Facile synthesis and characterization of PTA stabilized hydrophilic Au55 nanoparticles via a DEN–MPC method

Polyhedron 102 (2015) 469–478

Contents lists available at ScienceDirect

Polyhedron journal homepage: www.elsevier.com/locate/poly

Facile synthesis and characterization of PTA stabilized hydrophilic Au55 nanoparticles via a DEN–MPC method Jezreel Cloete 1, Selwyn F. Mapolie 2, Rehana Malgas-Enus ⇑ Department of Chemistry and Polymer Science, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa

a r t i c l e

i n f o

Article history: Received 21 March 2015 Accepted 11 September 2015 Available online 30 September 2015 Keywords: Water-soluble gold nanoparticles Dendrimer micelle Nanoparticle extraction PTA ligand Au55 nanoparticles

a b s t r a c t The synthesis of hydrophilic gold nanoparticles using a combined dendrimer templated (DEN) and monolayer protected cluster (MPC) method is reported. Initially, using the dendrimer as a template leads to well-ordered nanoparticle crystal growth (in the case of geometrical shell structure configurations), which can be controlled by using the desired dendrimer:gold ratio. The nanoparticles are then further stabilized by extraction with a hydrophilic ligand to form MPC’s. Using this combined DEN–MPC method, we managed to synthesize Au55 atom cluster nanoparticles with an average particle size of 1.5 ± 0.9 nm. In our attempts to synthesize nanoparticles of different sizes, it was clear that the dendrimer micelle to metal ratio not only plays a role in the formation of stable nanoparticles, but also in the specific atom cluster size. Complete geometrical shell configuration numbers, such as Au55 and Au13 (average size of 3.3 ± 1.0 nm), form stable monodisperse nanoparticles, whereas the attempted Au31 atom clusters were unstable, resulting in polydispersed nanoparticles with an average size of 9.4 ± 4.7 nm. The mechanism by which these nanoparticles form is discussed. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Various methods exist for the synthesis of nanoparticles, which include amongst others colloidal synthesis, impregnation, coprecipitation, and electrochemical methods [1–3]. Nanoparticles, however, have a propensity to self-assemble or agglomerate into larger structures [4]. As a result, stabilizers such as polymers, oxide supports, dendrimers and long carbon chain ligands have been utilized to prevent agglomeration [3]. The most versatile and simplistic manner in which to produce nanoparticles is considered to be colloidal synthesis [4]. This method typically involves the dissolution of an appropriate metal precursor, together with a surfactant. The metal precursor is then reduced by an appropriate reducing agent to generate the desired nanoparticles. However, using this protocol, nanoparticles with wide size distributions are often obtained, which has prompted further research with the aim of optimising surfactant concentrations, as well as the type and properties of the surfactant, in order to narrow these size distributions [5,6]. One such area of research which has been successful in producing highly monodisperse ⇑ Corresponding author. Tel.: +27 218082801. E-mail addresses: [email protected] (J. Cloete), (S.F. Mapolie), [email protected] (R. Malgas-Enus). 1 Tel.: +27 218089539. 2 Tel.: +27 218082722. http://dx.doi.org/10.1016/j.poly.2015.09.043 0277-5387/Ó 2015 Elsevier Ltd. All rights reserved.

[email protected]

nanoparticles is that of the utilization of dendrimers as templates for the synthesis thereof, whilst also subsequently acting as a cluster stabilizer. Dendrimers are obtained by controlled addition of repeating units to form regularly branched macromolecules emanating from a central core and terminating in a distinct number of functional end groups. As a result, these molecules possess highly distinct sizes, shapes and chemical compositions. Each successive addition of a repeating unit (or generation) increases the number of functional end groups and subsequent metal co-ordination sites. Most commonly, poly(amidoamine) (PAMAM) and poly(propyleneimine) (PPI) dendrimers have been shown to be effective as templates for preparing nearly monodispersed metal nanoparticles contained within the dendrimer interior. This occurs when transition metal ions disperse into the dendrimer interior from solution and coordinate to functionalities within the dendrimer. Depending on the reaction conditions and the nature of the peripheral functional end groups, coordination can be via the internal tertiary amine groups or a combination of tertiary amine and peripheral functional end groups. Coordination can also occur as a salt. A mixture of the three aforementioned coordination modes can also occur [7]. Reduction subsequently results in the formation of dendrimer encapsulated nanoparticles (DENs) which may form internal, external or mixed composite structures (Scheme 1) [1]. Because the dendrimer template is uniform in structure, each one contains the same number of interior, peripheral or

470

J. Cloete et al. / Polyhedron 102 (2015) 469–478

H3C

H3C

O O

NH

NH H3C

NH

N

NH O NH

O

NH O

H3C

O

NH

NH

NH O

NH

O

NH

O

N

O

CH3

NH O

H3C

O

H3C H3C

H3C

NH

CH3

N

NH

NH

NH O

NH

O

NH O

H3C

O

O N NH

NH O

NH

O

NH

H3C H3C

CH3

O

CH3

CH3

H3C

CH3

NH

N

N

NH

O

N

N

N

N

O

CH3

O

CH3

O

N

N

N NH

NH

N

O

N

NH

O

O

NH O

N

NH

N

CH3

H3C

H3C

CH3 O

N N

O NH

N

N

N N

N

H3C

CH3

O

N

N

NH O NH

NH NH

NH

NH

N

NH

O

CH3

NH O N

O

N

NH O

H3C

O

N

O NH

O

N

NH

N

CH3 O

N N

O

NH

N

O

H3C

CH3

O

N

N

NH O NH

NH NH

N

N

N

O NH

O

N N

O

H3C

H3C

CH3

NH O

CH3

CH3

CH3 H3C

H3C

CH3

CH3

CH3 H3C

H3C

H3C H3C

H3C

H3C

Scheme 1. Internal, external and mixed dendrimer nanocomposites (adapted from Ref. [7]).

combination of interior and peripheral coordination sites. As a result, an equal number of metal ions should coordinate within each dendrimer (as shown by binding studies) and hence the encapsulated nanoparticles that form after reduction should all contain approximately the same number of atoms [8]. This is particularly advantageous when targeting specific atom clusters, such as Aux for example; a dendrimer with a sufficient number of functional end groups resulting in x interior coordination sites would result in nearly monodispersed Aux atom clusters forming. Nanoparticle encapsulation is sustained predominantly by steric effects, but chemical interactions between the dendrimer and metal particle surface may also play a role [7,9]. In most papers it is reported that DENs are prepared in water, since the most commonly used dendrimers, PPI and PAMAM, are both water soluble. There are, however, a few reported cases of DENs being prepared in organic solvents [11]. In these cases the terminal groups of the dendrimer were modified with a hydrocarbon chain of a certain length (thereby forming a dendrimer micelle with a hydrophobic exterior and hydrophilic interior) which therefore renders it soluble in organic solvents. The synthesis of DENs in an organic medium therefore leads to metal encapsulation being driven by solubility (with metal salts diffusing to the hydrophilic interior of the dendrimer micelle) instead of interior amine coordination. Thus smaller generation dendrimers could be used to synthesize larger nanoparticle clusters since the cluster size would not be dependent on dendrimer generation. Here we report on the synthesis, extraction and characterization of Au nanoparticles from a hydrophobic dendrimer micelle template. Dendrimer micelle encapsulated nanoparticles (DENs) consisting of Au55, Au31 and Au13 cluster sizes were synthesised in an organic medium, after which the nanoparticles were extracted from the dendrimer micelle template into water using 1,3,5-triaza-7-phosphaadamantane (PTA) as a stabilizing ligand. This resulted in monolayer protected cluster (MPC) formation. Using this DEN–MPC method, we managed to synthesize hydrophilic Au55 nanoparticles with an average particle size of 1.5 ± 0.9 nm, which is much simpler and more economical than previously reported methods for Au55 cluster formation [35]. We also discuss the possible mechanisms by which these nanoparticles form.

Symochem and used without any further purification. PTA was synthesised according to the method of Daigle et al. [12]. 2.2. Characterization IR spectra were obtained using a Nicolet Avatar 330 FT-IR spectrometer. 1H and 13C NMR spectra were obtained on a Varian Innova 400 MHz NMR spectrometer in deuterated chloroform at a frequency of 299.74 MHz. UV–Vis absorbance spectra were obtained using a GBC 920 UV/VIS spectrometer and quartz cuvettes having a path length of 1.00 cm. The spectrum of either chloroform or water was used as the background, depending on the solution being analysed. TEM images were obtained using a FEI Tecnai G2 Field Emission Gun (FEG) TEM operating at 200 kV and having a 2.5 Å point-to-point resolution, 1.02 Å line resolution and 1.4 Å information limit. Samples were prepared by dropwise addition of the sample onto a carbon coated copper grid. Au concentrations were measured using a Spectro Acros ICP-OES spectrometer. 2.3. Preparation of G3 PPI dendrimer micelle (1) The dendrimer micelle was synthesised according the method of Niu and Crooks [10]. Characterization of the prepared dendrimer micelle was carried out using IR, 1H and 13C NMR spectra, with the results comparing well to those obtained in the literature (see Supplementary Information). 2.4. Preparation of Au DENs For the preparation of Au55, the G3 micelle (1) (0.01 g, 0.0018 mmol) was dissolved in chloroform (20 ml). HAuCl4 (0.032 g, 0.01 mmol) was then added to the micelle solution and stirred for 10 min. NaBH4 (0.01 g, 0.26 mmol) was dissolved in MeOH (0.2 ml) and added to the Au/micelle solution. The reduction reaction was carried out for 30 min. For the preparation of Au31 and Au13, the amount of HAuCl4 as well as the amount of NaBH4 and volume of solvent were kept the same as for Au55 synthesis, with the amount of micelle being used being varied. Thus for Au31 the amount of micelle used was 0.018 g (0.0033 mmol) and for Au13 it was 0.050 g (0.0091 mmol).

2. Experimental 2.5. Extraction of Au nanoparticles into an aqueous medium 2.1. Materials Reagent grade solvents, triethylamine and palmitoyl chloride were obtained from Merck Chemicals and used as received. Na2CO3, NaBH4 and HAuCl4 were obtained from Sigma–Aldrich and used as received. G3 DAB-PPI dendrimer was obtained from

1,3,5-Triaza-7-phosphaadamantane (PTA) (0.0065 g, 0.04 mmol) was dissolved in distilled water (20 ml) and added to the Au DENs solution. Extraction was carried out by stirring the mixture for up to 24 h with UV–Vis samples taken at 2, 4, 8, 16 and 24 h of extraction time.

471

J. Cloete et al. / Polyhedron 102 (2015) 469–478

3. Results and discussion For the synthesis of Au DENs, a third generation diaminobutane poly(propyleneimine) (DAB-PPI) dendrimer (which had been modified at the periphery with palmitoyl chloride in order to render it soluble in organic solvents), was used as a template. Previous reports of DENs prepared in an organic medium focused on fourth generation PAMAM dendrimers modified at the periphery with a C12 hydrocarbon chain [14]. In these PAMAM based dendrimer micelles, the metal ion strongly complexes with internal tertiary amine groups. In contrast, for a DAB-PPI dendrimer micelle, encapsulation is driven by the difference in metal ion solubility between the solvent and the dendrimer micelle interior and not through complexation [9,10]. After reduction, extraction of the weaker bound nanoparticles was thus thought to be potentially advantageous with regards to the concentration of gold being extracted into the aqueous phase. The extraction of these nanoparticles into water using 1,3,5triaza-7-phosphaadamantane (PTA) as a ligand also differed from previous reports in which other ligands such as glutathione and tiopronin, were employed [13,14]. Since the intended use for these nanoparticles is for application in catalysis, these ligands might hinder access of the substrate to the nanoparticle surface due to steric effects. Due to the compact structure of the PTA ligand compared to glutathione and tiopronin (Scheme 2), the extraction and stabilization of gold nanoparticles using PTA was evaluated. Gold-PTA coordination is well documented in the literature [15]. The extracted Au nanoparticle size, shape and concentration were monitored at intervals of 2, 4, 8, 16 and 24 h of extraction using UV–Vis spectroscopy, transmission electron microscopy (TEM) and inductively coupled plasma spectroscopy (ICP). The results indicate that the described synthesis and extraction process produces nearly monodisperse Au55 and Au13 monolayer protected clusters (MPCs), while in the case of Au31, nanoparticles exhibiting higher polydispersity are obtained. The Au55 nanoparticles were also the smallest (1.5 ± 0.9 nm) followed by Au13 (3.3 ± 1.0 nm), while those of Au31 are much larger (9.4 ± 4.7 nm) after 24 h of extraction. The Au DENs were prepared in chloroform by stoichiometric addition of HAuCl4 (0.1 mmol) to a vigorously stirred solution of (1). It has previously been demonstrated that DENs of a desired cluster size could be synthesised by utilising a dendrimer/AuCl4 ratio of, or close to, the desired size [14]. The concentration of micelle was therefore varied depending on whether Au55, Au31 or Au13 atomic cluster size was being aimed for (0.009, 0.017 or 0.045 lM respectively). Reduction with an excess of NaBH4 (0.26 mmol) resulted in a deep purple solution which is indicative of the presence of Au nanoparticles [14–16].

N

N

Upon addition of the Au salt to the micelle solution, the weak intensity absorption band of the micelle in solution (at approximately 290 nm) disappears, concomitant with the emergence of a strong peak in the region of 330 nm which is characteristic of the ligand-to-metal charge-transfer (LMCT) band of AuCl4 [17–19]. The observation of this band together with the absence of an absorption band at approximately 280 nm (which would indicate complexing between the dendrimer micelle and AuCl4 ) indicates the solvent driven encapsulation of the Au ions. After reduction, a strong plasmon resonance peak in the region of 540 nm materializes. This is characteristic of the presence of Au nanoparticles greater than two nanometres in diameter [20–22]. The disappearance of the AuCl4 LMCT band around 330 nm was indicative of complete reduction of the metal. Stirring the solution for 30 min proved sufficient for this to occur. Differences in the UV–Vis spectra of the reduction solutions of the aimed for Au55, Au31 and Au13 atomic cluster nanoparticles are an indication of the differences in the size, shape and stability of the initial nanoparticles formed within the dendrimer framework. After reduction of the Au salt, atoms are generated which then undergo self-nucleation [23,24]. The stability of atomic clusters formed after nucleation has been found to be dependent on electronic and geometric shell closing [25,26]. In terms of geometric stability, for Au13, twelve of the Au atoms form a geometrically closed icosahedral shell, with the thirteenth atom positioned at the centre, thus creating a stable structure. This represents the first geometric shell closing for both icosahedral and cuboctahedral structures [27]. Stable geometric cluster sizes then increase by the successive addition of further shells, which correspond to atom numbers 55, 147, 309 etc. (often referred to as ‘‘magic numbers”) [28,29]. This is shown in Scheme 3, where Au31 is included in order to demonstrate its incomplete shell geometry and therefore inferred instability. After nucleating, further particle growth occurs via seeding [29]. Depending on the reaction conditions, single crystal, twinned or multiply twinned seeds form and grow to form nanoparticles of various shapes. This is depicted in Scheme 4. Due to their more stable geometrical core atom shell configurations, the aimed for Au55 and Au13 atom clusters form seeds having distinct polyhedron and spherical shapes respectively. This is clearly depicted in Fig. 1. In contrast, for the aimed for Au31 atom cluster, the incomplete core shell hampers efficient seeding and results in anisotropic cluster seeds, which in turn form quite a number of irregularly shaped nanoparticles having modified Wulff constructions [30]. A hypothetical depiction of this occurring is given in Scheme 5. It has been reported that particle shape affects the wavelength at which UV–Vis bands appear [29]. The more irregular the shape

N

P N N N

P

P

N N N

P N

N

N

P N

N

NH2

HO

N

= 1,3,5-triaza-7-phosphaadamantane (PTA)

= H3C

NH O

O

or

HS

O NH

O

glutathione

OH

O SH

NH

COOH

tiopronin

Scheme 2. Possible differences in coordination of PTA compared to glutathione and tiopronin.

472

J. Cloete et al. / Polyhedron 102 (2015) 469–478

Scheme 3. Representation of the shell geometries of Au13, Au31 and Au55 clusters.

Scheme 5. Proposed formation of a nanoparticle having a hypothetical modified Wulff construction derived from an incomplete Au31 shell.

3.1. Extraction of Au nanoparticles from the dendrimer template

Scheme 4. Nucleation and particle growth via seeding, resulting in distinct polyhedron shaped particles (adapted from Ref. [24]).

of the particle, the higher the wavelength at which the plasmon resonance band occurs. There is thus a clear correlation between the UV–Vis spectra of the reduction solutions and the depicted particle shapes. The overwhelmingly spherical Au13 atom cluster nanoparticles result in the absorption band being found at the lowest wavelength, while the varying but distinct shapes of the Au55 atom cluster nanoparticles result in a slightly higher wavelength absorption band. The irregularly shaped Au31 atom cluster nanoparticles subsequently have this absorption band at the highest wavelengths. Furthermore, the difference in intensity of the absorption bands can be attributed to the sizes of the initially formed nanoparticles. The TEM results obtained show that the average DENs size for Au13 was found to be 3.9 ± 1.6 nm compared to 6.0 ± 1.8 nm for Au31 and 8.9 ± 3.1 nm for Au55 (Fig. 2). This corresponds to the lowest to highest absorption band intensity of Au13, Au31 and Au55 respectively.

After reduction, 1,3,5-triaza-7-phosphaadamantane (PTA) was dissolved in 20 ml distilled water and the resulting solution added to the reduction solution. The optimum ratio of Au to PTA ligand was only determined for the extraction of Au55 nanoparticles, but this ratio was also applied in the extraction of Au13 and Au31. The postulated mechanisms of extraction of nanoparticles from a hydrophilic dendrimer and a hydrophobic dendrimer micelle differ slightly. For extraction of a nanoparticle from the more common hydrophilic dendrimer, it is thought that the ligand penetrates the interior of the dendrimer and coordinates to the metal nanoparticle. This weakens the interactions between the dendrimer and the nanoparticle, and as more ligands bind the nanoparticle ultimately exits the dendrimer framework and is extracted into the organic phase [11,14]. For a hydrophobic dendrimer micelle, however, it is believed that the hydrophobic extremities are distorted at the organic/aqueous interface and in this way it exposes the hydrophilic interior containing the nanoparticles towards the aqueous phase. The hydrophilic ligands present in the aqueous phase can then easily bind to the nanoparticle surface and subsequently form a stable monolayer protected cluster (MPC) which is then transported into the aqueous layer [14].

Fig. 1. TEM images showing distinct polyhedron shaped DENs of Au55.

473

J. Cloete et al. / Polyhedron 102 (2015) 469–478

a

c

b

8.9 nm ± 3.1

6.0 nm ± 1.8

3.9 nm ± 1.6

Fig. 2. TEM images, histograms and particle size distributions of Au55, Au31 and Au13 DENs (a, b and c) respectively.

Since distortion at the interface has been proposed for a hydrophobic dendrimer micelle, it was therefore hypothesized that smaller amounts of ligand might be required to extract the nanoparticles as compared to the larger amounts of ligand needed to penetrate a hydrophilic dendrimer (as opposed to a hydrophobic dendrimer micelle). This hypothesis was tested by examining the extraction efficiency at various Au to ligand ratios. We started at the lowest possible ratios of Au to ligand in order to determine the least amount of ligand that would be necessary for optimal extraction. By starting at a ratio of 1:0.1 Au:PTA, and gradually increasing the amount of PTA to 1:1, a steady increase in the amount of Au extracted was observed. This however only continued up to a ratio of 1:0.4 Au:PTA, after which the amount of gold extracted decreased. Furthermore the use of higher Au:PTA ratios resulted in what appears to be the masking of the plasmon resonance peak in the UV–Vis spectrum by excess PTA [31]. The aforementioned trend is reflected in the UV–Vis spectra for samples extracted after 24 h and confirmed by ICP (Fig. 3).

Fig. 3. ICP showing optimum extraction of Au with varying amounts of PTA.

Zeta potential measurements of the extracted nanoparticles were performed using a Particulate Systems NanoPlus Zeta/Nano Particle Analyzer in order to ascertain the nature of the surface charge. It was found that with the optimum extraction results (as determined by ICP), the zeta potential was on average close to zero, with actual values fluctuating between +1 and 1. This would indicate that the particles are well dissolved within the aqueous layer with little to no charge difference between the fluid medium and nanoparticle surface. The slight charge differences which are observed are due to a minority of localized surface sites where Cl ions may have previously been attached after reduction. Nonetheless, the particles did not exhibit colloidal behavior which would have resulted in higher absolute values of the zeta potential. This in fact was observed where extraction was not optimum; higher zeta potential values approaching what would be characteristic of colloids were obtained in conjunction with larger average particle sizes. 3.2. Extraction and characterization of Au55 UV–Vis analysis in conjunction with ICP (Fig. 4) and TEM results (Fig. 5) shows an increase in the amount of gold extracted into the aqueous layer up to 16 h. After 24 h, the amount of gold extracted remains relatively unchanged. What is apparent from the TEM results is that there is a decrease in the nanoparticle size with increasing extraction time. The average particle size decreased from 10.5 ± 2.9 nm after 2 h of extraction to 1.5 ± 0.9 nm at the end of 24 h, which is close to the size of a single Au55 cluster [32,33]. As previously mentioned, it has been postulated that extraction occurs when the hydrophobic extremities of the micelle become distorted at the organic/ aqueous interface, exposing the hydrophilic interior where the Au nanoparticles have formed. The nanoparticles are now exposed and are able to come into contact with the ligand at the interface where the latter binds and in this manner renders the nanoparticle soluble in the aqueous phase. The proposed migration of Au nanoparticles is summarized in Scheme 6. Thus, applying the proposed mechanism, it can be reasoned that the initial larger particles which are formed, as observed by TEM, are as a result

474

J. Cloete et al. / Polyhedron 102 (2015) 469–478

Fig. 4. Relative Au concentrations in the aqueous layer after 2, 4, 8, 16 and 24 h of extraction for Au55, Au31 and Au13.

of Au being exposed by the micelle and being at the organic/ aqueous phase interface. Once exposed at the interface, seeded growth can more easily occur since the dendrimer micelle no longer offers a degree of protection from seeds coming into close proximity of each other. This appears to occur during the first 2 h of extraction, resulting in the observed nanoparticle size of 10.5 ± 2.9 nm after this time. Singly or multiply twinned seeds were observed to contain at least one twin defect. The strain energy caused by twin defects greatly increases as the seed grows in size. This in turn results in an increase of the free energy of the system. As a consequence, twinned seeds are only favoured thermodynamically at relatively small sizes. It therefore appears that after extraction, seed growth occurs to 10.5 ± 2.9 nm, after which they become unstable and revert to forming smaller seeds [29]. The regression to smaller seeds and stabilization thereof is most likely aided by what is referred to as the ‘‘divide-and-protect”

10.5 nm ± 2.9

8.4 nm ± 2.7

mechanism. Walter et al. proposed this mechanism for the capping of gold nanoclusters [34]. It proposes that a ligand and particular cluster of Au atoms will more likely form the nanoparticles which can obtain closed shell noble-gas configurations. It is thus possible that with time, the coordination of the ligands along and within the various twin defects can lead to more stable seeds, resulting in an increase of smaller, stabilized nanoparticles with time, as depicted in Scheme 7. The maximum concentration of Au in the aqueous phase using a Au:PTA ratio of 1:0.4 was found to be approximately 120 ppm after 24 h. This equates to approximately 12.0% of the Au initially reduced migrating into the aqueous layer. It therefore appears that the Au:PTA ratio in the aqueous layer resulting in a closed shell noble gas configuration and stable Au55 nanoparticles of size 1.5 ± 0.9 nm at the end of 24 h is approximately 1:6. Although the concentration of gold extracted is low, subsequent extraction from the same organic phase yields a further approximately 32 ppm of gold. The histogram and wide particle size distribution after 4 h of extraction, depicted in Fig. 5, is a clear indication of the intermediate sizes of particles formed before equilibrium on stabilization by PTA and formation of smaller nanoparticles is achieved. 3.3. Extraction and characterization of Au31 MPCs For the extraction of Au31, UV–Vis spectroscopy of the aqueous medium exhibits no plasmon resonance peak up to 4 h of extraction. From ICP it is evident that this is due to no (after 2 h) or very low concentrations (after 4 h) of Au being present in the aqueous layer (Fig. 4). It is only from after 8 h of extraction that nanoparticles are visible in TEM images and a plasmon resonance peak is detected by UV–Vis spectroscopy. The unstable nature of Au31 is clearly depicted in the TEM images after 8 h of extraction. Although UV–Vis spectroscopy shows a plasmon resonance peak, the TEM image shows distorted particles. This is the result of the energy of the electron beam affecting the unstable Au31

7.1 nm ± 4.1

4.2 nm ± 1.6

1.5 nm ± 0.9

Fig. 5. TEM images, histograms and particle size distributions of Au55 after 2, 4, 8, 16 and 24 h extraction (a, b, c, d and e respectively).

J. Cloete et al. / Polyhedron 102 (2015) 469–478

475

Scheme 6. Graphic representation of nanoparticle migration.

Scheme 7. Graphic representation of nanoparticle stabilization.

nanoparticles, causing them to distort. It is only after 16 h, where both concentrations of Au nanoparticles are high enough and it is possible for particles to be easily discerned, that it is possible to determine the average particle size and distribution. The instability of the Au31 nanoparticles is further reflected in the histogram and particle size distribution observed after 24 h of extraction; the particle size distribution widens from 16 to 24 h with an increase in the average particle size from 8.0 ± 2.4 nm to 9.4 ± 4.7 nm, thus clearly indicating that agglomeration has occurred. A possible reason for the above could be the ease, or lack thereof, with which migration occurs from the organic to the aqueous phase. Since Au31 is an unstable atom cluster, it takes some time for rearrangement to a more stable Wulff polyhedron cluster as well as regular twinned seeds to occur within the dendrimer micelle in the organic layer. This is clearly depicted in Fig. 6a and b where fused DENs having various atom stacking patterns are visible, thus indicating Wulff construction taking place. Once the more stable atom stacking arrangements are achieved (Fig. 6c), only then are ligands able to attach and render the nanoparticles soluble in the aqueous layer. Once migration

occurs, further rearrangement in order to form more stable nanoparticles continues. The result of this is seen by the distorted particles observed after 8 h of extraction. However as time progresses, more ligand stabilized nanoparticles are formed. The wide particle size distribution after 24 h as compared to after 16 h of extraction can be ascribed to the combining and continuous rearrangement of nanoparticles to form more stable atom cluster sizes in the organic and aqueous phases. After 16 h the nanoparticles which are stable enough to migrate from the organic phase are present in larger amounts. As time progresses, both newly migrated nanoparticles and those which have been further stabilized and agglomerated are present. 3.4. Extraction and characterization of Au13 MPCs The UV–Vis spectra at the various time intervals of extraction show only hints of a plasmon resonance peak for each time period. However ICP clearly indicates the presence of gold (Fig. 4), while TEM images show that the gold nanoparticles present should perhaps result in more pronounced plasmon resonance peaks being visible. This observation could be explained by considering the

476

J. Cloete et al. / Polyhedron 102 (2015) 469–478

Fig. 6. Rearrangement of Au31 DENs to form more stable nanoparticles.

Fig. 7. TEM image of Au13 DENs showing evidence of internal dendrimer nanocomposite formation.

physical appearance of the aqueous layer, together with what is further seen in the TEM images. Comparing the aqueous layers of Au55, Au31 and Au13, it was seen that both the aqueous layers of Au55 and Au31 are clear solutions (the light blue, almost colourless, Au55 solution is indicative of it containing extremely small nanoparticles, with the purple

Au31 solution indicative of the presence of larger nanoparticles) [20–22]. The aqueous layer of Au13 however clearly contained a suspension. In the TEM images it can be seen that most of the nanoparticles are surrounded by an organic layer. It therefore becomes clear that in addition to having PTA most likely coordinated to the nanoparticles, the latter are also surrounded by the micelle molecules in the aqueous layer. Because of this, and the lower concentration of Au extracted as compared to Au55 (Fig. 4), the plasmon resonance peak is not as strongly detected by UV–Vis spectroscopy. A possible explanation as to how the micelle could end up in the aqueous layer could be related to the concentration of micelle initially present in the organic phase. It must be remembered that the micelle concentrations were varied while the Au concentrations in the organic phase were kept constant in order achieve the desired Au atom cluster sizes. Thus Au13 would require the highest relative concentration of micelle in order to achieve the 1:13 micelle:Au ratio (as opposed to a 1:55 and 1:31 micelle:Au ratios for Au55 and Au31 respectively). Because of the higher concentration of micelle present, localized concentrations of nanoparticles are formed. TEM images of the DENs of Au13 provide evidence for this, with pockets of closely stacked nanoparticles clearly visible in Fig. 7. Extraction still occurs via the mechanism of micelle distortion at the interface. However in this case, the nanoparticle aggregates

Scheme 8. Graphic representation of excess micelle migrating into the aqueous layer along with nanoparticles.

J. Cloete et al. / Polyhedron 102 (2015) 469–478

are in contact with a large surface area of the micelle, while PTA coordinates to it simultaneously. This renders it water soluble and then draws the nanoparticle aggregate into the aqueous phase while still having the micelle attached. As a result, the dendrimer micelle migrates along with the nanoparticles into the aqueous layer. The mechanism is depicted in Scheme 8. The nanoparticles in the aqueous phase are now at an intermediate stage of no longer being completely encapsulated by the dendrimer micelle, after having been distorted, and being stabilized by the PTA ligand. Further seed growth occurs to a certain extent (up to 4.7 ± 1.4 nm from 4.0 ± 1.6 nm) before the previously mentioned ‘‘divide-andprotect” mechanism of ligand stabilization occurs, resulting in a final nanoparticle size of 3.3 ± 1.0 nm after 24 h. The average size of the DENs and initial extracted nanoparticle size of 3.9 ± 1.6 nm and 4.0 ± 1.6 nm are strikingly similar and it is therefore perhaps a further indication of the internal dendrimer nanocomposite having been formed and subsequent stabilization of the nanoparticles by the dendrimer micelle. Furthermore, the average size of the DENs and the nanoparticles extracted into the aqueous phase hint at the fact that the seeds are single crystals. A particle size of 4.0 nm corresponds to a cluster composed of in the region of 492 atoms and 7 complete shells starting from Au13. Seed growth in the aqueous layer adds only one more shell since 4.7 nm is in the region of 642 atoms and 8 complete shells starting from Au13. The final size of 3.3 nm obtained in the aqueous layer corresponds to a reduction to approximately 362 atoms and 5 complete shells [4]. The most stable MPC synthesised was that of Au55 atom clusters. These were found to be stable enough to exist at close to the size of a single 55 atom cluster. With nanoparticles formed from initial Au31 clusters, it was clear that rearrangement occurred in order to achieve a more stable shell configuration. The closer the starting geometrical configuration is to these ‘‘magic numbers”, the more stable the nanoparticles will be, albeit of larger size. The stability of the Au55 nanoparticles was further emphasised due to its observed propensity to reduce in size as extraction time increased. This is postulated to be due to the ligand/shell configuration satisfying the criteria of a closed shell noble gas configuration. This most likely leads to a ‘‘divide-and-protect” scenario, resulting in decreasing nanoparticle size as more ligand coordination takes place. It is thus evident, by observing the decrease in nanoparticle size with extraction time of Au55 clusters, that it might be possible to tailor particle size by stopping or continuing with extraction for certain lengths of time. Initially proposed synthetic routes to producing phosphine stabilized Au55 clusters are quite laborious, requiring anaerobic conditions and expensive diborane gas as a reducing agent [35]. Although subsequently developed methods are simpler, these still require inert conditions and several washing steps with various solvents [36]. The method described here in obtaining Au55 nanoparticles is therefore much simpler and more economical.

4. Conclusions In this work, we have shown that using a dendrimer as a template for nanoparticle synthesis initially aids in the formation of the desired nanoparticle cluster size (by controlling the dendrimer:metal ratio), however these initial nanoparticle seeds rapidly undergo growth within the dendrimer template leading to larger than expected nanoparticle sizes. By extracting these nanoparticles from the dendritic template using a hydrophilic ligand, we created stable monolayer protected clusters (MPCs) which gave the initially aimed for nanoparticle clusters in the case of Au55 after an extraction period of 24 h. In the case of Au13 a similar phenomenon was observed, however, due to the presence of

477

the dendrimer micelle in the aqueous phase, the nanoparticles were stabilized by the dendrimer as well as the PTA ligand which minimised agglomeration and the limited ‘‘divide and protect” mechanism, since the nanoparticle sizes remained stable at 4 nm. It was found that the initial dendrimer:metal ratio does play a role in the nanoparticle stability and subsequent extraction since the formation of nanoparticle clusters which do not result in a closed shell formation (magic numbers) leads to Wulff construction, as seen in the case of the Au31 nanoparticle synthesis. Wulff constructed nanoparticles form as a result of the anisotropic nature of the initially formed Au31 seeds, which leads to a continual rearranging in order to achieve a stable closed shell configuration. This continuous rearrangement and the consequent irregular-shaped nanoparticle, hampers ligand coordination to the nanoparticle surface, resulting in low extraction efficiencies. By combining the dendrimer templated method as well as the MPC method for nanoparticle synthesis, we have demonstrated a simpler and more economical procedure for Au55 nanoparticle synthesis. Acknowledgements We acknowledge the National Research Foundation for financial support and the University of the Western Cape Electron Microscope Unit for providing TEM imaging. We also thank Micromerics and Poretech, in particular Mr Charles Noakes, for the use of their Nanoplus instrument. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.poly.2015.09.043. References [1] X. Mu, D.G. Evans, Y. Kou, Catal. Lett. 97 (2004) 151. [2] P. Claus, A. Brückner, C. Möhr, H. Hofmeister, J. Am. Chem. Soc. 122 (2000) 11430. [3] K.-T. Wu, Y.-D. Yao, C.-R.C. Wang, P.F. Chen, E.-T. Yeh, J. Appl. Phys. 85 (1999) 5959. [4] Y. Xia, T.D. Nguyen, M. Yang, B. Lee, A. Santos, P. Podsiadlo, Z. Tang, S.C. Glotzer, N.A. Kotov, Nat. Nanotechnol. 6 (2011) 580. [5] D. Kumar, B.J. Meenan, I. Mutreja, R. D’Sa, D. Dixon, Int. J. Nanosci. 11 (2012) 1250023. [6] R.K. DeLong, C.M. Reynolds, Y. Malcolm, A. Schaeffer, T. Severs, A.J. Wanekaya, Nanotechnol. Sci. Appl. 3 (2010) 53. [7] K. Torigoe, A. Suzuki, K. Esumi, J. Colloid Interface Sci. 241 (2001) 346. [8] M.Q. Zhao, L. Sun, R.M. Crooks, J. Am. Chem. Soc. 120 (19) (1998) 4877. [9] L. Balogh, R. Valluzzi, K.S. Laverdure, S.P. Gido, G.L. Hagnauer, D.A. Tomalia, J. Nanopart. Res. 1 (1999) 353. [10] Y. Niu, R.M. Crooks, Chem. Mater. 15 (2003) 3463. [11] Y. Niu, R.M. Crooks, C.R. Chim. 6 (2003) 1049. [12] D.J. Daigle, T.J. Decuir, J.B. Robertson, D.J. Darensbourg, in: M.Y. Darensbourg (Ed.), Inorg. Synth., 32, John Wiley and Sons Inc., 1998, pp. 41–42. ch. 1. [13] R.M. Crooks, M. Zhoa, L. Sun, V. Chechik, L.K. Yeung, Acc. Chem. Res. 34 (2001) 181. [14] G.R. Newkome, C.D. Schreiner, Polymer 49 (2008) 1. [15] A.D. Philips, L. Gonsalvi, A. Romerosa, F. Vizza, M. Peruzzini, Coord. Chem. Rev. 248 (2004) 955. [16] M.R. Knecht, J.C. Garcia-Martinez, R.M. Crooks, Langmuir 21 (2005) 11981. [17] W. Jiang, D.B. Hibbert, G. Moran, J. Herrmann, Å.K. Jämting, V.A. Coleman, RSC Adv. 3 (2013) 7367. [18] M.V. Sujitha, S. Kannan, Spectrochim, Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 102 (2013) 15. [19] T.Y. Suman, S.R.R. Rajasree, R. Ramkumar, C. Rajthilak, P. Perumal, Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 118 (2014) 11. [20] L. Zhou, D.H. Russell, M. Zhao, R.M. Crooks, Macromolecules 34 (2001) 3567. [21] M.M. Alvarez, J.T. Khoury, T.G. Schaaff, M.N. Shafigullin, I. Vezmar, R.L. Whetten, J. Phys. Chem. B 101 (1997) 3706. [22] J.C. Garcia-Martinez, R.M. Crooks, J. Am. Chem. Soc. 126 (2004) 16170. [23] Y.G. Kim, S.K. Oh, R.M. Crooks, Chem. Mater. 16 (2004) 167. [24] V.K. LaMer, R.H. Dinegar, J. Am. Chem. Soc. 72 (1950) 4847. [25] Y. Xia, Y. Xiong, B. Lim, S.E. Skrabalak, Angew. Chem., Int. Ed. 48 (2009) 60. [26] S.N. Khanna, P. Jena, Phys. Rev. Lett. 69 (1992) 1664. [27] H. Qian, Y. Zhu, R. Jin, PNAS 109 (2012) 696.

478

J. Cloete et al. / Polyhedron 102 (2015) 469–478

[28] M. Gruber, G. Heimel, L. Romaner, J.-L. Bredas, E. Zojer, Phys. Rev. B 77 (2008) 165411. [29] A.L. Mackay, Acta Crystallogr. 15 (1962) 916. [30] G. Shafai, S. Hong, M. Bertino, T.S. Rahman, J. Phys. Chem. C 113 (2009) 12072. [31] C.L. Cleveland, U. Landman, M.N. Shafigullin, P.W. Stephens, R.L. Whetten, Z. Phys. D 40 (1997) 503. [32] G. Schmid, Chem. Soc. Rev. 37 (2008) 1909.

[33] G. Schmid, E. Emmrich, J.-P. Majoral, A.-M. Caminade, Small 1 (2005) 73. [34] M. Walter, J. Akola, O. Lopez-Acevedo, P.D. Jadzinsky, G. Calero, C.J. Ackerson, R.L. Whetten, H. Grönbeck, H. Häkkinen, PNAS 105 (2008) 9157. [35] G. Schmid, R. Pfeil, R. Boese, F. Bandermann, S. Meyer, G.H.M. Calis, J.W.A. van der Velden, Chem. Ber. 114 (1981) 3634. [36] W.W. Weare, S.M. Reed, G.M. Warner, J.E. Hutchison, J. Am. Chem. Soc. 122 (2000) 12890.