Preparation of platinum-on-carbon catalysts via hydrolytic deposition: Factors influencing the deposition and catalytic properties

Applied Catalysis A: General 449 (2012) 203–214

Contents lists available at SciVerse ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Preparation of platinum-on-carbon catalysts via hydrolytic deposition: Factors influencing the deposition and catalytic properties K.M. Kaprielova a,b , O.A. Yakovina a , I.I. Ovchinnikov a , S.V. Koscheev a , A.S. Lisitsyn a,∗ a b

Boreskov Institute of Catalysis SB RAS, Novosibirsk 630090, Russian Federation Novosibirsk State University, Novosibirsk 630090, Russian Federation

a r t i c l e

i n f o

Article history: Received 18 July 2012 Received in revised form 15 September 2012 Accepted 11 October 2012 Available online 23 October 2012 Keywords: Pt catalysts Carbon supports Catalyst preparation Hydrolytic deposition Pore-size effect

a b s t r a c t The deposition of Pt oxide from aqueous chloroplatinate solutions onto carbon supports (activated carbons, carbon blacks, carbon nanofibers) with the aid of Na2 CO3 and without adding the reductants or wetting agents was studied. It has been revealed that carbons accelerate greatly the hydrolysis of chloroplatinate, and the hydrolytic deposition can therefore interfere with the other preparation methods. Neutral pH was found optimum for heterogeneous nucleation and formation of smaller PtOx particles. High temperature (80 ◦ C) allowed depositions of 5–30 wt.% of Pt for a short time (1–3 h) but did not affect Pt dispersions (15–90% at surface concentration of Pt 0.5–30 ␮mol/m2 ). The supported particles could be further reduced in the liquid phase with minimal losses of the dispersion (Na-formate as reductant). The heterogeneous nucleation ensured a uniform distribution of Pt particles but provoked particle confinement in narrow pores. The possibility of influencing catalytic properties with the aid of the support, Pt loading and deposition conditions was demonstrated (hydrogenation of cycloalkenes as a model structure-insensitive reaction and oxidation of isopropyl alcohol as structure-sensitive one). © 2012 Elsevier B.V. All rights reserved.

1. Introduction The carbon-supported Pt catalysts demonstrate remarkable performance in a great variety of chemical processes, and hundreds of scientific papers on properties and synthesis of these catalysts are published every year. The overwhelming majority of the described syntheses are based on reductive deposition of Pt or impregnation of the support with a solution of Pt compound, followed by evaporation of the solvent and reduction in a flow of gaseous reductant [1–4]. A preliminary deposition of oxide species onto the support, being often used with other metals, is very rare in the preparation of Pt catalysts, probably due to the relative inertness of Pt(IV) compounds to hydrolysis. Several research groups have recently shown, however, that the hydrolytic deposition can be an effective way to highly dispersed Pt/C catalysts. Colloidal particles of PtO2 and mixed oxides could be obtained through hydrolysis of Pt chloride at 60–100 ◦ C in the presence of betaines [5,6] or polyvinylpyrrolidone (PVP) [7,8]. Reetz and Lopez [9] suggested that the colloidal Pt oxide could be stabilized by hydroxide ions and patented method of obtaining PtOx /C catalysts in the absence of organic stabilizers, through immobilization of preformed or in situ formed colloidal PtOx particles at high pH

∗ Corresponding author. Tel.: +7 383 3269529; fax: +7 383 3269529. E-mail address: [email protected] (A.S. Lisitsyn). 0926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apcata.2012.10.004

(Li2 CO3 as the base). Kvande et al. [10,11] attempted the deposition of Pt oxide onto carbon nanofibers (CNFs) under such conditions but found that a large amount of the metal precursor remained in the solution, and Pt loadings could be as low as 1.4 wt.% at the intended value of 20 wt.%; they achieved higher Pt loadings “with a twostep procedure where the precursor solution was initially heated without the presence of carbon”. Fang et al. proposed a monotonic increase of pH at the deposition stage and obtained a high loading of Pt oxide with a “urea-assisted” hydrolysis [12,13]; to what extent the smooth changes of pH contributed to the result is not clear, however. Another research group used a “redox-assisted” hydrolysis of Pt(IV) chloride complexes to prepare Pt/C catalysts [14–16], but the expediency of adding the reductant at the hydrolysis stage was not supported by corresponding data. The previous studies have not addressed the possible influence of the preparation variables on the state and properties of resulting PtOx particles, and the conditions were not always reported sufficiently in detail. All those studies only concerned an electrochemical application of the Pt/C catalysts obtained (fuel cells), and practically nothing is known regarding the potentialities of the method in the field of traditional catalysis. Reetz and Koch [5] supported Pt oxide colloids on Al2 O3 and tested the samples in reductive amination; per gram of Pt, the immobilized catalyst was only 5 times more active than bulk PtO2 (commercial Adams catalyst). Pan et al. [8] tested 1–3 nm PtO2 colloids (unsupported) in cyclohexene hydrogenation and obtained turnover frequencies

204

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

(TOF) 0.13 s−1 after optimal pretreatments; this is much less than should be expected for Pt catalysts under those conditions (∼7 s−1 at 30 ◦ C [17]). In the present study, the advantages and limitations of the hydrolytic deposition for preparation of both high- and low-loading Pt/C catalysts have been evaluated. Attention was focused on the role of the support and on the factors which influence has been poorly studied so far. We have tried to understand (i) what facilitates and what retards the deposition of Pt oxide; (ii) characteristic features of the catalysts being obtained by the hydrolytic method, including the Pt distribution over the support and in the porous structure, and (iii) to which extent and by what means one could govern the catalyst properties in the frameworks of the given method. The study was carried out with the conventional metal precursor (H2 PtCl6 ) and an ordinary base (Na2 CO3 ), and water served as a sole solvent (no additives). Carbons of different origin were used as the supports, and particular attention was given to activated carbons, taking into account their role in applied catalysis [18,19]. To evaluate contribution of different factors to catalytic properties, the Pt/C samples were tested in hydrogenation of cycloalkenes and in oxidation of isopropyl alcohol (IPA). These represent structureinsensitive and structure-sensitive reaction, respectively, and are well-studied and simple enough to serve as model test reactions.

2. Experimental 2.1. Reagents and catalyst supports Distilled water and chemicals of a reagent or analytical grade were used. Solutions of the metal precursor were obtained by dissolving commercial H2 PtCl6 ·6H2 O (Pt 37.66–37.82 wt.%). Concentration of Pt in the stock solutions was within 0.2–0.25 mol/L and dominant species was [PtCl6 ]2− , as evidenced by 195 Pt NMR spectra (main signal at ı 0 ppm and a small one at ı 505 ppm, characteristic of the partly hydrolyzed [PtCl5 (H2 O)]– complex). Commercially available activated carbons and carbon blacks have been used. The suppliers and other details are indicated in Supplementary Table S1. Norit carbons were supplied in a powder form (carbon particle size 5–100 ␮m, predominantly <50 ␮m). Granules of original Sibunit were disintegrated with an electrical grinder; fraction <50 ␮m was separated by sieving and used in the main experiments. To probe the potential effect of carbon particle size, coarsely ground Sibunit was also used (to be specified in the text). Agglomerated particles in original carbon blacks were gently crushed in a mortar prior to use. Separate experiments were done with herringbone and platelet carbon nanofibers; these supports

will be indicated in the text as CNFs1 and CNFs2, respectively. For their preparation, purification and characterization see Ref. [20]. Table 1 lists the codes by which other supports will be indicated in the text. These supports were used without additional purification. Sibunit carbons were of special grade and practically free of mineral admixtures (negligible ash residue after burning off), Norit carbons were purified by a manufacturer. To probe the potential effect of surface oxygen-containing groups, one of the Norit carbons was subjected to oxidative treatment with 7 N nitric acid at 90 ◦ C (described in detail in Supplementary Material, Section S1); the oxidized carbon is specified in the text as Nor-ox. 2.2. Hydrolytic deposition The reactor was made of glass and was ∼100 cm3 in volume. Heating was provided with water circulating through a jacket and stirring with a Teflon-coated magnetic bar. Vigorous stirring (1250 r.p.m.) was provided during addition of new components to aqueous suspensions and during heating the mixtures, when rapid processes could occur. To diminish catalyst grinding, the rest of the time stirring was kept at a level of 500–700 r.p.m., sufficient to lift up all amount of catalyst and make the reaction mixtures quasi-homogeneous. Typically, the reactor was charged with carbon (1 g) and water (50 ml) and the mixture stirred at ambient conditions for 15 min. After that, a portion of a stock solution of H2 PtCl6 with required amount of the metal precursor was diluted with water to 10 ml and added dropwise to the carbon suspension under vigorous stirring. Fifteen minutes later, a stoichiometric amount of Na2 CO3 was added (0.5 M solution), the slurry was heated to 80 ◦ C (∼8 ◦ /min) and kept at the final temperature for 1 h (longer times at high Pt loadings; to be specified). The suspension was filtered, and the catalyst was washed thoroughly with water. In the standard experiments, the amount of sodium carbonate initially added to the suspension corresponded to Eq. (1), the mole ratio between Na2 CO3 and H2 PtCl6 (further denoted by S/Pt) being equal three. During holding the suspension at final temperature, additional amounts of carbonate were added if pH measurements showed shortage of the base (pH < 7 after holding the suspension at final temperature for 0.5 h or more; at earlier times, pH could become acidic but further increase due to the evolution of CO2 ). For simplicity and to avoid misinterpretation, the ratio S/Pt = 3 will stay in the text for the standard conditions, which resulted in neutral mixtures at the end. (Actual value of the S/Pt ratio in such experiments altered with the support, ionic strength of the mixture and the source of carbonate and was in the range of 3.0–3.4.) In special experiments, the volume of reaction solution, concentrations and order of mixing of the ingredients, temperature and duration

Table 1 Specification of activated carbons and carbon blacks used as the catalyst supports. Typea

Code

SBET b

Smeso c (t-plots)

Smeso d (Pearce)

Vum e

Vum+sm e

Vmeso d (Pearce)

Vtot − Vum+sm (Vmeso )

Vtot f

Microporous activated carbon (Norit)g

Nor Nor-ox Nor2 Nor3 Sib Sib2 XC AB

951 858 1000 873 415 573 235 69

254 250 299 243 460 547 133 73

152 148 114 136 373 331 – –

0.12 0.13 0.15 0.11 0.03 0.03 0.01 0.01

0.33 0.28 0.33 0.30 –0.01 0.03 0.05 ∼0

0.36 0.34 0.21 0.29 0.85 0.56 – –

0.49 0.47 0.35 0.44 0.9 0.76 – –

0.82 0.75 0.68 0.74 0.93 0.79 – –

Mesoporous activated carbon (Sibunit) Furnace black (Vulcan XC72) Acetylene black a b c d e f g

Ultramicropores (um), supermicropores (sm) and mesopores (meso); specific surface areas S in m2 /g and volumes V in cm3 /g. By Brunauer–Emmett–Teller method (BET). From slope of t-plots at t within 0.48–0.64 nm. By Pearce method, with desorption branches of isotherms, for slit-like pores filled at P/Ps ≤ 0.95. From intercept of the ordinate axis in the back extrapolation of t-plots at t within 0.2–0.3 nm (Vum ) and 0.48–0.65 nm (volume of all micropores, Vum+sm ). Total pore volume; from pore filling with N2 at saturation (P/Ps > 0.995). Norit SX Ultra from two different batches (Nor and Nor2) and Norit GSX (Nor3); Nor-ox denotes Nor after oxidative treatment with nitric acid.

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

of separate stages could be varied. Non-standard conditions will be specified in the text. nH2 PtCl6 + 3nNa2 CO3 = Ptn (OH)m O2n−m/2 + 6nNaCl + (n−m/2)H2 O + 3nCO2 ↑

(1)

Liquid-phase reductions were performed with sodium formate, further denoted by F. As a rule, they were conducted at the same temperature as for the hydrolytic deposition, and the reactor was preliminary purged by Ar. Three-fold excess of formate (0.5 M solution; mole ratio F/Pt = 6) was added dropwise and the mixture stirred for 15 min. In special experiments, the F/Pt ratio, temperature and time could be varied (to be specified). Sample codes in the text will specify Pt content and the support used. For example, 5Pt/Nor2 indicates a sample with 5 wt.% of Pt supported on Nor2 carbon. The weight percent of Pt refers to total mass of sample, 100 × Pt/(Pt + C). Exact concentrations of Pt in solutions were determined with inductively coupled plasma (ICP) and Pt contents in solid catalysts by X-ray fluorescence (XRF) analyses. For rapid estimates of Pt remaining in solution during the hydrolytic deposition, a routine test with SnCl2 was used (described in Supplementary Material, Section S2). Concentration of chloride ions was determined with back-titration of Ag+ by Volhard method (e.g., Ref. [21]). Changes in pH during Pt depositions were monitored with a Delta 320 pH meter (Mettler Toledo). 2.3. Catalytic test reactions Conventional grease-free system supplied with vacuum, H2 , Ar and O2 lines, manometer and volumetric burette was used. The exchange of gases in the reactor was done through pumping. (IMPORTANT NOTE: The use of neat O2 and H2 connected to the same system enhances the risk of explosion; to ensure complete removal of an active gas from the reactor, a portion of an inert gas (Ar or He) should be given into the outgassed reactor and the pumping repeated.) The system allowed permanent monitoring of catalytic reaction through measurement of the volume of H2 or O2 consumed. A round-bottom reactor, approximately 100 cm3 in volume and equipped with a water-circulating jacket for maintaining constant temperature was used. Efficient stirring of the reaction mixtures was provided with a powerful magnetic stirrer (Heildolf MR3000) and a home-made Teflon-coated magnetic bar. 2.3.1. Cycloalkene hydrogenation Catalytic runs were conducted at 20 ◦ C and 1 bar of total pressure, in ethanol or isopropyl alcohol as solvent (to be specified). Commercial cyclohexene (Aldrich) and cyclooctene (Fluka) were distilled twice prior to use. Catalysts were taken in amounts of 5–50 mg, depending on expected rate of hydrogen consumption. The consumption rates were intended to be ≤6 cm3 H2 /min (achievable >30 cm3 /min), to exclude external diffusion limitations. To minimize internal diffusion limitations, the catalysts supported on activated carbons were subjected to grinding in a mortar; mild crushing was made to disintegrate particles of carbon-blacksupported samples. After placing the catalyst into the reactor, it was outgassed (slowly enough to prevent entrainment of fine catalyst particles with air as it leaves) and filled with hydrogen. (Remember about precautions when substituting air for H2 !) The solvent (9 ml) was added and the suspension was vigorously stirred for 2 min to convert the supported particles to fully reduced state. The run was started by addition of the substrate (cyclohexene, 0.4 ml, or cyclooctene, 0.5 ml; by syringe), and the rate of hydrogen consumption was followed volumetrically.

205

2.3.2. Oxidation of isopropyl alcohol Prior to testing, in situ reductive treatment was always provided to remove the strongly chemisorbed oxygen from the metal surface. After charging the reactor with catalyst, it was slowly outgassed and filled sequentially with H2 , water (9.5 ml) and IPA (0.55 ml). Stirring was provided and the reaction mixture held under H2 for 10 min. Stirring was then minimized and hydrogen carefully removed via pumping. After that, the reactor was filled with O2 . (Remember about precautions when substituting O2 for H2 and vice versa!) All catalytic runs were conducted at 30 ◦ C, 1 bar of total pressure, and started by switching on vigorous stirring. The course of the reaction was monitored by measuring O2 consumption. Precise value of final conversion of IPA was determined chromatographically, with the analysis being conducted immediately after completion of a run. Experimental details can be found in Supplementary Material, Section S3 and Figs. S1 and S2.

2.4. Other methods Chemisorption measurements were made with a pulse technique and CO as adsorbate. Procedures were similar to those in the previous studies [22,23] and are described in detail in Supplementary Material, Section S4. Before the chemisorption measurements, the Pt/C samples were reduced (or rereduced) in situ in flowing hydrogen at 100 ◦ C for 10 min. The fractions of metal atoms exposed (Ptsurface /Pttotal , DPt(CO) ) were calculated by assuming stoichiometry of adsorption CO/Ptsurface = 1. Under the conditions used, CO chemisorption took place very rapidly, and the DPt(CO) values can be considered close to those at equilibrium. The data reproducibility in repeated measurements with the same (fresh) catalyst was ±1.5 rel.% or better, and varying the temperature of reductive pretreatment near the standard value had little effect on the apparent dispersion; an example is given in Supplementary Table S2. The apparatus for temperature programmed reduction (TPR) was the same as for the chemisorption measurements (AutoChem II 2920, Micromeritics). A quartz tube was charged with a dry sample (typically, 100 mg at 5 wt.% of Pt) and flushed with Ar (25 cm3 /min, 15 min) and then with a 5%H2 /Ar mixture (25 cm3 /min) until the base line was stabilized (∼20 min). Heating rate during TPR was 10 ◦ /min. At the end of the experiment, position of baseline was checked, by directing the 5%H2 /Ar flow bypass to the reactor. The amount of H2 consumed was calculated using calibration of thermo conductivity detector (TCD) signal vs. concentration of H2 in the flow. Nitrogen adsorption/desorption was measured at 77 K (ASAP 2400, Micromeritics); prior to the measurements, the samples were degassed at 150 ◦ C. Corresponding t-plots were obtained with the use of a standard N2 adsorption isotherm for nonporous carbon black [24]. Transmission electron microscopy (TEM) images were obtained with a JEM-2010 electron microscope (JEOL; 200 kV), and size-distributions for Pt particles were found by counting the particles in different regions, at least 300 particles in each region. X-ray diffraction (XRD) patterns were recorded with a Thermo X’tra diffractometer (Thermo, Switzerland; Cu K␣ radi˚ X-ray photoelectron spectroscopy (XPS) data ation,  = 1.5418 A). were obtained with an ES-300 spectrometer (Kratos Analytical; an Mg anode source, h=1253.6 eV). The samples were dusted onto an adhesive tape, which was fastened to a sample holder. The energy scale was calibrated according to Au4f7/2 (84.0 eV), Ag3d5/2 (368.3 eV) and Cu2p3/2 (932.7 eV) lines of standard samples of gold, copper and silver foils. Relative concentration of the catalyst components was estimated from ratios of the line intensities with regard to the relative area sensitivity factors for the analyzed elements [25,26].

206

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

0,7

(a)

600

(b)

Nor

0,5

N2 ads., cm3 /g (as liquid)

Nitrogen adsorbed (cm3 /g, s.t.p.)

0,6

400

Nor

Sib 200

0,4

Sib 0,3

0,2

XC

0,1

AB

XC AB 0,0

0 0,00

0,25

0,50

0,75

1,00

0

2

4

6

8

10

Statistical thickness, t (Angstroems)

P/Ps

Fig. 1. Nitrogen adsorption/desorption isotherms (a) and corresponding t-plots (b) for the main supports.

3. Results and discussion 3.1. Characterization of catalyst supports Fig. 1 displays nitrogen adsorption/desorption isotherms and tplots for the main catalyst supports. The isotherms, t-plots, and size-distribution of mesopores for other carbons are presented in Supplementary Figs. S3–S5. Table 1 lists texture characteristics of the supports, as derived from the adsorption data (see Ref. [23] and footnotes to Table 1 for the data treatments). The isotherms and t-plots for Norits and Sibunits were typical of these types of activated carbons. Nor, Nor2 and Nor3 are distinguished by a pronounced microporosity, whereas Sib and Sib2 are essentially mesoporous (cf. Vum+sm and Vtot values in Table 1). The

downward swing in the t-plot for XC at t ∼ 0.35 nm (Fig. 1b) confirms microporosity of this carbon black [27] and indicates rather defective structure of its surface. In contrast, the acetylene carbon black (AB) is essentially nonporous, with the exception of the secondary porosity in mesopore region from the agglomeration of carbon particles. The nanofibers had BET surface of 105 m2 /g (CNFs1) and 250 m2 /g (CNFs2) and were essentially free of micropores [20]. As was to be expected, the treatment of Nor with nitric acid had only a weak impact on the texture of the carbon (Nor-ox in Table 1) but the concentration of carboxylic and other groups that produce CO2 during TPD was enhanced by a factor of 14 and that of CO-producing groups by a factor of 5 (Supplementary Table S3).

6

(a)

(b)

2,0

5

4 3

[Pt] (mmol/l)

[Pt] (mmol/l)

1,5

1,0 2

2

3

2

0,5 1

1

1

0

0,0 0

25

50

75

Time (min)

100

125

0

25

50

75

100

125

Time (min)

Fig. 2. Concentration of Pt remaining in solution vs. time of hydrolytic treatment. Conditions: (a) Nor2 as the support; stoichiometric amount of base (S/Pt = 3); 80 ◦ C (1) and 65 ◦ C (2); (b) XC as the support; 65 ◦ C; S/Pt = 2.5, 3 and 4.5 (1–3, respectively). Intended Pt loadings 5 wt.%.

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

3.2. Processes at the deposition stage

3.3. Chemical state and reducibility of deposited particles Fig. 4 displays Pt 4f XPS spectra of Pt/Nor samples at different stages of the preparation. The spectrum of 5Pt/Nor with adsorbed chloroplatinic acid (Fig. 4a) represented two doublets with binding energies (Eb ) Pt 4f7/2 of 74.2 and 72.2 eV, thus indicating the presence of Pt(IV) along with Pt(II) and/or Pt(0) species (Eb Pt 4f7/2 is near 74.5 eV for PtCl4 , PtO2 , Pt(OH)2 O, Pt(OH)4 , ≈72.2 eV for PtCl2 ,

+C

(a) 2

1 8

pH

30Pt/XC

6

4 0

50

100

150

200

250

Time (min) 10

(b)

+C,120 mg +C,10 mg +C,110 mg

8

5Pt/AB

pH

The activated carbons easily formed suspensions in water, but the hydrophobic carbon blacks could also be wetted upon addition of chloroplatinic acid and vigorous stirring. The adsorption capacities of carbons towards H2 PtCl6 were consistent with BET surface areas and fell in the order Nor > Sib ∼ XC > AB, being restricted by ∼2 wt.% of Pt in the case of Sib and XC, <0.5 wt.% for AB (Supplementary Table S4). A substantial part of the Pt complex was desorbed back to solution after addition of Na2 CO3 , and heating was needed for depositing all amount of Pt onto the support. Under the standard conditions used (80 ◦ C and stoichiometric amount of base, S/Pt = 3), the depositions of moderate amounts of Pt (≤15 wt.%) could be accomplished with all the supports used. The depositions of 5 wt.% of Pt could be completed for less than an hour and only took ∼2 h in the case of AB carbon. A predominant part of Pt left the solution for much shorter times (examples in Fig. 2), and the long treatments were only required to minimize the concentration of Pt resting in solution. With exception of AB carbon, large Pt loadings (30 wt.%) could also be provided for a rather short time (3 h), with full or almost full deposition of Pt species from the solutions (95–100%). The hydrolytic deposition was accompanied by a sharp drop in pH at the beginning of the process (Fig. 3). This took place even in the presence of the low-surface-area AB carbon (Fig. 3b), while carbon-free mixtures demonstrated small changes of pH (Fig. 3a, initial part of curve 2). The hydrolysis became slower and pH could stabilize at a high level if the amount of the carbon added was too small (Fig. 3b, curve 1, first addition of carbon). It follows that the carbon support strongly accelerates the hydrolysis of chloroplatinate and has therefore an important role in the deposition of PtOx particles. (Unfortunately, the pH measurements did not allow determining the exact rates, due to the buffering effect of HCO3 − and H2 CO3 which are produced during the hydrolysis.) Hydrolysis of Pt(IV) complexes in solution is a slow process, and formation of colloidal PtO2 particles is known to take hours/days at 50–70 ◦ C [5,6,9]. However, the presence of carbon provides the possibility of adsorption for the chloroplatinate, so that some of the Pt1 complexes appear fixed and closer to each other. This should facilitate the formation of Pt O Pt bonds. The inter-bonding should also be facilitated by the lower chemical potential of the PtOx particles on the carbon surface, due to the strong interaction between the particles and the support. This interaction is indeed strong; in opposite case, the particles would be incapable of retaining their high dispersion (Section 3.4) during the long-term hydrolytic treatments. The rapid hydrolysis of chloroplatinate in the presence of carbons may be of importance for the preparation of Pt/C catalysts by other methods. Particularly, many of the literature syntheses which are based on reductive deposition of Pt (with the aid of a reductant) employ alkaline agents but do not specify the sequence and time of adding the base and reducing agent to the suspension with the metal precursor. In fact, there could be hydrolytic deposition, e.g., during heating the suspensions prior to adding the reductant and/or in the course of a slow addition of reductant. The ratio between the hydrolytically and reductively deposited Pt could alter and so could the catalyst properties.

10

207

1

6

2

4

0

50

100

150

Time (min) Fig. 3. Changes in pH during hydrolytic treatments: (a) (1) all components were mixed and the mixture was heated to 80 ◦ C (final temperature at t = 0); (2) initially carbon-free mixture to which the carbon was added 80 min after starting the heating; (b) solution with the metal precursor and Na carbonate was held at 80 ◦ C for 0.5 h, and carbon was then added: (1) in two steps; (2) all quantity at once. Other conditions: H2 O 50 ml, carbon 120 mg (in total), S/Pt = 2.9.

PtO, Pt(OH)2 , Pt(CH3 )2 , and ≈71.5 eV for bulk Pt(0) [25,26]). The reduction of Pt(IV) on carbon surface is a known phenomenon [28,29]. The doublet for Pt(IV) became smaller after the hydrolytic treatment of the sample (Fig. 4b), probably because the treatment was made at elevated temperature. Titration of Cl− ions in filtrates by Volhard method showed that approximately 2 chlorine ligands were released to solution after the adsorption of H2 PtCl6 onto Nor and about 5.5 and 6 (in total) after the subsequent treatments with Na2 CO3 and Na formate, respectively. (The intensity of Cl 2p lines in XPS spectra was too low for quantitative assessment.) The doublet for Pt(IV), as well as Cl 2p lines in the XPS spectra, vanished after the reductive treatment (Fig. 4c) but the Pt 4f7/2 line was peaked at Eb 72.1 eV, which deviates from the value of 71.5 eV typical of metallic platinum. In order to understand whether it was caused by inadequate reduction conditions or should be ascribed to some other reasons, we studied the same samples with TPR (Fig. 5a). TPR pattern 1 in Fig. 5a is typical of the chloroplatinic acid on activated carbons (e.g., Refs. [30,31]). The peak at ∼200 ◦ C could be assigned to original acid and the peak at ∼130 ◦ C to complexes which were partly reduced, hydrolyzed, or converted into polynuclear species upon

208

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

(a)

(b)

68

72

76

80

68

84

72

76

80

84

Binding energy, eV

Binding energy, eV

(c)

(d)

30Pt/Nor

68

72

76

80

68

84

72

(a) Eb 4f7/2, eV Intensity, % C/Pt/O (atomic)

76

80

84

Binding energy, eV

Binding Energy, eV

(b)

72.0

74.1

67

33

100/0.22/5.5

72.7

74.1

85

15

100/0.24/7.4

(c)

(d)

72.1

71.6

100

100

100/0.29/5.7

100/4.3/11.2

Fig. 4. Pt 4f XPS spectra of Nor-supported samples: (a–c) 5Pt/Nor after adsorption of H2 PtCl6 (a), hydrolysis with Na2 CO3 (b; 80 ◦ C, 1.5 h) and reduction with Na formate (c; 80 ◦ C, 15 min); (d) 30Pt/Nor, after treatment with Na formate. Solid lines for deconvolution by Doniac–Suijic method.

interaction with the carbon surface. The high intensity of the peaks indicated high concentration of ionic Pt in the sample but, unfortunately, the exact value could not be determined, because of the uncertainty in the Pt(IV)/Pt(II) ratio and the hydrogen spillover onto support. The latter is characteristic of Pt/C catalysts with a high concentration of oxygen-containing groups on the carbon surface (e.g., Ref. [32]) and explains the excessive consumption of H2 in our case (see H2 (consumed)/Pt values in Fig. 5a and note that 1–2 mol of H2 per mole of Pt is only required for a full reduction of Pt(II)/Pt(IV) mixture). Anyway, only easily reducible species remained after the hydrolytic treatment (Fig. 5a, curve 2), and all the peaks attributable to ionic Pt species disappeared after the sample had been treated with formate at 80 ◦ C (Fig. 5a, curve 3). The negative peaks in the latter TPR pattern obviously originate from the H2 desorption from Pt(0) particles and the positive ones from the H2 spillover onto support, which overcompensates the H2

evolution at T > 200 ◦ C. Regardless of the support and Pt loading, all samples which went through the standard treatment with formate demonstrated analogous TPR patterns, and treatment with formate at ambient temperature could also be efficient (example in Fig. 5b). It follows that the hydrolytic deposition gives predominantly chlorine-free particles of Pt(II,IV) oxide/hydroxide, and the particles can be further reduced under mild conditions. The too high value of Eb Pt 4f7/2 in the spectrum in Fig. 4(c) is probably caused by the extremely small size of Pt particles in the given sample (Section 3.4.1). It is known that Eb Pt 4f7/2 for Pt(0) atoms is 2 eV higher than for bulk Pt metal [33], and the position of XPS lines for Pt(0) particles on activated carbons can be close to 72.0 eV (e.g., [34,35]). Supported Pd catalysts also show great variation in the binding energy for Pd(0) lines ( up to 1.5 eV, depending on the metal particle size; see Fig. 5 of review [36]). Pt/Nor samples with moderate Pt dispersions did display “normal” XPS spectra after the

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

(a)

0,20

209

(b)

1

H2,cons/Pt

H2 consumption rate (cm /min, s.t.p.)

(1) 3.3 (2) 2.3 (3) 0.3

2

0,15

(1) 1.25 (2) -0.15

0,10

3

3

0,10

H2 consumption rate (cm /min, s.t.p.)

H2,cons./Pt

0,05

0,00

3

0,025

Nor(S)

0,05

1

0,00

0,00 Nor(HCl)

- 0,025

2 0

100

200

300

Temperature, ºC

0

100

200

300

Temperature, ºC

Fig. 5. TPR patterns of 5Pt/C samples: (a) 5Pt/Nor, after the same treatments as in Fig. 4a–c (curves 1–3, respectively); in lower panel, TPR curves for bare support treated with Na2 CO3 or HCl; (b) 5Pt/AB after hydrolytic deposition (1) and subsequent treatment with Na formate at ambient temperature (2).

treatment with formate, Eb Pt 4f7/2 being equal to 71.6 eV (Fig. 4d and Supplementary Fig. S6). 3.4. Characteristic sizes of deposited particles 3.4.1. Dependence on the support and Pt loading As was to be expected, the metal dispersion (fraction of Pt atoms exposed; DPt(CO) in Table 2) was highest on the high-surface-area microporous activated carbons. The depositions of 5 wt.% of Pt onto Nor, Nor2 or Nor3 steadily resulted in DPt(CO) values within 80–90%, and TEM showed presence of tiny Pt particles, ≤1 nm, barely Table 2 Chemisorption data on Pt dispersion as a function of the support and Pt loading. No.

Support

Pt (wt%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Nor Nor Nor-ox Sib Sib Sib2 Sib2 XC XC AB AB AB AB CNFs1f CNFs2f

5 30 30 5 30 5 30 5 30 1 3 5 15 30g 30

DPt(CO) (%)a After deposition

After reduction

85 53-58b – 78 40 – – 75 33 65 49 40 25 14 24

82 46 30 58 35 (41c ) 62 (47c ) 67 22 (33d ) 62 37 31 (31d , 37e ) – – –

a Average data for each sample set (repeated preparations); hydrolytic deposition: 80 ◦ C, 1–3 h; S/Pt ∼ 3, final pH ∼ 7, degrees of deposition > 98%; reduction: 80 ◦ C, F/Pt = 6, 10–15 min. b Highest values for samples prepared with gradual addition of base (pH permanently near 7). c Short-time treatment with formate (80 ◦ C, F/Pt = 2, 1 min). d Treatment with formate at ambient temperature, 1 h (No. 9) and 15 min (No. 12). e Treatment in flowing hydrogen at 400 ◦ C (0.5 h). f SBET 105 m2 /g (CNFs1) and 250 m2 /g (CNFs2). g Degree of deposition 95% for 6 h.

distinguishable on the background of the support (Supplementary Figs. S7 and S8). The mean particle size increased with Pt loading but remained below 2 nm at 30 wt.% of Pt in the sample (Fig. 6a). After reduction of 30Pt/Nor with formate and/or H2 (200 ◦ C in the latter case), there were only weak and broad lines in the XRD pattern of the sample; the intense and sharp XRD lines, characteristic of large Pt crystallites, have only been detected for 30Pt/Nor-ox (on pre-oxidized support) (Supplementary Fig. S9). Fig. 7 displays Pt dispersion as a function of surface concentration of Pt on different carbons. One can see that the points for different carbons fall close to a common curve. It means that per a unit of surface area the number of the surface sites which are capable of nucleating does not differ much, despite the different origin of the carbons. Voropaev et al. [15] have recently reported analogous dependence for Pt/C catalysts which preparation included hydrolysis of chloroplatinate in the presence of sodium formate; neither in that nor relevant papers [14,16], the ratio between the reagents has been reported, however. Comparison of the dispersion values reported in Refs. [14–16] with the Pt dispersions obtained in the present study did not reveal advantages of the “redox-assisted” deposition over the straightforward one. At the same time, one always risks losing control of the hydrolytic deposition in the presence of reductant and obtaining a mixture of the hydrolytically and reductively deposited Pt. While the data in Fig. 7 show similarity between the carbons, there are also data which indicate some difference between them in the ability to nucleate and stabilize the PtOx particles. Particularly, there were elongated and collided particles in TEM images of 5Pt/AB (Fig. 6d), but the particles in 30Pt/XC looked small, uniform and isolated (Fig. 6c), despite the higher surface concentration of Pt in the latter case (9.4 ␮mol/m2 vs. 3.9 ␮mol/m2 for 5Pt/AB). The size-distribution of Pt particles in 30Pt/XC was narrower than in 30Pt/Sib and resembled that for 30Pt/Nor (cf. Fig. 6a–c), despite the lower specific surface area of XC (Table 1). We have to assume that defectiveness and functional coverage of the carbon surface also have a role. The acetylene blacks are most crystalline among all carbon blacks, and AB carbon should have minimum dislocations on the

210

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

Fig. 6. Characteristic TEM images of 30Pt/Nor (a), 30Pt/Sib (b), 30Pt/XC (c) and 5Pt/AB (d) after hydrolytic deposition with stoichiometric amount of base (non-reduced samples).

surface. The available ones give possibility of nucleating, but the grown PtOx particles have to be weakly bound to the surface and do not possess a perfect structure. In its turn, the activated carbons have very defective surface, but the reactive defects in these carbons are prone to oxidation in air [37]. The detrimental consequences of the surface oxidation have been clearly developed in the properties of 30Pt/Nor-ox – big supported particles appeared (Supplementary Fig. S9) and Pt dispersion decreased from 60% to 30% (Table 2, Nos. 2 and 3). The number of defects and the nature/concentration of oxygen-containing groups are probably

Fig. 7. Pt dispersion vs. Pt concentration on carbon surface after hydrolytic depositions with a stoichiometric amount of base (80 ◦ C, 1–3 h; Pt 1–40 wt.%); calculations with account of BET surface area of the supports.

finely balanced on the surface of XC carbon black. Kvande et al. found an inverse correlation between the intensity of O 1s lines in XPS spectra of CNFs and the achievable loading of Pt oxide on those supports; however, Vulcan XC72 enabled larger loadings despite the higher intensity of the O 1s line in the XPS spectrum [11].

3.4.2. Contribution from other factors Many experiments have been done to determine the most suitable conditions for the hydrolytic deposition. Owing to space limitations, the experimental data obtained are presented in Supplementary Material (Tables S5–S10). Two observations seem most important. First, higher temperatures and longer hydrolytic treatments ensured a more complete deposition but did not affect dispersion of the deposited particles, thus indicating that the interaction of the PtOx particles with the support is strong enough. In second, highest Pt loadings and dispersions were obtained with a stoichiometric amount of base, whereas excessive amounts resulted negatively. XC was the only carbon that allowed complete deposition of 30 wt.% of Pt from strongly alkaline solution (S/Pt 4.5, final pH > 9); however, such conditions decreased the dispersion of 30Pt/XC from 33 to 23%. The standard conditions used by us (80 ◦ C and final pH ∼ 7) differ from those in Refs. [9–11], where the depositions were conducted in alkaline solutions and at a lower temperature (60 ◦ C; Li2 CO3 , pH 9–10). Under those conditions, Kvande et al. could deposit 20 wt.% of Pt onto Vulcan XC72, but had small loadings on CNFs [11]. Reetz and Lopez deposited 32 wt.% of Pt onto Vulcan XC72 [9] but in two stages (24 h each). Those authors probably proceeded from the assumption that the PtOx particles form in the bulk of solution and saw it necessary to use mild conditions and high pH to stabilize the colloidal particles.

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

Our results give evidence, however, that carbon support plays an active role in the hydrolysis of the metal precursor (Section 3.2) and can be a sole but effective stabilizer. The high pH enhances negative charging of the Pt complex and the support, and this should lead to repulsion between them. It is an acidic rather than alkaline pH would be beneficial therefore for the heterogeneous nucleation of PtOx particles on the carbon surface. Interestingly, supporting the Pt oxide from strongly acidic solutions (H2 Pt(OH)6 in HNO3 ; impregnation method) gave more dispersed samples of PtOx /CNFs than the hydrolytic deposition onto the same supports; DPt(CO) values were 25% vs. 14% and 34% vs. 24% (data from Ref. [20] and Table 2, Nos. 14 and 15). As could be expected, the value of pH was less important at low Pt loadings. For example, the 5Pt/XC samples which were obtained with varying the S/Pt ratio as in Fig. 2b had similar dispersions. Low-loading Pt/Sib catalysts with Pt dispersions 55–80% could be easily obtained with an excess of the base, on coarsely ground supports or even on original carbon granules (Supplementary Table S7). Evidently, there are surface sites which interact with the metal precursor is strong enough to be affected by the deposition conditions, but the number of such sites is limited. Regardless of Pt loading, initial pH of the suspensions was of a minor importance, and the dispersions of the final PtOx particles were not sensitive to the base addition mode (whether gradually or all quantity at once). It is probably connected with the rapid decrease of pH at the very beginning of the process (Fig. 3). Only 30Pt/Nor samples showed somewhat higher dispersion if pH was permanently held near 7 all time during the hydrolytic deposition (Supplementary Table S6). The metal precursor also withstood treatment in carbon-free solutions at high pH and high temperature without detrimental consequences on the properties of final catalysts; the 30Pt/XC samples which were obtained in the two experiments in Fig. 3a showed equal dispersions (DPt(CO) 33%). In view of these data, the expediency of using a “hidden” base (to provide a gradual increase in pH, as proposed for the deposition of Pt oxide in Ref. [12]) seems questionable. 3.4.3. Consequences of reductive treatments The data in Table 2 allow one to compare Pt dispersions in initial samples and those after liquid-phase reduction. The data indicate a decrease in the dispersions by 5–30 rel.% after the treatment with Na formate, the decrease being usually stronger at higher surface concentration of Pt. Examination with TEM of the formatetreated 5Pt/Nor revealed that the sample still contained Pt particles ∼1 nm, and only separate larger particles appeared (Supplementary Fig. S8). At the same time, the sample with 5 wt.% of Pt on the low-surface-area AB carbon showed a considerable decrease in dispersion after treatment with formate both at elevated and ambient temperature; the supported particles took spherical shape and grew in size, and the number of aggregates also increased (TEM images in Supplementary Figs. S10 and S11). This corresponds to the observations by Reetz et al. [38]: electrochemical reduction led to severe agglomeration and enlargement of supported particles in PtOx /Vulcan-XC72 but to relatively small changes in a sample of PtOx /Printax-XE2, which was obtained on the carbon black with anomalous specific surface area (1000 m2 /g). The instability of supported metal particles in aqueous medium with a reductant is a known phenomenon [39–41]; both enlargement of Pt particles and their collision into clusters can take place [39]. To be sure in full reduction of PtOx particles, regardless of the support and Pt loading (Section 3.3), and to provide identical treatment for all samples, we employed reduction with an excess of formate at 80 ◦ C. As far as the metal dispersion is concerned, these standard conditions are not optimal for all samples, however. This issue is outside the scope of the paper, but one may consider the values given in parentheses in the last column of Table 2. Almost

211

Scheme 1. Places of location of PtOx particles grown on carbon surface and preformed in solution.

all samples (except for Pt/AB) retained initial dispersions during treatment with formate at ambient temperature, and all withstood treatment in flowing hydrogen even at a very high temperature. For example, Pt dispersion only decreased from 39.8 to 36.5% after the treatment of 5Pt/AB with H2 at 400 ◦ C (0.5 h).

3.5. The character of Pt distribution over the support According to XPS data (see the table attached to Fig. 4), the ratio between the intensities of Pt 4f and C 1s lines in the spectrum of 30Pt/Nor exceeds that for 5Pt/Nor by a factor of 15, although the difference in Pt loading is only by a factor of 6. This indicates an increased concentration of Pt near the top surface region in the high-loading sample. The sample with 5 wt.% of Pt was examined with XPS after grinding in a mortar. The grinding affected neither the line shapes nor the relative intensities IPt4f /IC1s (Supplementary Fig. S12). This indicated that the concentration and the state of Pt on the external surface of the support and inside the pore structure are practically the same in the low-loading catalyst. To obtain more convincing evidence, several samples of 5Pt/Nor2 and 5Pt/Nor3 with the deposited Pt oxide were divided into fractions <50 and >50 ␮m, and Pt content in the fractions was determined with XRF. The analysis showed similar Pt loadings, the enrichment in Pt of the finer fraction being only by a factor of 1.15–1.25. This confirmed the predominantly uniform distribution of Pt over the support at the given Pt loading. The slight enrichment with Pt of smaller carbon particles should be attributed to the higher specific surface area of such particles (SBET 930 m2 /g vs. 850 m2 /g for the <50 and >50 ␮m fractions of Nor3). The ratio between Pt contents in the finer and coarser fractions increased to 1.5, however, for a sample with 30 wt.% of Pt, thus confirming that Pt tended to concentrate near external surface of carbon particles upon increasing Pt loading. (If all Pt were deposited onto the external surface, its mass on a carbon particle would be proportional to square of the particle size, ␲d2 ; as the particle mass is proportional to (1/6)␲d3 , the Pt loading (wt.%) should be inversely proportional to the particle size.) It can thus be concluded that the rate of the hydrolytic deposition is not too high to prevent the penetration of the small and mobile Pt1 -complexes deep inside the pore network of the support. The less uniform distribution of Pt oxide in the high-loading samples is probably resulted from the limit in the appropriate sites on the carbon surface. When the available ones are occupied and the deposition slows down, homogeneous nucleation (in the bulk of solution) can take part. The oligomeric [PtCl6−l−m OHl (H2 O)m ]n n(m−2) species which forms in the solution meets obstacles, however, for entering the inner parts of carbon particles (size limitations) and has to locate in large pores and on the external surface of the support, as sketched in Scheme 1.

212

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

3.6. Pt concentration in different pores To estimate the distribution of Pt particles in the porous structure of the support, catalytic testing in cyclohexene hydrogenation has been used. This reaction is known to be structure-insensitive on Pt catalysts in a broad range of dispersions (10–100% [17]), so that the turnover frequencies TOFs should be constant and the rates per total Pt atoms TOFt directly proportional to Pt dispersion (TOFt = TOFs × DPt ), provided that there were no side-factors which affected the apparent activities. Before testing, the catalysts were reduced ex situ with sodium formate and pretreated in situ with H2 to eliminate the chemisorbed oxygen. They were also ground in a mortar, to minimize internal diffusion limitations. The hydrogenation of cyclohexene proceeded with a constant rate during a whole run (examples in Supplementary Fig. S13), and the initial rates could thus be determined with a high precision. Under the conditions used, the rates were directly proportional to catalyst mass, thus giving evidence for the absence of external diffusion limitations and poisons in the reaction mixture. Table 3 lists results of catalytic testing of the samples with 5 wt.% of Pt. One can see that TOFt values do not follow the dispersions and TOFs values can differ by a factor of 2.5, depending on the support used. The difference between the catalysts remained essentially the same after changing the solvent or duration of the reductive pretreatments, and no increase in TOFs values was observed for catalysts with lower Pt loadings (Supplementary Table S11). The latter confirmed absence of serious diffusion limitations for the chemical reaction. To have full confidence in the little role of the diffusion factor, separate catalysts have also been tested in hydrogenation of cyclooctene. It is less reactive than cyclohexene, and the diffusion limitations were not apparent even with initial catalysts (equal reaction rates before and after grinding the samples; examples in Supplementary Fig. S14). Nevertheless, the TOFs values for Nor, Nor2-, and Nor3-supported catalysts still differed by a factor of 2.5–3 from those for Pt/Sib (Supplementary Table S12). As the turnover frequencies in Table 3 were calculated on the basis of CO chemisorption data, the difference between the TOFs values might be attributed to uncertainty in the stoichiometry of CO adsorption, which could vary with metal particle size. However, the DPt(CO) values in Table 3 correlate with the DPt(TEM) values found with TEM, and the difference between the TOFs values is too strong to be accounted for by this factor. All indicates that (bulky) cycloalkene cannot enter some of the pores and contact metal particles therein. Analogous difference in specific activities has been detected previously for Pt and Pd catalysts which were prepared on microporous and mesoporous carbons through adsorption of H2 PtCl6 or H2 PdCl4 , followed by drying and reduction of the samples in a flow of H2 [22,23,42]. Both the adsorption of the chloride complex and the nucleation of PtOx particles should take place preferably in or nearby the narrowest pores, where the imperfection of the carbon surface reaches maximum.

Table 3 Characteristic dispersions of 5Pt/C catalysts and activities in cyclohexene hydrogenation.a Support Nor Sib XC AB

DPt(TEM) b (%) >80c 57 54 <37d

DPt(CO) (%)

TOFt (s−1 )

TOFs (s−1 )

82 58 66 31

3.5 5.3 4.7 3.3

4.3 9.1 7.1 11.0

a All catalysts after hydrolytic deposition and reduction with Na formate under standard conditions; catalytic runs: 20 ◦ C, H2 1 bar, cyclohexene 0.4 ml, ethanol 9 ml; TOFt values per all Pt atoms, TOFs per exposed ones, with account of DPt(CO) values. b As calculated from TEM data by formulae DPt(TEM) (%) = 108/dS (nm) [49]. c Pt particles predominantly ≤1 nm in size (poor contrast). d DPt(TEM) 37% without accounting aggregated particles.

Table 4 Impact of Pt loading and deposition conditions on specific activities of Pt/Nor catalysts in cycloalkene hydrogenation. No.

Pt loading (wt.%)

S/Pt

Hydrolysis time (min)a

DPt(CO) (%)

TOFs C6 (s−1 )b

TOFs C8 (s−1 )b

1 2 3 4

5 30 5 5

3 3 3 6

60 180 3 30

82 46 48 65

4.3 n.d.c 5.3 6.0

0.21 0.28 0.27 0.31

a Mixing the ingredients at ambient temperature, heating to 80 ◦ C (∼8 min) and holding at this temperature for indicated time; then reduction with Na formate (80 ◦ C, F/Pt = 6, 15 min). b Hydrogenation of cyclohexene (C6 ) and cyclooctene (C8 ) under standard conditions (20 ◦ C, 1 bar). c Not determined (too high reaction rate and internal diffusion limitations at the high Pt loading).

The microporous XC and Nor decrease the catalyst activity most strongly (Table 3), so that the proportion of micropores probably correlates with the number of the pores which are accessible for the metal precursor but which are sufficiently small to cause the confinement of final metal particles. Note that steric hindrances from the pore walls could make the major part of the metal surface inaccessible for organic molecules if even the molecules can enter a pore with the entrapped metal particle. 3.7. Impact of deposition conditions on catalytic properties With Nor as catalyst support, we investigated to what extent the confinement effect can be weakened with the aid of deposition conditions. Principal results are presented in Table 4. The turnover frequencies in the structure-insensitive hydrogenation appeared enhanced if the depositions were conducted with a large excess of base (Table 4, No. 4 vs. No. 1) and, especially, if NaOH was changed for Na2 CO3 (Supplementary Table S5). Certainly, the negative charging of the carbon surface at the high pH retards the penetration of the (negatively charged) Pt complex into narrow pores, and the PtOx particles have to form on more accessible places. Shortening the time which Pt complex had for entering the narrow pores also enhanced the accessibility of Pt particles (Table 4, No. 3; in the given case, the hydrolytic deposition was terminated soon after the beginning, and the resting part of Pt was rapidly deposited with the aid of Na formate). Nevertheless, these approaches have been only partly successful. The TOFs values still differed from those for Pt/AB in Table 3, and the enhanced accessibility of Pt was achieved for the expense of Pt dispersion (Table 4), so that there could even be a loss in catalyst productivity per total mass of Pt (TOFt = TOFs × DPt ). The nature of the catalyst support (non-porous or mesoporous carbons) is certainly a more effective means of resolving the problem. Interestingly, the specific activities of high-loading 30Pt/Nor catalysts also exceeded those for 5Pt/Nor (Table 4, No. 2 vs. No. 1; also Supplementary Table S6). This confirms that the proportion of Pt particles which nucleate on the more open places of the surface and in the bulk of solution increases upon increasing Pt loading (Section 3.5 and Scheme 1). This also predicts difference in catalytic properties between the catalysts being prepared through the in situ deposition of Pt oxide (as in the present work) and by supporting the preliminary obtained Pt oxide colloids (the two alternative variants described in patent [9]). We also estimated the possibility of influencing catalytic properties in structure-sensitive reactions, i.e., those which course depends on the metal particle size, the smallest one being not necessary optimum. IPA oxidation was chosen as a model reaction. It is well-studied on Pt catalysts [43–45], goes with 100% selectivity to acetone (pH remains constant and need not control), but highly dispersed Pt particles show stable performance if only the

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

213

Fig. 8. Catalytic properties of Pt/C samples in IPA oxidation as a function of the support (a), Pt loading (b), and hydrolysis time prior to reduction with formate (c). Conditions: H2 O 9.5 ml, IP 7 mmol, O2 1 bar, 30 ◦ C; all catalysts after liquid-phase reduction and grinding in mortar; standard conditions for hydrolytic depositions in cases (a) and (b); catalyst mass corresponds to 2.5 ␮mol (a and c) and 7.5 ␮mol (b) of Pt.

concentration of alcohol is high enough to prevent “overoxidation” of the metal [46,47]. As was to be expected, the oxidation of IPA proceeded with permanent decline in rate and could stop far before the conversion of IPA was completed (Fig. 8). Fig. 8a displays catalytic properties of samples which were prepared under standard conditions, had equal Pt loadings but differed in the support. One can see that the extremely dispersed 5Pt/Nor is distinguished drastically in the performance from the other catalysts. The majority of the tiny Pt particles in this catalyst are either incapable of catalyzing the alcohol oxidation or deactivate instantly in the process, so that the reaction rates are low from the very beginning of catalytic run. According to the data in Fig. 8b and c, such simple means as increasing the Pt loading on the support or shortening the hydrolysis time during the catalyst preparation are very effective for raising the catalyst efficiency. (Note that the catalyst mass in the experiments in Fig. 8b was varied inversely to Pt loading on the support; the mass of Pt in the reactor remained constant and the catalytic data reflect therefore real efficiencies.) There were probably two factors which acted in parallel. Both the high Pt loading and short hydrolysis time enhance the accessibility of the supported particles for the organic substrate (as found in the experiments on cyclohexene hydrogenation, Table 4), and the particles simultaneously grow in size (cf. DPt(CO) values in the Table attached to Fig. 8), thus becoming more stable in the alcohol oxidation. 3.8. The applicability domain of the method and catalysts The hydrogenation of cyclohexene on Pt catalysts has been thoroughly studied before, and the literature data predict turnover frequencies TOFs ≈ 5 s−1 at 20 ◦ C, 1 bar of H2 and gradientless conditions [17]. This corresponds to the catalytic activities in Table 3. For IPA oxidation on Pt catalysts, the TOF values were typically in the range of 50–500 h−1 at the temperatures near ambient, and IPA conversions did not exceed 90% [43–45]. The velocities in Fig. 8 are considerably higher than the literature values and the attainable conversions are well above 90% (curve

4 in Fig. 8b; also Supplementary Fig. S15). It means that the hydrolytic method enables one to have maximum possible rates in structure-insensitive reactions and to optimize the catalyst for a structure-sensitive one. We often used the hydrolytic deposition for practical purposes, and it gave good results on different scales. The uniform distribution of the hydrolytically deposited particles and the location of the particles preferably in smaller pores suggests better performances for the catalysts on finely ground and non-porous or mesoporous supports. It does not exclude using of microporous supports, however, especially if the catalytic process has to be conducted under severe conditions, where stability of the supported component to sintering can become of primary importance. We had fairly good results with PtRu/Nor and Ru/Nor catalysts in hydrogenation of adipic acid to hexanediol (concentrated aqueous solutions, Sn-promoted catalysts, 220–240 ◦ C, 80–100 bar). The hydrolytically prepared catalysts allowed reuse, resulted on almost quantitative yields and were not inferior in activity to the (Sn)Ru/C and (Sn)RuPt/C catalysts recently developed for the hydrogenation of carboxylic acids by Mitsubishi researchers (preparation via impregnation) [48].

4. Conclusions The results of the present study confirm the great capabilities of the hydrolytic method in producing the highly dispersed Pt nanoparticles. Under proper conditions, small Pt particles can be easily obtained both on hydrophilic and hydrophobic carbons, on finely and coarsely ground supports, by ordinary means, for a short time and without special additives. The new insights can be summarized as follows.

1) The hydrolysis of chloroplatinate is greatly accelerated in the presence of carbons. This enables rapid depositions of Pt oxide onto carbon supports and simultaneously indicates that the hydrolytic deposition can occur during synthesis of Pt/C catalyst

214

2)

3)

4)

5)

K.M. Kaprielova et al. / Applied Catalysis A: General 449 (2012) 203–214

by another method and can therefore interfere with the other preparation methods. The nucleation takes place predominantly on the carbon surface, thus resulting in tiny PtOx particles which are tightly bound to and evenly distributed over the carbon surface. At high Pt loadings, homogeneous nucleation in the bulk of solution can also occur, due to the limit in the surface sites for the heterogeneous nucleation, and Pt starts to concentrate near the top surface region. The size of deposited particles is primarily determined by the Pt loading and specific surface area of the support but is influenced by the surface chemistry. The structure defects facilitate but the oxygen-containing groups on the carbon surface impede the heterogeneous nucleation and formation of small PtOx particles. High pH weakens the interaction between the starting Pt complex and the support, thus retarding the deposition and worsening the Pt dispersions. High temperature and neutral pH ensure rapid depositions of Pt oxide regardless of the carbon origin and carbon particle size. The deposited PtOx particles demonstrate an easy reducibility and can be reduced directly in the liquid phase with minimal changes in dispersion. The formation of PtOx particles in the interior of the support raises the possibility of particle confinement in narrow pores, the degree of confinement being primarily dependent on the pore structure of the support. The accessibility of supported particles can be enhanced through shortening the hydrolysis time and weakening the interaction of the metal precursor with the support.

Acknowledgements We are thankful to T.Ya. Efimenko for nitrogen adsorption measurements, I.L. Kraevskaya, L.A. Sergeeva and N.P. Yatzko for chemical analyses, Dr. D.F. Habibulil for 195 Pt NMR spectra, Drs A.A. Bryliakova, E. Gerasimov and V.I. Zaikovskii for TEM images, Dr. O.A. Bulavchenko for XRD data, and Dr. A.E. Shalagina for a gift of CNFs. We are also grateful to Prof. S.T. Oyama and all reviewers, whose valuable comments helped in improving the manuscript. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apcata. 2012.10.004. References [1] K. Kinoshita, P. Stonehart, in: J.O’M. Bockris, B.E. Conway (Eds.), Modern Aspects of Electrochemistry, vol. 12, Plenum Press, 1977, pp. 183–266. [2] A.B. Stiles, Catalyst Manufacture. Laboratory and Commercial Preparations, Marcel Dekker, New York, 1983. [3] A.J. Bird, in: A.B. Stiles (Ed.), Catalyst Supports and Supported Catalysts, Butterworths, Boston, 1987 (Chapter 5). [4] L.R. Radovic, F. Rodriguez-Reinoso, in: P.A. Thrower (Ed.), Chemistry and Physics of Carbon, vol. 25, Marcel Dekker, New York, 1997, pp. 243–358. [5] M.T. Reetz, M.G. Koch, J. Am. Chem. Soc. 121 (1999) 7933–7934.

[6] M.T. Reetz, M. Lopez, W. Grunert, W. Vogel, F. Mahlendorf, J. Phys. Chem. B 107 (2003) 7414–7419. [7] B. He, Y. Ha, H. Liu, K. Wang, K.Y. Liew, J. Colloid Interface Sci. 308 (2007) 105–111. [8] R. Pan, Y. Wu, Q. Wang, Y. Hong, Chem. Eng. J. 153 (2009) 206–210. [9] M.T. Reetz, M. Lopez, US Patent 7244688 B2 (2007), to Studiengesellschaft Kohle mbH. [10] I. Kvande, S.T. Briskeby, M. Tsypkin, M. Ronning, S. Sunde, R. Tunold, D. Chen, Top. Catal. 45 (2007) 81–85. [11] I. Kvande, J. Zhu, T.-J. Zhao, N. Hammer, M. Ronning, S. Raaen, J.C. Walmsley, D. Chen, J. Phys. Chem. C 114 (2010) 1752–1762. [12] B. Fang, N.K. Chaudhari, M.-S. Kim, J.H. Kim, J.-S. Yu, J. Am. Chem. Soc. 131 (2009) 15330–15338. [13] B. Fang, M.-S. Kim, J.H. Kim, M.Y. Song, Y.-J. Wang, H. Wang, D.P. Wilkinson, J.-S. Yu, J. Mater. Chem. 21 (2011) 8066–8073. [14] J. Kaiser, P.A. Simonov, V.I. Zaikovskii, C. Hartnig, L. Jorissen, E.R. Savinova, J. Appl. Electrochem. 37 (2007) 1429–1437. [15] I.N. Voropaev, P.A. Simonov, A.V. Romanenko, Russ. J. Inorg. Chem. 54 (2009) 1531–1536. [16] N.P. Lebedeva, A.S. Booij, I.N. Voropaev, P.A. Simonov, A.V. Romanenko, Fuel Cells 9 (2009) 439–452. [17] R.J. Madon, J.P. O’Connell, M. Boudart, AIChE J. 24 (1978) 904–911. [18] E. Auer, A. Freund, J. Pietsch, T. Tacke, Appl. Catal. A 173 (1998) 259–271. [19] V. Calvino-Casilda, A.J. Lopez-Peinado, C.J. Duran-Valle, R.M. Martin-Aranda, Catal. Rev. Sci. Eng. 52 (2010) 325–380. [20] A.S. Lisitsyn, L.B. Okhlopkova, A.E. Shalagina, Z.R. Ismagilov, Carbon 46 (2008) 549–552. [21] A.V. Jain, N.C. Pradhan, A.K. Dalai, N.N. Bakhshi, Catal. Lett. 86 (2003) 221–227. [22] M. Gurrath, T. Kuretzky, H.P. Boehm, L.B. Okhlopkova, A.S. Lisitsyn, V.A. Likholobov, Carbon 38 (2000) 1241–1255. [23] L.B. Okhlopkova, A.S. Lisitsyn, V.A. Likholobov, M. Gurrath, H.P. Boehm, Appl. Catal. A 204 (2000) 229–240. [24] K. Kaneko, C. Ishii, M. Ruike, H. Kuwabara, Carbon 30 (1992) 1075–1088. [25] D. Briggs, M.P. Seah, Practical Surface Analysis, Wiley, New York, 1983. [26] Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer, Eden Prairie, MN, 1992. [27] V. Raghuveer, A. Manthiram, J. Electrochem. Soc. 152 (2005) A1504–A1510. [28] F. Coloma, A. Sepulveda-Escribano, J.L.G. Fierro, F. Rodriguez-Reinoso, Langmuir 10 (1994) 750–755. [29] H.E. van Dam, H. Van Bekkum, J. Catal. 131 (1991) 335–349. [30] M.A. Alvarez-Montero, L.M. Gomez-Sainero, J. Juan-Juan, A. Linares-Solano, J.J. Rodriguez, Chem. Eng. J. 162 (2010) 599–608. [31] C.D. Taboada, J. Batista, A. Pintar, J. Levec, Appl. Catal. B 89 (2009) 375–382. [32] M.C. Roman-Martinez, D. Cazoria-Amoros, A. Linares-Solano, C. SalinasMartinez de Lecea, Carbon 31 (1993) 895–902. [33] W. Eberhardt, P. Fayet, D.M. Cox, Z. Fu, A. Kaldor, R. Sherwood, D. Sondericker, Phys. Rev. Lett. 64 (1990) 780–783. [34] O.A. Scelza, I.M. Vilella, S.R. de Miguel, Chem. Eng. J. 114 (2005) 33–38. [35] E. Antolini, L. Giorgi, F. Cardellini, E. Passalacqua, J. Solid State Electrochem. 5 (2001) 131–140. [36] Z. Karpinski, Adv. Catal. 37 (1990) 45–100. [37] D.S. Cameron, S.J. Cooper, I.L. Dodgson, B. Harrison, J.W. Jenkins, Catal. Today 7 (1990) 113–137. [38] M.T. Reetz, H. Schulenburg, M. Lopez, B. Spliethoff, B. Tesche, Chimia 58 (2004) 896–899. [39] J.H. Vleeming, B.F.M. Kuster, G.B. Marin, F. Oudet, P. Courtine, J. Catal. 166 (1997) 148–159. [40] A. Douidah, P. Marécot, S. Labruquère, J. Barbier, Appl. Catal. A 210 (2001) 111–120. [41] M.T. Schaal, J. Rebelli, H.M. McKerrow, C.T. Williams, J.R. Monnier, Appl. Catal. A 382 (2010) 49–57. [42] N. Job, M.F.R. Pereira, S. Lambert, A. Cabiac, G. Delahay, J.-F. Colomer, J. Marien, J.L. Figueiredo, J.-P. Pirard, J. Catal. 240 (2006) 160–171. [43] J.W. Nicoletti, G.M. Whitesides, J. Phys. Chem. 93 (1989) 759–767. [44] R. Gläser, R. Josl, J. Williardt, Top. Catal. 22 (2003) 31–39. [45] R.L. Augustine, L.K. Doyle, J. Catal. 141 (1993) 58–70. [46] M. Besson, P. Gallezot, Catal. Today 81 (2003) 547–559. [47] T. Mallat, A. Baiker, Chem. Rev. 104 (2004) 3037–3058. [48] Y. Hara, K. Endou, Appl. Catal. A 239 (2003) 181–195. [49] J.L. Lemaitre, P.G. Menon, F. Delannay, in: F. Delannay (Ed.), Characterization of Heterogeneous Catalysts, Dekker, New York, 1984, pp. 299–365.