1 Nanoparticles Production by Supercritical Antisolvent Precipitation Una Interpretacion General

7/28/2019 1 Nanoparticles Production by Supercritical Antisolvent Precipitation Una Interpretacion General

1/13

J. of Supercritical Fluids 43 (2007) 126138

Nanoparticles production by supercritical antisolventprecipitation: A general interpretation

Ernesto Reverchon , Iolanda De Marco, Enza Torino

Universita degli Studi di Salerno, Dipartimento di Ingegneria Chimica e Alimentare, Via Ponte Don Melillo, 84084 Fisciano (SA), Italy

Received 19 December 2006; received in revised form 11 April 2007; accepted 30 April 2007

Abstract

Supercritical antisolvent micronization (SAS) has been used to obtain microparticles of several kind of materials, but the production of nanopar-

ticles have been observed and studied in some cases only. This work is focused on the systematic production of nanoparticles using SAS. Weperformed experiments on several compounds and different solvents at selected operating conditions, obtaining nanoparticles with mean diameters

ranging between 45 and 150 nm, thus demonstrating that nanoparticles production is a general characteristic of this process. Moreover, we found

a correlation between nanoparticles mean diameter and the reduced concentration of the starting liquid solution that can allow the prediction of

the mean diameter obtainable at fixed process conditions. Nanoparticles with mean diameters as small as 45 nm have been obtained, operating

at 150 bar, 40 C and xCO2 = 0.97; but, even smaller nanoparticles can be obtained operating at higher pressures. The mechanism that produces

nanoparticles in supercritical antisolvent precipitation has also been discussed.

2007 Elsevier B.V. All rights reserved.

Keywords: Nanoparticles; Supercritical antisolvent precipitation; Drugs; Catalysts precursors; Colouring matters; Polymers

1. Introduction

Some supercritical fluidsbased processeshave beenproposed

in the literature to produce micro and/or nanosized materials;

they all try to take advantage of the specific characteristics of flu-

ids at supercritical conditions, like the adjustable solvent power

and the gas-like diffusivity [19]. Among them, Supercritical

AntiSolvent (SAS) process is well known and has been used to

micronizeor to attempt themicronization of severalkind of com-

pounds [1,2]. SAS is also known with other acronyms: Aerosol

Solvent Extraction System (ASES), Solution Enhanced Disper-

sion by Supercritical fluids (SEDS), Supercritical AntiSolvent

with Enhanced Mass transfer (SAS-EM). The main difference

among these processes is in theinjection device: in thecase of the

SAS and ASES processes, the liquid solution is sprayed in the

precipitation chamber through a thin wall nozzle, in the case of

the SEDS process, the nozzle is coaxial; whereas, the SAS-EM

process utilizes a deflecting surface that vibrates at ultrasonic

frequencies to enhance the atomization of the solution.

Corresponding author. Fax: +39 089 964057.

E-mail address: [email protected](E. Reverchon).

The scientific literature shows that SAS treated materials

can range from nanoparticles to microparticles to large emptyparticles (balloons) [15]. The products can be amorphous or

semi-crystalline; but, crystalline particulates have also been

reported [1,2]. Most of the SAS produced powders range in the

micron-size region that has been the target of several studies:

many industrial applications require these particle dimensions

to obtainthe best process performance.For example, small parti-

cles in the 15m range with a narrow particle size distribution

are needed for applicationsin pulmonary delivery andcontrolled

release systems [10,11].

The production of nanoparticles is even more ambitious than

producing microparticles of controlled dimensions. Even the

definition of nanoparticles is debated; depending on the specific

field of interest of the various authors, more or less restric-

tive definitions have been proposed. Recently, Reverchon and

Adami [6], reviewing nanotechnological applications of super-

critical fluids, discussed the various nanoparticles definitions

and selected 200 nm as the maximum dimension for the defini-

tion of nanoparticles.

At nanodimensions, it is possible to produce explosives with

a higher potential; i.e., approaching the ideal detonation; colour-

ing matter with brighter colours; toners with a higher resolution;

polymers and biopolymers with improved functional and struc-

0896-8446/$ see front matter 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.supflu.2007.04.013
mailto:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.supflu.2007.04.013http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.supflu.2007.04.013mailto:[email protected]


2/13

E. Reverchon et al. / J. of Supercritical Fluids 43 (2007) 126138 127

tural properties; drugs with enhanced pharmaceutical activity or

that use different delivery routes and/or overcome human body

internal barriers. Metals, metal oxides and ceramic compounds

at nanodimensions can exhibit unusual strength and/or can be

used as fillers in nanostructured materials [6].

Thermolabile compounds are even more difficult to be pro-

cessed at nanodimensions; but, they can be obtained at mild

temperatures (3550 C) using SAS [15]. However, nanoparti-

cles haveonly been occasionallyfacedby theauthors that studied

this process. Reverchon et al. [1218] obtained nanometricparti-

cles of different compounds. For example, yttrium, gadolinium,

europium, samarium and neodymium acetates, that are precur-

sors of high temperature superconductors, were micronized and

the mean diameters of the obtained nanoparticles was of about

100 nm [1214]. Nanoparticles of zinc acetate, a catalyst precur-

sor, were also produced by SAS. Particles down to 30 nm, with a

mean diameter of 50 nm,were obtained at thebest operatingcon-

ditions [15]. Some pigments were also produced using dimethyl

sulfoxide (DMSO) and N-methyl 2-pyrrolidone (NMP) as the

liquid solvents: Disperse Red 60 nanoparticles, for example, atthe best operating conditions, reached a mean diameter of 50 nm

[16]. Wu et al. [19] micronized pigment Red 177 by precipita-

tion from DMSO and analyzed the influence of several process

parameters on particle size. Spherical nanoparticles down to

46 nm mean diameter were obtained, operating at 120 bar and

50 C. Tetracycline, an antibiotic, is an example of pharmaceuti-

cal compounds processed at nanodimensions; the mean particle

size was about 150 nm [17]. Chattopadhyay and Gupta [20]

SAS precipitated Fullerene (C60) nanoparticles from a toluene

solution. The experiments were performed operating in a SAS

batch mode and fullerene particles as small as 2963 nm were

obtained.Chattopadhyayet al. [21,22] used the SAS-EM processto produce Griseofulvin (antifungal, antibiotic) particles as low

as 130 nm and Lysozyme (enzyme) particles of about 190 nm.

Nanometric Lysozyme particles with a minimum mean diameter

of 180 nm were also produced, at the best operating conditions,

by Muhrer and Mazzotti [23], using GAS process. Snavely et al.

[24] produced Insulin (antidiabetic) nanoparticles by SAS with

the aid of an ultrasonic nozzle. They obtained a powder consist-

ing of physical aggregates of 50 nm spheres forming sponge-like

and cob-web-like structures that could be deagglomerated in

smaller units. Nanoparticles of some polymers and biopolymers

were also obtained: in the case of dextran, the particles showed

a mean diameter ranging between 125 and 150 nm [18]. Subra

and Jestin [25] obtained dextran particles with a mean diameterof 70 nm. Jarmer et al. [26] successfully produced nanoparticles

of polylactic acid (PLLA) with a semi-continuous antisolvent

process, injecting the solvent in a jet-swirl nozzle, designed to

enhance the mixing within a swirl chamber. Chang et al. [27]

consistently produced nanoparticles of metallocene catalyzed

cyclic olefincopolymer (mCOC), investigatingthe effect of SAS

process parameters on morphology and size of precipitated par-

ticles. These authors concluded that, when SAS was operated in

the supercritical region, nanoparticles of mCOC were produced.

From this analysis, we can conclude that, despite the indus-

trial interest, only a relatively small number of SAS works have

beenfocused on the productionof nanoparticles.Moreover, indi-

cations about the SAS process conditions required to obtain

nanoparticles and about the mechanisms that produce powders

with these characteristics are generally missing or connected

to the single experimentation presented and to the processed

material.

To contribute at a better knowledge of SAS applicability to

nanosizedmaterials, thescope of this work is to demonstrate that

the capability of producing nanoparticles is a general feature of

the SAS process and that it is possible to describe conditions of

theSAS parameters at which nanoparticles of controlled size and

distributions can be obtained. Literature data together with an

extensive SAS experimentation have been performed to assess

the possibility of obtaining general validity rules for nanoparti-

cles production. The mechanism that can produce nanoparticles

during SAS has also been investigated.

2. Materials, apparatus and methods

2.1. Materials

Yttrium, zinc, europium, gadolinium, samarium and neodim-

ium acetates, rifampicin, astemizole, nitrotriazole and polyvinyl

alcohol (PVA) were supplied by SigmaAldrich (Italy) and

have purities of 99.9%; cellulose acetate was kindly supplied

by British & American Tobacco (USA); Amoxicillin, Dextran-

40, Hyaluronic benzyl ester (HYAFF 11) and ampicillin were

bought by ICN Biochemicals (USA) and have purities higher

than 98%; N-(2-hydroxypropyl)methacrylamide (HPMA) was

produced at the University of Paris XIII [18]; Disperse Red 60,

Solvent Yellow 56 and Solvent Blue 35 (purities 99.9%) were

supplied by Sun Chemicals (USA); Inulin was kindly supplied

by Orafti (Belgium).Dimethyl sulfoxide (DMSO, purity 99.5%), Acetone (Ac,

purity 99.8%), N-methyl 2-pyrrolidone (NMP, purity 99.5%),

methyl alcohol (MeOH, purity 99.5%), ethyl acetate (EtAc,

purity 99.5%) and dichloromethane (DCM, purity 99.5%) were

supplied by SigmaAldrich (Italy). CO2 (purity 99%) was pur-

chased from SON (Italy).

The solubilities of the materials in the solvents used were

measured at room temperature and are reported in the third

column ofTable 1 . All materials were used as received.

2.2. Apparatus

The SAS laboratory apparatus we used consists of an HPLCpump equipped with a pulse dampener (Gilson, mod. 805) used

to deliver the liquid solution, and a diaphragm high-pressure

pump (Milton Roy, mod. Milroyal B) used to deliver supercriti-

cal CO2. A cylindrical vessel with an internal volume of 500 cm3

is used as the precipitation chamber. The liquid mixture is deliv-

ered to the precipitator, for most experiments, through a thin

wall 800m length and 200m diameter stainless steel nozzle;

but, also a 80m nozzle has been used [28]. A second col-

lection chamber located downstream the precipitator is used to

recover the liquid solvent. Further information have been given

elsewhere [12] and a schematic representation of the appara-

tus has been reported in Fig. 1. Typical liquid solution flow rates


3/13

128 E. Reverchon et al. / J. of Supercritical Fluids 43 (2007) 126138

Table 1

Mean diameters of the particles obtained by SAS at 150 bar, 40 C, xCO2 0.97

Solute Solvent Solubility (mg/mL) C(mg/mL) C/C0 d(nm) Mode (nm) S.D. (nm)

Yttrium Acetate (YAc) [12,14] DMSO 303

5 0.016 50 44 20

10 0.033 70 67 20

13.5 0.044 82 73 21

36 0.118 120 110 30

Europium Acetate (EuAc) [13] DMSO 59810 0.016 60 70 54

100 0.16 150 200 25

Gadolinium Acetate (GdAc) [13] DMSO 318

5 0.015 55 53 13

10 0.031 65 81 15

13 0.047 75 70 18

20 0.062 90 108 60

32 0.100 125 109 59

48 0.150 150 200 82

Samarium Acetate (SmAc) [12] DMSO 213

2 0.009 47 50 25

8 0.023 80 120 50

11 0.051 90 104 70

17.5 0.082 110 95 37

22 0.103 120 105 43

26 0.154 130 140 25

Neodimium Acetate (NdAc) [12] DMSO 202 5 0.024 67 65 11

Zinc Acetate (ZnAc) [15]DMSO (1) 530

5 0.0094 45 30 25

10 0.018 60 51 24

15 0.028 75 60 30

50 0.094 110 100 30

75 0.14 150 155 33

NMP (2) 478 10 0.016 55 55 20

Cellulose Acetate (Cell Ac) Ac 93 10 0.107 125 160 65

Nitrotriazole (NTO) DMSO 450 20 0.044 65 65 11

Rifampicin (Rifa) [30]

DMSO (1) 1223 0.024 70 70 83

10 0.081 115 105 90

MeOH (2) 273 10 0.044 60 55 30

EtAc (3) 120 5 0.041 70 60 33

DCM (4) 60 5 0.083 100 120 50

Amoxicillin (Amoxi) [17,31,35] NMP 195 20 0.102 118 150 72

Astemizole (Aste) DMSO 1105 0.045 95 80 45

10 0.090 115 105 48

Ampicillin (Ampi) [17] DMSO 480 15 0.031 45 100 60

Inulin (Inul) [18] DMSO 335 25 0.075 100 140 47

Dextran 40 (Dext 40) [18] DMSO 147

2.5 0.017 50 110 60

5 0.034 65 55 42

10 0.068 95 80 37

15 0.102 125 140 23

HYAFF 11 DMSO 247 10 0.04 95 85 53

PVA DMSO 236 10 0.042 60 75 30HPMA [18] DMSO 250 10 0.04 85 80 37

Disperse Red 60 (DR60) [16]

DMSO (1) 180

5 0.027 58 47 26

10 0.055 88 80 29

15 0.083 102 90 53

NMP (2) 150

5 0.033 93 85 53

10 0.066 105 60 30

15 0.1 112 97 57

Ac (3) 100 10 0.1 100 104 70

Solvent yellow 56 (SY56)NMP (1) 345 20 0.057 70 60 17

Ac (2) 207 20 0.096 115 99 38

Solvent Blue 35 (SB35)NMP (1) 90 5 0.055 70 75 21

Ac (2) 90 5 0.055 70 60 27


4/13


Table 1 (continued)

Solute Solvent Solubility (mg/mL) C(mg/mL) C/C0 d(nm) Mode (nm) S.D. (nm)

mCOCa [27]

Toluene (1) 435b

2.175 0.005 41.1 41.1 4.8

4.35 0.01 46.2 46.2 7.6

8.7 0.02 47.9 47.9 9.4

17.4 0.04 62.8 62.8 8.9

o-Xylene (2) 574b 5.74 0.01 50.3

m-Xylene (3) 574b 5.74 0.01 52.8 p-Xylene (4) 574b 5.74 0.01 53.5

THF (5) 440b 4.40 0.01 42.5

a For this data, the xCO2 is equal to 0.94.b Private communication.

rangedbetween 0.5 and2.0 mL/min andSC-CO2 flow rates were

correspondingly adapted to produce a XCO2 0.97. In each

experiment, liquid solvent was injected first, to obtain steady

state fluid phase concentration conditions in the precipitator;

then, the solution was delivered in quantities ranging from 50 to

about 300 mL, depending on the scope of the experiments and

on the quantity of powder that we want to collect.

The view cell SAS apparatus is similar to the previously

described one and differs only for the precipitator, that con-

sists of a stainless steel cylindrical vessel (375 cm3 i.v.) with

two quartz windows put along two longitudinal sections (NWA,

Germany). It is possible to visually observe the formation of

different phases and the macroscopic evolution of the precipita-

tion process from the liquid jet break-up to the precipitation of

particles [29,30].

The pilot plant used in this study is a closed-loop plant con-

sisting mainly of a CO2 storage vessel, a precipitator, a liquid

separator, two pumps, a heat exchanger, and a condenser. The

water-jacketed precipitator has an internal volume of 5.2 dm3

and a L/D ratio of 9.4. The liquid solution and SC-CO2 are fed

to the chamber through a tube-in-tube injection system (inter-

nal tube d= 3 mm; annulus d= 8.5 mm). The generation of small

liquid droplets is ensured by the presence of a 500-m nozzle

fitted on the tip of the internal tube [31]. A more detailed descrip-

tion of the pilot plant can be found in previous works [14,31]; a

photograph of the plant is reported in Fig. 2.

2.3. Methods

Samples of the precipitated powder, collected in different

points of the precipitation chamber, were observed using a Scan-

ning Electron Microscope (SEM) (Assing, mod. LEO 420).

SEM samples were covered with 250 A of gold using a sput-

ter coater (Agar, mod. 108A). Particle size (PS) and particle size

distributions (PSDs) were measured usingan imageanalysisper-

formed using Sigma Scan Pro software (Jandel Scientific), an

image processing program that counts, measures and analyzes

digital images; from about 700to 1000 particleswere considered

Fig. 1. Schematic representation of SAS apparatus: S1, CO2 supply; S2, liquid supply; B, refrigerating bath; P1P2, pumps; D, pressure dampener; CS, precipitation

vessel; M, manometer; TC, thermocouple; VM, micrometering valve; SL, liquid separator; BP, back-pressure valve; A, calibrated rotameter; CR, wet test meter.


5/13


Fig. 2. SAS pilot plant located at the University of Salerno.

in each PSD calculation. The number of particles was chosen

in agreement with the criteria used in the image analyses of

powders: 5001500 measured particles represent a good com-

promise between the time spent for the analysis and the accuracy

of results [32].

X-ray diffraction pattern (XRD) analyses were performed

using a Rigaku mod. RINT RAPID XRD apparatus to ascertain

if changes occurred in the crystal habit of the materials as a

consequence of the SAS process.

A SAS experiment begins by delivering supercritical CO2 tothe precipitation chamber until the desired pressure is reached.

Antisolvent steady flow is established; then, pure solvent is sent

through the nozzle to the chamber with the aim of obtaining

steady state composition conditions during the solute precipita-

tion. At this point, the flow of the liquid solvent is stopped and

the liquid solution is delivered through the nozzle. The experi-

ment ends when thedelivery of theliquidsolutionto thechamber

is interrupted. However, supercritical CO2 continues to flow to

wash the chamber for the residual content of liquid solubilized

in the supercritical antisolvent. If the final purge with pure CO2is not performed, the solvent condenses during the depressur-

ization and can solubilize or modify the powder. More details

have been given elsewhere [12].

3. Results and discussion

3.1. Conditions for successful SAS micronization

The prerequisites for successful SAS process are the com-

plete miscibility between the liquid solvent and the antisolvent

and the insolubility of the solute in the antisolvent (or, rather,

in the solution solventantisolvent formed in the precipitator).

Considering the binary system solventantisolvent, this condi-

tion is obtained at pressures larger than the mixture critical point

(MCP); it represents, in a pressure-composition plane at a fixed

temperature, the pressure at which only a single supercritical

phase can exist. However, it should be also considered that the

presence of a solute can modify the binary system vaporliquid

equilibria (VLEs), as a rule, moving the MCP of the ternary sys-

tem towards higher pressures than for the corresponding binary

one [16,29,33]. If the ternary system shows poorer solubility

when compared with the binary systems antisolvent + solvent

and antisolvent + solute, it is called non-cosolvency(antisolvent)

system [33]. The VLEs of a binary system and their hypothe-sized modifications due to the addition of a third component are

reported in Fig. 3.

Fig. 3. A possible qualitative modification of the VLEs of a binary (solid line)

system, when a third component (Z) is added (dotted line) at a given concentra-

tion.


6/13


Fig. 4. Experimental data for the binary system DMSOCO2 andfor theternary

system CefonicidDMSOCO2 in a pressure vs. CO2 molar fraction diagram.

VLEs modifications can be enhanced as the concentration

of the solute increases. Therefore, in the selection of the SAS

operating conditions, it could be not possible to consider that

the MCP of the ternary system is coincident to the one of the

binary system, except from the cases in which the concentration

of the liquid solution is very low. For example, in the case of the

system CefonicidDMSOCO2, we observed experimentally,

with a view cell, that VLEs of the system at low concentra-

tions are substantially the same of the binary one; whereas, at

higher concentrations, the VLEs change (see Fig. 4) [34]. In

SAS experiments performed using windowed precipitator, we

observed that the ternary system could be at subcritical condi-tions even when the corresponding liquidCO2 binary system is

supercritical.

For these reasons, it should be advisable to select pressure

relatively higher than the MCP pressure of the binary system, to

avoid the risk of working at subcritical conditions.

An initial analysis of the literature data on SAS process con-

ditions, at which it is possible to produce nanoparticles, allows to

approximately indicate following ranges: pressures between 120

and 180 bar and temperatures between 35 and 60 C [1218,27].

These operating conditions should assure that we work at super-

critical conditions at least for the binary solventantisolvent

system. Since we have several previous data on tests performed

at 150 bar, 40 C, the following experimentation has been ini-

tially performed at these process conditions; xCO2 = 0.97 has

been also selected to operate on the right of the MCP.

The new experimental results are in part obtained for com-

pounds different from those previously published; however, we

also performed some new experiments on materials that were

previously published, with the scope of producing the largest

data set possible. Indeed, in our previous works, frequently,

the major interest of the experimentation was the production of

micro- and sub-microparticles, not of nanoparticles. All data are

summarized in Table 1. In this table, the processed compounds,

the solvents used, the concentration, the mean particle size of the

powders obtained (d50), the mode and the standard deviations

of the distributions have been reported. In Table 1, also the sol-

ubilities of the selected compounds in the liquid solvent used to

perform SAS experiments are reported. This data has been sys-

tematically measured also for materials previously processed, to

obtain the same accuracy for all compounds. Indeed, in previous

works solubility data were in some cases approximated, because

they were measured to avoid the use of solutions that were too

near to the saturation value.It is possible to observe that many experiments have been

performed using DMSO as the liquid solvent; it is due to the

fact that it does not tend to produce strong interactions with

solutes. Therefore, the ternary system DMSOCO2solute fre-

quently tends to maintain VLE characteristics that are similar

to those of the corresponding binary system (DMSOCO2)

and is simpler to find the process conditions for successful

SAS.

Some examples of nanoparticles obtained operating at

150barand40 Careshownin Fig.5ah fordifferent solutes and

well illustrate the similarities observed among different solutes.

In all cases, quasi-spherical nanoparticles with a maximumdiameter smaller than 200 nm have been obtained, as shown in

Fig. 6ad, that report some PSDs calculated from SEM images,

as described in Section 2. These nanoparticles form a colloidal

system suspended in the supercritical solution and can tend to

form aggregates, as it can be expected by this kind of nanodis-

persions [36].

In some cases, however, it has not been possible to obtain

nanoparticles for selected couples soluteliquid solvent and, in

some other cases, the nanoparticles showed a further evolution

with the formation of solid bridges among groups of particles.

These latest systems have been reported in Table 2 and shown as

examples in Fig. 7a and b. They have not been further considered

in the following elaboration of the results.

Table 2

Coalesced nanoparticles obtained at 150 bar and 40 C

Solute Solvent Solubility (mg/mL) C(mg/mL) C/C0 Morphology

Prednisolone Ac 100 40 0.4 Needle-like particles

Salbutamol

[37]DMSO 15

5 0.333 Microparticles

10 0.667

Pentamidine DMSO 100 10 0.1 Agglomerated particles

Cefonicid [29,33] DMSO 150 10 0.067 Solid bridges among nanoparticles

Nalmefene EtOH 150 15 0.1 Solid bridges among nanoparticles

Polistyrene CHCl3 20 5 0.25 Solid bridges among nanoparticles

-Cyclodextrin DMSO 700 5 0.007 Solid bridges among nanoparticles


7/13


Fig.5. Nanoparticlesobtainedat 150bar, 40 C: (a) Yttrium acetate/DMSO, 5 mg/mL; (b) Gadolinium acetate/DMSO, 100mg/mL; (c) Amoxicillin/NMP, 20 mg/mL;

(d) Rifampicin/DMSO, 10 mg/mL; (e) Astemizole/DMSO, 10 mg/mL; (f) Dextran 40/DMSO,15 mg/mL; (g) HPMA/DMSO, 10 mg/mL; (h) SolventBlue 35/Acetone,

5 mg/mL.


8/13


Fig. 6. Particle size distributions of different materials SAS processed at 150 bar, 40C: (a) Solvent Blue 35; (b) Astemizole; (c) Rifampicin; (d) HPMA.

It is interesting to observe that nanoparticles size, fixed all

the other process parameters, largely depends on the concentra-

tion of the starting liquid solution. For concentrations larger

than those reported in Table 1, no more nanoparticles have

been obtained: sub-micro- and microparticles have instead been

obtained with a corresponding enlargement of the PSD. Sub-

micro- and microparticles also present a different morphology:

the particles are perfectly spherical (see Fig. 8a and b for exam-

ple); whereas, nanoparticles are only approximately spherical.

We also tested the possibility of producing nanoparticles bySAS at pressures and temperatures different than 150 bar, 40 C.

Assembling together the new results reported in this work and

some literature data, the pressure tested range between 100 and

180 bar at temperatures from 35 to 50 C. These further results

are summarized in Table 3.

X-ray analyses were performed on the untreated and on the

SAS precipitated particles in order to check their crystalline

patterns. The X-ray traces revealed that the SAS micronized

powders were amorphous, whereas the untreated materials show

crystalline patterns.

Some experiments have also been performed using the quartz

windowed precipitator. In the experiments in which nanoparti-

cles have been produced, the jet, at the exit of the injection

device, was not visually detectable and a single gaseous phase

was present in the precipitator.

3.2. Organization of the experimental evidences

To attempt an organization of the experimental evidences,

we have to consider that nanoparticles have been obtained for

several materials (Table 1) that belong to differentgroups: super-

conductor precursors, pigments, pharmaceuticals, polymers.Also different liquid solvents have been used, though the results

obtained using DMSO are prevalent, as previously explained.

It means that nanoparticles production does not depend on the

specific material or on the liquid solvent used: it is a general

feature of the SAS process. The characteristic that connects

all the results is that the pressure to produce nanoparticles is

above the MCP of the binary system liquidCO2: as a rule,

pressures at values much higher than the supercritical condition

of the selected system are required. We can summarize these

observations stating that completely developed supercritical

conditions are required. In contrast, pressures near the MCP

can be described as at near critical conditions.


9/13


Fig. 7. Coalescing nanoparticles obtained at 150 bar, 40C: (a) Cefonicid and

(b) Nalmefene HCl.

An increase of pressure produces smaller particles, as demon-

strated by the experiments performed at 180 bar. Temperature

dependence on nanoparticles production is not relevant; but,

seems that an increase of temperature produces an effect that

is opposite to that of an increase of pressure.

We performed the experiments using two different injection

devices: the first is a laser hole on a thin wall nozzle (differ-

ent diameters have been used), the second is a coaxial injection

device assisted by a liquid injection system (see Section 2); at

the best of our measurements, the injection device seems to have

no or negligible influence on nanoparticles size. Also Chang

et al. [27] performed experiments using two different injection

devices and found that the difference in diameter of mCOCnanoparticles was practically negligible at pressures higher than

100 bar. This observation does not mean that the characteristics

of the injector do not play a role in the SAS process; but that, at

the process conditions where nanoparticles have been observed,

this influence seems to be negligible.

The dependence on the concentration of the liquid solution

is evident: the increase of the solute concentration in the liq-

uid phase produces an increase of nanoparticles mean size (see

Table 1). We tried to report on a diagram the mean nanoparticles

diameter against the concentration of the liquid solution for three

of the tested compounds (Fig. 9). The mean diameter depends

somewhere linearly on the liquid solution concentrations, as

Fig. 8. Spherical submicro- and microparticles: (a) Samarium acetate and (b)

Cefonicid.

shown from the linear regression we added in this figure; but,each compound follows a different trend. The minimum mean

diameter is similar for the three compounds compared (Fig. 9).

It means that the dependence of particle size concentration is

Fig. 9. Mean diameter (d) vs. concentration for three different materials pro-

cessed at 150bar, 40

C.


10/13


Table 3

Experiments performed at different pressures and temperatures

Solute Solvent Solubility (mg/mL) P (bar) T(C) C(mg/mL) C/C0 dm (nm)

Cellulose acetate Ac 93 120 40 10 0.107 85

Yttrium acetate [12,14] DMSO 303

12040

13.5 0.044

200

150 82

160 50 120

Dextran 40 [18] DMSO 147

110

40 10 0.068

180

130 120

150 95

Samarium acetate [12] DMSO 213

100

40

15 0.070200

120 180

150 17.5 0.082 110

180 15 0.070 55

30 0.140 80

Amoxicillin [31,35] NMP 195

150 35

20 0.102

147

40118

180 58

Rifampicin [30] DMSO 122

120

40

10 0.081 150

20 0.164 17040 0.328 200

70 0.574 250

1503 0.024 70

10 0.081 115

180 20 0.164 40

PVA DMSO 236150

40 10 0.04260

180 50

demonstrated, but the results should be correlated in a different

way with the concentration.

To understand if the experimental results can be organized in

a more systematic arrangement,one of thepossiblehypotheses is

that nanoparticle size could not depend on the concentration but

on the ratio between the concentration of the liquid solution (C)

and its saturation concentration (C0); i.e., the reduced concen-

tration CR = C/C0. Therefore, we organized the mean diameter

data (d) obtained at 150 bar, 40 C in a diameter (d) against CRdiagram that is reported in Fig. 10. The mean particle diameters

for all compounds tested are fairly well described by a linear

dependence against CR with a correlation factor (R) of about

0.942. These results mean that (given the operating pressure,

temperature and xCO2 ) the diameter of the nanoparticles does

not significantly depend on the solute adopted. An explanationof this result is that the precipitation process could be driven by

the relative distance from the saturation conditions and by solid

nucleation and growth. The growth process is favoured by an

increase of solute concentration, since this phenomenon super-

imposes on the nucleation process. As expected, it also produces

wider particle sizedistributionsas the concentrationincreases, as

shown in this work. Theminimum diameter of nanoparticles the-

oretically obtainable can also been obtained by extrapolation of

the results in Fig. 10. It is about 45 nm. However, it is important

to remember that this is the mean diameter of the PSD; nanopar-

ticles as small as about 10 nm have been observed at low CR

values.

Fig. 10. Mean diameter (d) vs. reduced concentration (CR) for all the materials

processed at 150bar, 40 C: () YAc; () ZnAc-1; () ZnAc-2; () EuAc; ()

GdAc; ( ) SmAc;() Rifa-1; ( ) Rifa-2; () Rifa-3; ( ) Rifa-4; ()NTO;()

Dext40; () Aste; () Inul; () Hyaff; () CellAc; ( ) NdAc; ( ) PVA; ( )

Ampi; (+) HPMA; ( ) DR60-1; ( ) DR60-2; ( ) DR60-3; ( ) SY56-1; ( )

SY56-2; ( ) SB35-1; () SB35-2; ( ) Amoxi; ( ) mCOC-1; () mCOC-2;

( ) mCOC-3; ( ) mCOC-4; ( ) mCOC-5.


11/13


Fig. 11. Mean diameter (d) vs. pressure (P) for some materials at 40 C.

Using the diagram reported in Fig. 10, it can be also possible

to find the expected diameter for a material different from those

proposed in this work; fixed the SAS conditions, it is sufficient to

know the solubility in the given liquid solvent, calculate CR and

obtain as a result the expected mean diameter of nanoparticles

produced. The diagram in Fig. 10 can be, therefore, a very useful

tool to organize the experiments to produce nanoparticles of

predetermined dimensions.

Other experimental data are available at 40 C, as previously

reported in Table 3. These data in a mean particle diameter

against CR diagram produce a fair good linear correlation. The

decrease of nanoparticles diameter with pressure increase canbe observed in Fig. 11, where the experimental data reported

have been chosen for selected CR values approximately ranging

between 0.07 and 0.08.

This analysis of particle diameter data, thus, produces a gen-

eral definition, not only of the process conditions to be used to

produce nanoparticles by SAS; but, also of the expected diam-

eter of the particles that can be produced, that depends on the

relative distance of the concentration to the saturation of the

selected material in the liquid solvent. The percentage yield of

the SAS process, when nanoparticles are produced, is frequently

very high (higher that 90%) [31], since it substantially depends

on the saturation concentration of the solute in the fluid phase

formed in the precipitator. Since solutes that are not soluble orare only sparingly soluble in SC-CO2 are selected for this pro-

cess, it is expectedthat in a supercritical solution atxsolvent = 0.03

(at which we operate), only a small increase of solute solubility

will be obtained. From the point of view of powders recovered

in the precipitator, nanoparticles are easy to be collected, since

they are released from the colloidal suspension formed in the

precipitator as a fluffy powder substantially free of electrostatic

charges. Moreover, when we performed long pilot scale exper-

iments, we observed that the quantity of powder lost on the

walls of the precipitator was substantially constant, therefore,

its relative percentage decreased with the length of the experi-

ments.

3.3. Precipitation mechanism postulation

To explain the particles formation process in the jet of the liq-

uidsolution, it is possible to propose three differentmechanisms.

In the first one, the atomization of the liquid solution through the

injector results in the formation of droplets; the fast solubiliza-

tion of the SCF in the liquid solvent produces the formation of a

solid particle that can retain the shape of the originating droplet.

This mechanism is called one dropletone particle. The Weber

number

We =u2d

,

where is the liquid density, u the flux velocity, d the nozzle

diameter andis the surface tension, is the ratio between inertial

and surface forces and may be used to evaluate the droplet size

in sprays [38]. When the surface tension reduces, the Weber

number rapidly increases.

The second mechanism is a modification of the one

dropletone particle theory. In this case, the droplets are formedas described above; but, the rapid mass transfer of solvent

and antisolvent results in high supersaturation of the solute,

that causes the formation of several nuclei within the same

droplet. The result is the growth of several particles from one

droplet.

The third possible mechanism is that the surface tension

between the liquid and the antisolvent disappears at a time scale

smaller than the jet break-up of the liquid solution; therefore,

no droplets are formed and nucleation and growth of nanopar-

ticles could be the result of gas-like mixing; i.e., gas-to-particle

precipitation.

The interface between two miscible fluids at static equilib-riumat supercritical fluidconditionsshowsno significant surface

tension. But, in the case of theliquidjet injectionin the supercrit-

ical fluid phase, a time lag exists between the liquid injection and

the time at which equilibrium is obtained. The time evolution

of the interfacial tension between a liquid and a supercritical

fluid has been discussed by Lengsfeld et al. [39]. They found

that, for the CO2 + methylene chloride system, at 85 bar and

35 C; i.e., at the complete miscibility conditions, the transient

surface tension drops rapidly from approximately 2.5 mN/m at

the exit to 0.01 mN/m at about 1 m from the nozzle tip and

the Weber number based approach is no longer applicable. They

also verified these results injecting a jet of supercritical triflu-

orometane in supercritical carbon dioxide. For this system, thesurface tension driven droplet formation mechanismis obviously

not applicable. Sarkari et al. [40] also studied the evolution of

surface tension at the exit of the jet and found a time as short

as 2 ms for its complete elimination in supercritical CO2-based

systems. The velocity of reduction of the surface tension also

depends on the distance from the operating pressure from the

MCP. Moreover, near the MCP, small changes in the process

conditions can produce large changes in the equilibrium surface

tension.

Thus, the results we obtained are in favour of the gas mix-

ing precipitation mechanism: droplets are not formed at the

exit of the injector, but the liquid solution is almost instanta-


12/13


neously mixed in the gas phase, from which solids nucleate and

eventually grow.

This hypothesis of precipitation mechanism is thus supported

by the following experimental evidences:

The irregular spherical shape of the nanoparticles. If parti-

cles are generated from droplets, the surface tension confers

them a perfectly spherical shape and the resulting particles

obtained by droplet drying will be spherical too. The genera-

tion of the nanoparticles from a gaseous phase is compatible,

instead, with the irregular shape observed in SEMimages (see

Fig. 5ah).

As previously discussed, calculations from Lengsfeld et al.

[39] demonstrated that when SAS process is performed at

completely developed supercritical conditions, the time scale

of the surface tension disappearance in jets of miscible fluids

(solvent and antisolvent) determines the jet evolution as a gas

plume; i.e., no droplets are formed. During the experiments

performed usingthe windowed precipitator, the jet was practi-

cally not detectable when completely developed supercriticalconditions were obtained.

The dependence of nanoparticles diameter on CR also sup-

ports this mechanism of particles generation; indeed, Fig. 10

shows that particles diameter is correlated to nucleation and

growth mechanisms that are connected to the reduced con-

centration and their dimension depends on the distance from

the saturation value.

The position of the SAS operating point in the P-x diagram

is also consistent with the role played by the surface tension

in determining or not the formation of droplets. If the surface

tension of the liquid injected in the precipitator goes almost

instantaneously to zero, no droplets are formed. This pro-cess is faster, the more the process conditions are selected at

full developed supercritical conditions; i.e., as the pressure

increases. As long as the operating point goes to the vicinity

of the MCP, this process can compete with the formation of

droplets.

At near critical conditions, equilibrium surface tension can

be not zero or the time of its disappearance will be compara-

ble with jet break-up and small droplets can be produced: in

this case, the mechanism of powders formation becomes the

one dropletone particle. Therefore, the precipitation process

is also regulated by the position of the operating point with

respect to MCP.

A point that has to be clarified is how to determine the posi-

tion of the MCP. Frequently, the authors working on SAS use

as a reference the binary system solventantisolvent VLEs data;

thus, the MCP considered is the one of the binary system. How-

ever, in SAS precipitation, a ternary system is involved and

the presence of the solute can modify the VLEs of the corre-

sponding binary system. The most common effect could be the

movement of the MCP to higher pressures and an enlargement

of the two-phases region, caused by the reduction of affinity

between solvent and antisolvent due to the presence of a third

component. However, also the opposite effect could be observed

[33].

In some cases:

(a) when the solute concentration is very low;

(b) when thesolute is completely immiscible in thesupercritical

solution solvent + CO2;

the binary system data could be used as an approximation of

the real behaviour; but, in the other cases, they can suggest aposition of the MCP and SAS operating conditions, different

from that expected. As a consequence, SAS operation in the

two-phase region or at near-critical or subcritical conditions can

be performed and nanoparticles are not produced.

For example, at high values ofCR and for many of the mate-

rials proposed in Tables 1 and 2, microparticles with a perfect

spherical shape have been produced [13,14,18,2931,33,35] at

thesame SAS operating conditions, since theMCP of the ternary

mixture at higher solute concentration is continuously moving

towards higher pressures until the operating point is no more

at completely developed supercritical conditions. Sub-micro-

and microparticles generated by the one-droplet one-particle

mechanism are obtained.

For some solutes, solid bridges are formed among groups

of nanoparticles (see Table 2); we think that this phenomenon

could be due to the presence of small quantities of solvent in

the solid nanoparticles that, when these collided in the colloidal

suspension, formed a liquid bridge that upon drying produced

the solid connection among groups of them.

4. Conclusions

Nanoparticles production is a general feature of the SAS

process: the experimental results on more than 20 materials,

different liquid solvents and different apparatus confirm thispossibility.

Nanoparticles can be produced when the operating point is

at pressures far from the MCP of the ternary mixture. A gas to

solid precipitation occurs.

The results proposed can be used to predict when nanoparti-

cles are produced and which could be their expected diameters

on thebasisof their solubility in theselectedliquid solvent andof

the SAS operating conditions with respect to the ternary mixture

VLEs.

References

[1] E. Reverchon, Supercritical antisolvent precipitation of micro- and

nanoparticles, J. Supercrit. Fluids 15 (1999) 121.

[2] J. Jung, M. Perrut, Particle design using supercritical fluids literature and

patent survey, J. Supercrit. Fluids 20 (2001) 179219.

[3] Z. Knez, E. Weidner, Particles formation and particle design using super-

critical fluids, Curr. Opin. Solid State Mater. Sci. 7 (2003) 353361.

[4] A. Shariati,C.J. Peters, Recent developments in particle design using super-

critical fluids, Curr. Opin. Solid State Mater. Sci. 7 (2003) 371383.

[5] Y. Hakuta, H. Hayashi, K. Arai, Fine particle formation using supercritical

fluids, Curr. Opin. Solid State Mater. Sci. 7 (2003) 341351.

[6] E. Reverchon, R. Adami, Nanomaterials and supercritical fluids, J. Super-

crit. Fluids 37 (2006) 122.

[7] E. Reverchon, Supercritical-assisted atomization to produce micro- and/or

nanoparticles of controlled size and distribution, Ind. Eng. Chem. Res. 41

(2002) 24052411.


13/13


[8] P.J. Ginty, M.J. Whitaker, K.M. Shakesheff, S.M. Howdle, Drug delivery

goes supercritical, Mater. Today 8 (8) (2005) 4248.

[9] I. Pasquali, R. Bettini, F. Giordano, Solid-state chemistryand particle engi-

neering with supercritical fluids in pharmaceutics, Eur. J. Pharm. Sci. 27

(2006) 299310.

[10] P.G. Debenedetti, J.W. Tom, S.D. Yeo, G.B. Lim, Application of supercrit-

ical fluids for the production of sustained delivery devices, J. Control. Rel.

24 (1993) 2744.

[11] U.B. Kompella, K. Koushik, Preparation of drug delivery systems usingsupercritical fluid technology, Crit.Rev.Ther. DrugCarrier Syst. 18 (2001)

173199.

[12] E. Reverchon, G. Della Porta, A. Di Trolio, S. Pace, Supercritical antisol-

vent precipitation of nanoparticles of superconductor precursors, Ind. Eng.

Chem. Res. 37 (1998) 952958.

[13] E. Reverchon, I. De Marco, G. Della Porta, Tailoring of nano- and micro-

particles of some superconductor precursors by supercritical antisolvent

precipitation, J. Supercrit. Fluids 23 (2002) 8187.

[14] E. Reverchon, G. Caputo, I. De Marco, Role of phase behavior and atom-

ization in the supercritical antisolvent precipitation, Ind. Eng. Chem. Res.

42 (2003) 64066414.

[15] E. Reverchon, G. Della Porta, D. Sannino, P. Ciambelli, Supercritical anti-

solvent precipitation of nanoparticles of a zinc oxide precursor, Powder

Technol. 102 (1999) 127134.

[16] E. Reverchon, R. Adami, I. De Marco, C.G. Laudani, A. Spada, Dispersered 60 micronisationusingsupercritical fluidbasedtechniques, J. Supercrit.

Fluids 35 (2005) 7682.

[17] E. Reverchon, G. Della Porta, Production of antibioticmicro- and nanopar-

ticles by supercritical antisolvent precipitation, Powder Technol. 106

(1999) 2329.

[18] E. Reverchon, G. Della Porta, I. De Rosa, P. Subra, D. Letourneur, Super-

critical antisolvent micronization of somebiopolymers, J. Supercrit.Fluids

18 (2000) 239245.

[19] H.-T. Wu, M.-J. Lee, H. Lin, Nano-particles formation for pigment red 177

via a continuous supercritical anti-solvent process, J. Supercrit. Fluids 33

(2) (2005) 173182.

[20] P. Chattopadhyay, R.B. Gupta, Supercritical CO2-based production of

fullerene nanoparticles, Ind. Eng. Chem. Res. 39 (2000) 22812289.

[21] P. Chattopadhyay, R.B. Gupta, Production of griseofulvin nanoparticles

using supercritical CO2 antisolvent with enhanced mass transfer, Int. J.

Pharm. 228 (2001) 1931.

[22] P. Chattopadhyay, R.B. Gupta, Protein nanoparticles formation by super-

critical antisolvent with enhanced mass transfer, AIChE J. 48 (2002) 235.

[23] G. Muhrer, M. Mazzotti, Precipitation of lysozyme nanoparticles from

dimethyl sulfoxide using carbon dioxide as antisolvent, Biotechnol. Prog.

19 (2) (2003) 549556.

[24] W.K. Snavely, B. Subramaniam, R.A. Rajewski, M.R. Defelippis,

Micronization of insulin from halogenated alcohol solution using super-

critical carbon dioxide as an antisolvent, J. Pharm. Sci. 91 (9) (2002)

20262039.

[25] P. Subra, P. Jestin, Screeningdesign of experiment (DOE) applied to super-

critical antisolvent process, Ind. Eng. Chem. Res. 39 (2000) 41784184.

[26] D.J. Jarmer, C.S. Lengsfeld, T.W. Randolph, Manipulation of particle size

distribution of poly(l-lactic acid) nanoparticles with a jet-swirl nozzle dur-

ing precipitation with a compressed antisolvent, J. Supercrit. Fluids 27

(2003) 317336.

[27] S.-C. Chang, M.-J. Lee, H.-M. Lin, Nanoparticles formation for metal-locene catalyzed cyclic olefin copolymer via a continuous supercritical

anti-solvent process, J. Supercrit. Fluids 40 (2007) 420432.

[28] E. Reverchon, G. Dons, D. Gorgoglione, Salicylic acid solubilization in

supercritical CO2 and its micronization by RESS, J. Supercrit. Fluids 6 (4)

(1993) 241248.

[29] E. Reverchon, I. De Marco, Supercritical antisolvent micronization of

Cefonicid: thermodynamic interpretation of results, J. Supercrit. Fluids 31

(2) (2004) 207215.

[30] E. Reverchon, I. De Marco, G. Della Porta, Rifampicin microparticles pro-

duction by supercritical antisolvent precipitation, Int. J. Pharm. 243 (2002)

8391.

[31] E. Reverchon, I. De Marco, G. Caputo, G. Della Porta, Pilot scale

micronization of Amoxicillin by supercritical antisolvent precipitation, J.

Supercrit. Fluids 26 (2003) 17.

[32] M.-N. Pons, H. Vivier, V. Delcour, J.-R. Authelin, L. Pailleres-Hubert,Morphological analysis of pharmaceutical powders, Powder Technol. 128

(2-3) (2002) 276286.

[33] E. Reverchon, I. De Marco, Supercritical antisolvent precipitation of

Cephalosporins, Powder Technol. 164 (2006) 139146.

[34] S. Gentile, MSc Thesis, University of Salerno, Italy, 2005.

[35] E. Reverchon, G. Della Porta, M.G. Falivene, Process parameters and

morphology in Amoxicillin micro and submicro particles generation

by supercritical antisolvent precipitation, J. Supercrit. Fluids 17 (2000)

239248.

[36] Y.-P. Sun, M.J. Meziani, P. Pathak, L. Qu, Polymeric nanoparticles from

rapid expansion of supercritical fluid solution, Chem. Eur. J. 11 (2005)

13661373.

[37] E. Reverchon, G. Della Porta, P. Pallado, Supercritical antisolvent precip-

itation of Salbutamol microparticles, Powder Technol. 114 (1-3) (2001)

1722.

[38] A.H. Lefebvre, Atomization and Sprays, Hemisphere Publishing, Bristol,

PA, 1989.

[39] C.S.Lengsfeld, J.P. Delplanque, V.H. Barocas, T.W. Randolph, Mechanism

governing microparticle morphology during precipitation by a compressed

antisolvent: atomization vs. nucleation and growth, J. Phys. Chem. B 104

(2000) 27252735.

[40] M. Sarkari, I. Darrat, B.L. Knutson, Generation of microparticles using

CO2 and CO2-philic antisolvents, AIChE J. 46 (9) (2000) 18501859.

1 Nanoparticles Production by Supercritical Antisolvent Precipitation Una Interpretacion General

Documents

Transcript of 1 Nanoparticles Production by Supercritical Antisolvent Precipitation Una Interpretacion General