Article www.acsanm.org
Cite This: ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Ambient Oxidation of Ultrasmall Platinum Nanoparticles on Microporous Carbon Catalyst Supports Ritubarna Banerjee,† Donna A. Chen,‡ Stavros Karakalos,† Marie-Laure C. Piedboeuf,§ Nathalie Job,§ and John R. Regalbuto*,† †
Department of Chemical Engineering and ‡Department of Chemistry and Biochemistry, University of South Carolina, Columbia, South Carolina 29208, United States § Department of Chemical Engineering − Nanomaterials, Catalysis, Electrochemistry, University of Liège, Sart-Tilman, B-4000 Liège, Belgium
ACS Appl. Nano Mater. Downloaded from pubs.acs.org by 46.148.115.96 on 09/27/18. For personal use only.
S Supporting Information *
ABSTRACT: Ultrasmall platinum nanoparticles on carbon catalyst supports undergo oxidation in ambient air. A systematic survey of carbon supports with controlled pore structures was conducted to isolate the effect of pore structure on particle size and resistance to oxidation. It was found that carbon supports with higher microporosity gave larger metal particles at high metal weight loadings, whereas the particle sizes remained independent of pore size at low weight loadings. It is the large particles that are most stable to oxidation; particles of average size less than 1.4 nm are more than 80% oxidized, while particles greater than 1.8 nm are less than 30% oxidized. The effect of micropores on the stabilization of Pt to oxidation is thus indirect; the larger particles formed in the micropores at high metal loadings do not oxidize. At lower loadings, the pore environment plays no role in stabilizing the Pt nanoparticles to oxidation. KEYWORDS: micropores, ambient oxidation, carbon supports, platinum nanoparticles, particle stabilization
■
INTRODUCTION Porous carbon materials have long been used as supports in heterogeneous catalysis,1−4 and perhaps the most common catalytic metal supported on carbon is platinum. As a noble metal, platinum is normally considered to exist in catalysts in its metallic form. In some reactions, however, Pt oxides are thought to be more active than the metal; this has been cited for carbon monoxide oxidation, nitric oxide reduction, hydrocarbon hydrogenation, and methanol oxidation.5−9 An oxide layer is thought to form on the surface of 3−5 nm Pt clusters of a car catalyst under oxygen-rich conditions where the Pt3O4 phase has been identified as the active phase during carbon monoxide oxidation and nitric oxide reduction reactions, while the PtO2 phase is inactive.10 The propensity of supported Pt to oxidize appears to be a function of nanoparticle size and type of support.11−14 Oxidation is more prevalent for smaller nanoparticles (3 nm), which behave like bulk metallic platinum.13,14 In situ synchrotron studies by Miller et al.11 have shown that Pt nanoparticles are oxidized in ambient air and that the extent of oxidation differed for carbon versus alumina supports and was also size dependent; virtually all of the Pt existed as oxide for an average particle size of 1.2 nm on Al2O3, 67% on 2.1 nm © XXXX American Chemical Society
particles, and 33% on particle of average size 3.3 nm. On carbon nanotubes, 79% was oxidized for 1.6 nm average size. Other reports have detected Pt nanoparticle oxidation in ambient conditions using XPS.12,13 Our own recent study employed a high sensitivity X-ray diffractometer to identify the nanoparticulate oxide as Pt3O4, and scanning transmission electron microscopy (STEM) fast Fourier analysis of atomically resolved nanoparticles to accurately frame the oxidation as a function of size on an activated carbon support: below 1.5 nm, only the oxide phase exists; between 1.6 and 2.7 nm, nanoparticles are metal cores with oxide shells, and above 2.8 nm, are pure metal.15 Given that the extent of oxidation of Pt determines its catalytic activity, the oxidation is a function of particle size and support, and carbon pore structures can be finely tuned, we undertook a systematic survey of carbon supports with controlled pore textures to isolate the effect of carbon type and pore texture on particle size and degree of oxidation. A particular intention was to determine whether support Received: September 4, 2018 Accepted: September 13, 2018 Published: September 13, 2018 A
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials Table 1. Summary of Supports, Precursors, and Catalysts support
surface area (m2/g)
abbreviation
PZC
precursor
SEA pH
catalyst
oxidized VXC72 Timrex DarcoG60 carbon xerogels carbon xerogels
170 280 659 679 1162
C-170 C-280 C-659 CX-679 CX-1162
2.4 4 8 10 10.1
[Pt(NH3)4]2+ (PTA) [Pt(NH3)4]2+ (PTA) [Pt(Cl6)]2− (PHC) [Pt(Cl6)]2− (PHC) [Pt(Cl6)]2− (PHC)
carbon xerogels
1723
CX-1723
9.5
carbon xerogels
2234
CX-2234
KJ600 BP2000
1189 1474
C-1189 C-1474
12 12 2 2.8 2.8 2.8 2.9 2.9 2.9 2.9 2.9 2.9 2.5 2.5
2.7Pt/C-170 2.4Pt/C-280 10Pt/C-659 6Pt/CX-679 6.4Pt/CX-1162 10Pt/CX-1162 12.8Pt/CX-1162 17.3Pt/CX-1723 6.5Pt/CX-1723 6.5Pt CX-2234 10.3Pt CX-2234 16.4Pt/CX-2234 27Pt/C-1189 30Pt/C-1474
10
9.4 9.5
[Pt(Cl6)]2− (PHC)
[Pt(Cl6)]2− (PHC) [Pt(Cl6)]2− (PHC)
connected to a nitrogen and a carbon dioxide supply via a three-way valve, which allowed it to switch between both atmospheres. About 5 g of the powdery dry polymer gel was introduced in the tubular oven to first undergo a pyrolysis at 900 °C for 2 h under nitrogen as described above; subsequently, the atmosphere in the tube was modified by switching from the nitrogen supply to the CO2 bottle (Air Liquide N27), still at 900 °C (i.e., without cooling) and by keeping the same flow rate of 0.004 mol min−1. This temperature was maintained for different durations necessary for the activation process (3, 8, or 16 h). Finally, the cooling step was done under nitrogen atmosphere. The three carbon samples, reaching specific surface areas of 1162, 1723, and 2234 m2/g after treatment, obtained are denoted CX-1162, CX-1723, and CX-2234, respectively. The BET surface areas and pore size distributions were obtained using nitrogen adsorption−desorption isotherms at 77 K with a Micromeritics 2020 ASAP instrument. It has been mentioned in the literature that measurements of micropore volume by nitrogen adsorption may give erroneous isotherms due to the low temperature and pressures at which the adsorption takes place and the quadrupole moment of the diatomic nitrogen molecule that produces specific interactions at the gas−solid interface.19,20 For the carbons C-170, C280, and C-659, argon adsorption−desorption isotherms at 77 K were conducted to determine micropore volumes. For the carbon xerogels, the micropore and macro-meso pore size distributions and corresponding pore volumes were obtained from nitrogen desorption18 and mercury porosimetry techniques, respectively.21 The micropore volumes for the xerogels were calculated using the Dubinin−Radushkevich equation, whereas for all the other supports, the Harkin’s Jura thickness equation was used.22 Preparation of Carbon Supported Platinum Nanoparticles. Table 1 shows a summary of the carbon supported catalysts prepared by strong electrostatic adsorption (SEA).23−25 On the basis of the SEA protocol, a cationic precursor, tetraammineplatinum(II) hydroxide ([Pt(NH3)4]OH2, 99.999%) was chosen for the low PZC supports, while an anionic precursor, chloroplatinic acid (H2(PtCl6), 99.9%) was used for the high PZC supports. The required concentrations of the precursor solutions were then contacted with the supports for 1 h followed by filtration and subsequent drying in ambient air overnight and drying in an oven at 120 °C for 16 h. The resultant weight loadings were determined from the uptake of platinum ions in an inductively coupled plasma (ICP, PerkinElmer Optima 2000DV) and are tabulated in Table 1. The catalysts are denoted by their platinum weight loadings followed by support type (C, carbon and CX, carbon xerogel), followed by their surface area. For example, 2.7Pt/C-170 denotes 2.7 wt % Pt on 170 m2/g oxidized VXC72 carbon support. The dry impregnated samples were then reduced in a flowing 10% H2 (balance He) for 1 h with a ramp rate of 2.5 °C/min at 200 °C. Another set of 6% metal loading catalysts was prepared for all the xerogels and a 10% metal loading for the xerogels CX-1162 and CX-2234.
microporosity plays a role in stabilizing the particles against oxidation.
■
[Pt(Cl6)]2− (PHC)
EXPERIMENTAL SECTION
Materials. The supports and precursors used and the catalysts prepared along with their as-determined BET surface areas and point of zero charges (PZCs) are summarized in Table 1. Different types of high purity carbons were gathered: carbon blacks (VXC72, KJ600), activated carbons (DaroG60, BP2000), graphitic carbon (Timrex HSAG300), and carbon xerogels. Oxidized VXC72 (PZC = 2, BET area = 170 m2/g) was prepared by boiling VXC72 obtained from Cabot Corporation in nitric acid (>70%) at 90 °C for 3 h. The mixture was cooled to room temperature, filtered and washed with deionized water until the pH of the washing solution reached that of deionized water and was dried overnight at room temperature. Timrex HSAG300 (BET area = 236 m2/g) and Darco G60 (BET area = 659 m2/g) were obtained from Timcal and Cabot Corporation, respectively. The carbon xerogels were synthesized at the University of Liège, Belgium. The pristine carbon xerogel was prepared following the process described by Job et al.16 In brief, a 35 wt % resorcinol (Merck) aqueous solution was prepared and the pH was set by addition of powdery sodium carbonate. A 37 wt % solution of formaldehyde stabilized by 10−15 wt % methanol (Sigma-Aldrich) was then added in a molar ratio resorcinol/formaldehyde equal to 1:2 and the mixture was magnetically stirred for 1 h. The molar dilution ratio D = water/reactants was fixed at 5.7. The molar ratio between resorcinol and carbonate (R/C) was chosen equal to 2000. The obtained homogeneous gel precursor solution was then sealed in a 250 mL autoclaveable glass flask and aged for 3 days at 85 °C. After gelation and aging, drying was performed by placing the gel in an oven under vacuum (2 kPa) at 60 °C for 12 h and then at 150 °C for 30 h. The dry xerogel was then ground using a Fritsch planetary mill (Mono Mill P6) in agate jars with 1 cm diameter agate balls following the procedure described in a previous work to obtain homogeneous and reproducible particle sizes.17 The obtained powder was put in a quartz basket and introduced in a ceramic tubular oven to undergo a pyrolysis step at 900 °C, leading to a carbon xerogel. The pyrolysis was performed under nitrogen (Air Liquide Alphagaz 1, flow rate = 0.004 mol min−1) with the following temperature profile:1 ramp at 1.7 °C/min to 150 °C and hold for 15 min;2 ramp at 5 °C/min to 400 °C and hold for 60 min;3 ramp at 5 °C/min to 900 °C and hold for 120 min; and4 natural cooling to room temperature. This carbon xerogel pyrolyzed at 900 °C without any other treatment displays a specific surface area of 679 m2/g and is hereafter denoted CX-679. Physical Activation with CO2. The specific surface area of the carbon xerogel was increased by physical activation with CO2. This process is known to increase the micropore volume without modifying the meso/macroporosity.18 Activations were performed using CO2 in the same tubular oven as used for the pyrolysis. The oven was B
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials Catalyst Characterization. Powder X-ray diffraction (XRD) and scanning transmission electron microscopy (STEM) measurements were used to characterize the set of samples. XRD measurements were made using a Rigaku Miniflex-II equipped with D/teX Ultra silicon strip detector that can detect nanoparticles down to 0.8 nm.15,26 Diffraction patterns were recorded over a range of 20°−80° 2θ using Cu−Kα radiation (k = 1.5406 Å). XRD patterns were obtained for all the catalysts and compared to reference spectra using PDXL 2.0 (Rigaku Corporation) software. The carbon supports were background subtracted from the patterns and the metal and oxide peaks were deconvuluted as is previous work15 using Fityk 0.9.8 version Software.27 Z contrast images were obtained using an aberration-corrected JEOL 2100F STEM equipped with a 200 kV field emission gun and a double tilt holder for tilting the sample across a range of angles (±20°). High angle annular dark-field (HAADF) STEM images were acquired on a Fischione Model 3000 HAADF detector with a camera length such that the inner cutoff angle of the detector was 50 mrad.28 Sample preparation involved suspending the catalyst in isopropanol and depositing a drop of the suspension onto a holey carbon film attached to a Cu TEM grid. The images were recorded using Digital Micrograph software and particle size distributions were obtained by counting about 1000 particles on each sample. Volume average sizes (DV) were determined to compare with the sizes obtained from XRD. The volume average diameter is DV = ∑ nidi4/∑ nidi3 where ni is the number of particles with diameter di.29 XPS was used to probe the elemental composition and the chemical state of the air exposed platinum nanoparticles. XPS measurements were conducted using a Kratos AXIS Ultra DLD XPS system equipped with a monochromatic Al K source, a hemispherical analyzer, charge neutralizer, catalysis cell, and a load lock chamber for rapid introduction of samples without breaking vacuum.30 The base pressure in the XPS analysis chamber was 2 × 10−9 Torr before sample introduction and ≤2 × 10−8 Torr during experiments. Binding energies were set according to the C 1s peak, which was fixed at 284.8 eV. The fits of the XPS spectra were done after the subtraction of a Shirley background using an asymmetric line-shape-function of the XPS Peak 4.1 software. The obvious asymmetry of the Pt 4f peak was fit with a mixed Gaussian−Lorentzian function. The oxidation of the metal to the oxide was accompanied by a loss in symmetry of the Pt 4f peaks.31 The signal was deconvoluted using three doublets corresponding to metallic Pt (Pt 4f7/2 ≈ 71.1 eV), PtO (Pt 4f7/2 ≈ 72.3 eV), and PtO2 (Pt 4f7/2 ≈ 73.8 eV) .The maximum width (fwhm) of each component was held constant at 1.2 eV for Pt0, 1.7 eV for PtO, and 1.9 eV for PtO2.14 The platinum metal and oxide contents were determined from the area under the Pt, PtO, and PtO2 peaks, respectively. To evaluate the effect of increasing air exposure on the platinum nanoparticles, the xerogel supported catalysts were probed by XPS after the following treatments: (a) reduction in hydrogen for 1 h at 200 °C, (b) short time air exposure for 8−10 h, (c) long time air exposure for 1 week. In addition, the 6Pt/CX-679 catalyst was probed by XPS (a) after reduction in hydrogen for 1 h at 200 °C, (b) after short time air exposure, (c) after long time air exposure, and (d) after oxidation in oxygen for 1 h at 350 °C. The samples could be directly transferred from the XPS analysis chamber to the catalysis cell without exposure to air. In the catalysis cell, samples were exposed to a flow of pure gases (O2 and H2) at the above-mentioned conditions and XPS signal was collected after it was cooled down to room temperature.
Figure 1. (a) Pore size distributions of C-170, C-280, and C-659 with argon desorption. (b) Micropore size distributions of carbon xerogels with nitrogen desorption. (c) Meso and macropore size distributions of xerogels using mercury porosimetry. (d) Micropore volumes for carbons.
■
composed of micro-, meso-, and macropores. The micropore volumes of the carbons are plotted in increasing order in Figure 1d with C-280 having the lowest and CX-2234 having the highest micropore volumes. Two additional high surface area carbons, C-1189 (KJ600) and C-1474(BP2000), known to be microporous from the literature, have also been added to this figure for a more detailed study.32,33 Carbon xerogels are basically made up of microporous carbon nodules bonded together to form a 3D mesoporous material; the mesopores correspond to voids between the nodules.16 To determine the changes in pore structure of the
RESULTS AND DISCUSSION Microporosity of Supports. The pore size distributions and micropore volumes of the carbon supports are given in Figure 1. The corresponding adsorption isotherms have been provided in Figure S1 of the Supporting Information. Figure 1a shows that C-170, C-280, and C-659 are predominantly microporous (shown over a broader size range in Figure S2 in the Supporting Information) with a very small amount of mesopores, whereas Figure 1b and c show that the xerogels are C
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials xerogel nodules during the activation process, Hg porosimetry and N2 adsorption were used to obtain the distribution of microporous versus exterior surface area. Table S1 in the Supporting Information shows that the activation process leads to an increase of the total surface Sext (from 155 to 483 m2/g) due to the formation of micropores inside the nodules only. The external surface per volume of sample, SVext remains essentially constant (93 ± 3 m2/cm3). For the CX-2234 support, it was not possible to discriminate between the mercury that enters the pores and mercury that goes between the powder particles. Strong Electrostatic Adsorption of Pt onto the Carbon Supports. In earlier work,23−25 it has been shown that the Pt concentration of 180 ppm corresponds to about a 10% excess of one monolayer of Pt (1.6 μmol/m2) for the employed surface loading of 500 m2/L, while a PTA concentration of 312 ppm corresponds to about a 10% excess of one monolayer of Pt (0.84 μmol/m2) for a surface loading of 2000 m2/L. In the present work, 200 ppm of CPA and PTA solutions were used for the SEA with a surface loading of 500 m2/L for the CPA and 1000 m2/L for the PTA. All the carbons and the xerogels were found to have expected uptakes in the range of 0.7−0.8 μmol/m2 for the low PZC carbons and about 1.2−1.3 μmol/m2 for the high PZC xerogels.25 For clarity and brevity, results of three representative catalysts in the xerogel series, that is, 6Pt/CX-679, 12.8Pt/ CX-1162, and 17.3Pt/CX-1723, have been shown in the main paper. Characterization of all others, that is, 2.7Pt/C-170, 2.4Pt/C-280, 10Pt/C-659, 6.4Pt/1162, 10Pt/CX-1162, 6.5Pt/ CX-1723, 6.5Pt/CX-2234, 10.3Pt/CX-2234, and 16.4Pt/CX2234 can be found in the Supporting Information. Detection of Ambient Oxidation in Platinum Nanoparticles. After drying and reduction, all the samples were air exposed for a minimum of 2 weeks and then characterized with STEM (Figures 2 and S3) and XRD (Figures 3 and S4). From STEM images like those of Figure 2, particle size distributions were derived by counting at least 1000 particles; the summary of volume averaged particle size (for purposes of comparing with XRD data) and the standard deviations of all samples are given in Table 2. The smallest average particle size is 1.4 nm and largest is 3.9 nm. Tight size distributions were observed for all catalysts, typical of SEA preparations.23−25 XRD patterns for CX-679, CX-1162, and CX-1723 are given in Figure 3, which includes background subtracted and deconvoluted patterns in the insets. Fcc Pt peaks appear at 2θ of 39.76°, 46.24°, and 67.45° for (111), (200), and (220) reflections, respectively, whereas the Pt3O4 peaks are seen at 2θ values of 35.93°, 39.49°, 57.08°, and 59.64° for the (210), (211), (222), and (320) reflections. The support-subtracted signal after deconvolution could only be fit assuming reflections for a Pt3O4 phase. The absence of the second most intense Pt3O4 peak, the (110) at 22.5° 2θ, (48% of (210) intensity for a random, large sample) might be explained by the sampling depth problem discussed in the earlier work.15 As shown in Figure S5, the intensity of the support feature in the 20−30° 2θ range changes significantly and nonlinearly with Pt loading and this prevents the broad, low intensity (110) peak from being isolated in the subtracted patterns and leaves a shoulder in the 20−30° range in the deconvolutions.15 Metallic Pt is predominant in the deconvoluted patterns of the three samples with largest metal Pt particle size, the 17.3Pt/CX-1723 (Figure 3c), 10Pt/C-659 (Figure S4c), and
Figure 2. STEM images with particle size distribution histo-grams in the inset for (a) 6Pt/CX-679, (b) 12.8Pt/CX-1162, and (c) 17.3Pt/ CX-1723.
16.4Pt/CX-2234 (Figure S4i). To best fit the data, the fcc peaks in these samples were deconvoluted to reveal a bimodal size distribution. These are confirmed by the STEM particle size distributions seen in Figures 2c, S3c, and S3i, respectively. XRD size estimates for the Pt and Pt3O4 phases from the Scherrer equation are listed in Table 2. The sum of the Pt and Pt oxide sizes from XRD is consistently above the volume average STEM size. The explanation is that the Pt particles are not all Pt cores with Pt3O4 skins, in which case the sum of the D
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials
seen that all the catalysts have nanoparticles, which traverse this sensitive size domain; therefore, they would all contain a mixture of metal and oxide phases. XPS Analysis of Platinum Oxidation. XPS was used to directly determine the oxidation state of the platinum in the nanoparticles, which are all generally small enough that virtually all atoms are sampled. Hence, XPS yields the average bulk Pt oxidation state for samples such as these.12 Figure 5 shows the Pt 4f XPS peaks for the representative air exposed catalysts. The sample with the smallest particles (1.4 nm by STEM, Figure 5a) showed only a small amount of metallic Pt and significant amounts of Pt2+ and Pt4+. The medium-sized particle (1.9 nm by STEM, Figure 5b) showed much more Pt0 and less oxidized Pt, while the sample with the largest particle size (3.9 nm, Figure 5c) was mostly Pt0 with a little Pt2+ and Pt4+. This trend of increased oxidation with smaller size is fully consistent with the XRD and STEM FFT analyses. Pt 4f analyses of the remaining samples are given in Figure S6. A summary of the oxide content of all samples is given in Table 2. An additional XPS analysis was performed on the 6Pt/CX679 and on the CX-679 support without any metal to focus on the O 1s spectra with different in situ pretreatments: short time air exposed, long time air exposed, reduced, and oxidized in the XPS chamber. Here, short time air exposure refers to an 8−10 h air exposure and long time exposure refer to 1 week in ambient air. These results are shown in Figure 6. Figure 6a shows the Pt 4f peaks for the 6Pt/CX-679 sample. There is an expected positive binding energy shift in the Pt 4f peak positions with increased air exposures, the shift being maximum when it is oxidized in the catalysis cell (top spectrum of Figure 6a shown in red). A small positive binding energy shift (∼0.4 eV) is observed in the overnight air exposed sample (green) and a further shift in the longer air exposure (purple). Figure 7b constitutes a detailed study of the corresponding oxygen, O 1s peaks during this sequence. Figure 6b shows the O 1s peak for the different treatments as well as the constituent O 1s peaks after deconvolution. The O 1s peaks shifts in Figure 6b mirror the Pt 4f peak shifts in Figure 6a. Right after reduction, the oxygen content at 533 eV in the carbon depicts some of the nascent lattice oxygen and the oxygen content is seen to qualitatively increases with air exposure. In the short and long time air exposed samples, the O 1s spectrum contains two types of oxygen in the Pt/CX-679 catalyst, one is the oxygen peak at 533 eV that is in the carbon support and the second type, which is equivalent to that exclusively present in the 250 °C calcined sample at 531 eV. It was noted that after the calcination treatment in the pretreatment chamber, most of this sample had combusted away; the remaining mostly inorganic material giving rise to the single oxygen peak is undoubtedly Pt oxide. Thus, the second oxygen peak in the airexposed samples is attributed to bulk Pt oxide. Another control experiment was conducted on the CX-679 support without any metal shown in Figure 6c. A comparison between the oxygen spectrum in Figure 6b and c clearly shows that there is no increase in the oxygen signal on the bare support (green and purple lines in Figure 6c) in ambient with time, in contrast to the case with the Pt/CX-679 catalyst (green and purple lines in Figure 6b), which again suggests that the increased oxygen signal seen in Figure 6b after ambient air exposures arises from the oxidation of the platinum nanoparticles.
Figure 3. XRD Profiles with deconvoluted patterns in the inset for (a) 6Pt/CX-679, (b) 12.8Pt/CX-1162, and (c) 17.3Pt/CX-1723.
XRD estimates would agree with the STEM size. The explanation, as observed previously with a limited set of carbons,15 is that the smallest particles are pure Pt oxide. This is borne out by data represented in Figure 4, where fast Fourier transform analysis has been performed on individual atomically resolved particles. Previous FFT analysis15 revealed 0.23 and 0.25 nm d-spacings corresponding to the most intense reflections from the fcc Pt (111) of Pt and bcc (210) planes of Pt3O4, respectively. The FFT analyses of the particles shown in Figure 4 are consistent with those results; the smallest particle (1.5 nm) is pure oxide, the largest (2.7 nm) is pure metal, and the middle size (2.3 nm) contains both phases. On the basis of the STEM histograms in Figures 2 and S3, it is E
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials
Table 2. Summary of Micropore Volumes, XRD Sizes, STEM Volume Average Sizes, and Oxide Content of the Catalysts XRD sizes (nm) catalyst
total micropore volume (cm3/g)
Pt
Pt3O4
Pt+Pt3O4
2.7Pt/C-170 2.4Pt/C-280 10Pt/C-659 6Pt/CX-679 6.4Pt/CX-1162 10Pt/CX-1162 12.8Pt/CX-1162 17.3Pt/CX-1723 6.5Pt/CX-1723 6.5Pt/CX-2234 10.3Pt/CX-2234 16.4Pt/CX-2234 27Pt/C-1189 30Pt/C-1474
0.02 0.006 0.15 0.27 0.47 0.47 0.47 0.7 0.7 0.89 0.89 0.89 0.2 0.21
0.9 1.1 1.8 0.8 1.3 2.0 1.4 3.6 1.4 1.0 2.2 1.6
0.8 1.0 0.9 1.1 0.6 0.7 1.0 1.1 0.8 1.0 0.7 0.8
1.7 2.1 2.7 1.9 1.9 2.7 2.4 4.7 2.2 2.0 2.9 2.4
STEM volume average sizes (nm)
percentage of oxide
± ± ± ± ± ± ± ± ± ± ± ±
48 73 29 87 87 44 40 32 85 86 39 33
1.5 1.6 2.2 1.4 1.5 1.7 1.9 3.9 1.5 1.4 1.8 2.8 1.8 1.7
0.3 0.2 0.4 0.5 0.4 0.4 0.4 1.5 0.4 0.5 0.4 1.0
Figure 4. HRTEM images with inset FFT patterns for 10Pt/C-659 after air exposure.
It is further noted that the peaks seen for the 250 °C calcined metal-free support (top spectrum in Figure 6c) contain peaks at 531.7 and 533.5 eV, which are significantly shifted from the single peak of the completely oxidized Pt (top spectrum of Figure 6b). This sample did not significantly combust, as there was no catalytic metal present, and the oxygen that did add into the carbon surface appeared to be distinct from the oxide on the Pt nanoparticles. The platinum oxide contents were quantified from the XPS data in Figures 5 and S6 and are tabulated in Table 2. As reflected in Figure 5, as the particle size increases, the intensity of the Pt0 peak increases relative to the Pt2+ and Pt4+ peaks. The oxide content represents the sum of the Pt2+ and Pt4+ oxide phases. The oxide content for the metal nanoparticles ranges from ∼29% to ∼87%. To test the hypothesis that micropores may retard the oxidation in the platinum nanoparticles, the micropore volumes and STEM volume average sizes were also included in the table. Figure 7 is a variety of correlations of the data in Table 2 showing the variation of nanoparticle size with metal loading (Figure 7a), micropore volume (Figure 7b), and oxide content as determined from XPS as a function of nanoparticle size (Figure 7c) and micropore volume (Figure 7d). The general trend in Figure 7a is a sharp upturn in nanoparticle size with increased metal wt %, with the exception of the two high surface area carbons with low micropore volumes, C-1189 and C-1474. The largest two particle sizes are for the highly loaded, high microporosity supports (17.3Pt/CX-1723 and 16.4Pt/ CX-2234). This suggests a relationship between high micro-
Figure 5. Deconvoluted XPS spectra with STEM volume average sizes in the inset for (a) 6Pt/CX-679, (b) 12.8Pt/CX-1162, and (c) 17.3Pt/CX-1723. The purple line depicts Pt0, blue lines show Pt+2, and red lines refer to the Pt4+ peaks.
porosity and large particle size at high metal loading. This is seen more clearly in Figure 7b, in which particle size is plotted versus micropore volume. The sizes lie in a somewhat narrow range for most of the samples with the exception, again, of the two high metal loading, high microporosity catalysts. For the three sets of xerogel samples (square, circle, and diamond markers, respectively) at the highest micropore volumes of F
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials
Figure 6. XPS analysis of carbon oxidation. (a) Pt 4f spectra from 6Pt/CX-679 catalyst after in situ treatments (reduction in hydrogen at 200 °C, followed by short and long time air exposure, followed by oxidation in oxygen at 250 °C); O 1s spectra from (b) Pt/CX-679 after in situ treatments and (c) (Pt free) CX-679 support after in situ treatments. Detailed deconvolutions are shown on the right.
0.47, 0.70, and 0.89 cm3/g, the trend of increasing particle size with increasing metal loading is especially apparent. The variation of oxide content with nanoparticle size is shown in Figure 7c. There is a simple, clear trend of increasing oxide content with decreasing particle size, independent of support. Figure 7d, in which oxide content is plotted versus micropore volume, helps refine this dependence on weight loading and micropore volume, again seen most clearly for the xerogel (CX) series. The four samples with the highest oxide content are the xerogels with low metal loading, substantially below the precursor monolayer limit. The xerogel samples synthesized at loading closer to the monolayer limit are those which have higher metal loading (Figure 7a) and higher particle size (Figure 7b). In fact, the 10Pt/C-659 sample, with larger- than-average particles of 2.2 nm, is also near the precursor monolayer limit. In view of the above discussion, it is surmised that the primary variable determining the oxide content is the metal particle size. The role of the microporosity of the carbon on metal stabilization is only indirect; precursors adsorbed into micropores at high surface density (near monolayer coverage)
yield the largest particles, and it is the large size which stabilizes the particles toward oxidation. Figure 8a illustrates the postulated effect of particle size distribution at high metal loading. At high metal loadings, the micropores are so completely filled with metal precursors that, during reduction, the precursors experience not only side-toside interactions, but also overhead interactions. This gives rise to larger particles. In macropores and mesopores on the other hand, only lateral interactions occur during reduction, leading to smaller particles. Three-dimensional interactions would not occur in micropores at low metal loadings (Figure 8b), which explains why the effect is only seen with high metal loadings in microporous materials. Large particles formed in micropores may block a significant fraction of the micropores. This was explored by measuring the surface area and pore size distribution of a microporous, high metal loading sample (17.3 Pt CX-1723) before and after metal addition (Figure S7). The decrease in surface area after metal deposition was found to be about 57%, which is much higher than the standard decrease in surface area after loading G
DOI: 10.1021/acsanm.8b01548 ACS Appl. Nano Mater. XXXX, XXX, XXX−XXX
Article
ACS Applied Nano Materials
Figure 8. Illustration of nanoparticle formation in small and large pores of carbon at high and low metal loadings.
(∼20%), and the fraction of micropores decreased significantly. This supports the explanation of Figure 8.
■
CONCLUSIONS This study demonstrates the effect of pore size on nanoparticle generation from electrostatically adsorbed catalyst precursors. At high metal loadings (at monolayer capacity of the Pt precursor), microporosity has a deleterious effect on metal dispersion in that large particles are formed. Microporosity has only an indirect effect on the stabilization of Pt particles to ambient oxidation; it is the size of the Pt particles, which stabilizes the particles toward oxidation and not the micropores themselves. At lower metal loadings, there appears to be no difference in the stability toward oxidation of Pt nanoparticles in micro-, meso-, or macropores. A clear trend has been confirmed from XPS on the degree of oxidation with particle size. For particles ≤1.4 nm, the amount of oxide was >80%, while particles ≥1.8 nm showed
