the other two samples within statistical limits molecule further forms a hydrogen bond with another(Fig. 8). The formation of DH was confirmed by analyzing water molecule, which in turn forms a bifurcated hydrogenthe excess solid at the end of the solubility studies by DSC bond with oxygen at the 15th position and oxygen ofand TGA. Earlier studies with erythromycin (Allen et al., hydroxyl at the the 12th position. On the other hand, in the1978) and nedocormil zinc hydrates (Zhu et al., 1997) case of MH, water forms a single hydrogen bond withhave attributed such small differences in solubility between oxygen of the hydroxyl group at the 6th position (Fig. 10).hydrates to the difference in particle–particle interactions Formation of monohydrate was energetically less favorableand wettability of the crystalline materials. However, such than dihydrate and anhydrate was found to be least stabledifference in internal crystal lattice can only be revealed in molecular modeling studies. The existence of AZI in theusing single crystal XRD which is outside the scope of this more stable hydrate form can be anticipated due to thestudy. FT-IR spectra of DH and MH revealed distinct predominance of hydroxyl groups (potential hydrogendifferences corresponding to O–H stretching region (3400 bonding sites) compared to other macrolide antibiotics

21cm ) as shown in Fig. 9. The presence of a sharp high such as erythromycin and clarithromycin. Further, clari-21frequency peak at 3463 cm in the case of DH and CS is thromycin is reported to exist only in the anhydrous form,

indicative of the presence of ‘tightly bound’ water in the since one of the hydroxyl groups of erythromycin iscrystal lattice (Bettinetti et al., 1999), while the broad band replaced with a methoxyl group, while erythromycin exists

21at 3000–3500 cm in MH is due to the O–H stretching both in anhydrous and hydrated forms (Stephenson et al.,for self associated water that may be ‘loosely’ bound (Zhu 1997).

21et al., 1997). The difference at 1200 and 1600 cm can be In the case of erythromycin, it is unclear as to whetherattributed to the difference in the H-bonding of water to the water is present in stoichiometric (true hydrate) or

Fig. 9. FT-IR spectra of CS, DH and MH (arranged from top to bottom).


togram of CS at 9 and 14.98 u. The internal crystalstructure appears to be the same, as is evident from thesimilar enthalpy of fusion for all three samples (Tiwaryand Panpalia, 1999). CHN analysis showed that themolecular formula of both MH and DH was the sameindicating that the two differ only in their water content asfound from TGA and KFT. On the other hand, thecrystallinity of the anhydrous form formed after dehydra-tion in DSC/TGA pans could not be characterized as thematerial was hygroscopic and insufficient for characteriza-tion by XRD. Furthermore, as explained earlier the crystalhabits of all three samples were similar although, MHconsisted of a large proportion of irregularly-shapedcrystals, probably due to loosely bound water, compared toDH (Fig. 6). Further studies are required to explain thedifferences in XRD using other techniques such as variabletemperature XRD, pressure DSC (PDSC) and solid stateNMR to confirm the findings of this study. In addition,stability of the hydrates at different relative humidities willalso form a part of the detailed studies to be reported infuture.

4 . Conclusions

Azithromycin was found to exhibit pseudo-polymorphism and can exist as monohydrate anddihydrate. The anhydrous form of AZI seemed to beunstable since it converted to dihydrate on storage at roomtemperature. On the other hand, monohydrate in thepresence of moisture can convert to the more stabledihydrate form. Therefore, the most stable form of AZI isdihydrate. It is important to select the appropriate form ofAZI and also control the moisture levels during variousprocessing operations involved in the formulation of soliddosage forms. In addition during the selection of excipient,it is necessary to ensure the excipients do not have aninfluence on the moisture content of AZI, which in turncan induce inter conversion of one form to anotherFig. 10. Stereoview of molecular model of DH and MH (from top to

bottom). (anhydrous to dihydrate or vice versa). Future studies withsingle crystal XRD, pressure DSC (PDSC), variable tem-perature XRD (VTXRD) and solid state NMR will providemore insight into the stability of different forms of AZI.

nonstoichiometric amounts (pseudohydrate), as the waterbound in erythromycin (Bauer et al., 1985) can easily belost at temperatures as low as 408C. Unlike erythromycin,the water seems to be more strongly bound in AZI as the A cknowledgementswater is found to be lost only at temperatures around1008C as is found in TGA. On the other hand, the absence We are grateful to Dr M.E. Sobhia, for her help withof a separate dehydration peak in DSC thermograms in DH modeling calculations. Vikas Grover, Technical Assistantand MH, indicates that water may be bound stoichio- CIL, NIPER is acknowledged for his support in thermalmetrically within the crystal lattice. Powder X-ray diffrac- analysis. The assistance of Dr A.K. Ghosh, Polymertograms are shown in Fig. 11 and it was found that both Division of IIT, New Delhi, India in conducting HSMDH and MH are isostructural, but differs from the diffrac- studies is gratefully acknowledged.


Fig. 11. X-Ray diffractograms of different forms of AZI. From top to bottom, diffractograms of CS, DH and MH.

Jeffery, G.A., 1969. Water structure in organic hydrates. Acc. Chem. Res.R eferences2, 344–352.

Khankari, R.K., Grant, D.J.W., 1995. Pharmaceutical hydrates. Thermoch-Allen, P.V., Rahn, P.D., Sarapu, A.C., Vanderwielen, A.J., 1978. Physical im. Acta 248, 61–79.

characterization of erythromycin. Anhydrate, monohydrate and Khankari, R.K., Law, D., Grant, D.J.W., 1992. Determination of waterdihydrate crystalline solids. J. Pharm. Sci. 67, 1087–1093. content in pharmaceutical hydrates by differential scanning

Bauer, J., Quick, J., Oheim, R., 1985. Alternate interpretation of the role calorimetry. Int. J. Pharm. 82, 117–127.of water in the erythromycin structure. J. Pharm. Sci. 74, 899–900. Morris, K.R., Griessa, U.J., Eckhardt, C.J., Stowell, J.G., 2001. Theoret-

Bettinetti, G., Mura, P., Sorrenti, M., Faucci, M.T., Negri, A., 1999. ical approaches to physical transformations of active pharmaceuticalPhysical characterization of picotamide monohydrate and anhydrous ingredients during manufacturing process. Adv. Drug Deliv. Rev. 48,picotamide. J. Pharm. Sci. 88, 1133–1139. 91–119.

Byrn, S., Pfeiffer, R., Ganey, M., Hoiberg, C., Poochikial, G., 1995. Powell, M.J.D., 1977. Restart procedures for the conjugate gradientPharmaceutical solids: a strategic approach to regulatory considera- method. Math. Progr. 12, 241–254.tions. Pharm. Res. 7, 945–954. Stephenson, G.A., Stowell, J.G., Toma, P.H., Pfeiffer, R.R., Byrn, R.,

Dunn, C., Barradell, L.B., 1996. Azithromycin—A review of its pharma- 1997. Solid-state investigations of erythromycin A dihydrate: struc-cological properties and use as 3-day therapy in respiratory tract ture, NMR spectroscopy and hygroscopicity. J. Pharm. Sci. 86, 1239–infections. Drugs 51, 483–505. 1244.

Gandhi, R., Kaul, C.L., Panchagnula, R., 2000. Validated HPLC method Tiwary, A.K., Panpalia, G.M., 1999. Influence of crystal habit onfor in-vitro analysis of azithromycin using electrochemical detection. trimethoprim suspension formulation. Pharm. Res. 16, 261–265.J. Pharm. Biomed. Anal. 23, 1073–1079. USP, 1995. In: USP/NF 23/18. The United States Pharmacopoeial

Han, J., Suryanarayanan, R., 1997. Applications of pressure differential Convention, Rockville, MD, pp. 152–153.scanning calorimetry in the study of pharmaceutical hydrates. I. Vippagunta, S.R., Brittain, H.G., Grants, D.J.W., 2001. Crystalline solids.Carbamazepine dihydrate. Int. J. Pharm. 157, 209–218. Adv. Drug Deliv. Rev. 48, 3–26.

Holgado, M.A., Fernadez-Hervas, M.J., Rabasco, A.M., Fini, A., 1995. Zhu, H., Padden, B.E., Munson, E.J., Grant, D.J.W., 1997. PhysiochemicalCharacterization study of a diclofenac salt by means of SEM and characterization of nedocromil bivalent metal salt hydrates. 2. Nedoc-fractal analysis. Int. J. Pharm. 120, 157–167. romil Zinc. J. Pharm. Sci. 86, 418–428.

Documents

Characterization of Azithromycin hydrates