Drug Development and Industrial Pharmacy, 35:283–296, 2009Copyright © Informa UK, Ltd.ISSN: 0363-9045 print / 1520-5762 online DOI: 10.1080/03639040802282896

283

LDDIFlocculation of Polymer Stabilized Nanocrystal Suspensions to Produce Redispersible Powders

FLOCCULATION NANOCRYSTAL SUSPENSIONSXiaoxia Chen, Michal E. Matteucci, Connie Y. Lo, and Keith P. JohnstonDepartment of Chemical Engineering, University of Texas at Austin, Austin, TX, USA

Robert O. Williams IIICollege of Pharmacy, University of Texas at Austin, Austin, TX, USA

Aqueous suspensions of crystalline naproxen nanoparticles,formed by antisolvent precipitation, were flocculated with sodiumsulfate, filtered, and dried to form redispersible powders for oraldelivery. The particles were stabilized with polyvinylpyrrolidone(PVP K-15) and/or poly(ethylene oxide-b-propylene oxide-b-ethyleneoxide) (poloxamer 407). The yield of the drug in the powder wastypically 92–99%, and the drug loading was reproducible towithin 1–2%. The filtration process increased the drug loading byup to 61% relative to the initial value, as unbound surfactant wasremoved with the filtrate. Upon redispersion of the dried powder,the average particle size measured by light scattering was compa-rable to the original value in the aqueous suspension prior to floc-culation, and consistent with primary particle sizes observed byscanning electron microscopy (SEM). For 300-nm particles, up to95% of the drug dissolved in 2 min. The dissolution rate was cor-related linearly with the specific surface area calculated from theaverage particle diameter after redispersion. The redispersion ofdried powders was examined as a function of the salt concentra-tion used for flocculation and the surfactant composition and con-centration. Flocculation followed by filtration and drying is anefficient and highly reproducible process for the rapid recovery ofdrug nanoparticles to produce wettable powders with high drugloading and rapid dissolution.

Keywords salt flocculation; nanoparticles; naproxen; antisolventprecipitation; dissolution

INTRODUCTIONIt is estimated that more than one-third of the compounds

discovered by the pharmaceutical industry are poorly watersoluble (Lipinski, 2002; Radtke, 2001). The bioavailability ofclass II drugs is often dependent upon the dissolution rate ofthe drug in the gastrointestinal tract (Oh, Curl, & Gl, 1993).Dissolution rates may be increased by reducing the particlesize to increase the surface area, and by coating drug particles

with hydrophilic surfactants to enhance wetting and salvation(Chen, Vaughn, Yacaman, Williams, & Johnston, 2004;Rogers et al., 2003; Sarkari et al., 2002).

Poorly water-soluble drugs may be formulated with submi-cron features to enhance dissolution rates (Rabinow, 2004).However, the recovery of nanoparticles (<500 nm) from aque-ous suspension is a formidable challenge, particularly if it isdesired to achieve the original particle size upon redispersion inaqueous media, for example, in oral drug delivery. Budesonideparticles with a size range from 1 to 10 μm were filtered with0.8-μm pore size polycarbonate membranes (Ruch &Matijevic, 2000). Filtration becomes much more challenging,however, for particle sizes below 1 μm. Ketoconazole, itracon-azole, and ibuprofen micronized particles have been recoveredby spray drying (Rasenack & Muller, 2002). For particles pro-duced by antisolvent precipitation, that is mixing of organicand aqueous solutions, the organic solvent may be strippedfrom the aqueous suspension prior to spray drying to minimizeOstwald ripening (Elder et al., 2003; Liu, Kathan, Saad, &Prud’homme, 2006).

Antisolvent precipitation is a widely used process to prepareinorganic and organic particles (Falk, Randolph, Meyer, Kelly,& Manning, 1997; Jarmer, Lengsfeld, & Randolph, 2004;Randolph, Randolph, Mebes, & Yeung, 1993) including nano-particles of poorly water-soluble drugs (Elder et al., 2003;Johnson & Prud’homme, 2003; Rasenack & Muller, 2002;Ruch & Matijevic, 2000). In this process, a poorly water-soluble drug with or without a polymeric surfactant(s) is dis-solved in a water-miscible organic solvent, such as methanol,ethanol, tetrahydrofuran (THF), and acetonitrile. The organicsolution is then mixed with an “antisolvent,” usually an aque-ous solution containing a surfactant(s) by a confined impingingjets (CIJ) mixer (Johnson & Prud’homme, 2003), controlledprecipitation (Elder et al., 2003; Rogers et al., 2004) sonication(Ruch & Matijevic, 2000), or direct addition (pouring) of theantisolvent into organic drug solution (Rasenack & Muller,2002). Supercritical fluid antisolvents can be used to achieve a

Address correspondence to Keith P. Johnston, Department ofChemical Engineering, University of Texas at Austin, Austin, TX78712, USA. E-mail: kpj@che.utexas.edu

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

284 X. CHEN ET AL.

great deal of control over particle morphology, as a result ofrapid two-way diffusion (Dixon, Johnston, & Bodmeier, 1993;Palakodaty & York, 1999; Randolph et al., 1993; Yeo, Lim,Debenedetti, & Bernstein, 1993). Upon mixing, the supersatu-ration and concentration of polymeric steric stabilizers may becontrolled to manipulate the nucleation and growth of drugparticles (Jarmer et al., 2004; Matteucci, Hotze, Williams III,& Johnston, 2006; Rogers et al., 2004). With sufficient super-saturation, and arrested growth by surfactant stabilization, itbecomes possible to form suspensions of submicron particlesin the aqueous solution.

The objective of this study was to recover nanoparticlesof poorly water-soluble drugs produced by antisolvent pre-cipitation at low temperatures from 0 to 22°C by floccula-tion and filtration, and to examine their morpholgy anddissolution rates. The temperature for antisolvent precipita-tion is shown to have a large effect on the particle size dis-tribution in the suspension. The particles were separatedfrom the solvent by flocculation with a concentrated saltsolution, to desolvate and collapse the polymeric steric sta-bilizers on the particle surface. The large flocculated parti-cles were recovered from the aqueous solution by filtrationand were vacuum dried. The surfactant composition andstructure, and the type and concentration of salt in the drugsuspension were optimized to produce large, loose flocs thatcould be redispersed into pure water readily after filtrationand drying. The increase in drug loading (wt. drug/totalweight excluding salt) upon filtration and the reproducibil-ity in the composition and yield of the powder were exam-ined. Another goal was to achieve particle sizes afterredispersion similar to those in the original suspension priorto flocculation. The dissolution rate of naproxen powderproduced by this technique was compared with that of iden-tical aqueous drug suspensions dried by lyophilization. X-raydiffraction was used to investigate the crystallinity ofnaproxen. Optical microscopy and scanning electron micros-copy (SEM) were used to characterize the morphologies ofthe naproxen flocs in the suspension and after drying. Theconcentration of residual salt in the dried samples was mea-sured by conductivity and shown to be far below the toxiclimit.

Compared to other solvent removal techniques, the use offlocculation and filtration offers several advantages. The flocsmay be filtered much more rapidly and efficiently than theoriginal nanoparticle suspension. Rapid filtration reduces thetime for the growth of the primary particles in the concentratedprecipitate. Particle growth is a potential problem in this stepas the stabilizers on the particle become less solvated by wateras the filtrate is removed, and the presence of organic solventenhances Ostwald ripening (Liu et al., 2006). The filtration canbe operated at low temperatures more easily than in the case ofspray drying and other solvent evaporation techniques. Forexample, drying at high temperatures may lead to undesirableparticle growth (Elder et al., 2003). Separation by filtration

avoids challenges in evaporation, for example, for solventssuch as ethanol that form azeotropes with water. Whereas itmay take days for lyophilization, the time for flocculation andfiltration will be shown to be on the order of minutes. Finally,the loading of the drug can be increased in the filtration stepsince the dispersed particle phase contains a higher fraction ofdrug than the continuous phase, as soluble surfactant stabilizersare removed with the filtrate.

EXPERIMENTAL SECTION

MaterialsNaproxen, ketoconazole, polyvinylpyrrolidone (PVP K-15,

Mw = 10,000), and poly(ethylene oxide-b-propylene oxide-b-ethylene oxide) (poloxamer 407) with a nominal molecularweight of 12,500 and a PEO/PPO ratio of 2:1 by weight werepurchased from Spectrum Chemical Mfg. Corp. (Gardena, CA,USA). Sodium sulfate (anhydrous) and sodium carbonate(anhydrous) were obtained from Fisher Scientific Company(Fairlawn, NJ, USA). Sodium phosphate tribasic dodeca-hydrate was from EM Science (Gibbstown, NJ, USA). HPLCgrade acetonitrile and methanol were obtained from EM Indus-trial Inc (Gibbstown, NJ, USA).

Antisolvent PrecipitationA schematic of the antisolvent process is shown in Figure 1.

The naproxen solution in methanol or ethanol was fed by aHPLC pump through a 1-m-long 1/16 in. o.d. × 0.030 in. i.d.stainless-steel tube. The organic solution was sprayed to form acylindrical jet roughly the inside diameter of the tube withoutatomization into precooled aqueous surfactant solution. Theaqueous surfactant solution was contained in a 250-mL glasscylinder submerged in a water/ethylene glycol bath controlledto 3°C. The tip of the stainless-steel tube was submergedapproximately 4 cm under the surface of the aqueous solution.To enhance the mixing between organic phase and aqueousphase, a magnetic stir bar was placed inside the glass cylinderand stirred at a fixed rate to form a vortex. Unless specifiedelsewhere, 7% (wt/vol) naproxen with or without surfactantwas dissolved into methanol. The organic solution was thensprayed into 50 mL aqueous solution at 5 mL/min for 1.4 minto yield a suspension concentration of 10 mg/mL. After spray-ing for a required time to produce the desired suspension con-centration, the suspension was analyzed within 5 min todetermine the particle size by static light scattering withMalvern Mastersizer-S (Malvern Instruments Ltd., Malvern,UK). The suspension was sonicated in the Mastersizer to breakup the aggregates until a stable particle size distribution wasobtained. The particle size of the same suspension was alsomeasured after 5 min sonication with a powerful sonicator(Branson Sonifer 450, Branson Ultrasonics Corp., Danbury,CT, USA) in an ice/water bath at an output control of 7 and30% duty cycle.

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

FLOCCULATION NANOCRYSTAL SUSPENSIONS 285

Recovery of Naproxen Nanoparticles by Salt Flocculation Followed by Filtration and Vacuum Drying

After particle size measurement, another suspension pro-duced at exactly the same experimental conditions was soni-cated in an ice/water bath at an output control of 7 and 30%duty cycle for 5 min. Then, a given volume of 20% (wt/vol)sodium sulfate was added into the suspension and mixed thor-oughly with a spatula. The suspension was left at room temper-ature for 3 min to form large flocs. The flocs were filtered withP2 filter paper (Fisher Scientific, Fair lawn, NJ, USA) undervacuum (–27 in. Hg). Gentle stirring was necessary in the first3 min of filtration to prevent forming a dense precipitate cake,which would increase resistance to filtration and result in along filtration time. The filtration was continued for 10 s afterno more water droplets were formed at the tip of the ceramicfunnel. The precipitate was placed into a vacuum oven anddried overnight at room temperature and a vacuum of –30 in. Hg.To check the reproducibility of this process, for each formula-tion, triplicate samples were prepared.

The precipitate weight was determined after vacuum drying.A known amount of this dry powder, about 5 mg, was dis-solved into 50 mL acetonitrile/water mixture, 50:49 (vol/vol).The naproxen loading was determined by measuring the naproxenconcentration by HPLC (Shimadzu, LC-10AT VP, Kyoto,Japan). The salt concentration was determined from the electri-cal conductivity as described below. The surfactant composi-tion in the final powder was calculated by difference given thetotal precipitate weight, the drug loading, and the salt concen-tration. The drug recovery was calculated from the HPLC mea-surement and the total amount of drug fed to the suspension inthe antisolvent process. The surfactant recovery was calculatedwith the same method, given the surfactant composition in thedry powder. Wide-angle X-ray scattering was employed todetect the crystallinity of naproxen. Cu Kα1 radiation with a

wavelength of 1.54054 Å at 40 kV and 20 mA from a PhilipsPW 1720 X-ray generator (Philips Analytical Inc., Natick,MA, USA) was used. The samples were well-mixed to mini-mize the effects of preferred orientation. The reflected intensitywas measured at a 2θ angle between 5° and 45° with a step sizeof 0.05° and a dwell time of 1 s.

Particle SizeParticle size distributions based on volume fraction were

measured for the original antisolvent suspension prior to floc-culation and for the dried powders after redispersion and soni-cation with laser light scattering (Mastersizer-S, MalvernInstruments Ltd., Malvern, UK). Approximately 5 mL of thesuspension with a concentration of 10 mg drug/mL water wasdiluted with 500 mL distilled water, to produce a light obscura-tion in the desired range of 10–30%. The amount of drug in thewater was several fold above the solubility limit. In controlexperiments, samples were stirred in the Mastersizer for up to10 min, and there was very little change in the particle size dis-tribution. To study the redispersibility of the dry powders,about 100 mg dry powder was suspended into 500 mL distilledwater to produce an obscuration in the range 10–30%. After 1min, the particle size distribution was measured. Ultrasoundwas used in the measurement to break up any aggregated parti-cles. The uncertainty in mean particle size for different spraysat the same experimental conditions was about 10–15%.

Measurement of Residual Salt Concentration in Flocculated Samples by Conductivity

The residual concentration of sodium sulfate was measuredby conductivity. The conductivities of a series of standard con-centration of sodium sulfate solutions in acetonitrile/watermixture (vol/vol = 50:50) were measured with a conductivity

FIGURE 1. Process for producing pharmaceutical powder by antisolvent precipitation, flocculation with salt, filtration, and vacuum drying.

Filtrate

Retentate

Salt

ParticleSlurry

OrganicSolution

AqueousSolution

WaterBath

ParticleSlurry

Surfactant

SolventDrug

Vacuumoven

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

286 X. CHEN ET AL.

probe with cell constant of 1 cm–1 (Model 3252, YSI Inc.,Yellow Springs, OH, USA). For salt concentrations rangingfrom 0.003 to 0.05 mg/mL, a linear standard line was obtainedwith a correlation coefficient of 0.9999. A small amount, about5 mg, of dried naproxen powder was dissolved in the same aceto-nitrile/water mixture to yield a salt concentration in the linearrange of the standard curve to determine the conductivity.

Morphology of the Flocs in the Suspension and After Drying

Optical microscopy and SEM were used to visualize themorphology of the naproxen flocs in the suspension and in thedried powder. A drop of antisolvent suspension flocculatedwith sodium sulfate solution was placed onto a microscopeslide (25 × 75 mm2, Erie Scientific Co. Portsmouth, NH, USA)and carefully covered with a cover glass (22 × 22, Fisher Sci-entific, Pittsburg, PA, USA). The morphology of naproxenflocs in the suspension was examined with an optical microscope(Axioskop 2 plus, Carl Zeiss Vision GmbH, Jena, Germany).Another suspension produced with the same experimental con-ditions was filtered with a P2 filter paper under vacuum for11 min. The dried powders were mounted on an aluminum cyl-inder using double adhesive carbon conductive tabs (Ted PellaInc., Redding, CA, USA) and coated with Au for 25 s using aPelco Model 3 sputter-coater under an Ar atmosphere. A HitachiS-4500 SEM (Hitachi Instruments Inc., Irvine, CA, USA) at anaccelerating voltage of 10 kV with a secondary electron detec-tor was used to obtain digital images of the samples.

Dissolution TestThe dissolution of naproxen powder was tested in pure

water using a USP Apparatus II (Vankel 7000, Vankel Tech-nology, Cary, NC, USA) at 50 rpm. All dissolution tests wereconducted at sink conditions, in this case at 15% of its solubil-ity in water. Antisolvent powders containing approximately 2 mgnaproxen were added to 900 mL pure water at 37°C. Aliquotsof the dissolution medium (5 mL) were sampled at 2, 5, 10, 20,and 30 min. The aliquots were filtered through 0.45-μmsyringe filters and 2 mL of each sample was diluted with 0.1 mLacetonitrile before analysis. Naproxen concentrations weremeasured using HPLC (Shimadzu, LC-10AT VP, Japan). Dis-solutions were repeated in duplicate or triplicate for the powderaliquots, and the average deviations are reported in the dissolu-tion figures.

RESULTS AND DISCUSSION

Effect of Temperature on Drug Particle Size in Antisolvent Precipitation

The effect of temperature on the particle size distribution inthe aqueous suspensions produced by antisolvent precipitationwas determined. A mixture of 5% (wt/vol) naproxen and 2%

(wt/vol) poloxamer 407 in ethanol solution was sprayed into50 mL 3% (wt/vol) PVP K-15 at flow rate of 1 mL/min for 5 min.The final suspension concentration was 5 mg/mL. The organicsolution was at room temperature. As shown in Figure 2, themean particle size doubled from 270 nm at 0°C to 540 nm at30°C. At temperatures higher than 30°C, the mean particle sizeincreased markedly and reached 9.3 μm at 60°C. Several fac-tors may lead to larger particles at high temperature, includingdegree of supersaturation, nucleation rate, crystal growth rate,interparticle attractive interactions between surfactant chains,and Ostwald ripening. The solubility of drug in the organic–water mixture increases with temperature. Thus, for a givendrug concentration in the feed, the supersaturation decreaseswith temperature. The lower supersaturation lowers the nucle-ation rate. The smaller number of nuclei may be expected toproduce larger particles for a given drug concentration in thefinal suspension. Also, the diffusion rate of drug molecules tothe surface of the growing particles and the kinetics of additionof drug molecules to the surface increase with temperature.Furthermore, Ostwald ripening or diffusion of molecules fromsmaller crystals to larger crystals is also favored by higher sol-ubility at higher temperature. The solvation of ethylene oxidegroups in the poloxamer tails becomes stronger as temperaturedecreases due to hydrogen bonding between water and theether oxygen (Blankschtein, Thurston, & Benedek, 1986;Chen, Young, Sarkari, Williams, & Johnston, 2002). Greatersolvation will improve the repulsive forces between ethyleneoxide blocks on approaching particles to improve steric stabili-zation. Based on these experiments, all other experiments inthis paper were performed at or below room temperature.

Compositions of the FormulationsAs shown in Table 1, four systems were studied. In system A,

only 10% (wt/vol) PVP K-15 was used in the organic phase

FIGURE 2. Temperature effect on particle size of naproxen suspensionsproduced by antisolvent precipitation. Organic phase: 5% (wt/vol)naproxen + 2% (wt/vol) poloxamer 407 in ethanol; aqueous phase: 50 mL3% (wt/vol) PVP K-15; flow rate Q = 1 mL/min; t = 5 min; suspensionconcentration = 5 mg/mL.

0123456789

10

0 10 20 30 40 50 60Temperature (°C)

D (v

,0.5

) afte

r son

icat

ion

(μ m

)

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

FLOCCULATION NANOCRYSTAL SUSPENSIONS 287

and there was no surfactant in the aqueous phase. This systemwas designed for long-term stability since PVP K-15 has a highglass transition temperature which enhances drug stability bydecreasing the mobility of drug molecules. In system B, 10%(wt/vol) poloxamer 407 was used in the organic phase, as it hasbeen shown to be an effective stabilizer (Chen, Lo, Sarkari,Williams, & Johnston, 2006; Chen et al., 2002; Matteucciet al., 2006) and 3% (wt/vol) PVP K-15 was included in theaqueous phase to raise the Tg of the final powder. To attempt toraise the loading of the drug, in system C the poloxamer 407concentration was reduced to 5% (wt/vol) and in system D, thePVP K-15 concentration was reduced to 1% (wt/vol). Thedrug/poloxamer 407/PVP K-15 ratios in the final suspensionbased on the amounts fed from both organic and aqueous phaseduring spray time are also given. The total solid weight andtotal drug loading (weight of drug/total weight excluding salt)are listed for each system.

Particle Size of Naproxen in Aqueous SuspensionsAs shown in Tables 2 and 3, aggregated naproxen nanopar-

ticles were produced during antisolvent precipitation at both22 and 3°C. With only PVP K-15 in the organic phase andpure water in the aqueous phase (system A, drug/poloxamer407/PVP K-15 ratio = 1:0:1.4), the particle sizes were large inthe original suspensions after spraying at either 22 or 3°C. Evenafter sonication in the laser light scattering chamber, the meanparticle size was still undesirably large, as shown in Table 2.After the original suspensions were stirred with a magnetic stirbar for 20 h at room temperature and then sonicated, the meanparticle sizes became extremely small, well below 0.4 μm atboth temperatures. More surfactant may adsorb to the particlesduring stirring and aid breakup of the aggregates.

In order to attempt to reduce the aggregation of the particlesat 3°C, 5% (wt/vol) poloxamer 407 was added to the aqueousphase to modify system A. The mean particle size of the origi-nal suspension was 15.9 μm. After 6 min sonication in theMastersizer, the mean particle size decreased to 0.29 μm. Inthis case, the aggregates broke up into small particles withoutthe need to stir the solution for 20 h. Because the copolymerpoloxamer 407 has hydrophobic moieties that adsorb ontohydrophobic drug particle surfaces and two separate hydro-philic blocks, the adsorption of poloxamer 407 with PVP K-15

may be expected to provide greater steric repulsion and looseraggregates. The looser aggregates may be broken up more eas-ily by sonication on the basis of the smaller particle sizes.

The particle sizes for the formulations in addition to systemA are presented in Table 3. The results for the suspensionsprior to flocculation are shown in the second and third col-umns. To avoid the need for stirring for long periods of time tobreak up aggregates, the organic phase stabilizer was changedto poloxamer 407 for all of the systems in Table 1 except sys-tem A. In addition, PVP K-15 was utilized in the aqueousphase to achieve a sufficiently high Tg for the final powder. Forsystems B and C, the mean particle sizes were extremely small,below 0.5 μm, in the original suspension even without sonica-tion. After sonication, the particle size decreased only a smallamount. However, when the PVP K-15 concentration in theaqueous phase was lowered to 1% (wt/vol) in system D, themean particle size of the original suspension increased to 9.8 μm.These aggregates readily broke up upon sonication. After soni-cation the mean particle size was 0.29 μm for systems B–D,although the D (v, 0.9) was 7.0 μm for system D. Without son-ication, a higher concentration of 3% (wt/vol) PVP K-15 in theaqueous phase was helpful in forming the nanosuspension. Insystems B and C, the overall concentration of PVP K-15 wasmuch higher than that of poloxamer 407, unlike the case forsystem D. The higher overall stabilizer concentration may havebeen required to provide enough steric stabilization to preventaggregation of naproxen nanoparticles.

Polymer-Salt Cloud Point Concentrations and Flocculation of Naproxen Nanoparticles with Various Salts

Sterically stabilized dispersions may be flocculated byreducing the solvency of the dispersion medium for the stabi-lizing moieties to induce the onset of instability. Pelton (1988)reported that the critical flocculation temperature (CFT) corre-sponds to the cloud point temperature of the stabilizing poly-mer. A few reports have described the influence of inorganicsalts on the cloud point behavior of poloxamers (Bahadur, Li,Almgren, & Brown, 1992; Bahadur, Pandya, Almgren, Li, &Stilbs, 1993; Pandit, Trygstad, Croy, Bohorquez, & Koch,2000) and PVP (Güner & Ataman, 1994; Salamova & Rzaev,1996; Sekikawa, Hori, Arita, Ito, & Nakano, 1978). As shownin Figure 3, the cloud point temperatures of PVP 44,000 and

TABLE 1 Systems Investigated

Systems Organic Phase Aqueous PhaseDrug/Poloxamer/PVP

Ratio in the SuspensionTotal Solid Weight (g)

Original Drug Loading (%)

A 10% PVP K-15 Pure water 1/0/1.4 1.22 41.2B 10% poloxamer 407 3% PVP K-15 1/1.4/3 2.72 18.5C 5% poloxamer 407 3% PVP K-15 1/0.7/3 2.36 21.3D 5% poloxamer 407 1% PVP K-15 1/0.7/1 1.36 37.0

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

TA

BL

E 2

Pa

rtic

le S

ize

of N

apro

xen

Susp

ensi

ons

Prod

uced

by

Ant

isol

vent

Pre

cipi

tatio

n V

ersu

s T

ime

for

Syst

em a

With

out a

ny A

dded

Sal

t

Tem

pera

ture

T =

22°

CT

= 3

°C

Susp

ensi

onD

(v, 0

.1/0

.5/0

.9)

w/o

Son

icat

ion

(μm

)

D(v

, 0.1

/0.5

/0.9

) Aft

er

Soni

cati

on w

ith

Mas

ters

izer

(μm

)So

nica

tion

T

ime

(min

)D

(v, 0

.1/0

.5/0

.9)

w/o

Son

icat

ion

(μm

)

D(v

, 0.1

/0.5

/0.9

) A

fter

Son

icat

ion

with

M

aste

rsiz

er (

μm)

Soni

cati

on

Tim

e (m

in)

Ori

gina

l sus

pens

ion

in D

I w

ater

7.5/

15.9

/28.

42.

4/5.

3/7.

27

6.2/

12.6

/153

.12.

3/4.

5/6.

315

Aft

er 2

0 h

stir

ring

at r

oom

te

mpe

ratu

re4.

5/6.

4/11

.00.

15/0

.36/

2.1

36.

1/12

.1/1

9.7

0.11

/0.2

9/0.

885

Ori

gina

l sus

pens

ion

in 5

%

(wt/

vol)

pol

oxam

er 4

076.

9/15

.9/4

2.4

0.10

/0.2

9/3.

36

Org

anic

pha

se: 7

% (

wt/v

ol)

napr

oxen

+ 1

0% (

wt/v

ol)

PVP

K-1

5 in

met

hano

l.

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

TA

BL

E 3

Pa

rtic

le S

ize

of N

apro

xen

in th

e O

rigi

nal S

uspe

nsio

n B

efor

e an

d A

fter

Son

icat

ion,

and

Aft

er F

locc

ulat

ion,

Fil

trat

ion,

Vac

uum

Dry

ing,

an

d R

edis

pers

ion

into

Pur

e W

ater

Syst

ems

D(v

, 0.1

/0.5

/0.9

) in

the

Ori

gina

l Sus

pens

ion

(μm

)

D(v

, 0.1

/0.5

/0.9

) A

fter

5 m

in

Soni

cati

on (

μm)

V o

f Sa

lt (

mL

)C

salt in

Su

spen

sion

(M

)

D(v

, 0.1

/0.5

/0.9

) Aft

er

Red

ispe

rsio

n in

to

Wat

er (

μm)

D(v

, 0.1

/0.5

/0.9

) Aft

er

Red

ispe

rsio

n an

d So

nica

tion

(μm

)t s

on. (

min

)R

elea

se in

2

min

(%

)

A4.

5/6.

4/11

.00.

15/0

.36/

2.1

175

1.10

16.5

/91.

6/24

3.0

0.12

/0.3

2/3.

243

27.3

A20

01.

1329

.4/1

55.7

/447

.10.

22/2

.02/

11.6

625

.5B

0.14

/0.4

8/95

.20.

11/0

.29/

1.01

100

0.94

0.10

/0.3

0/6.

80.

10/0

.27/

3.07

195

.4B

125

1.01

0.11

/0.3

1/1.

870.

11/0

.28

/1.1

31

96.4

B15

01.

060.

23/0

.48/

19.3

0.11

/0.3

2/7.

721

74.8

C0.

11/0

.30/

3.8

0.12

/0.2

9/0.

6812

51.

010.

11/0

.31/

2.57

0.10

/0.2

5/0.

791

94.9

D0.

14/9

.8/1

73.9

0.10

/0.2

9/6.

9512

51.

010.

28/2

.65/

22.3

0.19

/0.3

8/4.

102

49.5

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

290 X. CHEN ET AL.

poloxamer 407 decrease linearly with increasing concentra-tions of sodium sulfate. At a given temperature, the polymerprecipitates as the salt concentration increases. In Figure 3, theconcentration of PVP was 0.74% (wt/vol) on the basis of a par-tial specific volume of 0.952 cm3/g for PVP (Pleštil & Al,2001). The concentration of poloxamer 407 was 5 mg/mL or0.5% (wt/vol). PVP contains N–C = O units on the lactam ringsthat hydrogen bond with water as observed with viscometricmeasurements (Guven & Eltan, 1981) and spectrophotometricstudies (Turker, Guner, Yigit, & Guven, 1990). The decreaseof the Huggins constant with the addition of inorganic salt intoPVP aqueous solution indicates a loss in water–PVP H-bonding,and thus in PVP hydration, leading to precipitation (Salamova& Rzaev, 1996). For the same reasons, salt also precipitatesPEO homopolymers (Khoultchaev, Kerekes, & Englezos, 1997a;Khoultchaev, Kerekes, & Englezos, 1997b; Pang & Englezos,2002a; Pang & Englezos, 2002b) and of PEO-containing non-ionic surfactants such as poloxamers (Bahadur et al., 1992;Bahadur et al., 1993; Pandit et al., 2000).

The reciprocal of cloud point temperatures, 1/Tcp of twofractionated and two unfractionated PVP samples have beenshown to be linear in (Güner & Ataman, 1994). Witha lower molecular weight of 10,000 PVP, more sodium sulfateis required to precipitate the PVP K-15 used in this study thanfor the PVP 44,000 shown in Figure 3 (Sekikawa et al., 1978).

The stabilizing moieties collapse at the onset of the cloudpoint where the solvent is worse than a θ-solvent, as a result ofthe large number of interactions of the hydrophilic groups(Napper, 1983). Steric repulsion becomes weak and the stabi-lizing chains interact with each other leading to sticky Brown-ian collisions and flocculation. Some free surfactant might alsoprecipitate out of solution and interact with flocculating parti-cles. Because the particle size of coated naproxen was typically

larger than 200 nm and the molecular weight of stabilizingpolymers was not more than 12,500, the steric stabilization ofthe polymer chains does not completely screen the van derWaals attraction between the particles (Napper, 1983).

To more fully understand the effect of salinity on floccula-tion, the minimum salinity to desolvate polymer was measuredfor a 1.26% (wt/vol) concentration of either PVP K-15 orpoloxamer 407. The solvent was a mixture of methanol(12.6%, vol/vol) and water (87.4%, vol/vol) with the samepolymer composition as the system A suspension. A concen-tration of 1.06 M sodium sulfate was needed to precipitate PVPfrom solution at room temperature in the methanol/water mix-ture while only 0.98 M was needed for the case of methanol-free water. For pure poloxamer 407, 0.71 M sodium sulfateconcentration was needed for the methanol/water mixture,while 0.58 M was needed without methanol. Apparently, at thesame concentration and temperature, poloxamer 407 was pre-cipitated much more easily with sodium sulfate than was PVPK-15, which is consistent with Figure 3. Less salt is needed todesolvate poloxamer 407 than PVP (Mw = 44,000) at 25°Cfrom pure water. This difference would be expected to be evenlarger for PVP K-15 as a greater salt concentration would berequired to collapse the lower molecular weight (Sekikawaet al., 1978). The salinities required to precipitate the polymers,despite the presence of methanol, are in the same orderexpected from the cloud points. Methanol also raises the cloudpoint temperature for poloxamine 908, a copolymer containingpolyethylene oxide moieties (Na, Yuan, Stevens, Weekley, &Rajagopalan, 1999).

A much lower salt concentration is required for a divalentsulfate to precipitate these polymers relative to a monovalentanion (Pandit et al., 2000; Salamova & Rzaev, 1996). Threemultivalent inorganic salts which are known to produce a largereduction in the cloud point temperature of PVP (Salamova &Rzaev, 1996) and poloxamer 407 (Pandit et al., 2000), sodiumcarbonate, sodium sulfate, and sodium phosphate, were used toattempt to flocculate polymer-stabilized naproxen nanoparti-cles. Large flocs were formed with sodium sulfate, and theywere readily filterable. Solutions of 20% (wt/vol) sodium car-bonate and 10% (wt/vol) sodium phosphate were also testedfor flocculation. In each case, when a small volume of the saltsolution was added to the suspension to produce a molarity lessthan 0.04 M, a clear solution was formed as the naproxen parti-cles dissolved in the basic media. When 0.4 mL sodium car-bonate or 1 mL sodium phosphate solution was added to 25 mLof the suspension, the pH of the solution increased to 11.3 forCO3

2– and 11.9 for PO43– due to the acid–base hydrolysis. The

solubility of naproxen increases from 0.0159 mg/mL in purewater to 196.7 mg/mL at a pH higher than 8 (Chowhan, 1978),which is well above the pKa of naproxen. This solubility ismuch higher than the naproxen concentration in the suspen-sion, consistent with our observation of dissolution. Based onthese results all of the flocculation experiments were per-formed with sodium sulfate.

FIGURE 3. Cloud point temperature of PVP (Salamova & Rzaev, 1996)and poloxamer 407 (Pluronic F127) (Pandit et al., 2000) at various sodiumsulfate concentrations in water.

10

20

30

40

50

60

70

80

90

100

0 0.2 0.4 0.6 0.8Sodium sulfate concentration (mol/L)

Clo

ud p

oint

tem

pera

ture

(°C

)

Pluronic F127

PVP 44, 000

11 2

M w/

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

FLOCCULATION NANOCRYSTAL SUSPENSIONS 291

Dissolution RateThe dissolution rates of the dried naproxen powders are

shown in Figure 4A for System B and for various formulationsin Figure 4B. The dissolution rates were reproducible as shownby the error bars that represent the average deviation, definedby . The dissolution rates were very high forsystems B and C, and significantly slower for systems A andD. As shown in Figure 4A, approximately 100% naproxen wasdissolved in 2 min for the suspension dried by lyophilization.For samples flocculated with 100 or 125 mL salt solution, thedissolution rates of naproxen were only slightly slower than forthe lyophilized ones. The flocculated sample of system B wasstored with desiccant under vacuum at room temperature. After1 month, the dissolution rate of this sample did not change, asshown in Figure 4B. The differences in these dissolution rateswill now be analyzed in terms of the particle size distributions.

Particle Redispersibility into Water after Flocculation Followed by Filtration and Vacuum Drying

The dry powder was redispersed into pure water, and theparticle sizes are shown in the sixth and seventh columns ofTable 3. For system A flocculated with a salt concentration of1.10 or 1.13 M, the filtration time was approximately 8 minand the dried naproxen particles redispersed to form a verycoarse suspension upon stirring without sonication. The meanparticle size of the dried particles was 91.6 μm at 1.10 M saltconcentration, but decreased to 0.32 μm after 3 min sonication.With a higher salt concentration of 1.13 M, the mean particlesize of naproxen decreased from 155 μm without sonication to2.02 μm after 6 min sonication. Because the mean particle sizewas only 0.29 μm after flocculation and filtration, but beforedrying, the particle growth must have taken place during thevacuum drying step. The slow dissolution rates in system A areconsistent with the large particle sizes upon redispersion with-out sonication.

To study the reversibility of flocculation, a suspension floc-culated at 1.13 M salt concentration was filtered with a 0.45-μmWhatman® nylon membrane filter paper (Whatman Interna-tional Ltd., Maidstone, England) for 43 min. The particle sizeof redispersed naproxen after drying was 17.1 μm after sonica-tion. Apparently, if the particles are allowed to remain floccu-lated for more than 40 min, redispersion is no longer complete.Perhaps slow diffusion of the stabilizer away from the high-energy zones that are created by the close approach of theparticles leads to irreversible aggregation (Napper, 1983). Incontrast, the filtration times were on the order of 10 min for theother experiments.

For system B, with either 0.94 or 1.01 M salt, the dried par-ticles could be redispersed into water readily with particle sizessimilar to the original suspensions. Consequently, the dissolu-tion rates were extremely rapid with 96% release in 2 min. Asthe salt concentration was increased to 1.06 M, the mean sizeof redispersed particles increased modestly to 0.48 μm with a

(1/ )n x x−∑

FIGURE 4. (A) Effect of salt concentration on dissolution rate offlocculated, filtered and vacuum-dried naproxen nanoparticles produced byantisolvent precipitation (system B). Conditions were the same as in Tables3 and 4. As indicated, the suspension was lyophilized in one case. (B) Effectof stabilizers on dissolution rate of flocculated, filtered, and vacuum-driednaproxen nanoparticles produced by antisolvent precipitation. Conditionswere the same as in Tables 3 and 4. (C) Correlation between dissolution rateand specific surface area of flocculated, filtered, and vacuum-dried naproxenparticles (R2 = .97).

0

20

40

60

80

100(A)

0 10 20 30Time (min)

Rel

ease

(%)

0.94 M salt (1) (n = 2)

0.94 M salt (2) (n = 2)

1.01 M salt (n = 3)

1.06 M salt (n = 2)

Suspensionlyophilized (n = 3)

Bulk naproxen (n = 3)

0

20

40

60

80

100

(B)

0 10 20 30Time (min)

Rel

ease

(%)

System A + 1.10 M salt(n = 2)

System A + 1.13 M salt(n = 2)

System D + 1.01 M salt(n = 2)

System C + 1.01 M salt(n = 2)

System B + 1.01 M salt(original, n = 3)

System B + 1.01 M salt(after 1 month, n = 3)

0

20

40

60

80

100

(C)

0 5 10 15 20Specific surface area calculated from particle size ( µm–1)

Intia

l dis

solu

tion

rate

(% re

leas

ed in

2 m

in)

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

292 X. CHEN ET AL.

high D (v, 0.9) of 19.3 μm. Likewise, the dissolution ratedecreased to 75% in 2 min. Here, 1 min sonication wasneeded to reduce the particle size to the original value prior toflocculation. Higher salt concentrations appeared to producetighter flocs after drying, requiring sonication to fully breakthem up, whereas sonication was not needed for the lower saltconcentrations.

For system C, particles flocculated at 1.01 M salt concentra-tion could be redispersed easily after drying to achieve a simi-lar particle size as in the original dispersion. Consequently, thedissolution rate was extremely fast as in the case for the twolowest salinities in system B. However, for system D, with alow PVP K-15 concentration in the aqueous phase, the meanparticle size of the redispersed dried powders without sonica-tion was extremely large, 2.7 μm. The size decreased most ofthe way to the original value prior to flocculation after 2 minsonication. The dissolution rate was much slower than in sys-tems B and C. Therefore, the dissolution rate is more closelyrelated to the aggregate size than the size of the particles aftersonication.

For all of the examples in Table 3, the dissolution rate wascorrelated closely with the particle size of the redispersed pow-der without sonication. Assuming spherical particles, thespecific surface area of naproxen particles was calculated by,S = 6/D, where D is the average diameter of the particles fromlight scattering shown in the sixth column of Table 3. Asshown in Figure 4C, a straight line with a correlation coeffi-cient of 0.97 was obtained between initial dissolution rate(% released in 2 min) and specific surface area of naproxenparticles after redispersion. This correlation is consistent withNoyes–Whitney equation, where the dissolution rate is propor-tional to the specific surface area of drug particles.

Reproducibility of Salt Flocculation Process on Precipitate Weight, Drug Loading, Salt Concentration, Surfactant Concentration, Drug, and Surfactant Yield

The filtration time and properties of the precipitate includ-ing naproxen loading (drug/total solids, wt/wt), drug yield(fraction in precipitate vs. total amount fed), surfactant yield(fraction in precipitate), and filtration selectivity ([g precipitatedrug/g filtrate drug]/[g precipitate surfactant/g filtrate surfac-tant]) for four systems flocculated with various concentrationsof sodium sulfate are shown in Table 4. Compared to the origi-nal naproxen loadings in Table 1 for the suspensions, thenaproxen loadings in the precipitate were higher, since somefree or nonadsorbed surfactant was removed with the filtrate.They were higher despite the presence of salt in Figure 4, rela-tive to no salt for the drug loadings in Table 1. For system A,even at a salt concentration of 1.10 M, the resulting flocs werenot stable. Stirring or pouring of the suspension broke up thefragile flocs. Thus, the precipitate weight was not reproducible,relative to the other systems. The poor flocculation was due toa salinity too close to the critical flocculation salinity, 1.06 M

with 12.6% (vol/vol) methanol in the suspension. When saltconcentration was increased to 1.13 M, larger flocs formed andconsequently the drug and surfactant yields increased.

For system B, a control was performed in which the suspen-sion was filtered without adding any salt. Without flocculationby salt, only 0.050 g precipitate was recovered as the particleswere too small for the filter. At salt concentration of 0.47 M,only a small amount of flocculation occurred, and the drug par-ticles plugged up the filter paper in 5 min. As the salt concen-tration increased from 0.94 to 1.01 to 1.06 M, the drug andsurfactant yields increased significantly. A drug yield higherthan 92% was obtained at a salt concentration higher than1.4 times the critical flocculation salinity of poloxamer 407.The naproxen loading decreased a small amount with salinityand was about double the original loading. The relative devia-tion (average deviation/mean value) of the precipitate weightdecreased from 5.5% with 0.94 M salt to 0.9% with 1.01 or1.06 M salt. The increase of reproducibility of precipitateweight and drug yield with salt concentration likely indicatesmore stable, stronger and larger aggregates that could be fil-tered more effectively. An increase in aggregate size with anincrease in the distance above the cloud point temperature hasalso been observed for clay particles stabilized with PEOformed at higher temperature above the cloud point (Pang &Englezos, 2002a). The increase in the flocculation efficiencywas attributed to an increase in the tendency for the PEO tophase separate and adsorb onto the clay particles (Pang &Englezos, 2002a).

As the poloxamer 407 concentration was lowered twofoldin system C relative to system B, for a given salinity, the drugloading increased from 36.7 to 54.1%, yet the dissolution rateremained extremely high. The drug loading was over 2.5 timesthat of the original loading in the suspension. The drug yieldincreased to 97.6% while the surfactant yield decreased signif-icantly. Furthermore, the particle size changed very little.Given these beneficial results, the PVP K-15 concentration wasreduced by a factor 3 in system D relative to system C and thedrug yield increased to nearly 100%. The concentrations ofdrug and surfactant were similar, while the salt concentrationdecreased to 6.4%.

The selectivity for drug particles over surfactants ([g precip-itate drug/g filtrate drug]/[g precipitate surfactant/g filtrate sur-factant]) in the filtration process was very high and usuallyquite reproducible as shown is the last column of Table 4. Thehigh deviation for system C was due to high drug yield, andthus the very small amount of drug in the filtrate. The excellentreproducibility in nearly all of the properties in Table 4 formost systems makes the flocculation/filtration concept relevantfor practical application.

Based on dissolution rate and drug recovery, systems B andC, flocculated at 1.01 M salt concentration, may be expected toproduce the highest drug bioavailability. The recommendeddaily dose of naproxen in adults for rheumatoid arthritis,osteoarthritis, and ankylosing spondylitis is 500–1,000 mg

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

TA

BL

E 4

Pr

oper

ties

of th

e Pr

ecip

itat

e A

fter

Vac

uum

Dry

ing

Incl

udin

g Se

lect

ivit

y fo

r D

rug

Ver

sus

Sur

fact

ant

Syst

ems

Csa

lt in

Su

spen

sion

(M

)Fi

ltrat

ion

Tim

e (m

in)

Prec

ipat

ion

Wei

ght (

g)D

rug

Loa

ding

(%

)

Salt

C

once

ntra

tion

(%, w

t/wt)

Surf

acta

nt

Con

c.

(%, w

t/wt)

Dru

g Y

ield

(%

)Su

rfac

tant

Y

ield

(%

)Se

lect

ivity

fo

r Fi

ltrat

ion

A1.

108.

60.

5083

± 0

.099

4 (n

= 3

)69

.0 ±

0.2

(n

= 3

)25

.2 ±

0.3

(n =

3)

5.8

± 0.

4 (n

= 3

)70

.2 ±

14.

0 (n

= 3

)4.

4 ±

0.2

(n =

3)

94.5

± 7

3.9

A1.

137.

70.

6552

± 0

.003

0 (n

= 2

)66

.3 ±

0.7

(n

= 2

)19

.6 ±

0.7

(n

= 2

)14

.1 ±

1.4

(n

= 3

)86

.9 ±

1.3

(n

= 3

)12

.9 ±

1.2

(n

= 3

)45

.5 ±

6.8

B0.

9416

.51.

0768

± 0

.059

6 (n

= 3

)38

.2 ±

0.8

(n

= 3

)10

.6 ±

1.4

(n

= 3

)51

.2 ±

1.1

(n

= 3

)79

.7 ±

1.4

(n

= 3

)24

.1 ±

0.9

(n

= 3

)12

.4 ±

0.5

B1.

0111

1.25

23 ±

0.0

113

(n =

4)

36.7

± 0

.7

(n =

4)

10.5

± 0

.5

(n =

4)

52.9

± 0

.5

(n =

3)

91.8

± 1

.4

(n =

3)

29.9

± 0

.3

(n =

3)

28.8

± 8

.5

B1.

0611

.51.

4825

± 0

.013

9 (n

= 3

)31

.8 ±

0.7

(n =

3)

9.3

± 0.

5 (n

= 3

)58

.9 ±

1.0

(n

= 3

)94

.1 ±

1.7

(n

= 3

)39

.4 ±

1.0

(n

= 3

)28

.9 ±

12.3

C1.

0112

0.90

21 ±

0.0

017

(n =

2)

54.1

± 0

.5

5 (n

= 2

)8.

8 ±

0.0

(n =

2)

37.1

± 0

.7

(n =

2)

97.6

± 1

.5

(n =

2)

18.0

± 0

.3

(n =

2)

316.

1 ±

206.

55

D1.

0114

0.89

3055

.7 ±

0.5

(n

= 2

)6.

4 ±

0.2

(n =

2)

37.9

(n

= 2

)99

.4

(n =

2)

39.6

(n

= 2

)25

6.8

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

294 X. CHEN ET AL.

(Fuller & Sajatovic, 1999). According to the ratio of sodiumsulfate to naproxen in systems B or C, the daily dose of sodiumsulfate from these powders would be 81–286 mg. This amountis much less than the commercial daily dose from OCL®, asaline laxative (Abbott), in which 1.29 g sodium sulfate wasused, indicating that the sodium sulfate would not present tox-icity limitations.

Morphology of Flocculated Naproxen Nanoparticles by Optical Microscope and SEM

The morphology of naproxen nanoparticle flocs of system Bflocculated at 1.01 M salt concentration in the suspension and afterdrying is shown in Figure 5A and B. As shown in Figure 5A,naproxen flocs with sizes ranging from 5 to 150 μm wereformed in the suspension. These flocs were larger than the

2 μm limit for P2 filter paper and thus more than 90% ofthe naproxen was recovered. As shown in Figure 5B (leftpanel), a floc with size larger than 30 μm was formed duringflocculation and drying. This particle, however, was composedof primary nanoparticles of approximately 300 nm, shown onthe right, consistent with the size measured by light scattering.

X-Ray DiffractionX-ray diffraction was used to analyze the crystallinity of the

dry powders. As shown in Figure 6, the largest naproxen peakscoincided with those of bulk poloxamer 407. The naproxenpeaks were much smaller and broader in the physical mixturesof the various formulations as well in the dried powders pro-duced by antisolvent precipitation. However, the naproxenpeaks at 2q = 22.6° and 23.9°, especially in the samples with

FIGURE 5. (A) Microscopic picture of naproxen flocs of system B in suspension at salt concentration of 1.01 M. (B) and (C) SEM pictures of naproxen flocsafter filtration and vacuum drying. Same sample as in A.

(B) (C)

(A)

50 μm

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

FLOCCULATION NANOCRYSTAL SUSPENSIONS 295

high drug loading produced by flocculation, were evident.Thus, crystalline naproxen was formed whether the suspensionwas dried by lyophilization or flocculation followed by filtra-tion and vacuum drying. The presence of sodium sulfate in thefinal powder is also apparent in the diffraction pattern.

CONCLUSIONSCrystalline naproxen submicron particles were formed by

antisolvent precipitation in the presence of polymeric stericstabilizers. The nanocrystals were successfully recovered fromaqueous suspensions by flocculation with sodium sulfate fol-lowed by filtration and vacuum drying. The flocculation wasoften reversible as the particle size upon redispersion in water,without sonication, was on the order of only 300 nm and com-parable to the original particle size in the aqueous suspensionprior to flocculation. The size measured by light scattering wasconsistent with the primary particle size in the powders mea-sured by SEM. Less salt was needed for flocculation when thestabilizer was poloxamer 407, with hydrophilic ethylene oxidegroups, than for PVP K-15, with N–C=O units on lactam rings,consistent with the trends in the cloud point phase equilibriacurves. The dissolution rate was linearly correlated with thespecific surface area calculated from the average particle diam-eter after redispersion without sonication. Extremely rapid dis-solution, up to 95% of the powder in 2 min, was achieved for300-nm particles. The salt concentration could be optimized tocontrol the flocculation to balance the drug yield versus load-ing with excellent reproducibility in both properties (averagedeviation ranged in 1–2%). The drug loading of flocculatedsamples was enhanced in the filtration step by up to 61%relative to the initial value, due to removal of surfactant withthe filtrate. The yield of the drug in the powder was typically92–99% and the drug loading varied by only 1–2%. The residualsodium sulfate concentration in the dried powders, typically

10% or less was far below toxic limits based on the recom-mended dose of naproxen. The flocculation/filtration processsimplifies the drying step relative to lyophilization and spraydrying. Flocculation followed by filtration and drying is anefficient and highly reproducible process for the rapid recoveryof drug nanoparticles to produce wettable powders with highdrug loading and high dissolution rates.

ACKNOWLEDGMENTSThe authors would like to gratefully acknowledge the finan-

cial support from The Dow Chemical Company (Midland, MI,USA), partial support from the STC Program of the NationalScience Foundation under Agreement No. CHE-9876674, theWelch Foundation and from the Process Science and Technol-ogy Center at the University of Texas.

REFERENCESBahadur, P., Li, P., Almgren, M., & Brown, W. (1992). Effect of potassium

fluoride on the micellar behavior of Pluronic F-68 in aqueous solution.Langmuir, 8, 1903–1907.

Bahadur, P., Pandya, K., Almgren, M., Li, P., & Stilbs, P. (1993). Effect ofinorganic salts on the micellar behavior of ethylene oxide-propylene oxideblock copolymers in aqueous solution. Colloid Polym. Sci., 271, 657–667.

Blankschtein, D., Thurston, G. M., & Benedek, G. B. (1986). Phenomenologi-cal theory of equilibrium thermodynamic properties and phase separation ofmicellar solutions. J. Chem. Phys., 85, 7268–7288.

Chen, X., Lo, C. Y., Sarkari, M., Williams, R. O., & Johnston, K. P. (2006).Ketoprofen nanoparticles in gels by evaporative precipitation into aqueoussolution. Am. Inst. Chem. Eng. J., 52, 2428–2435.

Chen, X., Vaughn, J. M., Yacaman, M. J., Williams, R. O., & Johnston, K.P. (2004). Rapid dissolution of high potency danazol particles producedby evaporative precipitation into aqueous solution. J. Pharm. Sci., 93,1867–1878.

Chen, X., Young, T. J., Sarkari, M., Williams, R. O., & Johnston, K. P. (2002).Preparation of cyclosporine A nanoparticles by evaporative precipitationinto aqueous solution. Int. J. Pharm., 242, 3–14.

Chowhan, Z. T. (1978). pH-solubility profiles of organic carboxylic acid andtheir salts. J. Pharm. Sci., 67, 1257–1260.

Dixon, D. J., Johnston, K. P., & Bodmeier, R. A. (1993). Polymeric materialsformed by precipitation with a compressed fluid antisolvent. AIChE J., 39,127–39.

Elder, E. J., Hitt, J. E., Rogers, T. L., Tucker, C. J., Saghir, S., Svenson, S., &Evans, J. C. (2003). Particle engineering of poorly water soluble drugs bycontrolled precipitation. Polym. Mater. Sci. Eng., 89, 741.

Falk, R., Randolph, T. W., Meyer, J. D., Kelly, R. M., & Manning, M. C.(1997). Controlled release of ionic compounds from poly(L-lactide) micro-spheres produced by precpitation with a compressed antisolvent. J. Control.Release, 44, 77–85.

Fuller, M. A., & Sajatovic, M. (1999). Drug Information Handbook for Psychi-atry 1999–2000. Hudson, OH: Lexi-Comp.

Güner, A., & Ataman, M. (1994). Effects of inorganic salts on the propertiesof aqueous poly(vinylpyrrolidone) solutions. Colloid & Polym. Sci., 272,175–180.

Guven, O., & Eltan, E. (1981). Molecular association in aqueous solutions ofhigh-molecular-weight poly(N-vinyl-2-pyrrolidone). Makromol. Chem., 11,3129–3134.

Jarmer, D. J., Lengsfeld, C. S., & Randolph, T. W. (2004). Nucleation andgrowth rates of poly(L-lactic acid) microparticles during precipitation with acompressed-fluid antisolvent. Langmuir, 20, 7254–7264.

Johnson, B. K., & Prud’homme, R. K. (2003). Engineering the direct precipita-tion of stabilized organic and block copolymer nanoparticles as unique com-posites. Polym. Mater. Sci. Eng., 89, 744–745.

FIGURE 6. X-ray diffraction of flocculated, filtered, and vacuum-dried orlyophilized naproxen particles produced by antisolvent precipitation.Conditions were the same as in Tables 3 and 4.

0

2,000

4,000

6,000

8,000

10,000

12,000

5 15 25 35 452θ

Inte

nsity

Bulk naproxen

Lyophilized system B

Flocculated system B (125 mL salt)

Flocculated system A (175 mL salt)Sodium sulfate

Physical mixture

Bulk PVP K-15

Bulk Pluronic F127

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.

296 X. CHEN ET AL.

Khoultchaev, K. K. H., Kerekes, R. J., & Englezos, P. (1997a). The role ofpolyethylene oxide-water solution phase behavior in the retention of fibrefines and clay. Can. J. Chem. Eng., 75, 161–166.

Khoultchaev, K. K. H., Kerekes, R. J., & Englezos, P. (1997b). Temperature-dependent behavior of polyethylene oxide in papermaking suspensions.AIChE J., 43, 2353–2358.

Lipinski, C. (2002). Poor aqueous solubility—an industry wide problem indrug delivery. Am. Pharm. Rev., 5, 82–85.

Liu, Y., Kathan, K., Saad, W., & Prud’homme, R. K. (2006). Ostwald ripeningof β-carotene nanoparticles. Phys. Rev. Lett., 98, 036102.

Matteucci, M. E., Hotze, M. A., Williams III, R. O., & Johnston, K. P. (2006).Drug nanoparticles by antisolvent precipitation: Mixing energy versus sur-factant stabilization. Langmuir, 22, 8951–8959.

Na, G. C., Yuan, B. O., Stevens, H. J., Weekley, B. S., & Rajagopalan, N.(1999). Cloud point of nonionic surfactants: Modulation with pharmaceuti-cal excipients. Pharm. Res., 16, 562–568.

Napper, D. H. (1983). Polymeric Stabilization of Colloidal Dispersions.New York: Academic Press.

Oh, D., Curl, R., & Gl, A. (1993). Estimating the fraction dose absorbed fromsuspension of poorly water soluble compounds in humans: A mathematicalmodel. Pharm. Res., 10, 264–270.

Palakodaty, S., & York, P. (1999). Phase behavioral effect on particle forma-tion processes using supercritical fluids. Pharm. Res., 16, 976–985.

Pandit, N., Trygstad, T., Croy, S., Bohorquez, M., & Koch, C. (2000). Effect ofsalts on the micellization, clouding, and solubilization behavior of pluronicF127 solutions. J. Colloid Interface Sci., 222, 213–220.

Pang, P., & Englezos, P. (2002a). Kinetics of the aggregation of polyethyleneoxide at temperature above the polyethylene oxide-water cloud point tem-perature. Colloids Surf A: Physicochem. Eng. Asp., 204, 23–30.

Pang, P., & Englezos, P. (2002b). Phase separation of polyethylene oxide(PEO)-water solution and its relationship to the flocculating capability of thePEO. Fluid Phase Equilib., 194–197, 1059–1066.

Pelton, R. H. (1988). Polystyrene and polystyrene-butadiene latexes stabilizedby poly(N-isopropylacrylamide). J. Polym. Sci. A, 26, 9–18.

Pleštil, J., & Al, E. (2001). Small-angle neutron scattering study of onion-typemicelles. Macromol. Chem. Phys., 202, 553–563.

Rabinow, B. E. (2004). Nanosuspensions in drug delivery. Nat. Rev. Drug Dis-cov., 3, 785–796.

Radtke, M. (2001). Pure drug nanoparticles for the formulation of poorly solu-ble drugs. New Drugs, 3, 62–68.

Randolph, T. W., Randolph, A. D., Mebes, M., & Yeung, S. (1993). Sub-micrometer-sized biodegradable particles of poly(L-lactic acid) via the gasantisolvent spray precipitation process. Biotechnol. Prog., 9, 429–435.

Rasenack, N., & Muller, B. W. (2002) Dissolution rate enhancement by in situmicronization of poorly water soluble drugs. Pharm. Res., 19, 1894–1900.

Rogers, T. L., Gillespie, I. B., Hitt, J. E., Fransen, K. L., Crowl, C. A., Tucker,C. J., Kupperblatt, G. B., Becker, J. N., Wilson, D. L., Todd, C., Broomall,C. F., Evans, J. C., & Elder, E. J. (2004). Development and characterizationof a scalable controlled precipitation process to enhance the dissolution ofpoorly water-soluble drugs. Pharm. Res., 21, 2048–2057.

Rogers, T. L., Nelson, A. C., Sarkari, M., Young, T. J., Johnston, K. P., &Williams, R. O. (2003). Enhanced aqueous dissolution of a poorly water sol-uble drug by novel particle engineering technology: Spray-freezing into liq-uid with atmosphere freeze-drying. Pharm. Res., 20, 485–493.

Ruch, F., & Matijevic, E. (2000). Preparation of micrometer size budesonideparticles by precipitation. J. Colloid Interface Sci., 229, 207–211.

Salamova, U. U., & Rzaev, Z. M. O. (1996). Effect of inorganic salts on themain parameters of the dilute aqueous poly(vinylpyrrolidone) solutions.Polymer, 37, 2415–2421.

Sarkari, M., Brown, J., Chen, X., Swinnea, S., Williams, R. O., & Johnston, K. P.(2002). Enhanced drug dissolution using evaporative precipitation intoaqueous solution. Int. J. Pharm., 243, 17–31.

Sekikawa, H., Hori, R., Arita, T., Ito, K., & Nakano, M. (1978). Application ofthe cloud point method to the study of the interaction of polyvinylpyrroli-done with some organic compunds in aqueous solution. Chem. Pharm. Bull.,26, 2489–2496.

Turker, L., Guner, A., Yigit, F., & Guven, O. (1990). Spectrophotometricbehavior of poly(vinylpyrrolidone) in aqueous and nonaqueous media. I.Colloid Polym. Sci., 268, 337–344.

Yeo, S. D., Lim, G. B., Debenedetti, P. G., & Bernstein, H. (1993). Formationof microparticulate protein powders using a supercritical fluid antisolvent.Biotechnol. Bioeng., 41, 341–346.

Dru

g D

evel

opm

ent a

nd I

ndus

tria

l Pha

rmac

y D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y D

alho

usie

Uni

vers

ity o

n 07

/11/

14Fo

r pe

rson

al u

se o

nly.