Characterization and photothermal properties of Au-Pd Core Shell Nanoparticles:
Figure 1A illustrates the colloidal nanoparticles used in this first part of the study. They consist of Au nanospheres (Au NS) with a diameter of 67 nm, and Au@Pd core-shell nanoparticles (CS-NPs) with an Au core of 67 nm and Pd shells of ~ 2 and ~ 4 nm in thickness. The NPs were prepared using a seed-mediated growth strategy.36 This method starts with the synthesis of small (seed) Au NS and the subsequent growth to the desired size, as described in the Materials and Methods sections. Two different shell thicknesses were obtained by varying the ratio of Au core and Pd salt. The three colloids are stabilized in water by means of cetyltrimethylammonium chloride (CTAC) capping. Figure 1B shows transmission electron microscopy (TEM) images of the obtained colloidal NPs. Au NSs are spherical, while the Au@Pd are more faceted. The spatial distribution of the two metals measured by Energy Dispersive X-Ray Spectroscopy (EDX) is presented as an inset in the bottom panel of Fig. 1B. While the signal of Au is confined to the core region, the Pd signal is uniformly distributed around the entire particle, as expected for a homogeneous coverage. Particle size histograms were obtained from TEM images and are shown in Fig. 1C. The median diameter of the Au NS is (66.6 ± 0.2) nm with a standard deviation of \(\sigma\) =1.3 nm, while the two CS-NPs colloids have median sizes of (71.5 ± 0.1) nm with a standard deviation of \(\sigma\)=1.2 nm and (73.8 ± 0.2) nm with a standard deviation of \(\sigma\) =1.7 nm. These values correspond to median Pd thicknesses of 2.4 ± 0.3 nm, and 3.6 ± 0.3 nm. For simplicity, we name them Au67 NS, Au67@Pd2 and Au67@Pd4. These parameters were employed to simulate the optical properties of the NP, as described in section S1 of the supplementary materials (SM). Figure 1D shows the numerically calculated absorption cross section spectra for different Pd thicknesses. The Pd shell leads to significant damping of the plasmon resonance of the Au core, reducing its amplitude and broadening its bandwidth. In addition, a small blue shift of the resonant frequency is predicted. In previous reports, a blue shift is observed for CS-NPs, 36 and a redshift for Au nanorods coated with Pd on the tips.61 Fig. 1E shows the normalized experimental extinction spectra of the three synthesized colloids. The predicted broadening of the resonances due to the Pd shell is clearly observed. However, instead of a blue shift, the spectra exhibit slightly red-shifted resonances, which could also be attributed to the larger concentration of CTAC used as the stabilizing agent of the CS-NPs. 62
In order to study optical and photothermal properties at the single particle level, arrays of well-separated individual NPs were fabricated on glass substrates using optical printing, as schematically shown in Fig. 2A.63–65 Exemplary dark-field images of the optically printed grids are displayed in Fig. 2B. Having ordered arrays of isolated NPs facilitates the automation of data acquisition and allows the collection of larger datasets. It must be noted that the printing process does not alter the stability or composition of the NPs, as further discussed in section S2 of the SM. Figure 2C shows representative normalized scattering spectra for each type of NP. CS-NPs present a redshift and a broadening of their resonance with respect to the Au NS cores (see section S3 of the SM for further details). Because PL emission can provide insights into plasmon decay processes, it is relevant to investigate the emission properties of the NPs under study. Figure 2D shows PL spectra of the types of NPs, excited with CW laser excitation at 532 nm. This wavelength is suited to excite the plasmon resonance of three kinds of NPs. In agreement to reports on other NPs, we found that the shape of the PL spectra follows the characteristics of the scattering one. 66–69 Interestingly, the amplitude of the PL emission decreases with increasing Pd thicknesses. Qualitatively, this decrease in PL emission is a result of the Pd damping the plasmon resonance of the Au core.61 For quantitative analysis, the Stokes emission quantum yield \({QY}^{S}\) was calculated as the ratio between the total Stokes PL emission and the absorbed photons\({QY}_{PL}^{S}=\frac{{\int }_{543}^{\infty }PL\left(\lambda \right) d\lambda }{{\sigma }_{abs}\left(\lambda =532nm\right) {I}_{exc}}\), where \({I}_{exc}\) is the excitation irradiance (see section S4 of the SM for further details). In addition, the resonance quality is quantified as Q = \(\frac{{\text{E}}_{\text{r}\text{e}\text{s}}}{{\Gamma }}\) where \({\text{E}}_{\text{r}\text{e}\text{s}}\) is the resonance energy and Γ its full width at half maximum calculated from the scattering spectra. Figure 2E shows the measured \({QY}^{S}\) versus Q for the three NPs, showing a positive correlation. A similar trend was reported for Au nanorods and ascribed to enhanced emission due to an increased density of photonic states (Purcell effect). 61, 70–72
After the optical characterization of the individual NPs, their photothermal response was measured using hyperspectral Anti Stokes (AS) thermometry with a CW 532 nm laser. The technique, as introduced by Barella et. al. 53, allows single particle photothermal characterization by raster scanning a laser over the NP, as schematized in Fig. 3A. In this way, heating and PL excitation are performed simultaneously with the same beam. Throughout the scanning, the relative position between the beam and the NP changes, leading to different excitation irradiances and hence, different steady state temperatures. In such a manner, a set of temperature-dependent PL emission spectra is collected. Figure 3B, shows three PL spectra for different irradiance levels for an illuminated Au67@Pd2 CS-NP. Processing the set of acquired PL spectra allows for finding the photothermal coefficient\(\beta\), defined as
\({ T}^{NP}=\beta {I}_{exc}+{T}_{0}\)
|
(1)
|
where \({ T}^{NP}\)and \({T}_{0}\) are the temperature of the particle in the presence and absence of light, respectively. For details on data processing, see section S5 of SM.
It must be noted that this method assumes that the PL emitting object has a homogeneous temperature\({ T}^{NP}\). This condition is fulfilled by Au NS Under CW illumination,3 and is also true for Au@Pd CS-NPs, as we demonstrate in the following. Figure 3D shows the calculated temperature increase \(T\left(r\right)-{T}_{0}\) versus the radial coordinate \(r\) for a Au67@Pd2 NP immersed in water under illumination at 532 nm with an irradiance of \({I}_{exc}\)=1 mW/µm2 (see Fig. 3C for the definition of the parameters and section S6 of SM for a full derivation of the calculus). The temperature is practically constant inside the CS-NP, except for a small drop \({\nabla T}_{Au-Pd}\) at the Au/Pd interface, shown in the inset of Fig. 3B. The temperature variations inside the CS-NP are \({10}^{-3}\) times smaller than the temperature increase of the NP surface. Hence, the NP can be described by a single, uniform temperature\({ T}^{NP}\). This is a consequence of the high thermal conductivities of Au and Pd with respect to water, and the high electronic thermal conductance of the Au/Pd interface that allows efficient heat transfer through electron-electron scattering.73 In addition, because Au and Pd have similarly high volumetric electron-phonon couplings, electrons and phonons equilibrate rapidly within each metal. For this reason, \({ T}^{NP}\) refers indistinctly to the electronic or lattice temperature. (See Section S7 for a quantitative comparison between lattice and electron gas temperatures). Thus, AS thermometry retrieves the photothermal coefficient of the entire CS-NPs.
Figure 3E shows histograms of the measured photothermal coefficients for the three systems under study. The median value obtained for the Au NS is \({\beta }_{Au67}\) is (51 ± 1)\(\frac{K{ \mu m}^{2}}{mW}\), in line with the one reported by Barella et. al. for 64 nm Au NS.53 Interestingly, a significant reduction of \(\beta\) is observed for the CS-NPs. The obtained median values of the Au67@Pd2 and Au67@Pd4 were \({\beta }_{Au67@\text{P}\text{d}2}\) = (38 ± 1) \(\frac{K{\mu m}^{2}}{mW}\) and \({\beta }_{Au67@\text{P}\text{d}4}\) = (35 ± 1) \(\frac{K{\mu m}^{2}}{mW}\), respectively. The standard deviation of the three measurements is \(\sigma\) = 8\(\frac{K{\mu m}^{2}}{mW}\). Figure 3F (solid lines) shows the temperature increase versus excitation irradiance following Eq. 1, corresponding to the median \(\beta\) of each type of NP. In dashed lines, the maximum irradiances used in each experiment are shown. The maximum temperature reached by the NPs was around 100°C. It must be noted that at these temperatures, the NPs are stable and no changes in their scattering or PL spectrum were observed during or after the measurements (see section S2 of the SM).
An analytical model was developed to calculate the temperature \({ T}^{NP}\)of a NP surrounded by a media of thermal conductivity \({\kappa }_{3}\) and supported on a substrate with thermal conductivity\({ \kappa }_{4}\). Details can be found in section Materials and Methods and sections S6-S8 of the SM. The temperature \({ T}^{NP}\) is given by
\({ T}^{NP}=f\left(\frac{1}{{4\pi \kappa }_{3}a}+\frac{{R}_{2-3}^{th}}{4\pi {a}^{2}}\right){\sigma }_{abs}{I}_{exc}+{ T}_{0}\)
|
(2)
|
where \({R}_{2-3}^{th}\) is the Kapitza interfacial thermal resistances between the shell and the surrounding media, and \(f=\left(1-\frac{{\kappa }_{4}-{\kappa }_{3}}{2\left({\kappa }_{4}+{\kappa }_{3}\right)}\right)\) is a factor that accounts for the role of the substrate in the heat dissipation. The solid black lines in Fig. 3E correspond to the photothermal coefficient\(\beta\) predicted by Eq. (2) for Au@Pd CS-NPs as a function of the Pd shell thickness, for NPs on glass, surrounded by water and illuminated at 532 nm, as in the experiments. The calculation requires several thermodynamical constants for which accurate experimental values are scarce. This is the case for example for the Kapitza resistance between Pd and water \({R}_{Pd-Water}^{th}\). For this reason, we have included a maximum and a minimum calculated value, to represent the large dispersions of available data in the literature. A list of used parameters is shown in Table 2 in Section Materials and Methods. The calculations reasonably predict the experimental trends. However, it must be noted that for Au@Pd, half of the measured CS-NPs had a value below the predicted range. This could be due to several factors. i) An overestimation of the absorption cross sections: The simulations (shown in Fig. 1B and Figure S1) predict a 10 nm blue shift in the resonant frequency of CS-NPs, which is not observed experimentally. However, considering that the resonances are broad, a 10 nm detuning of the spectrum versus the excitation wavelength only modifies the absorption cross sections by less than 2% ii) The influence of the surfactant CTAC in the thermal resistance of the Pd-water interface.74 iii) an overestimation of the factor \(f\) accounting for the effect of the substrate in heat dissipation. For a spherical NP immersed in water on a glass substrate, \(f\) takes a value of 0.875. However, Au@Pd CS-NPs are faceted, as shown in Fig. 1B, and can present a larger contact area with the substrate, enhancing heat dissipation. Thermal simulations estimate a value of \(f\)= 0.843 for faceted NPs, which is 4% smaller than its spherical counterpart (see section S12 of SM for details). To summarize, accurate theoretical predictions require a complete knowledge and modeling of every geometrical boundary, also including the liquid-substrate interface. Having a precise description of all these factors is challenging, making most predictions only approximate, and reinforcing the need for methods able to measure the temperature of nanoscale objects in their operation environments (i.e., in situ).
Effect of morphology on photothermal properties of Au-Pd Nanosystems:
The functionalities of bimetallic nanostructures are not only determined by the material composition, but also by the spatial distribution of the constituents.36 In the following, the effect of morphology on light-to-heat conversion for different Au-Pd bimetallic structures is investigated. Au NS 60 nm diameter cores were combined with Pd in two different configurations: i) assembled with 5 nm spherical Pd satellites NPs (named Au60-Pd-sat) and ii) coated with a homogeneous (1.8\(\pm\)0.3) nm Pd shell (named Au60@Pd2). Figure 4A shows an illustration of the three synthetized systems, as well as their corresponding TEM images. The average number of satellites per NP of the Au60-Pd-sat system was estimated from TEM images to be\(⟨{N}_{s}⟩=160\pm 30\). The approximated volume of Pd per Au core is \({V}_{Pd}=(1.0\pm 0.2)\times {10}^{4} n{m}^{3}\) for Au60-Pd-sat and \({V}_{Pd}=(2.4 \pm 0.4)\times {10}^{4} n{m}^{3}\) for Au60@Pd2. This means that the amount of Pd is on the same order of magnitude for both configurations. Further characterizations of the NPs are presented in section S10 of the SM. Figure 4B shows the normalized extinction spectra of the three studied systems. Again, adding Pd to the Au cores results in plasmon damping and a decrease in the resonance quality, although larger damping was observed for the CS-NP. Arrays of single Au NS 60 nm and Au60@Pd2 CS-NPs were fabricated through optically printing on glass substrates. Instead, the Au60-Pd-sat were deposited by a drop-casting method, because the solution was not stable enough for the optical printing process. Figure 4C shows representative single particle PL emissions when excited with laser light at 532 nm in a water environment. The lower Q factor is reflected in a drop in the single particle PL emission. The decrease in PL emission is significantly larger in the Au60@Pd2 CS-NPs compared to the Au60-Pd-sat, indicating a larger plasmon damping. It must be noted that the Pd satellites interact weekly with light (see section S11 of SM) and absorb only a minor fraction of the incoming light (altogether less than 10%, when placed around the core).(35) Hence, the PL emission of the Au60-Pd-sat system is mostly emitted by the Au core.
Next, the photothermal coefficient of each system was determined using hyperspectral AS thermometry. The resulting histograms are presented in Fig. 4D. The median photothermal coefficient for the Au 60 NS cores was \({\beta }_{Au60}\) = (47 ± 1) \(\frac{K{\mu m}^{2}}{mW}\) with a standard deviation of 8\(\frac{ K{\mu m}^{2}}{mW}\). Remarkably, there’s no significant difference with the median value for the core-satellites \({\beta }_{\text{A}\text{u}60-\text{P}\text{d}-\text{s}\text{a}\text{t}}\)= (46 ± 1)\(\frac{K{\mu m}^{2}}{mW}\), with a larger standard deviation of 13\(\frac{K{\mu m}^{2}}{mW}\). If the PL is mostly emitted by the Au, the measured \(\beta\) corresponds to the temperature of the core. By contrast, a significant reduction of the photothermal coefficient is observed for the CS-NPs, with a median of \({\beta }_{\text{A}\text{u}60@\text{P}\text{d}2}\) = (29 ± 1) \(\frac{K{\mu m}^{2}}{mW}\) and a standard deviation of 5\(\frac{K{\mu m}^{2}}{mW}\), in line with the results presented in the previous section of this work. Overall, the experiments of this section point out that the presence of an interface between the two metals is crucial and dictates the photothermal response of the bimetallic nanostructures. The direct contact between Au and Pd enhances surface damping of the Au plasmon, leading to a poorer quality factor and lower absorption.