Preferential Solvation Stabilization for Hydrophobic Polymeric

Jul 2, 2005 - Issei Takeuchi , Tomoyoshi Takeshita , Takaaki Suzuki , Kimiko Makino. Colloids and Surfaces B: Biointerfaces 2017 160, 520-526 ...
0 downloads 0 Views 317KB Size
Download PDF
J. Phys. Chem. B 2005, 109, 13877-13882

13877

Preferential Solvation Stabilization for Hydrophobic Polymeric Nanoparticle Fabrication Jun-Ying Xiong,† Xiang-Yang Liu,*,‡ Shing Bor Chen,† and Tai-Shung Chung† Department of Chemical and Biomolecular Engineering, National UniVersity of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore, and Department of Physics, National UniVersity of Singapore, 2 Science DriVe 3, Singapore 117542, Singapore ReceiVed: NoVember 24, 2004; In Final Form: June 1, 2005

Preferential solvation of polymer molecules and strong EPD-EPA (EPD, electron pair donor; EPA, electron pair acceptor) interaction between solvent and nonsolvent molecules were found to be of great significance in the fabrication of two kinds of aromatic polyimide (AP) nanoparticles. Surfactant free yet stable AP nanoparticles were prepared using a liquid-liquid phase separation method. The stability of the AP nanoparticles can be achieved by the solvation multilayer resulting from a solvation stabilization chain in the form of nonsolvent f solvent f AP (a f b denotes that component b is solvated by component a). The significance of this stabilization chain was identified by many comparative experiments using different types of molecular probes. On the other hand, the formation of AP nanoparticles was found to be governed by a nucleation process and therefore the particle size is controlled by the nucleation rate. A very high level of supersaturation can be attained under the intensive local motions induced by ultrasound, resulting in a very high nucleation rate. This effect was found to be extremely useful in the fabrication of sub-50 nm AP nanoparticles.

1. Introduction Polymeric nanoparticles are polymer-based matrixes ranging in size from 1 to 1000 nm.1-5 Among the technologies developed to prepare polymeric nanoparticles,1-24 two types of methods are conventionally used. One is the dispersion of preformed polymers, including microphase inversion,3 nanopreciptation,11,12 solvent evaporation,20,21 spontaneous emulsification/diffusion,22 salting out,23 and spray drying.24 The other is the polymerization of monomers in dispersed emulsions.3,5,14,15 These two types of technologies share at least one limitation: almost all the approaches require surfactant to achieve a stable dispersion, in particular, dispersions of nanoparticles in the sub100 nm range. However, the removal of surfactant without affecting the stability is an extremely difficult task, if not impossible.3,15 In the fabrication of hydrophobic polymeric nanoparticles, some strong repulsive interactions have to be introduced among nanoparticles to obtain a kinetically stable dispersion.25 Most reported works25,26 utilized either electrostatic or steric repulsion to achieve this purpose. Although the solvation effect is thought to be helpful for obtaining stable dispersions, to date, no work has been reported concerning the important role of the solvation in the stabilization. On the contrary, some researchers claimed that solvation force, due to its short range nature, may not be of any practical importance in the preparation of stable dispersions.27 People have found that hydrophobic nanoparticles of small molecules could be prepared by the decrease in solubility as a result of the addition of nonsolvent.28 For example, sulfur nanoparticles can be fabricated by pouring sulfur/hot alcohol solution into water. However, since alcohol is dissolved into * To whom correspondence should be addressed. E-mail: phyliuxy@ nus.edu.sg. † Department of Chemical and Biomolecular Engineering. ‡ Department of Physics.

the bulk water and no longer serves as a true solvent for sulfur, the presence of a peptizing agent is usually necessary for the preparation of a stable dispersion.28 When a polymer solution is mixed with a large amount of nonsolvent, a high supersaturation will result in a liquid-liquid phase separation in solutions.29 During the liquid-liquid phase separation, a polymer rich phase and a polymer poor phase will form. When the concentration of the polymer rich phase exceeds the solidification concentration, the polymer rich phase will gel into a solid phase. Therefore, polymeric particles can be produced in this way. However, without the presence of surfactants, the polymeric particle dispersions are normally not stable. On the basis of the thermodynamics principles, polymeric particles will tend to aggregate to facilitate the whole system to achieve its minimum free energy.29 To date, it still remains a big challenge to prepare surfactant free yet stable polymeric nanoparticles. Although the theoretic study of solvation forces still remains in its infancy, people have already recognized that solvation force can have indirect effects on long range interactions.30 This finding evokes the reconsiderations concerning the importance of the solvation stabilization. We notice that, for the polymer (P)/solvent (S)/nonsolvent (NS) mentioned earlier, the main reason for the particle aggregation is that the solvent is dissolved into the bulk nonsolvent and no longer serves as a true solvent for the polymer. In such a case, polymeric particles will be exposed to nonsolvent molecules and tend to aggregate due to the hydrophobic interactions.31 However, if the P-S interaction wins over the S-NS interaction, polymeric particles will be selectively solvated by the solvent. This is so-called preferential solVation32 (also referred to as selectiVe solVation31b). From the viewpoint of colloid science, to obtain a stable particle dispersion, one should push the position corresponding to the minimum of the interaction free energy curve far from the particle surface in order to cause the attractive van der Waals’ interaction to be less than kT (k, Boltzmann’s constant; T, temperature).30 For the particles selectively solvated by the

10.1021/jp0446480 CCC: $30.25 © 2005 American Chemical Society Published on Web 07/02/2005

13878 J. Phys. Chem. B, Vol. 109, No. 29, 2005

Xiong et al. TABLE 1: Stability Tests for P/S/NS Combinations Using Different Polymersa P

S

NS

stability

P84 (AP)

NMP EtOH stable (no precipitation was observed; particle size was ∼65 nm and remained stable for at least 25 days Matrimid (AP) NMP EtOH similar to above paricle size: ∼27 nm polystyrene (PS) NMP EtOH unstable (precipitation occurred immediately regardless of PS/NMP concentration; no nanoparticle formed) ultrason E1010 NMP EtOH unstable (white, almost opaque (PSf) sample was obtained immediately; as-prepared particle size was ∼140 nm; precipitation was observed after 3-4 h) Udel A-300 (PES) NMP EtOH same as above (PSf) Ultem 1010 (PEI) NMP EtOH same as above (PSf) a P, polymer; S, solvent; NS, nonsolvent. All of the samples were prepared using the BT method.

Figure 1. (a) Chemical structure of P84; (b) chemical structure of Matrimid; (c) schematic presentation of P84 nanoparticle preparation. FT, forward titration method; BT, backward titration method; UEBT, ultrasound enhanced BT method; AP, aromatic polyimide (P84 or Matrimid); S, solvent (NMP); NS, nonsolvent (ethanol); PR, polymer rich phase (particle phase); CP, continuous phase in equilibrium.

solvent, the single solvation shell may still be too thin to achieve this goal. If the solvent molecules constituting the solvation shell can be further solvated by the nonsolvent, a solvation stabilization chain in the form of NS f S f P can form. The solvation multilayer induced by this solvation chain may have great significance in the fabrication of stable polymeric nanoparicles. However, currently, this still remains a kind of speculation without any experimental support, although a few works32-34 have just slightly remarked upon solvation stabilization31b,32,34 and polymers in mixed solvents32,33 separately. Three challenges then arise. (1) How can we find suitable P/S/NS combinations to form the proposed stabilization chain? (2) How can we verify the significance of this stabilization chain? (3) How can the particle size be tuned in a nonsolvent dominated solvent/ nonsolvent mixture? The aim of this study is to utilize preferential solvation stabilization to fabricate surfactant free yet stable polymeric nanoparticles. First, we will select suitable P/S/NS combinations in which the solvation stabilization chain may form. Second, the significance of the stabilization chain will be identified by abundant comparative experiments. Finally, three methods (including one ultrasound integrated method) will be developed and their effectiveness in tuning nanoparticle size will be investigated. 2. Experimental Section The experiments were performed in two ways denoted as the forward titration (FT) method and the backward titration (BT) method, based on the moving directions of the composition locus in the ternary phase diagram (Figure 1). As an extension of the

BT method, the so-called ultrasound enhanced backward titration (UEBT) method (Figure 1), ultrasound was introduced into the nanoparticle formation system. Two aromatic polyimides, P84 (supplied by Lenzing, Austria, Figure 1) and Matrimid (supplied by Ciba Geigy, U.S.A., Figure 1), were used as typical polymers. Other polymers used in this study are listed in Attachment 1 of the Supporting Information. NMP (N-methyl-2-pyrrolidone, g99.5%, Merck-Schuchardt, Germany) and ethanol (>99.7%, supplied by Hayman, U.K.) were used as the typical solvent and nonsolvent, respectively. The specifications of all of the other solvents and nonsolvents are listed in Attachment 2 of the Supporting Information. To get comparable experiment results, the same final compositions (2 mL of 2 wt % AP/NMP solution + 20 mL of ethanol) were used for all three methods. The morphology of the AP nanoparticles was observed using a JEOL JSM 6700 F field emission scanning electron microscope (FESEM) at 20 kV. The volume-average particle size and the particle size distribution were obtained by a Malvern high performance particle sizer (HPPS). Utilizing a patented optical system called NIBS (noninvasive backscatter), the HPPS is able to measure the size of dispersions up to 20 vol % directly. 3. Results and Discussion 3.1. Selection of Suitable Polymer/Solvent/Nonsolvent Combinations. To promote the formation of a solvation multilayer through a NS f S f P solvation chain, the first issue is to find suitable P/S/NS combinations in which the P-S interaction wins over the S-NS interaction. Unfortunately, there are too many possible combinations for us to conduct a complete survey. To simplify the sieving process, we fixed the solvent and nonsolvent to be NMP and ethanol (EtOH), respectively, and only examined the polymers available within our lab. These polymers included aromatic polyimide (AP), polyetherimide (PEI), polysulfone (PSf), polyethersulfone (PES), and polystyrene (PS). In membrane science, NMP is a common solvent for these polymers, while ethanol often serves as the common nonsolvent. Results are listed in Table 1. Finally, we found two aromatic polyimides, P84 and Matrimid, which resulted in very stable nanoparticle dispersions. Dispersions obtained from the FT, BT, and UEBT methods are shown in Figure 2. It was found that, without any surfactant, no detectable precipitation occurred and particle size distribution remained almost unchanged in 25 days (Figure 2). Preferential solvation of the AP nanoparticles can be qualitatively analyzed using the quasi-lattice quasi-chemical (QLQC)

Preferential Solvation Stabilization

J. Phys. Chem. B, Vol. 109, No. 29, 2005 13879 TABLE 2: SPP-SB-SA Values for the Solvents and Nonsolvents Used in This Studya solvent

formula

SPP

SB

SA

Type I: Strong EPD with High Polarity, High Basicity, and Low Acidity NMP C5H9NO 0.970 0.613 0.024 DMP C3H7NO 0.954 0.613 0.031 DMAC C4H9NO 0.970 0.650 0.028 DMSO C3H6NO 1.0 0.647 0.072 Type II: Solvents with Medium Polarity and Low Acidity acetone C3H6O 0.881 0.475 0.0 dichloromethane CH2Cl2 0.876 0.178 0.040 chloroform CHCl3 0.786 0.071 0.047 THF C4H8O 0.838 0.591 0.0 Type III: Solvents with Low Polarity and Low Acidity n-hexane C6H14 0.519 0.056 0.0 toluene C7H8 0.655 0.128 0.0 cyclohexane C6H12 0.557 0.073 0 Figure 2. P84 nanoparticle size distribution by volume. The final composition is kept the same for all three methods (2 wt % P84/NMP solution + 20 mL of ethanol).

model.32 On the basis of the principle of the QLQC model, AP nanoparticles and NMP and ethanol molecules are all treated as particles. All of the particles are assumed to be distributed on the sites of a quasi-lattice of the dispersion which is characterized by the lattice parameter Z. This parameter specifies the number of neighbors each particle has and is independent of the nature of the particles. If we denote AP, NMP, and ethanol as component P, S, and NS, respectively, the local mole fraction of NMP around AP (xLS ) can be written as

Type IV: Strong EPA with Medium Polarity and High Acidity methanol CH4O 0.857 0.545 0.605 ethanol C2H6O 0.853 0.658 0.400 n-propanol C3H8O 0.847 0.727 0.367 n-butanol C4H10O 0.837 0.809 0.341 t-butanol C4H10O 0.829 0.928 0.145 a SPP, solvent polarity/polarizability; SB, solvent basicity; SA, solvent acidity.

xLS ) 1/[1 + (NNS-NS/NS-S)1/2 exp(∆eP-S-NS/2kT)] (1) ∆eP-S-NS ) (∆solvG(P,S) - ∆solvG(P,NS))/ZNA

(2)

where k denotes Boltzmann’s constant, T is temperature, NNS-NS is the number of the nearest neighboring NS-NS pairs, NS-S is the number of the nearest neighboring S-S pairs, NA is Avogadro’s number, ∆solvG(P,S)and ∆solvG(P,NS) are the molar Gibbs energies of solvation of AP in NMP and in ethanol, respectively. For a system with fixed polymer/solvent/nonsolvent compositions, both Z and NNS-NS/NS-S are constant and independent of the nature of polymers. Therefore, for different polymers, xLS is determined by ∆eP-S-NS. Since the dissolution of AP in NMP is a thermodynamically favorable process while the dissolution of AP in ethanol is a thermodynamically very unfavorable process, according to eq 1, a very negative value of ∆eP-S-NS will result in an NMP dominant solvation layer around the aromatic AP nanoparticles. 3.2. Significance of the Solvation Stabilization Chain. NMP is a good electron pair donor (EPD) solvent due to the presence of its lone electron pairs.31 Ethanol is a good electron pair acceptor (EPA) nonsolvent.31 Therefore, a strong EPD-EPA interaction arises between NMP and ethanol molecules. On the other hand, resultant stable AP nanoparticle dispersions have already supported that preferential solvation of AP molecules occurred in the ethanol dominated NMP/ethanol mixture. Considering the hydrogen bonding formed among ethanol molecules, the stabilization mechanism can be further presented as preferential solvation

strong EPD-EPA interaction

hydrogen bonding

AP 798 NMP 798 ethanol 798 ethanol (EPD) (EPA) (bulk)

Figure 3. SPP-SB-SA plot of the solvents and nonsolvents used in this study.

Many scales can be used to quantify the strength of EPD and EPA, that is, the donicity and acceptivity of solvents.31b,c In this study, we accepted the scale proposed by Javier C. et al. because they provided property parameters for a very broad range of solvents, ranking in increasing order of polarity.31c As shown in Table 2, the polarity, donicity, and acceptivity of solvents/nonsolvents can be quantitatively characterized by solvent polarity/polarizability (SPP), solvent basicity (SB), and solvent acidity (SA) values, respectively. All of the solvents or nonsolvents used in this study then can be classified into four types: (1) type I, strong EPD with high polarity, high basicity, and low acidity; (2) type II, solvents with medium polarity and low acidity; (3) type III, solvents with low polarity and low acidity; and (4) type IV, strong EPA with medium polarity and high acidity. High polarity, medium polarity, and low polarity are defined in Figure 3. Currently, theoretic frameworks of the preferential solvation32 and the solvation stabilization are not well-established yet.30 Therefore, in this study, we investigated the significance of the solvation stabilization chain through abundant comparative experiments. Various molecular probes were used to substitute one element in the stabilization chain every time. First, to examine the importance of the preferential solvation, we used polystyrene, a polymer with a weakly polar structure, to replace

13880 J. Phys. Chem. B, Vol. 109, No. 29, 2005 TABLE 3: Stable AP Nanoparticle Dispersions Prepared from AP/Strong EPDs/Strong EPAsa P

S (strong EPD)

NS (strong EPA)

stability

P84 P84 P84 P84 P84 P84 P84 Matrimid

NMP DMF DMAC DMSO NMP NMP NMP NMP

ethanol ethanol ethanol ethanol methanol n-propanol n-butanol ethanol

stable; d ∼ 65 mn stable; d ∼ 81 mn stable; d ∼ 69 mn stable; d ∼ 73 mn stable; d ∼ 67 mn stable; d ∼ 62 mn stable; d ∼ 47 mn stable; d ∼ 27 mn

a

AP, aromatic polyimides (P84 or Matrimid). d is the volumeaverage particle size, stable for at least 21 days.

Xiong et al. TABLE 6: Unstable Samples Prepared from P84/Medium EPD (THF) or Weak EPDs (CH2Cl2, CHCl3)/Ethanola P

S (medium or NS (strong weak EPD) EPA)

Matrimid

CH2Cl2

EtOH

Matrimid

CHCl3

EtOH

Matrimid

THF

EtOH

stability unstable (serious precipitation occurred immediately) unstable (serious precipitation occurred immediately) unstable (serious precipitation occurred immediately)

a Matrimid was used instead of P84, since P84 cannot be dissolved in THF, CH2Cl2, and CHCl3, which are all of medium polarity. Matrimid cannot be dissolved in acetone.

TABLE 4: Unstable Samples Prepared from P84/NMP/ non-EPAs with Low Polarity P

S (strong EPD) NS (non-EPA)

P84

NMP

toluene

P84

NMP

cyclohexane

P84

NMP

n-hexane

stability unstable (flocculation occurred soon after preparation) unstable (precipitation occurred immediately) not mixable with P/S solution

TABLE 5: Unstable Samples Prepared from P84/NMP/ Non-EPAs (Acetone, THF) or Weak EPAs (CH2Cl2, CHCl3) with Medium Polarity (Similar Results Were Obtained When NMP Was Replaced by Other Strong EPDs Such as DMSO, DMAC, and DMF) P

S (strong NS (non-EPA or EPD) weak EPAs)

P84

NMP

acetone

P84

NMP

CH2Cl2

P84

NMP

CHCl3

P84

NMP

THF

stability unstable (floating flakes can be observed soon after preparation) unstable (precipitation was observed 28 h after preparation) unstable (precipitation occurred immediately) unstable (flocculation was observed 15 h after preparation)

aromatic AP. We found that complete precipitation occurred immediately regardless of the concentration of the PS/NMP solution (Table 1). This can be interpreted by eqs 1 and 2. Preferential solvation stabilization requires that the local solvent fraction of the solvation layer exceeds a certain value. According to eqs 1 and 2, this means a negative ∆solvG(P,S) value, that is, a strong P-S interaction, is necessary. However, polystyrene, as a polymer with a weakly polar structure, interacts only weakly with NMP. Most NMP molecules therefore dissolved into the bulk ethanol, and therefore, the preferential solvation effect is too weak. Second, to investigate the importance of the strong EPD-EPA interaction between the solvent and the nonsolvent, many S-NS pairs were investigated. On the basis of the polarity, basicity, and acidity of the solvents/nonsolvents used, these pairs can be classified into five groups: (1) strong EPDstrong EPA (Table 3); (2) strong EPD-non-EPA with low polarity (Table 4); (3) strong EPD-non-EPA/weak EPA with medium polarity (Table 5); (4) medium/weak EPD with medium polarity-strong EPA (Table 6); and (5) medium/weak EPD with medium polarity-non-EPA with low polarity (Table 7). According to the results presented in Tables 3-7, it turns out that, to achieve stable nanoparticle dispersions, besides a high polarity, a high basicity and a high acidity are also required for solvent and nonsolvent, respectively. This facilitates the solvent molecules in the preferential solvation shell (S f AP) to be further solvated by nonsolvent molecules (NS f S) and then form a complete stabilization chain (NS f S f AP). The results from Tables 4-7 support that, when either the strong EPD or strong EPA is replaced by another type of solvent or

Figure 4. Evolution of P84 particle size with time. The final composition is kept the same for both tests (2 wt % P84/NMP solution + 20 mL of butanol).

nonsolvent, the stabilization chain will be weakened and cannot stabilize nanoparticle dispersions. The significance of the solvation stabilization chain can also be supported by another example. To verify the existence of the solvation stabilization chain, we used two isomers, n-butanol and tert-butyl alcohol, to replace ethanol. As shown in Figure 4, P84 nanoparticles resulting from n-butanol are very stable: particle size remains almost constant in the testing period. However, P84 nanoparticles resulting from tert-butyl alcohol are unstable: the particle size kept increasing in the testing period. This may be due to the following reasons. First, as shown in Table 2 and Figure 3, although both tert-butyl alcohol and n-butanol are medium polar nonsolvents, the SA value of tertbutyl alcohol (0.145) is much less than the SA value of n-butanol (0.341), leading to much a weaker S-NS interaction. Second, compared with tert-butyl alcohol, n-butanol is a chain-type molecule with high molecular regularity. A thicker and denser solvation multilayer may therefore form using n-butanol as the nonsolvent, leading to a highly stable dispersion. On the contrary, bulky pendent groups of tert-butyl alcohol hinder the regular arrangement of tert-butyl alcohol molecules. Therefore, the external part (NS f S) of the solvation stabilization chain (NS f S f AP) is very weak. As a result, an unstable P84 dispersion was obtained. 3.3. Temperature Effect on the Stability of AP Nanoparticles. We use the nanoparticle dispersion obtained from P84/ NMP/EtOH through the BT method as a typical example. We examined the stability of this system at 4, 10, 21 (room temperature), 40, 60, and 80 °C to cover the full feasible range of the nonsolvent (ethanol). The duration of the observation was fixed at 4 days. It was found that the dispersion was quite stable at relatively low temperatures (4-40 °C), while at high temperatures (60 and 80 °C), obvious precipitation occurred in 2 days. These results may be interpreted as follows. On one hand, with the increase of temperature, both the preferential solvation of polymer molecules (S f AP) and the S-NS

Preferential Solvation Stabilization

J. Phys. Chem. B, Vol. 109, No. 29, 2005 13881

TABLE 7: Unstable Samples Prepared from P84/Medium EPD (THF) or Weak EPDs (CH2Cl2, CHCl3)/Non-EPAs P

S (medium or weak EPD)

NS (non-EPA)

stability

Matrimid Matrimid Matrimid

CH2Cl2, CHCl3, or THF CH2Cl2, CHCl3, or THF CH2Cl2, CHCl3, or THF

n-hexane cyclohexane toluene

unstable (serious precipitation occurred immediately) unstable (serious precipitation occurred immediately) unstable (serious precipitation occurred immediately)

interaction (NS f S) were weakened, leading to a weakened stabilization chain (NS f SAP) and thus a lower energy barrier. On the other hand, higher temperature facilitated a much higher collision frequency of nanoparticles so that the possibility to overcome the lowered energy barrier was significantly increased. The above two reasons took effect together and resulted in evident precipitations under high temperatures. 3.4. Tuning of AP Nanoparticle Size. To tune polymeric nanoparticle size, the immediate idea we may have is to change the P/S concentration so as to obtain nanoparticles with different sizes. Using the combination of P84/NMP/EtOH as the typical example, we investigated the concentration effect on particle size through the BT method. It was found that when the concentration was decreased from 3 to 2 wt % and further 1 wt %, the volume-average diameter of nanoparticles decreased from 79 to 65 nm and further to 63 nm, respectively. This result implies that although smaller nanoparticles can be obtained through lowering the P/S concentration, the effectiveness of this approach is quite limited. Meanwhile, with the decrease of the P/S concentration, the mass of nanoparticles per unit volume also decreased. To effectively tune AP nanoparticle size in a broad range, we need to develop other approaches. To explore other approaches to control AP nanoparticle size, a good understanding of the formation mechanism is required. Concerning the size tuning of the AP nanoparticles, a way to control the energy barrier of the nanoparticle formation needs to be identified. Since the concentration of the polyimide/NMP solution is only 2 wt %, liquid-liquid phase separation can only occur through a nucleation controlled mechanism in the presence of a large amount of nonsolvent.29,35 Figure 2 shows that the nanoparticles obtained with all three methods are round and regular. This particle morphology indicates that the surface free energy plays an important role in the nanoparticle formation, which is a typical feature of a nucleation controlled process. On the basis of the nucleation theory,36 the radii of the critical nuclei (rc), the nucleation barrier (∆G*), and the nucleation rate (J), which is the number of critical nuclei generated per unit time-volume, can be expressed as

rc ) 2Ωγcf/kT ln(1 + σ) ∆G* )

16πγ3cfΩ2 3[kT ln(1 + σ)]2

J ) B exp(-∆G*/kT)

the nucleation barrier (∆G*) by establishing a high supersatuation. In this study, we tuned the supersatuation by taking three different kinetic paths (Figure 5a and b). In the FT method, involving the addition of nonsolvent (ethanol), the composition of the whole system moves along a straight line (AB in Figure 5a). AP nanoparticles are obtained when the composition of the whole system exceeds the binodal line. In this case, the supersaturation (σFT in Figure 5a) is quite low. A low nucleation rate results in the production of rather large particles with quite an extensive size distribution (100-300 nm) (Figure 2). In the BT method, the supersaturation is generated in a different way. When a drop of P84/NMP solution is introduced into the ethanol, stirring causes it to break up into small droplets very quickly. Because the viscosity of the polymer solution is much higher than that of the surrounding nonsolvent, the outflux of the solvent from the droplet is greater than the influx of the nonsolvent, leading to an upward locus in the phase diagram37 (AB in Figure 5b). Fast interdiffusion between droplets and the surrounding ethanol results in a high supersaturation in the droplet domain (σBT in Figure 5b), leading to quite a high

(3) (4) (5)

where B is a kinetic parameter and is constant for a given system, Ω is the volume of the growth unit, γcf is the surface free energy between the nuclei and the mother phase. σ is the supersaturation (σ ) (C0 - C*)/C*, where C0 is the polymer concentration of the ternary system before the phase separation (e.g., point B in Figure 5a) and C* is the polymer concentration of the corresponding continuous phase in equilibrium (e.g., point B′ in Figure 5a). Since C* is very small and almost a constant, σ can be well represented by a line segment connecting the phase before and after phase separation (e.g., BB′ in Figure 5a). For a nucleation controlled process, the particle size and the size distribution are controlled by the nucleation rate. According to eqs 2-5, to achieve a high nucleation rate, we need to lower

Figure 5. Schematic presentation of polyimide nanoparticle formation: (a) FT method; (b) BT and UEBT methods. (c) Phase behavior during AP nanoparticle formation. AP, polyimide (P84 or Matrimid); S, solvent (NMP); NS, nonsolvent (ethanol); ST, supersaturated state; PR, polymer rich phase (particle phase); PP, polymer poor phase (continuous phase); CP, continuous phase in equilibrium;

13882 J. Phys. Chem. B, Vol. 109, No. 29, 2005 nucleation rate. In this situation, we obtained quite small nanoparticles with a narrower size distribution (30-100 nm) (Figure 2). However, the BT method has a serious drawback: when a drop of polymer solution is broken into many droplets, these droplets move along with the convection flow. Since the relative motion between droplets and convection flow is rather weak, the mass transfer boundary layer around the droplets becomes increasingly thicker with time. This inhibits the fast interdiffusion process that is necessary for a higher supersatuation. To overcome this serious drawback, ultrasound was introduced into the system (Figure 1). Ultrasound can induce very intensive relative motion near the tip region. Such intensive local motion not only breaks up the introduced polymer solution into very fine droplets instantly, but it also greatly enhances the mass transfer between the droplets and the surrounding nonsolvent. In this way, an extremely high degree of supersaturation (σUEBT in Figure 5b) can be achieved, leading to extremely small polyimide nanoparticles (10-50 nm) (Figure 2). Compared with published works on polyimide particle fabrication,38-41 our technology is much simpler. Utilizing selective solvation together with an EPD-EPA interaction, we can produce sub-50 nm polyimide nanoparticles in the absence of surfactants. If suitable solvent and nonsolvent combinations are available, hopefully our technology can be extended to other polymers. 4. Conclusions In summary, preferential solvation stabilization can have great significance in the fabrication of stable hydrophobic polymeric nanoparticles. Surfactant free yet stable P84 and Matrimid nanoparticles were prepared using a liquid-liquid phase separation method. Systematic comparative experimental study supports that the solvation stabilization chain in the form of NS f S f P plays the key role in the stabilization. On the other hand, the formation of AP nanoparticles is governed by a nucleation controlled process. Three methods (the FT, BT, and UEBT methods) were used to obtain P84 nanoparticles of different sizes. A very high level of supersaturation can be attained under the intensive local motions induced by ultrasound, resulting in a very high nucleation rate. This effect is extremely useful in the fabrication of sub-50 nm polyimide nanoparticles. Acknowledgment. We gratefully acknowledge the financial support from NUS (Grant No. R-279-000-108-112) and from A*star funding (Project No. 0221010036). We thank Miss O. H. Lily for her support in experiments and Dr. Da-Wei Li for many fruitful discussions. Supporting Information Available: Graphic showing some other polymers used in this study and table showing the specifications of some other solvents and nonsolvents used in this study. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Couvreur, P.; Dubernet, C.; Puisieux, F. Eur. J. Pharm. Biopharm. 1995, 41, 2-13. (2) Yang, L.; Alexandridis, P. Curr. Opin. Colloid Interface Sci. 2000, 5, 132-143. (3) (a) Zhang, G. Z.; Niu, A.; Peng, S. F.; Jiang, M.; Tu, Y. F.; Li, M.; Wu, C. Acc. Chem. Res. 2001, 34, 249-256. (b) Zhao, Y.; Liang, H. J.; Wang, S. G.; Wu, C. J. Phys Chem. B 2001, 105, 848-851.

Xiong et al. (4) Harmia, T.; Kreuter, J.; Speiser, P.; Boye, T.; Gurny, R.; Kubis, A. Int. J. Pharm. 1986, 33, 187-93. (5) Soppimath, K. S.; Aminabhavi, T. M.; Kulkarni, A. R.; Rudzinski, W. E. J. Controlled Release 2001, 70, 1-20. (6) Merkel, T. C.; Freeman, B. D.; Spontak, R. J.; He, Z.; Pinnau, I.; Meakin, P.; Hill, A. J. Science 2002, 296, 519-522. (7) Xu, Z. K.; Xiao, L.; Wang, J. L.; Springer, J. J. Membr. Sci. 2002, 202, 27-34. (8) He, Z. J.; Pinnau, I.; Morisato, A. Desalination 2002, 146, 11-15. (9) Nunes, S. P.; Peinemann, K. V.; Ohlrogge, K.; Alpers, A.; Keller, M.; Pires, A. T. N. J. Membr. Sci. 1999, 157, 219-226. (10) Joly, C.; Goizet, S.; Schrotter, J. C.; Sanchez, J.; Escoubes, M. J. Membr. Sci. 1997, 130, 63-74. (11) Ezpeleta, I.; Irache, J. M.; Stainmesse, S.; Chabenat, C.; Gueguen, J.; Popineau, Y.; Orecchioni, A.-M. Int. J. Pharm. 1996, 131, 191-200. (12) Duclairoir, C.; Irache, J. M.; Nakache, E.; Orecchioni, A.-M.; Chabenat, C.; Popineau, Y. Polym. Int. 1999, 79, 327-333. (13) Lee, J. H.; Jung, S.-W.; Kim, I.-S.; Jeong, Y.-I.; Kim, Y,-H.; Kim, S.-H. Int. J. Pharm. 2003, 251, 23-32. (14) Fine particles: synthesis, characterization, and mechanisms of growth; Sugimoto, T., Ed.; Marcel Dekker Inc.: New York, 2000. (15) (a) Fitch, R. M. Polymer Colloids: A ComprehensiVe Introduction; Academic Press: San Diego, CA, 1997. (b) Polymeric Dispersions: Principles and Applications; Asua, J. M., Ed.; Kluwer Academic: Boston, MA, 1997. (16) Wu. C.; Fu. J.; Zhao. Y. Macromolecules 2000, 33, 9040-9043. (17) Kim, I. S.; Kim, S. H. Int. J. Pharm. 2001, 226, 23-29. (18) Gao, J.; Frisken, B. J. Langmuir 2003, 19, 5212-5216. (19) Colloidal Drug DeliVery Systems; Kreuter, J., Ed.; Marcel Dekker: New York, 1994. (20) Klier, J.; Tucker, C. J.; Kalantar, T. H.; Green, D. P. AdV. Mater. 2000, 12, 1751-1757. (21) (a) Kietzke, T.; Neher, D.; Landfester, K.; Montenegro, R.; Gu¨ntner, R.; Scherf, U. Nat. Mater. 2003, 2, 408-412. (b) Landfester, K.; Montenegro, R.; Scherf, U.; Gu¨ntner, R.; Asawapirom, U.; Patil, S.; Neher, D.; Kietzke, T. AdV. Mater. 2002, 14, 651-655. (22) Niwa, T.; Takeuchi, H.; Hino, T.; Kunou, N.; Kawashima, Y. J. Controlled Release 1993, 25, 89-98. (23) Allemann, E.; Gurnay, R.; Doelker, E. Int. J. Pharm. 1992, 87, 247-253. (24) Mu, L.; Feng, S. S. J. Controlled Release 2001, 76, 239-254. (25) Hunter, R. J. Foundations of colloid; Science Oxford University Press: New York, 2000. (26) Fitch, R. M. Polymer Colloids: A ComprehensiVe Introduction; Academic Press: San Diego, CA, 1997. (27) Napper, D. H. Polymeric Stabilization of Colloidal Dispersions; Academic Press: San Diego, CA, 1983. (28) Hartman, R. J. Colloid Chemistry; Sir Isaac Pitman: London, 1948. (29) Kamide, K. Thermodynamics of polymer solutions: phase equilibria and critical phenomena; Elsevier Science Publishers B. V.: New York, 1990. (30) Garbassi, F.; Morra, M.; Occhiello, E. Polymer Surfaces: From Physics to Technology; John Wiley & Sons: Chichester, U.K., 1998. (31) (a) Atkins, P. W. Physical Chemistry; Oxford University Press: Oxford, U.K., 1998. (b) Reichardt, C. SolVents and SolVent Effects in Organic Chemistry; Wiley-VCH: New York, 1988. (c) George, W., Ed. Handbook of solVents; ChemTec: Toronto, 2001. (32) Marcus, Y. SolVent Mixtures: Properties and SelectiVe SolVation; Marcel Dekker: New York, 2002. (33) Soria, V.; Figueruelo, J. E.; Abad, C.; Campos A. Macromol. Theory Simul. 2004, 13, 441-452. (34) Israelachvili, J. Acc. Chem. Res. 1987, 20, 415-421. (35) Koros, W. J.; Fleming, G. K. J. Membr. Sci. 1993, 83, 1. (36) (a) Liu, X. Y. J. Chem. Phys. 2000, 112, 9949. (b) Sato, K., Nakajima, K., Furukawa, Y., Eds. AdVances in Crystal Growth Research; Elsevier: Amsterdam, The Netherlands, 2001. (c) Xiong, J. Y.; Liu, X. Y.; Sawant, P. D.; Chen, S. B.; Chung, T. S.; Pramoda, K. P. J. Chem. Phys. 2004, 121, 12626. (37) Matsuura, T. Synthetic membranes and membrane separation processes; CRC Press: Boca Raton, FL, 1994. (38) Lin, T.; Stickney, K. W.; Rogers, M.; Riffle, J. S.; Mcgrath, J. E.; Marand, H.; Yu, T. H.; Davis, R. M. Polymer 1993, 34, 772. (39) Nagata, Y.; Oonishi, Y.; Kajiyama, T. Polym. J. 1996, 28, 980. (40) Brock, T.; Sherrington, D. C. J. Mater. Chem. 1991, 1, 151. (41) Hasegawa, M.; Horie, K. Prog. Polym. Sci. 2001, 26, 259.