III. RESULTS AND DISCUSSION

Based on previous reports of metal aerogels formed from sol–gels and nanofoams using conventional aqueous based reducing agents, we tested a range of noble metal salt types and concentrations using fixed concentrations of DMAB, sodium borohydride (NaBH₄), and sodium hypophosphite monohydrate (Na₂HPO₂·H₂O). Unlike other studies that examined the role of reducing agents in noble metal nanofoam synthesis in using excess reducing agent by volume, here we used a 1:1 (v/v) ratio of noble metal salt solution to reducing agent solution. ^{Reference Ameen, Rajasekharan and Rajasekharan31} Upon addition of reducing agent, salt solutions immediately turned black, and effervescent gas evolution was observed consistent with numerous previous reports using similar reducing agents to synthesize noble metal nanostructures. In the cases where gels formed, a nascent gel structure was observed within seconds to minutes upon addition of the reducing agent. Rapid gas evolution compacted the gel toward the upper volume of the solution. Synthesis reactions then proceeded until gas evolution ceased, after which the gels were rinsed in deionized water.

Based on observations of the synthesis reactions, gel formation proceeds through five steps depicted in a proposed mechanism in Figs. 1(a)–1(e). First, nucleation of nanoparticles quickly initiates upon addition of reducing agent. Second, continued reduction of metal ions results in growth of the nanoparticles. Third, vigorous hydrogen gas evolution, as a product of the reduction mechanism, compacts and densifies the enlarging nanoparticles toward the top of the solution. Fourth, the densified nanoparticles confined toward the top of the reduction solution then fuse with neighboring particles, forming an extended three-dimensional structure. Continued gas evolution does not appear to further compact the gel once it forms. Fifth, within the pores of the nascent gel, remaining salt ions are depleted through reduction by the aqueous reducing agent and hydrogen gas. The particle fusion and further ion reduction is believed to account for the strengthening of the gel. After which the surface free energy of the gel structural elements arrives at equilibrium. Note that capping agents or linker molecules used in other nanoparticle and aerogel synthesis studies were not used in this work, although the influence of reaction products is not dismissed as having possible capping agent-like effects. Further, in the absence of solution buffers, there may be a further influence of pH changes.

FIG. 1.

Proposed mechanism of noble metal gel formation. Synthesis proceeds via (a) initial nucleation of nanoparticles, (b) rapid growth of nanoparticles, (c) densification of nanoparticles due to hydrogen gas evolution, (d) fusion of nanoparticles, and finally (e) surface energy minimization and equilibration of resulting gel.

To determine if there was a threshold salt concentration for gel formation, a fixed reducing agent concentration of 0.1 M was used to reduce a range of salt concentrations of 100, 50, 25, 10, 5, 1, 0.5, and 0.1 mM at a 1:1 (v/v) with reducing agents. This volume ratio appeared to facilitate the densification and confinement needed for the rapid gel formation described above. To determine the influence of the type of noble metal salt used on gel formation, a number of different Pd and Pt salts were used, whereas only [AuCl₄]⁻ was used to synthesize gold gels. The different combinations of salts, concentrations, and reducing agents tested are summarized in Table I, along with the minimum concentration at which contiguous gel formation was observed. Gold ions reduced easily and formed gels with DMAB and NaBH₄ with salt concentration thresholds of 10 mM and 5 mM, respectively, though the DMAB gels presented a smoother, more well-formed appearance. While NaBH₄ was the most effective reducing agent for both [PdCl₄]²⁻ and [Pd(NH₃)₄]²⁺ with a 5 mM salt concentration threshold for gel formation, DMAB reduction resulted in gels for both salts above 25 mM salt concentration. Likely due to their thermodynamic stability, square planar platinum salts [PtCl₄]²⁻ and [Pt(NH₃)₄]²⁺ resulted in poor gel formation with all reducing agents tested, but with some gelation with 0.1 M salt and 0.1 M NaBH₄. [PtCl₆]²⁻ reduction was facile with all reducing agents tested. While gels were observed for [PtCl₆]²⁻ at 25 mM and higher with DMAB reduction, the gels were not stable monolithic gels but rather broke apart into nanofoam fragments upon handing of the solutions. No monolithic gels were seen for [PtCl₆]²⁻ reduced in NaBH₄ and sodium hypophosphite. Ethanol was added to platinum salt solutions for a 1:1 (v/v) ethanol to deionized water solvent. Ethanol was selected based on previous use in platinum nanoparticle synthesis to destabilize surface energy and drive aggregation. ^{Reference Wang and Tseng43} With the 50% ethanol solvent, [PtCl₆]²⁻ formed gels above a 10 mM salt concentration with DMAB as the reducing agent but did not form gels with NaBH₄ or sodium hypophosphite. Despite its strong reduction potential (E _o) of −1.65 V (versus SHE), sodium hypophosphite was the least effective reducing agent for the salts tested, only forming gels with [PdCl₄]²⁻ at 50 mM and higher.

TABLE I.

Concentration thresholds for gel formation for different combinations of salt type and reducing agents.

Of all the salt, concentration, and reducing agent combinations tested, the following resulted in consistent, well-formed monolithic aerogels: [AuCl₄]⁻ reduced with 0.1 M DMAB, [PdCl₄]²⁻ reduced with 0.1 M NaBH₄, and [PtCl₆]²⁻ in 50% ethanol reduced with 0.1 M DMAB. The salt concentration range of 100, 50, 25, 10, and 5 mM for these synthesis combinations with 0.1 M reducing agent is shown in the photos in Fig. 2. Distinct gels are seen at higher concentrations, with gel size decreasing at lower concentrations until a single gel is no longer observed or no visible reduction product at all. Gold gels reduced with DMAB transitioned in color from reddish-gold at 100 mM salt solution to a black appearance at 10 mM and below as seen in Fig. 2(a). Pd and Pt gels appeared black for all synthesis concentrations in Figs. 2(b) and 2(c). The resulting gels are significantly compacted relative to the total solution volume.

FIG. 2.

Noble metal gel synthesis across a range of salt concentrations from right to left of 100, 50, 25, 10, and 5 mM. (a) [AuCl₄]⁻ reduced with 100 mM DMAB. (b) [PdCl₄]²⁻ reduced with 100 mM NaBH₄. (c) [PdCl₆]²⁻ prepared in 50% ethanol, reduced with 100 mM DMAB.

After complete reduction, gels were rinsed in deionized water for 24–48 h. After rinsing, gels were rapidly frozen at −80 °C and freeze dried to achieve the final aerogel structures. Gold, palladium, and platinum aerogels are shown in the SEM images in Fig. 3. For the SEM images and all subsequent discussion, Au aerogels were synthesized from 0.1 M [AuCl₄]⁻ and 0.1 M DMAB, Pd aerogels were synthesized from 0.1 M [PdCl₄]²⁻ and 0.1 M NaBH₄, and Pt aerogels were synthesized from 0.1 M [PtCl₆]²⁻ in 50% ethanol and reduced with 0.1 M DMAB. All reactions were performed at a 1:1 (v/v) of salt solution to reducing agent. SEM images indicate porous networks with contiguous metal ligaments for all three metals. Aerogel ligament diameter and pore sizes seen in Fig. 3 were analyzed with ImageJ. Gold aerogels appear to have smooth, fused elements ranging in diameter from 18 to 280 nm (average = 63.7 ± 36.0 nm), with large macropores ranging in size from 50 to 600 nm (average for meso- and macropores = 148.8 ± 121.5 nm). Palladium and platinum aerogels differ from gold in that they present a “string-of-beads” fused nanoparticle appearance with a rougher surface than gold. Palladium ligaments are 12–65 nm in diameter (average = 34.5 ± 9.5 nm) with macropores ranging from 50 to 340 nm (average for meso- and macropores = 82.7 ± 57.7 nm). Platinum ligaments range from 13 to 60 nm in diameter (average = 29.7 ± 9.0 nm), with macropores from 50 to 470 nm (average for meso- and macropores = 98.9 ± 83.7 nm). Based on SEM images, the pore structure for all three metals ranges from meso- to macroporous with the inclusion of very large macropores, likely due to the hydrogen gas evolution during synthesis. The possible effect from gas evolution during synthesis may account for the large standard deviation in pore sizes approaching the magnitude of average pore diameter for Au, Pd, and Pt aerogels. The color transition from red-gold to black for Au gels reduced with DMAB at 10 mM salt concentration and below correlates to structural differences observed in SEM images. Compared to gold aerogel structures with large, smooth, fused elements seen in the SEM images in Figs. 3(a) and 3(b), a string-of-beads structure forms for Au aerogels at a 10 mM salt concentration and below (Fig. S1). The smaller feature size of gold gels synthesized with lower salt concentrations corresponds to the structural features seen in palladium and platinum SEM images in Figs. 3(c)–3(f) and likely accounts for the optical color change observed in the photo images in Fig. 2.

FIG. 3.

Scanning electron micrographs of (a, b) gold, (c, d) palladium, and (e, f) platinum aerogels.

Noble metal aerogel densities were determined by measuring the freeze dried products’ volume and mass. Gold, palladium, and platinum aerogels synthesized from 0.1 M salts had densities of 0.54, 0.065, and 0.055 g/cm³ with corresponding porosities of 97.2%, 99.5%, and 99.7%, respectively. Given the low densities, the aerogels on the order of 1 cm in length are seen floating on water in Fig. 4. These aerogels were synthesized in microfuge tubes and they assumed the general diameter of the tube in its final shape but were smaller relative to the starting reaction volume. This synthesis method is more rapid than sol–gel and other techniques in terms of gel formation time within seconds to minutes. However, in the absence of capping agents and much longer coalescence time, definitive shape control is not yet a feature of this particular approach. It should be noted that gels dried at ambient temperature collapsed under capillary forces and resulted in free-standing films with a similar porous morphology to freeze dried aerogels, but with increased bulk material density.

FIG. 4.

(a) Gold, (b) palladium, and (c) platinum aerogels floating on water.

To determine the aerogel material composition, XRD was performed. XRD spectra of the noble metal aerogels are shown for a 2θ range of 20°–70° in Fig. 5. XRD spectra for gold, palladium, and platinum aerogels synthesized from 0.1 M salt precursors were matched to Joint Committee on Powder Diffraction Standard (JCPDS) patterns 00-004-0784, 03-065-6174, and 03-065-2868, respectively. No oxide or other material phase peaks were observed. Gold aerogel peaks were indexed to 38.185°, 44.393°, and 64.578°. Palladium peaks were indexed to 40.145°, 46.695°, and 68.176°. Platinum peaks were indexed to 39.755°, 46.234°, and 67.454°. All aerogels had XRD peaks corresponding to 100, 200, and 220 Miller indices and had a $Fm\bar{3}m$ space group for a face centered cubic (FCC) atom arrangement. Crystallite size determined using full width half-max fitting of the 100 peaks indicated crystallite sizes of 43.6, 45.1, and 51.9 nm for Au, Pd, and Pt, respectively. These crystallite sizes generally correspond to the range of ligamentous element thicknesses determined from SEM image analysis.

FIG. 5.

XRD spectra for platinum (top), palladium (middle), and gold (bottom) aerogels.

Nitrogen physisorption was performed on Au, Pd, and Pt aerogels. Adsorption–desorption isotherms shown in Figs. 6(a), 6(c), and 6(e) are type IV according to IUPAC standards, indicative of mesoporous materials (2–50 nm). The monolayer-multilayer adsorption at the initial portion of the isotherms and the lack of a plateau at high pressures suggest the presence of macroporous content in the aerogels, which correlates with pore sizes determined from SEM images. This multi-scale porosity is similar to noble metal aerogels formed using sol–gel processes. ^{Reference Walsh, Arcelli, Ikoma, Tanaka and Mann22} The H3 hysteresis loops are similar for Pd and Pt aerogels in Figs. 6(c) and 6(e). H3 loops are often associated with particle aggregates which also correlate with the fused particle structures seen in the SEM images in Fig. 3. The gold aerogel hysteresis loop in Fig. 6(a) displays a combination of H2 and H3 loop characteristics. H2 hysteresis arises when the pore structure is not well-defined, and with pore blocking for a wide distribution of pore neck widths. ^{Reference Sing, Everett, Haul, Moscou, Pierotti, Rouquerol and Siemieniewska42} The combination of H2 and H3 hysteresis indicates the influence of open network effects in the Au aerogels where the network of predominantly macropores as seen in SEM imagery does not completely fill with pore condensate, whereas network effects seem less dominant in Pd and Pt gels. ^{Reference Thommes, Kaneko, Neimark, Oliver, Rodriguez-Reinoso, Rouquerol and Sing44} These hysteresis loops appear consistent with the pore structures seen in SEM imagery in Figs. 3(a) and 3(b) in that Au aerogel samples are comprised of large, smooth, fused particles with a relatively open internal pore network, while the Pt and Pd aerogels appear more similar to agglomerated powders with randomly arranged strands of particles.

FIG. 6.

Nitrogen adsorption–desorption isotherms and pore size distribution with cumulative pore volume for (a, b) gold, (c, d) palladium, and (e, f) platinum aerogels.

The BJH model for desorption curves was used to calculate pore size distribution. The cumulative pore volume (cm³/g) and pore size distribution (dV/dd) are shown for Au, Pd, and Pt in Figs. 6(b), 6(d), and 6(f), respectively. The pore size distributions indicate predominantly mesoporous length scales between 2 and 50 nm. The relatively small macroporous content seen is attributed to the open pore network not amenable to pore filling, especially in the case of gold. Palladium and platinum do exhibit slightly greater macroporous content, likely due to the smaller pore structure compared to Au aerogels as seen in SEM imagery. A multi-point BET model was applied to the adsorption isotherms to determine the specific surface areas of 3.06, 15.43, and 20.56 m²/g for Au, Pd, and Pt aerogels. These values are in the same range, although slightly lower than previous reports for noble metal aerogels synthesized using sol–gels from preformed nanoparticles. ^{Reference Ding, Chen and Erlebacher28} To examine the correlation of the BET determined specific surface area to the nanostructures observed in SEM imagery, the equivalent diameters were determined. The corresponding equivalent diameters are 85.4, 33.1, and 13.6 nm for Au, Pd, and Pt. These idealized structure diameters generally correlate with the feature size trend of average ligament diameters determined with SEM image analysis, with Au having the largest, and Pt having the smallest features.

With a relatively high specific surface area indicated by nitrogen physisorption measurements, the solvent accessible surface area was assessed with EIS and CV to determine the feasibility of using the gels for catalytic, energy storage and conversion, and sensor applications. Gels were tested in a frequency range of 1 MHz – 1 mHz, with each representative EIS spectra shown in Fig. 7(a). Au, Pd, and Pt gels exhibit good capacitive behavior at the low frequency range, with Pd and Pt exhibiting a slightly greater non-ideal capacitive character than Au as seen in the near vertical lines at high imaginary components of impedance. The high frequency response for the gels is shown in the inset of Fig. 7(a). The depressed semicircle with slight neck before transitioning to predominantly capacitive behavior at lower frequencies is observed in some porous materials. ^{Reference Barrett, Joyner and Halenda38,Reference Liu, Zhang, Zhan, Long, Qiao, Yang and Ling45} The difference in the equivalent series resistance seen in the real impedance component intercepts is normally associated with solution ion resistance, intrinsic material resistance, and electrode-sample contact resistance. ^{Reference Gamby, Taberna, Simon, Fauvarque and Chesneau46} Given that the supporting electrolyte is the same for all three metal gels, the difference in equivalent series resistances is attributed to a variation in contact resistance between the gel samples and the platinum wire working electrode and possible differences in intrinsic material resistance. At high frequency, the semicircle diameter is normally associated with charge transfer resistance or the internal material resistance in the case of capacitors in the absence of faradaic reactions. ^{Reference Farma, Deraman, Awitdrus, Talib, Omar, Manjunatha, Ishak, Basri and Dolah47} The semicircle diameters are estimated to be 3.88, 7.15, and 2.07 Ω for the Au, Pd, and Pt gels.

FIG. 7.

(a) Electrochemical impedance spectra for gold, palladium, and platinum gels performed in 0.5 M KCl versus Ag/AgCl reference electrode. [inset of (a)] High frequency EIS spectra from (a). (b) Specific capacitance (C _sp) for gels calculated from low frequency EIS spectra in (a). [inset of (b)] Low frequency C _sp values. (c) RLC fitted model for EIS spectra.

A number of studies have examined the relationship of electrochemical impedance spectra with the pore structure of porous materials. ^{Reference Barrett, Joyner and Halenda38,Reference Song, Jung, Lee and Dao48} TLMs for electrochemical response were applied by de Levie for porous electrode systems. ^{Reference Brunauer, Emmett and Teller37} In general, transmission line behavior results in a 45° phase angle at high to medium frequencies before turning more vertical at the knee of the curve with increasing capacitive behavior at lower frequencies. Keiser, Beccu, and Gutjahr subsequently demonstrated that the average pore structure could be estimated for porous electrodes and applied specific pore geometries to the TLM which resulted in high frequency semi-circles for certain pore geometries. ^{Reference Barrett, Joyner and Halenda38} Given that at high frequencies ions do not penetrate as deeply into pores as at lower frequencies and that many porous materials have a distribution of pore sizes, Song further extended the TLM to account for both internal pore resistance and pore size dispersion. ^{Reference Gamby, Taberna, Simon, Fauvarque and Chesneau46} Bisquert developed a restricted diffusion model to account for mass transport diffusion resistance, normally accounted for by a Warburg element for semi-infinite diffusion, where the pore size imposes constraints on the development of the double layer. ^{Reference Bisquert49,Reference Bisquert50}

Using the TLM with restricted diffusion elements, an equivalent circuit to model the ion diffusion through the noble metal aerogel network pores was fit to each representative EIS spectra. The measured EIS spectra in Fig. 7(a) for Au, Pd, and Pt aerogels were fit with the equivalent circuit model shown in Fig. 7(c). Fitted data are plotted in Fig. S2. The TLM represents the movement of ions through the network of macro-, meso-, and micropores by restricted diffusion. This model is based on a modified Randles circuit consisting of parallel and serially connected resistors (R), capacitors (C) or constant phase element (CPE), and restrictive diffusion elements (Z _diff) through the non-homogenous porous network within the aerogel. ^{Reference Bisquert, Garcia-Belmonte, Bueno, Longo and Bulhoes51} A CPE is incorporated into the model to account for non-ideal capacitive behavior observed in porous networks attributed to pore size distribution. ^{Reference Gamby, Taberna, Simon, Fauvarque and Chesneau46} The element R1 represents the high frequency resistance of the electrolyte at the interface of the aerogel. The parallel combination of CPE or C elements with serially connected R and Z _diff represents the double layer capacitance, charge transfer and intrinsic material resistance, and restricted diffusion of ions through the multiple layers of hierarchically distributed pores through the internal macro-, meso-, and microporous network. A number of studies modeling cylindrical pore networks result in complex plane plots with a 45° line at high frequency turning to vertical at lower frequencies. For other non-cylindrical arbitrary shaped pores, models suggest semi-circles at high frequency. ^{Reference Barrett, Joyner and Halenda38} The shape of Au, Pd, and Pt gel EIS spectra conform to the modeled complex plane impedance plots from Keiser, Beccu, and Gutjahr for large, pear-shaped pores with small openings. This large pore structure is similar to that seen in SEM images for Au aerogels, as well as the H2 physisorption hysteresis in the Au aerogels attributed to pore blocking with a wide distribution of neck widths. ^{Reference Sing, Everett, Haul, Moscou, Pierotti, Rouquerol and Siemieniewska42}

Specific capacitances were calculated from the EIS spectra using the following equation:

where f is the frequency, Z″ is the imaginary component of impedance, and m is the mass of the sample. Specific capacitance versus frequency is plotted in the inset of Fig. 7(b) and used to extrapolate the specific capacitance of the gels from the lowest frequencies. From these data, specific capacitances of 2.18, 4.13, and 4.20 F/g are estimated for Au, Pd, and Pt gels. As seen in the inset of Fig. 7(b), as scan frequencies decrease and approach 1 mHz, only the gold gel indicates a leveling off in specific capacitance with the decrease in scan rate, thus fully assessing the gel’s capacitance. Palladium and platinum gel capacitances do not exhibit a plateau approaching 1 mHz suggesting even greater capacitance at lower scan rates. Complete access to the capacitance of the smallest meso- and micropores is limited to all but very low frequencies, below 1 mHz for Pd and Pt gels, whereas the larger macroporous structure in the gold gels allows more complete surface access for capacitance within pores down to 1 mHz scan rates. To estimate the specific surface area of the gels from electrochemical measurements, a nominal range of capacity per area from 20 to 40 µF/cm² have been used. ^{Reference Lukaszewski, Sosko and Czerwinski52} The capacity per area of a metal is dependent on electrode potential, electrolyte type and concentration, and electrode surface structure. Therefore, an average capacitance per area of 30 µF/cm² was used as a nominal value for metal surface capacitance. ^{Reference Kornyshev and Irbakh53} Specific surface areas from EIS data are estimated to be 7.27, 13.77, and 14.00 m²/g for Au, Pd, and Pt. These specific surface area estimates from EIS correlate with the trend in surface area from gas physisorption BET measurements from Fig. 6 with gold having the lowest and palladium having the highest estimated values.

CV was performed using the same electrochemical cell configuration as for EIS. Voltammogram scan rates of 50, 25, 10, 5, and 1 mV/s are shown in Figs. 8(a)–8(c) for Au, Pd, and Pt. All gels display double layer capacitance behavior across the tested voltage windows as indicated by the characteristic rectangular shaped curves. Pd and Pt gels exhibit slight hysteresis at the voltage range limits likely attributable to hydrogen adsorption–desorption just below 0.1 V and metal oxidation-reduction just above 0.4 V (versus Ag/AgCl). The specific capacitance can be calculated from the CV curves using the following equation:

where i and v are the current and potential in the CV test (A and V), µ is the scan rate (V/s), m is the mass of active materials (g), and ΔV is the potential window of discharge (versus Ag/AgCl). Specific capacitance (C _sp) values for Au, Pd, and Pt gels calculated from the 1 mV/s scan rate are 2.67, 7.99, and 5.12 F/g. Estimating the specific surface area using the nominal value of 30 µF/cm², the same as for EIS capacitances, yields 8.90, 26.63, and 17.07 m²/g. A comparison of specific surface areas estimated from EIS and CV capacitances and BET measurements is shown in Table II. In general, there is a correlation between specific surface area estimates determined from the three experimental techniques for each metal aerogel. Specific capacitances determined with EIS and CV are in close agreement in the case of gold with 2.18 and 2.67 F/g, respectively. Palladium gel specific capacitances from EIS and CV are 4.13 and 7.99 F/g with the CV determined value nearly twice as large. The EIS and CV determined specific capacitances are also close for Pt gels with 4.20 and 5.12 F/g, respectively. The specific surface areas estimated from EIS and CV measurements using an average metal capacitance per area value of 30 µF/cm² indicate a reasonable correlation with BET determined specific surface areas. BET, EIS, and CV determined specific surface areas were 3.06, 7.27, and 8.90 m²/g for Au; 15.43, 13.77, and 26.63 m²/g for Pd; and 20.56, 14.00, and 17.07 m²/g for Pt. These specific surface area values are close to or less than a factor 2 of each other. Given the range of metal capacitances in general of 20–40 µF/cm², the specific surface areas estimated from electrochemical measurements are sensitive to the value selected. Using the limits of the metal capacitance per area range, the electrochemically determined specific surface areas remain within a factor of 2–4 from BET surface area measurements.

FIG. 8.

Cyclic voltammetry at scan rates of 50, 25, 10, 5, and 1 mV/s for (a) gold, (b) palladium, and (c) platinum gels. Voltage windows are (a) −0.2 to 0.2 V and (b, c) 0.1–0.4 V (versus Ag/AgCl).

TABLE II.

Aerogel specific surface areas estimated from EIS and CV capacitances and BET.