It is well known that the melting point of Au nanoparticles (NPs), and other metals, decreases with decreasing NP size, where the NP melting temperature (Tm,NP) relative to the bulk metal (Tm,B) is proportional to 1/diameter (1/d). 1,2 Many thermal processes occur below Tm, such as sintering 3, mercaptoethanesulphonic acid-stabilized 1.5 and 1.0 nm diameter Au NCs began ligand loss and sintering at 242 and 245 o C, respectively. 17 Others showed that the presence of halides greatly affected the thermal stability, 21 shape transformations occurred for ~10 nm oleylamine-coated Au NCs at 165-250 o C, 10 and 2.2 nm diameter dodecanethiol-stabilized Au NCs grew to 4.8 nm in the presence of bromide and acid at room temperature over 5 hours. 20 In these examples, understanding the size-dependent thermal properties of the clean Au NPs/NCs was not possible because the strongly-coordinated ligands and other additives played a major role as evidenced by the greater thermal stability for smaller Au NCs 17 and low thermal stability for 10 nm Au NCs in the presence of bromide. 10 The size-dependent melting of clean Au NPs and NCs has been the subject of many studies for well over 100 years now. 23 Accordingly, there have been numerous theoretical studies modeling the size-dependent melting temperature of clean Au by (nano)thermodynamic methods, 2,24,25 cohesive energy calculations using different potential functions, 26,27 Monte Carlo simulations, 28 molecular dynamics simulations, 29,30 and bond-order-length-strength (BOLS). 31 Tm,NP is generally related to 1/d 31 and also depends on the specific shape (truncated octahedral, icosahedral, etc.), 32 crystal structure, 27 the support that the NP/NC is attached to, 33 the proportion of edge and corner atoms, 34 and the environment (vacuum, air, gases). 15 These different factors make it difficult to accurately predict the thermal behavior for all cases, but there have been attempts at universal melting equations. 35 The sintering of clean Au NPs/NCs depends on the initial size dispersity, 36 dispersion, 36 metal-support interaction, 37 and metal-reactant interaction, 38 showing that size is not the only important factor governing the thermal properties. Theoretical melting models agree better with experimental melting temperatures for Au NPs above 5 nm in diameter, deviating significantly for Au NCs below about 2 nm in diameter or 500 atoms. 23,39 Most theoretical studies compare predicted melting points to the experimental values of Buffat and Borel, 1 Dick et al., 40 Sambles, 41 and Castro et al. 39 Common experimental techniques for measuring the melting temperature of Au NPs include microscopic, 39,42 differential scanning calorimetry (DSC), 17 thermal gravimetric analysis (TGA), 17,40 electron scattering, 43 and electron diffraction, 1 ranging from larger Au nanostructures that are 10s of nm down to ~2 nm diameter Au NCs. 39,40,42 Experimental measurements below 2 nm are rare.
In this work, we used electrochemistry to monitor the thermal properties of weakly-stabilized 0.9, 1.6 and 4.1 nm diameter Au NPs/NCs and compared the results to theory and previous experimental studies. One major advantage of our method is that the stabilizers used allowed the synthesis of Au NCs down to 0.9 nm diameter and size characterization by anodic stripping voltammetry (ASV) directly on conductive supports. 44,45 This enabled size characterization before and after exposure to various temperatures to determine thermal stability. This is not possible with strongly-coordinated ligands, such as thiols, because they inhibit the ASV analysis and thermal properties. Our group previously used ASV for analyzing metal NP/NC size, 46 SA/V, 46 aggregation state, 47 composition and atomic arrangement, 48 and size stability under various conditions. 13,45 Our approach allows for fast, low cost, and highly sensitive size analysis of metal NCs/NPs attached to applicable conductive support surfaces in normal environments. 49 In contrast, electron microscopy in vacuum suffers from potential electron beam effects, high cost, low sample throughput, unrealistic vacuum environment, and limited types of support surfaces.
The thermal stability of metal NPs is of major scientific interest due to their unique size-dependent properties and potential applications that require their thermal stability, such as catalysis. 15 EXPERIMENTAL Chemicals and Materials. Sodium borohydride, (3-aminopropyl)triethoxysilane (≥ 98.0%), 2-propanol (ACS reagent) were purchased from Sigma Aldrich. Gold salt (HAuCl4•3H2O) was synthesized from metallic Au (99.99%) in our lab. Acetone, methanol and ethyl alcohol (ACS/USP grade) were purchased from Pharmco-AAPER. Trisodium citrate salt, potassium perchlorate (99.0-100.5%), potassium bromide (GR ACS), perchloric acid (60%) and triphenyl phosphinosulfonate (TPPS, >90%) were purchased from Bio-Rad laboratories, Beantown Chemical, EMD, Merck and TCI respectively. Sodium hydroxide pellets were purchased from Fisher Scientific. Tetrakis(hydroxymethyl)phosphonium Chloride (THPC, 80% solution in water) was purchased from ACROS ORGANICS. Indium-tin-oxide (ITO)-coated and fluorine-doped tin oxide (FTO)-coated glass slides (CG-50IN-CUV, Rs = 8-12 Ω) were purchased from Delta Technologies Limited (Loveland, CO).
Synthesis of TPPS-stabilized 0.9 ± 0.2 nm Diameter Au Nanoclusters (NCs). We synthesized 0.9 ± 0.2 nm diameter Au NCs by following the protocol originally developed by Yao and co-workers and later adopted in our group. 1,2 Briefly, 11.5 mL of a methanol solution containing a mixture of HAuCl4•3H2O (0.5 mmol) and TPPS sodium salt (0.75 mmol) was placed in a glass scintillation vial. Next, 3 mL of freshly prepared ice-cooled 0.2 M NaBH4 solution was added under vigorous stirring. The immediate appearance of a brown color in the solution indicated the formation of 0.9 nm diameter Au NCs and the NCs were attached to electrode surfaces after stirring for 15 min. Synthesis of THPC-stabilized 1.6 ± 0.4 nm Diameter Au NCs. We synthesized 1.6 ± 0.4 nm diameter THPC-stabilized Au NCs using the synthesis protocol reported by our group previously 3 and originally reported by Duff and co-workers. 4 Briefly, 400 µL of the reducing agent THPC (200 µL of 80% THPC diluted to 16.66 mL of nanopure water) was added to a glass vial containing 15.5 mL of nanopure water followed by the addition of 500 µL of 0.2 M NaOH solution with constant stirring. After 2 min of stirring, 660 µL of 25 mM HAuCl4•3H2O was added.
Immediately an orange-brown color formed in solution after the addition of HAuCl4•3H2O, indicative of small Au NCs. These NCs were used within a couple of hours after synthesis and stable for at least 24 hours. Synthesis of 4.1 ± 0.7 nm Average Diameter Citrate-Stabilized Au NPs. We synthesized citrate-stabilized 4.1 ± 0.7 nm average diameter Au NPs as described by our group previously [5][6][7][8][9] and originally reported by Murphy and co-workers. 10 In this protocol, a solution mixture of 0.5 mL of 10 mM HAuCl4•3H2O and 0.5 mL of 10 mM trisodium citrate was added to 18.5 mL of water followed by the addition of 0.6 mL of ice-cold 0.1 M NaBH4 at once with rapid stirring for 2 hr. After the addition of NaBH4, the solution color turned from colorless to orange immediately and eventually converted to a red color within 5 min, indicating the formation of Au NPs. These NPs were used for further experiments within 24 hours. Preparation of THPC-stabilized 4.1 ± 0.7 Diameter Au NPs. We prepared THPC-coated 4.1 nm diameter Au NPs by a ligand exchange method reported recently by Gulka and coworkers. 11 Briefly, 500 µL of 100 µM THPC in water was added to 10 mL of citrate-coated 4.1 nm Au NPs synthesized by the method already described. After addition of THPC, the Au NPs immediately change from red to blue, indicating that the NPs become aggregated. The NP solution reverts back to a red color again after 24 hours, which indicates that the NPs spontaneously de-aggregate back to individual, well-separated NPs. After that, NPs were attached to glass/ITO/APTES electrodes by directly soaking the electrode in the Au NP solution. The SA/V ratio of the NPs was calculated following the method previously reported by our group to confirm that the size was ~4 nm. The CVs were scanned from -0.2 V to 1.6 V in 0.1 M HClO4 at a scan rate of 0.1 V/s. The ASVs were scanned from -0.2 V to 1.4 V at a scan rate of 10 mV/s in 0.1 M KClO4 plus 10 mM KBr solution. The ASV conditions were optimized for good size analysis based on the measured oxidation peak potential (Ep) as reported by our group previously.
UV-Vis Characterization. Ultraviolet-visible spectrophotometry (UV-Vis) was performed using a Varian instrument, Cary 50 Bio-spectrophotometer. UV-Vis spectra of as-prepared Au NPs were obtained in aqueous solutions of different sizes from 300-800 nm using water as the blank.
For studying the effect of thermal treatment, the Au NPs were attached to glass/ITO/APTES electrodes, heated for different times and temperature, and analyzed by UV-Vis using glass/ITO/APTES as the blank. images were obtained using a 200 kV FEI Tecnai F20 operated in TEM mode. NPs were attached to silica-coated Au TEM grids by directly soaking the APTES functionalized grid in the NP solution for 4-5 min. The grids were then rinsed with water and dried under N2 before imaging.
General Procedure. In this work, we synthesized different-sized Au NPs/NCs (0.9 ± 0.2, 1.6 ± 0.4 and 4.1 ± 0.7 nm average diameter) following the protocols reported in our previous publications (See supporting information for all experimental details). 13,44,46,50
We monitored the thermal stability of 0.9, 1.6 on theoretical work by Plieth and our previous experimental studies, 13,51 the peak at 0.63 V correlates to ~3.6 nm diameter Au NPs, indicating that some of the 0.9 nm Au NCs transformed into this larger size NPs by a thermal ripening mechanism. The peak current increased at 0.63 V and the peak at 0.26 V completely shifted to about 0.34 V upon heating up to 100 o C (Figure 1A).
The peak at 0.34 V likely corresponds to a stable intermediate size between 0.9 nm and 1.6 nm diameter and can be considered a different size or structural transition 8 that occurs at ~90 o C. The peak at 0.34 V completely disappeared after heating to 130 o C or higher, leaving only one peak in the ASV near 0.63 V. This indicates full transformation of all Au NCs from 0.9 nm to a size between 0.9 and 1.6 nm (first transition) and finally to 3. We observed similar behavior for 1.6 nm Au NCs when heated from room temperature to 200 o C. At 25 o C, the ASV of 1.6 nm Au NCs showed one peak at 0.47 V (Figure 1B). As the NCs were heated to 80 o C, a small shoulder peak appeared at 0.66 V, indicative of an increased size in the Au NPs. The peak current at 0.66 V gradually increased while that at 0.47 V gradually decreased with increasing temperature as occurred with 0.9 nm Au NCs (Figure 1B and Figure S1). At 130 o C, a small peak remained but it was slightly below 0.47 V. The presence of smaller Au NCs than the original size is consistent with an Ostwald Ripening process. 36 At 150 o C, one single peak appeared at 0.66 V along with complete disappearance of the original peak at 0.47 V.
This indicated the complete size transformation of 1.6 nm Au NCs to ~4 nm Au NPs. Upon further heating up to 200 o C for 30 min (Figure S1), the Ep does not increase further. Some broadening of the peak occurred, indicating that another size transformation may have been starting.
In contrast to 0.9 nm and 1.6 nm Au NCs, the ASV of 4.1 nm diameter Au NPs showed one single peak with Ep at 0.71 V even after 30 min of heating up to 400 o C (Figure 1C), This is due to the greater temperature stability (higher Tm,NP) of larger Au NPs compared to smaller Au NCs. Next, we studied the effect of thermal treatment on 1.6 nm Au NCs as a function of heating time at different temperatures for up to 60 minutes by ASV. We heated the surface-attached 1.6 nm diameter Au NCs to 70 o C for up to 60 min, where the main peak sharpened, and a minor shoulder peak appeared at ~0.65 V. This is consistent with a size increase due to some thermal instability at this low temperature over longer time (Figure S1A, blue plot). After 10 min of heating at 80 o C, a clear shoulder peak at 0.65 V emerged. After 30 min, the shoulder peak emerged into a well-defined peak at ~0.68 V (Figure S1B, pink graph). The peak at ~0.68 V along with the original peak at 0.47 V indicates that some of the 1.6 nm diameter Au NCs remained stable. The peak current at 0.68 V relative to that at 0.47 V increased after 60 min of heating, consistent with more NCs transforming into bigger sizes with time (Figure S1B, black graph). Figure S1C shows ASVs following the heating of glass/ITO/APTES/Au NPs (1.6 nm) at 100 o C for 10, 30 and 60 min. After 10 and 30 min, peaks exist at 0.47 and 0.68 V (blue and pink plots), where the peak at 0.47 V sharpened. After 60 min of heating, the peak at 0.68 V became dominant but not all Au NCs transformed in size (black plot). Heating at 130 o C led to an even more dominant peak at 0.68 V (Figure S1D) until the peak at 0.47 V completely disappeared at 150 o C after 30 and 60 min (Figure S1E), indicating complete thermal size transformation of the 1.6 nm Au NCs. Figure S1F shows were not visible by SEM before heating (Figure S2A) but became visible (Figure S2B) with a diameter of 5.7 ± 0.9 nm when heated at 400 o C for 60 min (Figure S2C), consistent with the ASV in Figure S1F. The SEM images of citrate-coated 4.1 nm Au NPs before (Figure S2D) and after thermal treatment at 200 and 400 o C (Figure S2E-F) showed no significant change in size after heating. This is generally consistent with the ASV, although ASV showed a small shift and broadening of the peak (Figure 1C). The ASV may be potentially more sensitive than SEM at this size. We monitored the effect of heating 1.6 nm Au NCs attached to glass/ITO/APTES by UV-Vis spectroscopy. Figure S3 shows a featureless spectrum for the 1.6 nm Au NCs characteristic of light scattering by the sub 2 nm diameter Au NCs, where there is no localized surface plasmon resonance (LSPR) absorbance peak. After heating from 100 o C up to 400 o C, a LSPR peak developed in the 500-550 nm region, consistent with NCs above 2 nm in diameter. The peak increased and red shifted as the temperature increased, indicative of larger size NPs or aggregates, consistent with the ASV and SEM data on thermal ripening.
We next used the ASV data to quantify the size transition temperature (Tt) for the different sized Au NPs as they transform from their original size to a more stable larger size upon heating. This can be determined from ASV since the Ep starts at one value, indicative of the NP/NC size, and then shifts to a more positive value upon heating due to size transformation to a larger size. We plotted the ratio of the current from the final transformed peak (ip,f) to the current from the inital peak (ip,i) as a function of temperature for the different sized Au NPs/NCs (Tables S1 andS2). For example, the Ep shifts from 0.26 V to 0.63 V for the 0.9 nm diameter Au NCs, which correlates to a size transformation from 0.9 nm diameter to ~3.6 nm diameter NPs. A plot of ip,f/ip,i, or i0.63V/i0.26V, as a function of temperature produced a sigmoidal curve, where Tt is defined as the inflection point. For this analysis, we ignored the transition from Ep = 0.26 to Ep = 0.34 V for 0.9 nm Au NCs since we only had one data point there, but we approximate Tt for that first transition is ~90 o C. The same plot was generated for 1.6 nm Au NCs and 4.1 nm Au NPs using the ratios of i0.66V/i0.47V and i0.80V/i0.72V, respectively, versus temperature as shown in Figure 2A. The transition is from 1.6 nm to 3.5-4.0 nm and 4.1 nm to 15-20 nm diameter, respectively. We fit the sigmoidal plots for 0.9, 1. 52 homogenous growth (HOG), 31 liquid nucleation and growth (LNG), 31 and liquid shell model (LSM). 31 The experimental values agree best with LDM 52 and the nanothermodynamics 25 models for the 4.1 and 1.6 nm Au NCs, but Tt is larger than expected for the 0.9 nm Au NCs for all models.
One potential issue with our size-dependent thermal studies is that the 4.1 nm diameter Au NPs are stabilized by citrate while the 0.9 nm and 1.6 nm diameter Au NCs are stabilized by THPC and TPPS, respectively. To rule out the ligands as the main reason for the different thermal properties, we compared ASV and SA/V data at different temperatures for citrate-and THPC-stabilized 4.1 nm diameter Au NPs following the work of Sharma et al. 46 to determine their size at different temperatures. The ASV peak of the THPC-stabilized 4.1 nm Au NPs was slightly broader and had a small positive shift in Ep (Figure S4) along with smaller SA/V ratios (Table S3) compared to the citrate-stabilized 4.1 nm Au NPs due to increased Au NP diameter, but the slightly lower stability with THPC was not significant enough to explain the large differences between 0.9 nm/1.6 nm Au NCs and the 4.1 nm Au NPs. Exploring the role of different ligands and additives on the sizedependent thermal properties by ASV will be the focus of a future study.
Comparison to Literature. Previous melting and sintering studies involved Au NPs/NCs coated with various stabilizers or those that had clean Au surfaces. Within these classes there are also unsupported and supported Au NPs/NCs, where the supported ones can be close-packed assemblies, well-separated assemblies, or individual NPs/NCs. We do not compare our work to Au NCs stabilized with strongly-coordinated ligands, such as thiols, since we have well separated, electrode-supported Au NPs/NCs with weak ligand stabilizers that do not significantly alter the physical properties of the Au based on the expected decreasing Ep values and decreasing thermal stability with decreasing size. 44,45,49 Accordingly, we compare our work to experimental and theoretical studies of clean Au NCs in the literature. As shown in Figure 2C, our Tt values are generally in the expected range when comparing to theoretical Tm,NP values based on thermodynamic calculations of clean, unsupported Au NPs/NCs (see also Figure S5). Table S4 shows the expected Tm,NP for the same size Au NPs we studied using these thermodynamic models and different molecular dynamics (MD) simulations. The experimental Tt value for 4.1 nm Au NPs is a little lower than expected and that for 0.9 nm diameter Au NCs is higher than expected.
Our results match best the Tm,NP for 4.1 nm and 1.6 nm diameter Au NPs/NCs calculated using the nanothermodynamics 25 and LDM 52 models. All thermodynamic models fail for 0.9 nm Au NCs and some also fail for 1. 53 and Rey et al. 54 are 128 o C and 140 o C, respectively, which is very close to our Tt value for 0.9 nm Au NCs. In contrast, MD calculations by Soulé de Bas on Au7, Au13, and Au20 predict higher thermal stability for these clusters, near the Tm,b and even above Tm,b for the Au7 cluster. 55 Breakdowns in thermodynamic models are expected for NCs below about 500 atoms or 2.0 nm. 37,39 nm diameter Au NC on a tungsten tip using field emission microscopy (FEM). 39 While this is larger than our value, it is interesting that a slightly larger Au NC of 2.3 nm diameter had a lower Tm,NP of 260 o C and that the Tm,NP for clusters below about 2.5 nm diameter all had similar values. 37,39 This is consistent with our 0.9 nm Tm,NP being close to 1.6 nm, except our values are lower.
Clearly more experiments and theory are needed for Tm,NP of metal NCs < 2 nm.
It is likely that our Tt measurement reflects the temperature for thermal sintering or ripening and is lower than the true Tm,NP (solid to liquid transition). 6 The Hüttig and Tammann temperatures for Au are 401 K (128 o C) and 668 K (395 o C), respectively. 6 At the Hüttig temperature (~0.3Tm,b), atoms at defects become mobile, where atoms in the bulk become mobile at the Tammann temperature (~0.5Tm,b). 6 We expect the Hüttig and Tammann temperatures to be size-dependent and the Tm,NP may even be the same as the Hüttig temperature for Au NCs that contain mostly surface defect atoms, such as the Au13 NC, where 12 out of 13 atoms are at the surface (Garzon et al. calculated the Tm,NP to be 0.3Tm,b for Au13). Au NPs on amorphous carbon at 250 o C in oxygen for 2 hours. 15 Size-selected Au4, Au6, Au13, and Au20 NCs deposited onto amorphous carbon grew in size at room temperature by Ostwald ripening over a several month time period, where the Au atom diffusion and ripening kinetics were greater for the larger Au NCs. 57 Nelli et al. 58 modeled the ripening by coalescence for 1.6 nm diameter Au147 NCs occurring at 127-227 o C and Wan et al. 59 calculated the onset for Ostwald Ripening at 327 o C for 1 nm diameter Au NCs on TiO2(110). Triphenylphosphine (TPP)-stabilized Au11 NCs on a mesoporous silica support increased in size from 0.8 nm to 1.9 nm at 200 o C over 16 hours by cluster migration and aggregation. 7,60 While Ostwald ripening 36,37 and NP coalescence-based Smoluchowski ripening depend on many different factors, both processes can occur in ambient conditions at temperatures well below the Tm,NP and are important to consider.
We characterized the thermal properties of weakly-stabilized 0.9 nm, 1.6 nm, and 4.1 nm diameter Au NCs/NPs using ASV. A decrease in thermal stability with decreasing Au NC size shows that the weak stabilizers do not strongly alter the size-dependent thermal properties of the Au, which allows us to assess the properties of the metal without dominant ligand-metal interactions. The onset for size ripening of TPPS 0.9 nm, THPC 1.6 nm, and citrate 4. The ASV data was consistent with the appearance of a LSPR band in the UV-Vis spectrum and selected size analysis by SEM with temperature. The thermal stability is generally consistent with melting point depression and ripening models observed previously in the literature. Some thermodynamic models predict 4.1 nm and 1.6 nm Au NPs/NCs well, but there is a clear breakdown for the 0.9 nm Au NCs, as it has much greater thermal stability than expected by thermodynamics. The lower than expected thermal stability for 1.6 nm and 4.1 nm Au NPs is likely due to the fact that ASV measures the sintering/ripening temperature, which occurs lower than the Tm,NP. Our results on 0.9 nm Au NCs is most consistent with the MD simulations of Garzon et al. 53 placing the Tm,NP at about 0.3Tm,b. It is also qualitatively consistent with the experimental results of Castro et al., who showed similar thermal behavior for Au NCs below about 2.5 nm. 39 More work will be needed to improve theoretical models in this size range, while ASV can clearly find great use to determine the experimental temperature for size transformation/melting as a function of NP/NC size, support surface, ligand stabilizer, and metallic composition. ASV analysis is high throughput, low cost, and highly sensitive to size. The thermal properties of sub-1 nm NCs are very difficult to analyze under real conditions attached to normal electrode supports using other methods.
Details of the experimental procedures, ASV of 1.6 nm Au NCs heated for different time and temperature, SEM images of 1.6 and 4.1 nm Au NPs on glass/ITO/APTES after heating at different temperatures, UV-vis spectra of Au NCs before and after heating at different temperatures, ASV of 4.1 nm citate-and THPC-coated Au NPs upon heating, and plot of theoretical melting point of Au NPs as a function of diameter for several different thermodynamic models up to 30 nm diameter with our experimental data plotted for comparison are provided in Figures S1 through S5, respectively. Individual peak currents for electrooxidation of Au NCs/NPs from ASV at two different potentials upon 30 min heating at different temperatures are provided in Tables S1 andS2. Integrated Au oxidation charge under the peaks in CV and ASV, calculated SA/V ratios, and calculated NP size for 4.1 nm citrate-coated Au NPs after ligand exchange with THPC and heating are provided in Table S3. Comparisons of our Tt values to theoretical melting points and experimental melting points in the literature for the same sizes are provided in Table S4 and S5, respectively.