Separation and Purification of Nanoparticles in a Single Step

Dec 29, 2009 - (22) reported an elegant method for Au-NP synthesis and recovery using a catanionic surfactant stabilized ME. ..... method also has not...
0 downloads 10 Views 2MB Size
pubs.acs.org/Langmuir © 2009 American Chemical Society

Separation and Purification of Nanoparticles in a Single Step Martin J. Hollamby and Julian Eastoe* School of Chemistry, University of Bristol, Cantocks Close, Bristol BS81TS, U.K.

Angela Chemelli and Otto Glatter Department of Chemistry, Karl-Franzens University, Heinrichstr. 28, A-8010 Graz, Austria

Sarah Rogers and Richard K. Heenan ISIS Facility, STFC Rutherford Appleton Laboratory, Didcot, Oxfordshire OX11 0QX, U.K.

Isabelle Grillo Institut Max-von-Laue-Paul-Langevin, BP 156-X, F-38042 Grenoble Cedex, France Received November 6, 2009. Revised Manuscript Received December 11, 2009 Reversed-micelle synthesis has been used to generate CTAB-stabilized gold (Au-NPs) and silver nanoparticles (Ag-NPs). By inducing a phase transition and subsequent separation of the background supporting microemulsion, it has been possible to extract and purify the NPs from the reaction medium. After addition of excess water, the NPs concentrate into an upper octane-rich phase, with impurities and reaction debris (in particular CTAB) partitioning into the water-rich lower phase. UV and 1H NMR showed that 82% of the original mass of Au-NPs can be purified from the excess CTAB and other salt impurities. The concentrated and purified NPs can be dried down, by solvent removal, and then redispersed in octane. Using the complementary techniques small-angle neutron and X-ray scattering (SANS and SAXS), the structures of microemulsions both with and without nanoparticles prior to separation, and in both upper and lower phases after separation, have been elucidated. The approach has also been applied to the synthesis and recovery of silver nanoparticles, but on a larger scale. This new approach compares favorably with existing methods as it uses no additional organic solvents, has a low-energy demand, and requires no specialist surfactants. The new advance here is that by using a colloidal system to prepare and support the nanoparticles as a structured solvent, a simple soft purification method becomes accessible, which is otherwise impossible with a normal molecular solvent.

Introduction Using water-in-oil (w/o) microemulsions (MEs) and reversed micelles (RMs) as media for nanoparticle (NP) synthesis1,2 remains a very popular method, having one major (often unspoken) drawback. In order to generate sufficient quantities of the NPs, significant amounts of surfactant are required (typically 2-3 orders of magnitude higher in concentration than the reactants for NPs). Such surfactant impurities are difficult to remove, and while calcination can be employed as a purification step, this often affects strongly particle size, being also an inappropriate treatment for certain nanomaterials, such as metals. Furthermore, extreme heat treatment burns off stabilizing surfactant layers, preventing ready redispersion of the nanoparticles in nonaqueous media. Gold nanoparticles (Au-NPs) are of particular current academic and industrial interest due to high efficiency in greener oxidation catalysis3-6 and biomolecule detection.7,8 In such *Corresponding author. E-mail: [email protected]. (1) Pileni, M.-P. Nat. Mater. 2003, 2, 145. (2) Eastoe, J.; Hollamby, M. J.; Hudson, L. Adv. Colloid Interface Sci. 2006, 128-130, 5. (3) Haruta, M. Chem. Rec. 2003, 3, 75. (4) Hughes, M. D.; Xu, Y.-J.; Jenkins, P.; McMorn, P.; Landon, P.; Enache, D. I.; Carley, A. F.; Attard, G. A.; Hutchings, G. J.; King, F.; Stitt, E. H.; Johnston, P.; Griffin, K.; Kiely, C. J. Nature 2005, 437, 1132. (5) Goodman, D. W. Nature 2008, 454, 948. (6) Campbell, C. T. Science 2004, 306, 234. (7) Famulok, M.; Mayer, G. Nature 2006, 439, 666. (8) Service, R. F. Science 2005, 308, 1099a.

Langmuir 2010, 26(10), 6989–6994

applications, pure NPs are usually required. Because of the low size-dependent melting point of Au-NPs,9,10 calcination is inappropriate, and hence for RM/ME prepared Au-NPs other purification methods must be employed. Existing approaches include centrifugation,11 precipitation by antisolvent,12-14 flocculation by use of a photolyzable surfactant,15,16 or temperatureinduced separation.17 Unfortunately, many of these techniques suffer from the disadvantages of being time-consuming and highly energy or solvent intensive, and they are often ineffective in removing inorganic and small-molecule impurities.18 Considering that one major application of Au-NPs is in greener oxidation catalysis, the use of such techniques is less than ideal. This difficult issue of removal of inherent impurities has led to a shift in focus (9) Dick, K.; Dhanasekaran, T.; Zhang, Z.; Meisel, D. J. Am. Chem. Soc. 2002, 124, 2312. (10) Jiang, Q.; Zhang, S.; Zhao, M. Mater. Chem. Phys. 2003, 82, 225. (11) Coelfen, H. Polym. News 2004, 29, 101. (12) Zhang, R.; Liu, J.; He, J.; Han, B.; Wu, W.; Jiang, T.; Liu, Z.; Du, J. Chem.;Eur. J. 2003, 9, 2167. (13) Chen, D.-H.; Wu, S.-H. Chem. Mater. 2000, 12, 1354. (14) Salabat, A.; Eastoe, J.; Mutch, K. J.; Tabor, R. F. J. Colloid Interface Sci. 2008, 318, 244. (15) Vesperinas, A.; Eastoe, J.; Jackson, S.; Wyatt, P. Chem. Commun. 2007, 3912. (16) Salabat, A.; Eastoe, J.; Vesperinas, A.; Tabor, R. F.; Mutch, K. J. Langmuir 2008, 24, 1829. (17) Zhang, R.; Liu, J.; Han, B.; He, J.; Liu, Z.; Zhang, J. Langmuir 2003, 19, 8611. (18) Dahl, J. A.; Maddux, B. L. S.; Hutchison, J. E. Chem. Rev. 2007, 107, 2228.

Published on Web 12/29/2009

DOI: 10.1021/la904225k

6989

Article

away from the ME method to prepare Au-NPs to other reactions taking place in aqueous media19,20 or surfactant-free synthesis21 where the yields are often better and impurities fewer. However, in both cases the synthesized Au-NPs typically still require an excess of added stabilizing ligands to prevent agglomeration in nonaqueous solvents, which are important reaction media for various applications. Aside from stability in such nonpolar media, one generally accepted benefit of using the ME-NP synthesis method is a good size and shape control under ambient conditions.1,2 However, it is clear that in order for MEs to become viable systems for synthesis of such high value nanoparticles, new purification and recovery methods must be found. Quite recently, Abecassis et al.22 reported an elegant method for Au-NP synthesis and recovery using a catanionic surfactant stabilized ME. Reducing temperature effected a phase separation, after which Au-NPs partitioned strongly into an octane-rich upper phase, with excess surfactant concentrating in the lower aqueous phase. That nice work, while being quite specific and employing an unusual surfactant, highlighted the potential benefits of the use of a structured colloidal fluid as a supporting medium to facilitate particle recovery after synthesis. Instead of changing temperature, adding an excess of water into a w/o microemulsion can also induce phase separation; in the simplest case two distinct phases are formed, one comprising mostly oil and the other dominated by water. This phenomenon is employed here to concentrate and purify gold and silver nanoparticles. Surfactant, cosurfactant, and salt impurities are shown to concentrate in lower aqueous-rich phases, while the surfactantcoated NPs remain dispersed in upper oil-rich phases. This appears to represent a simpler and more universally applicable way to separate the unwanted parent ME components and thus purify NPs. The hydrophobic surfactant-coated NPs are more soluble in the upper oil-rich phase, whereas the free residual surfactant molecules partition strongly into the aqueous-rich phase: the NPs can then be separated from excess surfactant and hydrophilic reactants. Appropriate well-known hydrophilic surfactant candidates are CTAB or SDS, both being capable of forming w/o microemulsions with added cosurfactants. Here, Au-NPs have been synthesized in CTAB reversed micelles with butanol as coadditive.23 On addition of excess water (1:1 volume ratio water:microemulsion), the Au-NPs concentrate into the upper octane-rich phase, and impurities (in particular, excess CTAB) remain in the water-rich lower phase, as confirmed by UV-vis and 1H NMR spectroscopy. In addition, using the complementary techniques of small-angle neutron and X-ray scattering (SAXS, SANS), structures of the microemulsions have been determined, both with and without nanoparticles prior to separation, and in both upper and lower phases after separation. Furthermore, this separation method has also been applied to synthesis and recovery of silver nanoparticles (Ag-NPs). The benefits of using water to achieve the separation are apparent: as the greenest possible solvent, it represents a clean method for particle recovery. In addition, CTAB could be easily recovered from the water and reused after purification by recrystallization.24 (19) Turkevich, J.; Stevenson, P. C.; Hillier, J. Discuss. Faraday Soc. 1951, 11, 55. (20) Lin, S.; Li, M.; Dujardin, E.; Girard, C.; Mann, S. Adv. Mater. 2005, 17, 2553. (21) Schulz-Dobrick, M.; Sarathy, K. V.; Jansen, M. J. Am. Chem. Soc. 2005, 127, 12816. (22) Abecassis, B.; Testard, F.; Zemb, T. Soft Matter 2009, doi: 10.1039/ b816427d. (23) Chen, F.; Xu, G.-Q.; Hor, T. S. A. Mater. Lett. 2003, 57, 3282. (24) Chaimovich, H.; Bonilha, J. B. S.; Politi, M. J.; Quina, F. H. J. Phys. Chem. 1979, 83, 1851.

6990 DOI: 10.1021/la904225k

Hollamby et al. Table 1. Compositions of the Samples and Systems Referred to as A-E in the Texta ID

description

A B

water/CTAB/butanol/octane KAuCl4/water/CTAB/butanol/octane; [KAuCl4] = 0.0072 mol dm-3 C Au-NPs/water/CTAB/butanol/octane; [Au-NPs]b = 0.0036 mol dm-3 or 0.71 g dm-3 D lower phase (water-rich) of separated system E upper phase (oil-rich) of separated system a Note that for microemulsion samples A-C w = [water]/[CTAB] = 10 and the volume ratio of butanol:octane was kept at 1:4. b Assuming 100% conversion of Au3þ to Au0.

Experimental Section Chemicals. (1-Hexadecyl)trimethylammonium bromide (CTAB, 98%, Alfa-Aesar), potassium tetrachloroaurate (KAuCl4, 99.99%, Alfa-Aesar), silver nitrate (AgNO3, 99þ%, SigmaAldrich), sodium borohydride (NaBH4, 99%, Sigma-Aldrich), D2O (99.9 atom %, Sigma-Aldrich), and butanol (butan-1-ol, 99þ%, Sigma-Aldrich) were purchased and used without further purification. Octane (98%) was purchased from Sigma-Aldrich and purified by passing through chromatographic silica (Silica 60A, chromatography grade, Fisher Scientific). H2O was of ultrahigh purity (Elga Maxima or Millipore Milli-Q Plus system). Nanoparticle Synthesis and Recovery. Two microemulsions were prepared at w = [H2O]/[CTAB] = 10 with water (D2O was used for the samples for SANS analysis), CTAB, butanol, and octane, one containing KAuCl4 at an aqueous concentration of 0.1 mol dm-3 (overall concentration = 0.0072 mol dm-3) and the other containing NaBH4 at an aqueous concentration of 0.30 mol dm-3 (overall concentration = 0.0216 mol dm-3). The concentration of CTAB was 0.40 mol dm-3, the volume ratio of butanol: octane was 1:4, and the volume of both microemulsions was 5 mL. Upon dropwise mixing of the two microemulsions, the orange color (AuCl4-) was observed initially to disappear, and eventually a ruby-red dispersion of Au-NPs was formed. Assuming full conversion (likely due to the large excess of reductant employed), the concentration of gold is overall 0.0036 mol dm-3 (= 0.71 g dm-3). Samples were allowed to age for 1 week prior to attempting the separations. At this point, equal volumes of microemulsion containing Au-NPs (sample C) and water (D2O was used for the samples for SANS analysis) were mixed together and allowed to separate. The upper phase was removed using a pipet and was stable for many weeks. The same process was repeated to make Ag-NPs, but on a larger scale (50 mL of each precursor microemulsion was prepared) and using AgNO3 in place of KAuCl4. In this case, both precursor microemulsions (containing AgNO3 and NaBH4, respectively) appeared clear and colorless. Upon dropwise mixing of the two microemulsions, a yellow-brown color appeared, indicating the formation of nanoparticles. The concentrations employed were exactly the same as those for the Au-NPs. The purification method was repeated in the same way (after aging), albeit also on a larger scale (adding 100 mL of water to 100 mL of the microemulsion containing Ag-NPs). The upper phase, removed using a pipet, was stable for many weeks. Sample Nomenclature. For the gold-containing samples, which make up the majority of the reported work, sample identification letters from A to E have been employed: Table 1 lists the different samples.

Results and Discussion Gold Nanoparticle Synthesis. The visual appearance of the “blank” microemulsion (sample A), the microemulsion containing AuCl4- (sample B), and the microemulsion containing AuNPs (sample C) are shown inset in Figure 1. The NaBH4containing microemulsion appears colorless, similar to the blank Langmuir 2010, 26(10), 6989–6994

Hollamby et al.

Figure 1. UV spectra of samples B and C. Inset: visual appearance of samples A, B, and C.

sample. Upon mixing, a color change from yellow (due to AuCl4-) to ruby red was observed. This change was quantified by UV spectroscopy; disappearance of the peak around 400 nm and commensurate appearance of a strong absorbance band centered on 525 nm strongly suggests the presence of Au-NPs (Figure 1).25 The change happens rapidly, as evidenced by UV-vis stopped flow (Figure S1) at λ = 512 nm (chosen due to its proximity to the UV maximum); initially, absorbance is seen to decrease (in the first 20 ms), either due to an initial reduction of Au3þ to Auþ (colorless) or due to the formation of Au0 nuclei which are initially too small to permit a resonance band. After this incubation period, the absorbance at λ = 512 nm increases quickly, reaching a maximum in 35 ( 3 ms. It should be noted that this is a quicker reaction than seen in other works25 and, on the basis that intermicellar exchange takes place of order on the microsecond time scale,26 is likely to be essentially diffusion limited. Phase Separation and Purification. Addition of equal volumes of water and sample C caused a clean phase separation into two coexisting phases: an octane-rich upper phase and a more viscous water-rich lower phase. The middle image in Figure 2i shows the visual appearance of the sample after separation. The same liquid-liquid type phase separation was observed for the pure microemulsion alone, in the absence of the Au-NPs. Using 1 H NMR, the compositions with respect to surfactant, cosurfactant, and solvents of both phases were determined (see Supporting Information). The upper phase (sample E) was found to consist mainly of octane (96 wt %) with a small quantity of butanol (4 wt %); the quantity of CTAB and water was too low to be detected above the background (see Figure S2). Conversely, the lower phase (sample D) comprises 61 wt % water, 20 wt % octane, 9 wt % butanol, and 10 wt % CTAB (or 0.27 mol dm-3). These approximate compositions are subject to an error of order (1.0 wt %. Even so, this quantity of CTAB represents almost all of that initially present in sample C taking into account the added water and respective volumes of the two phases; so the analysis is self-consistent. Significantly, the Au-NPs partitioned strongly, exhibiting a preference for the upper phase as evidenced by the observed colors of the two phases: dark ruby red for the upper phase and light pink for the lower phases (middle vial, Figure 2i). This difference in concentration was quantified by UV-vis spectroscopy (Figure 3). The absolute values for absorbance were used to (25) Abecassis, B.; Testard, F.; Spalla, O.; Barboux, P. Nano Lett. 2007, 7, 1723. (26) Fletcher, P. D. I.; Howe, A. M.; Robinson, B. H. J. Chem. Soc., Faraday Trans. 1 1987, 83, 985.

Langmuir 2010, 26(10), 6989–6994

Article

Figure 2. (i) Visual appearance of sample C (left-hand vial), the resulting separated sample after addition of equal volumes water and sample C, comprising systems D and E (middle vial), and Au-NPs from sample E redispersed in octane (right-hand vial). Representative TEM image of (ii) Au-NPs from upper phase of sample C and (iii) Au-NPs from sample E. Scale bars on TEM images are 500 A˚.

Figure 3. UV spectra of samples and systems C, D, and E from Figure 2; the precursor Au-NP containing microemulsion and the two separated phases. Inset: calculated concentrations of Au-NPs from analysis of UV spectra.

approximate the concentration of gold nanoparticles in samples D and E (sample C is calculated on the basis of 100% conversion Au3þ to Au0), assuming Beer-Lambert behavior27 (inset in Figure 3). The concentration of gold in sample E is twice that in the initial homogeneous sample C and 18 times that in the coexisting lower aqueous phase (sample D). The volume fractions of the upper and lower phases are of order 0.2 and 0.8, respectively (middle vial, Figure 2i). Given these values and the concentrations from Table 1, it can be calculated that ∼82% of the mass of Au-NPs from the original ME þ Au-NPs system (sample C) is partitioned in the top phase after separation. To clarify this, over 80% of the mass of Au-NPs contained in sample C can be separated and also purified from any excess CTAB. This purified phase can then by dried down by solvent removal and the Au-NPs redispersed in fresh octane, as shown in the right-hand vial (Figure 2i). At first sight this preference for the nonaqueous phase is surprising, considering the well-known tendency of CTAB to (27) Rance, G. A.; Marsh, D. H.; Khlobystov, A. N. Chem. Phys. Lett. 2008, 460, 230. (28) Nikoobakht, B.; El-Sayed, M. A. Langmuir 2001, 17, 6368. (29) Rennie, A. R.; Lee, E. M.; Simister, E. A.; Thomas, R. K. Langmuir 2002, 6, 1031.

DOI: 10.1021/la904225k

6991

Article

Figure 4. SANS and SAXS profiles for samples A-C at 25 °C. Lines on SANS plots represent fits to the data using a model representing a Shultz distribution of polydisperse spheres with S(Q) approximated as a hard-sphere function. Lines in the SAXS plot represent fits generated by GIFT analysis.

form highly stable bilayers on negatively charged surfaces.28-31 It is likely that the mixed octane and butanol organic solvent disfavors the formation of CTAB bilayers through a solvophobic-type effect, therefore permitting the effective hydrophobization of the surface. It is perhaps likely that any residual gold nanoparticles in the lower phase are stabilized by a CTAB bilayer, although this has not been investigated. TEM images, shown in Figure 2, display nanoparticle size and shape both before (sample C) and after (sample E) purification. Both TEM images appear very similar, consisting of a polymorphic collection of spheres, triangles and short rods, with a wide length-scale range from approximately 100 to 400 A˚. Small-Angle Scattering (SAS). Microemulsion Samples A-C. Complementary small-angle scattering (SAS) techniques with neutrons (SANS) and X-rays (SAXS) were used to investigate the structures of samples A-E. In general, in smallangle scattering, the intensity I(Q) ∼ nV2(ΔF)2, where n is the number of particles, V is particle volume, and ΔF is the difference in scattering length density contrast between particle and solvent medium. By combining SANS and SAXS, it is possible to examine different structural components of interest. The SAXS contrast is proportional to atomic number (Z), whereas for SANS the associated scattering cross section bcorr is isotope dependent and essentially uncorrelated with Z. In fact, Au-NPs are all but invisible to SANS at the concentrations employed but can be detected by SAXS. A more detailed discussion of SAS analysis is provided in the Supporting Information, but the important results are summarized here. SANS and SAXS profiles for samples A-C (30) Pastoriza-Santos, I.; Perez-Juste, J.; Liz-Marzan, L. M. Chem. Mater. 2006, 18, 2465. (31) Sui, Z. M.; Chen, X.; Wang, L. Y.; Xu, L. M.; Zhuang, W. C.; Chai, Y. C.; Yang, C. J. Physica E 2006, 33, 308.

6992 DOI: 10.1021/la904225k

Hollamby et al.

are shown in Figure 4. The log-log plots for samples A and B exhibit classic microemulsion droplet scattering with I(Q) ∼ Q0 at low Q, switching over to I(Q) ∼ Q-4 at high Q and a peak in the mid-Q range. The data have been interpreted using two different methods (model fitting for SANS: generalized indirect Fourier transform (GIFT) for SAXS) and give average droplet radii for both samples A and B as Rav,SANS = 15 A˚ and Rav,SAXS = 21 A˚ (see Table S3 and Figure S5). The SANS arises from the D2O droplet core, whereas SAXS is more sensitive to the electron density contrast step at the surfactant head groups (especially the bromide ions), so one might expect a difference between reported radii from analyses of SANS and SAXS data. However, this alone cannot fully account for the difference in sizes. The droplet radii from SAXS analyses agree with previously reported values of e.g. R = 23 A˚ for water/CTAB/chloroform/isooctane32 and R = 22 A˚ for water/AOT/n-heptane,33 whereas the value extracted from the SANS fitting appears to be a little low. It should be noted that model fitting is highly sensitive to polydispersity, concentration, and interparticle interactions; samples A-C are all concentrated systems with a high proportion of cosurfactant (see Supporting Information). It therefore is not surprising that an indirect analysis method (GIFT) is more reliable in this case (more detailed discussion to be found in Supporting Information). Although the D2O contrast SANS profiles for samples A-C are essentially identical, a clear difference is seen in the SAXS from samples A and B compared to the profile for system C. The increase in SAXS intensity for sample C is likely to be due to the nanoparticles, and data fitting (Figure S6 and discussion in the Supporting Information) confirms the presence of a bimodal distribution of microemulsion droplets of similar dimensions to those of samples A and B alongside a small concentration of large particles, presumably being Au-NPs. The average NP radius (52 A˚, Figure S6) is at the low end of particle dimensions as observed by TEM (Figure 2), highlighting problems of crosscomparison between imaging techniques, which require sample treatment and preparation, and direct scattering methods which examine the sample in situ. Separated Samples D and E. SANS and SAXS data from sample D (Figure 5) are dominated by large peaks. The separation was carried out using D2O to introduce SANS contrast against the remaining H-components in the system. Recalling the compositional analysis by NMR, sample D (61 wt % water, 39 wt % surfactant þ organics) is likely to be a concentrated system comprising micelles swollen by oil or possibly bicontinuous34,35 or vesicle36,37 structures. Although robust, model fitting to the SANS profiles was hampered by the lack of a priori structural knowledge and by the presence of a repulsive S(Q), it was however possible to process the SAXS data using GIFT, to generate pair distance distribution functions (Figure S7). The size distribution is very broad, in line with highly polydisperse large oil-swollen aggregate droplets (Rav = 59 A˚, maximum dimension = 150 A˚). The most significant scattering results come from sample E, which exhibits low SANS intensity but high SAXS intensity (Figure 5): the profile was fitted to a simple form factor model (32) Lang, J.; Mascolo, G.; Zana, R.; Luisi, P. L. J. Phys. Chem. 1990, 94, 3069. (33) Nave, S.; Eastoe, J.; Heenan, R. K.; Steytler, D. C.; Grillo, I. Langmuir 2000, 16, 8741. (34) Zemb, T. C. R. Chim. 2009, 12, 218. (35) Silas, J. A.; Kaler, E. W. J. Colloid Interface Sci. 2003, 257, 291. (36) Jian, X.; Ganzuo, L.; Zhiqiang, Z.; Guowei, Z.; Kejian, J. Colloids Surf., A 2001, 191, 269. (37) Singh, M.; Ford, C.; Agarwal, V.; Fritz, G.; Bose, A.; John, V. T.; McPherson, G. L. Langmuir 2004, 20, 9931.

Langmuir 2010, 26(10), 6989–6994

Hollamby et al.

Article

Figure 6. (i) UV-vis spectra of the Ag-NPs before and after purification, showing characteristic silver NP peak at ∼405 nm. (ii) Phase separation with a dark upper phase containing Ag-NPs in equilibrium with a larger coexisting phase having a much lower AgNP concentration and approximate volumes of the phases.

and in the separated upper phase is approximately a factor of 5, suggesting that the process is at least as efficient with Ag-NPs as with gold. Figure 5. SANS and SAXS profiles for samples D and E (lower and upper phases of the separated samples) at 25 °C. The line in the upper phase SANS profile represents a form-factor fit to the data using a Shultz polydisperse spheres model, and the lines in the SAXS profiles represent fits generated by GIFT analysis.

describing a Schultz distribution of polydisperse spheres (see Supporting Information). Fitted parameters were SF = 5.4  10-7, Rav = 20 ( 4 A˚, and polydisperisty σ/R = 0.2. The extremely low scale factor value (see Supporting Information) indicates the concentration of droplets is low (if the contrast step is assumed to be from pure H-octane to pure D2O, volume fraction = 10-4). The conclusion is that the SANS is indicative of only a very small number density of small water droplets, in agreement with the NMR chemical analysis. On the other hand, the high SAXS intensity is consistent with UV-vis spectra (Figures 1 and 3), showing the obvious presence of Au-NPs. The pair distance distribution function shows an envelope of particle sizes (in agreement with TEM) with an average particle radius (assuming spherical particles) of ∼45 A˚, which is similar to the value obtained by GIFT analysis of sample C. Combining these points with the relatively high intensity SANS seen for sample D, and results from UV and NMR, the obvious conclusion is that Au-NPs have partitioned strongly into the upper phase and been largely purified from the CTAB and background microemulsion droplets. Silver Nanoparticles. This separation process was repeated with larger scale volumes and for different nanoparticles, Ag-NPs. After formation, the color change indicated the presence of silver nanoparticles, as confirmed by UV-vis spectroscopy (Figure 6i): the peak around 400 nm is indicative of surface plasmon resonance bands of Ag0-NPs.38,39 The water-induced separation technique was repeated as for the Au-NPs, giving similar results (Figure 6ii). The difference in absorption maxima (and therefore concentration) between Ag-NPs in the microemulsion (38) Ji, M.; Chen, X.; Wai, C. M.; Fulton, J. L. J. Am. Chem. Soc. 1999, 121, 2631. (39) Petit, C.; Lixon, P.; Pileni, M.-P. J. Phys. Chem. 1993, 97, 12974.

Langmuir 2010, 26(10), 6989–6994

Conclusions A new approach to separation, purification, and recovery of high value gold and silver nanoparticles is reported. The NPs are synthesized in a supporting CTAB-stabilized w/o microemulsion. Phase separation of these microemulsions, induced by addition of excess water drives a strong preferential partitioning of the NPs into an upper coexisting phase, accompanied by a redistribution of the excess CTAB surfactant into the lower aqueous phase. Purified NP samples can then by dried down by solvent removal and redispersed in other solvents such as pure octane. Structural studies, by SANS and SAXS, alongside UV-vis and 1H NMR spectroscopic analyses, indicate that the upper phase contains almost exclusively nanoparticles, whereas the lower aqueous phase is dominated by oil-swollen normal curvature micelles. It should be pointed out that although the results reported here are broadly similar to those recently published by Abecassis et al.,22 the systems in this current study employ a common surfactant (CTAB) and require no fine temperature control. The benefits of using water to achieve NP separation are apparent: as the greenest possible solvent, it represents a clean method for particle recovery. In addition, it should be noted that, if required, surfactant can be recovered from the water and reused after purification.24 This new method also has notable advantages over other existing recovery techniques,11-17 as it uses no additional organic solvents, has a low-energy demand, and requires no specialist surfactants. While use of self-assembly systems as reaction template media is quite common, the significant advance here is manipulation of the phase behavior of the supporting “colloidal solvent” to result in recovery and purification of valuable nanoparticles, in this case simply by adding excess water. Such flexibility to induce separation phase transitions is not so readily available if a primitive molecular solvent is used as supporting medium (e.g., water or oil). Therefore, there appear to be notable benefits of using colloidal self-assembly fluids as reaction and extraction media for nanoparticles. It is envisaged that the approach described here has potential for widespread applications in the further exploitation of nanoparticle technology. DOI: 10.1021/la904225k

6993

Article

Acknowledgment. M.J.H. thanks Kodak and School of Chemistry at the University of Bristol for the provision of a PhD scholarship. The ISIS-STFC Neutron Scattering Facility (formerly CCLRC) and the ILL are thanked for the provision of beam-time and grants towards consumables and travel. The EPSRC is thanked for provision of funding under Grants EP/ C523105/1 and EP/F020686.

6994 DOI: 10.1021/la904225k

Hollamby et al.

Supporting Information Available: Experimental detail of characterization techniques; supporting discussions of compositional analysis (1H NMR) and small-angle scattering (SAS) measurements; supplementary figures (stopped flow analysis, PDDFs); fit parameters table for SANS data. This material is available free of charge via the Internet at http:// pubs.acs.org.

Langmuir 2010, 26(10), 6989–6994