Sunjie
Ye
ab,
Felix
Benz
c,
May C.
Wheeler
a,
Joseph
Oram
a,
Jeremy J.
Baumberg
c,
Oscar
Cespedes
a,
Hugo K.
Christenson
a,
Patricia Louise
Coletta
b,
Lars J. C.
Jeuken
d,
Alexander. F.
Markham
b,
Kevin
Critchley
a and
Stephen D.
Evans
*a
aSchool of Physics and Astronomy, University of Leeds, Leeds, LS2 9JT, UK. E-mail: s.d.evans@leeds.ac.uk
bLeeds Institute for Biomedical and Clinical Sciences, University of Leeds, Leeds, LS9 7TF, UK
cNanoPhotonics Centre, Cavendish Laboratory, University of Cambridge Cambridge, CB3 0HE, UK
dSchool of Biomedical Sciences, University of Leeds, Leeds, LS2 9JT, UK
First published on 21st June 2016
Hollow metallic nanostructures have shown potential in various applications including catalysis, drug delivery and phototherapy, owing to their large surface areas, reduced net density, and unique optical properties. In this study, novel hollow gold nanoflowers (HAuNFs) consisting of an open hollow channel in the center and multiple branches/tips on the outer surface are fabricated for the first time, via a facile one-step synthesis using an auto-degradable nanofiber as a bifunctional template. The one-dimensional (1D) nanofiber acts as both a threading template as well as a promoter of the anisotropic growth of the gold crystal, the combination of which leads to the formation of HAuNFs with a hollow channel and nanospikes. The synergy of favorable structural/surface features, including sharp edges, open cavity and high-index facets, provides our HAuNFs with excellent catalytic performance (activity and cycling stability) coupled with large single-particle SERS activity (including ∼30 times of activity in ethanol electro-oxidation and ∼40 times of single-particle SERS intensity, benchmarked against similar-sized solid gold nanospheres with smooth surfaces, as well as retaining 86.7% of the initial catalytic activity after 500 cycles in ethanol electro-oxidation). This innovative synthesis gives a nanostructure of the geometry distinct from the template and is extendable to fabricating other systems for example, hollow-channel silver nanoflowers (HAgNFs). It thus provides an insight into the design of hollow nanostructures via template methods, and offers a versatile synthetic strategy for diverse metal nanomaterials suited for a broad range of applications.
Gold “nanoflowers” represent specialized gold nanostructures with large numbers of branches that give the overall appearance of a flower. Owing to the presence of sharp tips acting as “hot spots” for localized near-field enhancements, as well as their high surface area-to-volume ratios, nanoflowers could be more effective in SERS,16–18 photothermal conversion19 and drug loading20 than nanoparticles with smooth surfaces. The synthesis of branched and flower-shaped nanometals is a recent development,17 and it is unprecedented and even more challenging to fabricate gold nanoflowers with an open hollow structure, that can integrate the advantages of the internal open cavity and external multiple tips/branches.
Despite a series of reported approaches, the fabrication of metal hollow nanostructures via a rational, facile and cost-effective method remains a challenge. To date, template-directed approaches have been demonstrated to be most efficient for the preparation of hollow structures. These typically include the formation of nanocrystals on the surface/in the cavity of templates and the removal of the templates afterwards.21,22 As an extended template method, the galvanic replacement reaction adopts the pre-synthesized metal nanostructures as sacrificial templates to generate another metal nanostructure with complementary morphology.23 However, these methods tend to have some drawbacks including low efficiency, high cost, complicated preparation and/or removal processes for the templates, which limit their practical applications.24,25
A practical templating method for nanostructure synthesis should ideally meet the following key criteria: (1) the convenient availability of well-defined templates in large quantity at low cost, (2) the precise control of the nanocrystal growth on the surface/in the cavity of the templates, and (3) a reliable and simple process to selectively remove the templates, without compromising the structural integrity of the final nanostructures.26 Additionally, templates are usually elegantly designed to exploit the differences in spatially distributed nucleation for constructing complex plasmonic nanoarchitecture.27
Here we describe the first preparation of novel hollow Au nanoflowers (HAuNFs) with an open cavity in the center and multiple branches on the outer surface. A facile one-step approach has been developed employing a bifunctional template of MO–FeCl3 nanofiber. The 1D template not only contributes to the formation of an open hollow channel by a threading template effect, but also promotes the generation of spiky features by inducing the anisotropic growth of Au crystal around the template. The catalytically induced auto-degradation of the template enables its easy removal. This method for synthesizing hollow-channel nanostructure is not restricted to Au, but can be readily extended to other metals, e.g. Ag.
Furthermore, owing to the synergy of favorable structural/surface features including sharp edges, open cavity and high-index facets, our HAuNFs have shown excellent catalytic performances (activity and cycling stability) and large single-particle SERS activity (e.g. ∼30 times of activity in ethanol electro-oxidation and ∼40 times of single-particle SERS intensity, benchmarked against similar-sized solid gold nanospheres with smooth surface).
Different stages of the growth of the HAuNFs (synthesized with 150 mM AA) were studied by electron microscopy (EM). The MO–FeCl3 nanofiber template has been previously employed for the preparation of conducting polymer nanotubes (e.g. polypyrrole).28,29 The nanofiber dispersion (Fig. 1B) was prepared by simply dissolving FeCl3·6H2O in 5 mM methyl orange solution (for details see ESI†). The nanofibers have a rod shape with a width of (74 ± 22) nm (ESI: Fig. S1†).
After the nanofiber templates were added to the HAuCl4 solution, the strong interactions (the coordination between Au(III) and azobenzene in MO molecules,30 as well as the electrostatic interaction between AuCl4− ions and Fe3+) lead the AuCl4− to be absorbed onto the MO–FeCl3 nanofiber (see ESI†). Following the introduction of the reducing agent AA, Au nucleates on the surface of the template to generate the primary Au nanocrystal (Fig. 1C).31 The template hinders isotropic growth of the nanocrystal, thus inducing anisotropic growth of the primary nanocrystals, from which the branches protrude and extend in length, forming the observed flower-shape (Fig. 1D).16 The as-formed AuNFs in turn act as a catalyst towards the degradation of MO in the presence of FeCl3, excess reducing agent AA and O2 in the solution, resulting in the disassembly of the MO–FeCl3 nanofiber template, leaving an open hollow channel in the center of Au NF (The mechanism of template degradation is discussed in detail in ESI.†). The scanning electron microscopy (SEM) images (Fig. 1E and inset showing the connected channel from different orientations) of two partially formed AuNFs, which share one connected hollow channel, demonstrate the growth of the Au crystal around the elongated threading template. The SEM image of the products before purification by centrifugation (ESI: Fig. S2†) shows no trace of nanofiber, which corroborates the auto-destruction of the template. These experimental observations agree well with the schematic illustration shown in Fig. 1A.
The formation of the HAuNFs and the auto-degradation of the template were also indicated by a change in the optical properties. As can be seen from Fig. 1F, the solution of free MO molecules and the MO–FeCl3 dispersion displayed a peak at 465 nm and 505 nm respectively. Shortly after the addition of AA to the mixture of MO–FeCl3 and HAuCl4, the solution turned from an orange color to a ruby color, and a new absorption peak was observed around 530 nm (Fig. 1G), which overlapped with the peak of MO–FeCl3 template. These observations indicate the generation of the primary Au nanocrystals, corresponding to Stage I in the HAuNF formation. As the growth continued, the solution color gradually changed from ruby to purple, and a plasmon band centered at a longer wavelength (∼650 nm) appeared (Fig. 1H), intensified and red-shifted, suggesting the anisotropic growth of the Au nanocrystals to form the branches and correlating well with Stage II. Afterwards, the absorption spectrum displayed a shoulder around 470 nm, which can be ascribed to free MO molecules, due to the disassembly of the MO–FeCl3 template. With the reaction proceeding, the solution color turned from purple to blue and the absorption around 470 nm weakened until it finally disappeared, suggesting the auto-degradation of the template corresponding to Stage III. Noticeably, the long wavelength peak (600–700 nm) assigned to the HAuNFs showed no further red-shift after the peak of MO appeared, revealing that the template destruction occurred after the completed formation of the nanoflower structure. The absorption peak of MO/MO–FeCl3 was absent in the spectrum of the supernatant after centrifugation of the reaction solution (ESI: Fig. S3A†), providing more evidence for the template degradation. To verify that the as-formed HAuNFs catalyzed the degradation of MO–FeCl3 template in the presence of the excess AA, we added purified HAuNFs to the mixture of MO–FeCl3 template and AA solution, the absorption assigned to MO/MO–FeCl3 decreased and then disappeared within 12 min (ESI: Fig. S3B†). In contrast, the absorption of MO–FeCl3 template dispersions was almost unaltered after 3 h, in the absence of either AA or HAuNFs.
In order to investigate the dual roles of the template in the formation of both the hollow channel and the multiple branches on the outer surface, a control experiment for synthesizing Au nanoparticles was performed by adding the AA solution to the HAuCl4 solution in the absence of template. The obtained Au nanoparticles showed similar size distribution of (210 ± 65) nm (Fig. 2F) to that of HAuNFs (Fig. 2H), but they were spherical in shape (ESI: Fig. S7†), exhibited smoother surfaces (Fig. 2E), and had an absorption peak at 550 nm (Inset of Fig. 2F). These results indicate that the templates are critical for the formation of a flower-like morphology, due to the following two effects: (1) physically restricting the nanocrystal growth in a particular direction and hence leading to the anisotropic growth; (2) increasing the local concentration of Au precursor by adsorbing AuCl4− ions, thus enhancing the supersaturation, which is responsible for the formation of the branches.31,32
The surface topography of the HAuNFs is dependent on the concentration of the reducing agent ascorbic acid (AA). Fig. S8A(ESI†) shows that the HAuNFs synthesized with 50 mM AA exhibit a “meatball” shape, of which the branches have the lengths of 5–10 nm and the tip angles >90°. When 100 mM AA was used, the tips of obtained HAuNFs (shown in Fig. 2B) become longer (10–15 nm) and sharper (70°–90°). The use of higher concentration of AA (150 mM) changes the product morphology to sea urchin-like, spike-coated structures, with the spike lengths of 20–30 nm and the tips <50° (ESI: Fig. S8B†), suggesting the length and sharpness of the spikes on the nanoflower surface increase with higher reducing agent concentration, likely caused by the secondary nucleation induced by the larger amount of reducing agent.32,33 The longer spikes and increased roughness of branched gold nanoparticles have been demonstrated to result in the red-shift of SPR.34–36 Similarly, in our work, the absorption peak of the HAuNFs was tuned from 610 nm in the visible region to 690 nm in the near infrared (NIR) region, by increasing the AA concentration from 50 mM to 150 mM (Fig. 2G), while the size distribution did not show discernible variation (Fig. 2H). The NIR absorption suggests that these HAuNFs have potential uses for NIR-based photothermal therapy and enhanced photoacoustic imaging.37
High resolution transmission electron microscopy (HR-TEM) measurements of the HAuNFs (synthesized with 150 mM AA) showed that the spikes had edges and corners (Fig. 3A, ESI: S9A and S10A†), and they exhibit zig-zag features on the borders, demonstrating the presence of low-coordinate atomic steps. Fig. 3B, S9B and S10B† show the atomic steps on the branch surface with a series of (100) and (111) sub-facets. These small facets are combined to form the high-index facets, for examples, (211) and (311) planes composed of (111) terraces and (100) steps, and (411) planes composed of (100) terraces and (111) steps, matching the corresponding atomic models (Fig. 3C, S9C and S10C†) well.38–40 The projection angle (ESI: Fig. S11†) between the surface and the {100} facets confirmed the presence of {211}, {311} and {411} facets.41 In addition, it can be seen from Fig. 3D–F that, a high density of dislocations, steps and kinks are exposed on the HAuNF spikes, which have been reported to be the basis of high catalytic activity in gold catalysts.42,43
The above results show that our HAuNFs integrate several beneficial and attractive morphological/structural properties, on the nano- and atomic-scale, for their use as nanocatalysts, including large surface area, high degree of roughness, a large number of active sites (edges and corners),13,44 high-index facets,45,46 a high density of dislocations, steps and kinks on the branch tips,43 “clean surface” which is free from surfactants conventionally involved in fabricating other types of multiple-branched Au nanostructures and hence more accessible for the reactants.47 Compared with the solid counterpart, the inner surface of HAuNFs increases the total surface area by ∼18.3% (The details of estimating the surface area are shown in ESI.†), the open cavity facilitates the diffusion of reactants to the inner surface, thus effectively activating its catalytic activity,12–14 and the confinement (cage) effect enables the interior cavity to act as a nanoreactor building up the concentrations of the species involved in the reaction.44 Owing to the synergy of these characteristics, our HAuNFs are expected to possess excellent catalytic effect, which was indeed displayed in the aforementioned process in the template degradation stage.
![]() | (1) |
Fig. 4A shows the catalytic degradation of MO with time at 20 °C, and it can be observed that the absorption peak at 465 nm from MO decreases gradually as the reaction proceeds in the presence of HAuNFs (ESI: Scheme S2† depicts the mechanism of catalytic degradation of MO by NaBH4, and Fig. S12† shows the time-dependent UV-vis spectra of MO, in the presence of NaBH4 and AuNSs), and complete degradation occurs within 100 s. The plot of ln[C0/Ct] versus time for these spectra is linear, demonstrating that the reaction follows first-order kinetics. At the concentration of 8 mg l−1, the HAuNFs and AuNSs have respective kapp of 3.4 × 10−2 s−1 and 1.7 × 10−3 s−1. We further measured the kapp in the reaction systems with different concentration of nanocatalysts (HAuNFs or AuNSs), finding that the kapp is proportional to the nanocatalyst concentration, (M, mg l−1). Therefore, we derived the slope of the line (k1) from eqn (2), which reflects the intrinsic catalytic activity.49
![]() | (2) |
![]() | ||
Fig. 4 Catalytic properties of HAuNFs methyl orange degradation: (A) extinction spectra recorded at different reaction time points using HAuNFs (concentration: 8 mg I−1) as catalyst, indicating the disappearance of 465 nm peak owing to the degradation of MO; (B) plots of ln(Ct/C0) versus reaction time for the MO degradation (concentration of HAuNFs/AuNSs: 8 mg l−1); (C) plots of the apparent rate constants (kapp) as a function of the mass concentration (M); (D) conversion within 100 s (Table S2† shows the data for calculating conversions) and kapp against the number of successive cycles. (Concentration of HAuNFs: 8 mg I−1). 4-Nitrophenol reduction: (E) the extinction spectra recorded at different reaction time points using HAuNFs (concentration: 8 mg I−1) as catalyst, showing the decrease in intensity for the peak at 400 nm associated with 4-nitrophenolate as the reduction reaction proceeded; (F) plot of ln(Ct/C0) versus reaction time for the 4-NP reduction (concentration of HAuNFs/AuNSs: 8 mg I−1); (G) plots of the apparent rate constants (kapp) as a function of the mass concentration; (H) conversion within 120 s (Table S3† shows the data for calculating conversions) and kapp against the number of successive cycles (concentration of HAuNFs: 8 mg I−1). The concentration of gold nanoparticles (HAuNFs and AuNSs) in solution was determined using an atomic absorption spectrometer (AAS). |
The slope (k1) related to HAuNFs and AuNSs were determined to be 3.99 × 10−3 and 2.18 × 10−4 l s−1 mg−1, which shows that the HAuNFs are ∼18 times more catalytically active than the AuNSs for this reaction.
In addition to activity, stability and reusability are also essential factors and desired properties for the practical applications of catalysts.50,51Fig. 4D shows the plots of kapp and conversion versus the number of successive MO degradation that repeatedly employed the HAuNFs (concentration: 8 mg I−1) as catalyst. Notably, the HAuNF catalyst retained its high activity and achieved high conversions even after 10 cycles.
In the case of the reduction of 4-nitrophenol (4-NP) by NaBH4, we investigated the kinetic process by monitoring the intensity of the absorption peak at 400 nm (assigned to 4-nitrophenolate ions in the mixture of 4-NP and NaBH4). After the addition of HAuNFs, a new peak at 315 nm of the product 4-aminophenol (4-aminophenolate) appeared, and the absorption peak at 400 nm gradually dropped and then disappeared after 120 s. The gas bubbles of H2 evolved by the decomposition of NaBH4 may impede absorbance measurements during the reaction, resulting in the absence of isosbestic points in Fig. 4E (and Fig. S13†).52–54 Because the peak at 400 nm dominated for most of the time, it is reasonable to derive the concentration of 4-NP ions from absorbance at 400 nm and thus to investigate the reaction kinetics.49 The HAuNFs also showed excellent activity (k1 = 3.12 × 10−3 l s−1 mg−1 for HAuNFs, while k1 = 4.62 × 10−4 l s−1 mg−1 for AuNSs), stability and reusability (Fig. 4H showed that no significant loss was observed in the conversion or Kapp after 10 successive cycles). It is noteworthy that, compared with the recently reported solid Au nanostars,55 our HAuNFs enjoy ∼1.7 times of catalytic activity in this model reaction (Table 1).
We also examined the electrocatalytic performances of our HAuNFs by adopting electro-oxidation of ethanol as a model reaction. Fig. 5A shows cyclic voltammetry (CV) curves of the ethanol oxidation activities of HAuNFs and AuNSs in a 0.1 M KOH with 0.5 M ethanol solution. The mass normalized peak current density (376.3 mA mg−1) of ethanol oxidation on the HAuNFs in the forward direction scan was ∼30 times of that of AuNSs (11.7 mA mg−1) and ∼3.4 times of that of commercially available Pt black (0.11 A mg−1)41 when measured under the same conditions, demonstrating the superior electrocatalytic activity of our HAuNFs towards ethanol oxidation. It is further noted that the catalytic activity of the HAuNFs was also higher than that of recent state-of-the-art Au-based nanocatalysts, such as Au NPs supported on activated carbon,56 and Au dendritic nanostructures with hyperbranched architectures.57 Furthermore, our HAuNFs with spatially separated spikes drastically suppress the activity loss caused by the undesirable agglomeration of Au active sites, which therefore contribute to the high electrocatalytic stability.58 As shown in Fig. 5B, after 500 cycles, 86.7% of the initial catalytic activity was still maintained.
The dark-field microscopy image of the HAuNFs on the Si substrate (inset of Fig. 6A) shows that, the individual nanoparticles were well separated from each other, with interparticle distances much larger than the dimension of each particle. As a result, interparticle plasmonic coupling was negligible and should have no contribution to the Raman enhancement. The large interparticle distances allowed us to focus the laser beam on one particle each time, using a confocal Raman microscope to collect single-particle SERS (spSERS) signals.
SERS spectra were collected from 50 individual particles for each sample (HAuNFs or AuNSs), and the average SERS spectra of a monolayer of BPT adsorbed on individual HAuNFs or AuNSs are shown in Fig. 6A. The Raman features of BPT molecules were detectable at single nanoparticle level for both HAuNF and AuNS (The broad Raman mode around 940 cm−1 originates from the used silicon substrate). The spectral features are in good agreement with earlier experimental and theoretical work.59 Strikingly, the SERS intensity on individual HAuNFs was ∼40 times higher compared with that on AuNSs.
We further estimated the enhancement factors (EFs) by comparing SERS signals to normal Raman intensities obtained from neat BPT powders (ESI: Fig. S14†). Two intense Raman bands at 1100 cm−1 (C–H stretching mode) and 1589 cm−1 (phenol ring C–C stretching mode) were chosen for EF estimation (The detailed method for estimating the EFs is described in ESI.†). It can be seen from Fig. 6B that, the phenol ring C–C stretching showed larger enhancements than the C–H stretching mode, most likely due to the stronger coupling between the transition dipole moment and local electric field.34 The EFs were estimated to be on (or close to) the order of 105 for HAuNFs, while on the order of 103 for AuNSs.
Among the structural features of HAuNFs, the open cavity makes the inner walls accessible, and the facing inner walls form nanogaps of several tens of nanometers or even sub-10 nm, greatly enhancing the plasmon electromagnetic field.60,61 Moreover, the increased surface area and roughness, tips, edges, intra-particle gaps and nanoscale porosity between the spikes of the HAuNFs (shown in Fig. 6C) all contribute to significant electromagnetic field enhancements, thus affording strong spSERS signals.62,63
The optical absorption of these HAuNFs can be modulated to NIR region by tailoring the topology/roughness, without enlarging the dominant dimension of the nanoparticles, providing them with the benefits of both smaller size and NIR absorption desirable for future in vivo applications. Moreover, our HAuNFs enjoy a large surface area, rich edges and corners, an open hollow cavity, high-index facets and a clean surface, which synergistically lead to the excellent (electro) catalytic performances and SERS activity.
It is envisaged that these HAuNFs can be utilized as effective catalysts and SERS probes, and hold potential uses as drug carriers and photothermal conversion agents. We also envisage that our novel synthesis protocol is generic and provides an insight into the room-temperature aqueous phase synthesis of a variety of hollow metal nanomaterials for diverse future applications.
(2) Since the catalytic activity of HAuNFs was independent of their concentration, we tracked the variation in their catalytic activity by repeated addition of new MO.64 After each cycle of reaction proceeded for 100 s, another 10 μl of MO aqueous solution (5 mM) was added to the reaction system.
(3) Following the protocol in (1), the catalytic activities of HAuNFs/AuNSs at different concentrations were measured by adding 10 μl dispersions of HAuNFs/AuNSs (200 μg ml−1, 400 μg ml−1 or 1600 μg ml−1, respectively corresponding to 2 mg l−1, 4 mg l−1 and 16 mg l−1 in the reaction system).
(2) Since the catalytic activity of HAuNFs was independent of their concentration, we tracked the variation in their catalytic activity by repeated addition of new 4-NP.64 After each cycle of reaction proceeded for 120 s, another 10 μl of 4-NP aqueous solution (20 mM) was added to the reaction system. This step was repeated ten times to investigate the stability of catalyst particles.
(3) Following the protocol in (1), the catalytic activities of HAuNFs/AuNSs at different concentrations were measured by adding 10 μl dispersions of HAuNFs/AuNSs (200 μg ml−1, 400 μg ml−1 or 1600 μg ml−1, respectively corresponding to 2 mg l−1, 4 mg l−1 and 16 mg l−1 in the reaction system).
Footnote |
† Electronic supplementary information (ESI) available: Fig. S1–S14, Tables S1–S3 and Schemes S1–S4. See DOI: 10.1039/c6nr04045d. The data presented in this article are openly available from the University of Leeds data repository http://dx.doi.org/10.5518/66. |
This journal is © The Royal Society of Chemistry 2016 |