Enhancement of catalytic activity of Pd-PVP colloid for direct H2O2 synthesis from H2 and O2 in water with addition of 0.5 atom% Pt or Ir

Takashi Deguchi a, Hitoshi Yamano b, Sho Takenouchi§ b and Masakazu Iwamoto *a
aResearch and Development Initiative, Chuo University, 1-13-27 Kasuga, Bunkyo-ku, Tokyo 112-8551, Japan. E-mail: iwamotom@aoni.waseda.jp
bChemical Resources Laboratory, Tokyo Institute of Technology, 4259-R1-5 Nagatsuta, Midori-ku, Yokohama 226-8503, Japan

Received 13th September 2017 , Accepted 3rd January 2018

First published on 4th January 2018

Catalytic activity of Pd-polyvinylpyrrolidone (PVP) colloid for the reaction of H2 and O2 to H2O2 in the presence of H2SO4 and NaBr was greatly enhanced with the addition of a very small quantity of Pt or Ir. In contrast to ordinary alloy catalysts, the activity of Pd was doubly increased with the addition of only 0.5 atom% Pt or Ir, although the selectivity for the H2O2 formation somewhat decreased. Addition of Ru, Rh, or Au affected little the activity. Kinetic analyses of the reactions over modified Pd-PVP indicated that the high H2-activating abilities of Pt and Ir increased the H2–O2 reaction rate, which was supported by DFT study. In contrast, H2O2 hydrogenation rates were almost constant irrespective of Pt or Ir addition, indicating little participation of the hydrogen activated on Pt or Ir in H2O2 destruction. Pt-modified Pd-PVP showed a discontinuous change in catalytic activity in the time dependency measurement, resulting in higher H2–O2 reaction rates and lower H2O2 selectivity. Lower partial pressure of hydrogen under the reaction conditions or higher temperatures in the absence of H2 accelerated the change. Structures of the Pt- and Ir-loaded Pd-PVP catalysts were suggested to change during the reaction; the Pt or Ir atoms would initially be located at coordinatively more unsaturated sites of the Pd particles and move into the surface Pd lattice to be stabilized. It is shown that the amount of H2O2 produced reaches several wt% when estimated using the Pt- and Ir-loaded Pd-PVP catalysts under an O2 partial pressure of 360 kPa.

1. Introduction

H2O2, a clean oxidizing agent, is currently manufactured indirectly from H2 and O2 by the so-called anthraquinone process, which requires large scale instruments and a lot of energy due to its complexity. Direct synthesis of H2O2 has been intensively studied expecting simpler processes by means of various methods. They are categorized into three representative methods according to their principles; catalytic reactions over noble metals in the liquid phase, electrochemical reactions on electrode catalysts of fuel cells and non-catalytic chain reactions in plasma.1 The catalytic method has been most extensively studied from the practical as well as the theoretical point of view,1–6 and the last two have made notable progress in relatively recent years.7–14

With respect to the catalytic method, bimetallic alloy catalysts have widely been studied in order to improve the catalytic performance. The combination of Pd and Au has been intensively studied as a candidate catalyst that may be put to practical use.15–28 Several DFT calculations explained the origin of the observed high performance of Pd–Au mainly from alloy formation.29–35 A synergy effect between Pd and Pt was also reported to enhance the H2O2 formation.36–39 Similar effects were recently observed for Pd–Ru and Au–Ru.40 Combinations of Pd and base metals such as Ni, Sn and Te were also reported.41–43

Most supported bimetallic nanoparticle catalysts were prepared by reducing mixed precursor metal salts on inorganic or organic carriers,36,44 while colloid catalysts were obtained by reducing the precursor metal salts dissolved in water in the presence of polyvinylpyrrolidone (PVP).20 The contents of the second metal were generally high in the H2O2 synthesis chemistry: for instance, 30 atom% of Au and 70 atom% of Pd afforded a maximum H2O2 formation rate,20 and addition of 5 atom% Pt to a Pd catalyst resulted in a 2.5-fold increase in the H2O2 formation rate and a small decrease in the selectivity.36 Detailed structures of the alloy catalysts have not been clarified yet, but it is supposed that Pd and Pt probably formed segregated domain structures based on the phase diagram.45 In addition, solid-solution structures of alloy nanoparticles of Pd and Au, or Pd and Pt, where the metals are homogeneously mixed at the atomic level, could be formed under limited conditions.46,47 The findings suggested agglomeration of the second metals in the Pd-based bimetallic catalysts reported so far. We were interested in the effects of the second component dispersed atomically on the Pd surface. To solve the problems, preparation of atomically dispersed Pd-metal bimetallic systems needs to be achieved.

We previously reported the kinetics of the H2O2 synthesis over the Pd-PVP colloid catalyst in the presence of H+ and Br.48 The colloid catalysts are convenient for study on the catalysis without the influence of supports, although their industrial use might be difficult. It would widely be recognized that supporting treatment of the second metals onto a supported Pd catalyst such as Pd/carbon and Pd/Al2O3 results in their deposition not only on the Pd particles but also on the surface of the support. Since the colloid catalysts have no support, we can expect atomic level dispersion of the second metals upon their loading in very small amounts through in situ reduction of their precursor compounds. Very small amounts (0.1–1.0 atom%) of Pt, Ru, Rh, Ir, or Au were here added to the Pd-PVP particles by in situ reduction of the precursor compounds. Because the average particle size of the parent Pd-PVP was 3.6 nm,48 the number of Pd atoms in a particle was approximately 2000 and the number of the 0.5 atom% heteroatoms was around 10 in the particle. It is reasonable to assume atomic dispersion of the heteroatoms on the Pd particle under these conditions.

Kinetic analysis of direct H2O2 synthesis is rather troublesome due to two side reactions, concurrent direct H2O formation and consecutive H2O2 destruction to H2O, which proceed in parallel with the main H2O2 formation. Conventional experiments include measurements of the H2O2 concentration in the liquid phase by titration or colorimetry and determination of the amount of consumed H2 by gas chromatography to calculate the reaction rate of H2 and the selectivity to the H2O2 formed. Frequent sampling of the liquid phase, however, inconveniently perturbs the reaction system making it difficult to obtain stationary data. We proposed an improved experimental method as will be described in the Experimental and data processing section. Rates of H2 consumption and H2O2 formation could be directly obtained based on mass balance calculations without disturbing the reaction system. The H2O2 concentration could be monitored by integration of the H2O2 formation rate. A data processing method for kinetic analysis of the three reactions was also proposed.

Another problem for the kinetic analysis comes from the insufficient rates of mass transfer of gas components into the liquid phase.2,49,50 In particular, the mass transfer rate of H2 should be considered, because the partial pressure of H2 is usually much lower than that of O2. Reaction rates cannot be analyzed correctly without quantitative evaluation of the mass transfer rate of H2 or the concentration of H2 in the liquid phase. We actually measured the mass transfer coefficient of H2 in the reactor, and calculated the H2 concentration in the liquid phase.

2. Experimental and data processing

2.1. Materials

Pd-PVP, Ru-PVP, Rh-PVP, Ir-PVP, Pt-PVP, and Au-PVP colloids were provided by Tanaka Kikinzoku Kogyo in the form of 4 wt% aqueous solutions and used as catalysts as received. The average diameters were around 3.6,48 1.5, 4.0, 2.0, 2.0 (ref. 51) and 9 nm, respectively. To study the addition effect of a precious metal on the Pd catalyst, an aqueous solution of RuCl3·nH2O (Kanto Chemical), RhCl3·3H2O (Wako Pure Chemical Industries), IrCl3·nH2O (Kanto Chemical), Pt(NH3)4(NO3)2 (Tanaka Kikinzoku Kogyo), or HAuCl4 (Kishida Chemical) was added into the Pd-PVP solution and then the solution underwent H2 activation treatment (written later).

2.2. Apparatus and reaction procedure

Fig. 1 shows the semi-batch reactor of this study schematically. The reaction was performed in a 300 ml flat bottom separable flask made of Pyrex glass with an inner diameter of 80 mm equipped with a magnetic stirrer, a gas feed nozzle, a gas outlet nozzle, and a thermometer. The reaction procedures using the colloid catalysts were essentially identical to those described in the previous paper.48 Prescribed amounts of water, the colloid solution, and the second metal precursor solution (when necessary) were introduced into the flask. H2 gas of 50 sccm was flowed at 303 K for 40 min to activate the catalyst and to reduce the second metal precursor to deposit the metal on the Pd surfaces. Subsequently, N2 gas of 50 sccm was introduced for 10 min to purge H2 gas, the additives (NaBr and H2SO4) were added in the form of an aqueous solution, and the total volume of the solution was adjusted to 300 mL with water. The typical amount of Pd used in the catalytic run was 8.33 mg L−1. A mixture consisting of H2, O2, and N2 gases was continuously passed into the reactor at ambient pressure. The flow rate of each gas was regulated using a high precision mass flow controller (10, 20, and 20 sccm, respectively, under standard conditions, and varied as needed), and calibrated using a soap-film flow meter before the reaction. The outlet gas composition was analyzed periodically (every 10 minutes) by gas chromatography using N2 as an internal standard to determine the consumption rates of H2 and O2, rH2 and rO2 (mmol L−1 h−1), according to eqn (1) and (2). Here fH2, fO2, fN2, xH2, xO2, and xN2 denote the feed rates of H2, O2, and N2 (mmol h−1) and the mole fractions of H2, O2, and N2 in the outlet gas. The numeral 0.3 on the right sides is the liquid volume (L). Note that the gas composition was often within the flammable range.
rH2 = (fH2fN2xH2/xN2)/0.3(1)
rO2 = (fO2fN2xO2/xN2)/0.3(2)

image file: c7cy01890h-f1.tif
Fig. 1 Outline of a semi-batch H2–O2 reaction system.

The data in the initial unsteady period resulting from the dead volume of the reactor were treated in the following way. Using the gas phase dead volume (VG) and the rate of gas flow out of the reactor (FG), the corrected initiation time (t0) was approximated using t0 = VG/FG (in general, t0 was approximately 5 min). Because the measured gas composition became nearly stationary after an elapse time of 20 min, the corrected initial gas composition was determined by linear extrapolation of the analyzed compositions of the three successive samples collected after time t = 20 min to time t = t0.

2.3. Data processing for evaluation of selectivity and destruction rate

The data processing for the rate analyses was performed by the method described in our papers48,50 and the ESI.51 From the material balance of H2 and O2, the reaction rates of H2 and O2, rH2 and rO2, are related to the H2O2 and H2O accumulation rates using eqn (3) and (4), in which [H2O2] and [H2O] denote the concentrations of H2O2 and H2O (mmol L−1). Here, changes in the volume of the reaction mass are neglected. Eqn (5) can be derived from eqn (3) and (4).
rH2 = d[H2O2]/dt + d[H2O]/dt(3)
rO2 = d[H2O2]/dt + (1/2)d[H2O]/dt(4)
d[H2O2]/dt = 2rO2rH2(5)

The amount of H2 consumed, image file: c7cy01890h-t1.tif (mmol L−1), and [H2O2] were calculated by numerical integration of rH2 and d[H2O2]/dt, respectively, at each gas sampling point image file: c7cy01890h-t2.tif. The concentration of H2O2 at the end of the reaction was determined separately with a UV-vis absorption method using a titanium sulfate solution ([H2O2]UV). The correlation between [H2O2]fin and [H2O2]UV for all data, where [H2O2]fin was the final calculated value of [H2O2], is summarized in Fig. 2. The values of |[H2O2]fin – [H2O2]UV| divided by the overall H2 consumed at the final sampling point image file: c7cy01890h-t3.tif (>10 mmol L−1) fell within the error range of 2.5% on average. The excellent agreement of the data showed the accuracy of the above calculation method using d[H2O2]/dt.

image file: c7cy01890h-f2.tif
Fig. 2 Correlation between H2O2 concentrations determined by colorimetry ([H2O2]UV) and by integration of d[H2O2]/dt ([H2O2]fin).

Two H2O2 selectivities, Se and Sf, are defined. Se, the effective selectivity of H2O2, is the proportion of the amount of H2O2 to the consumed H2, and determined using eqn (6) at each sampling. (Se)UV represents the value of Se calculated using [H2O2]UV and image file: c7cy01890h-t4.tif as shown by eqn (7).

image file: c7cy01890h-t5.tif(6)
image file: c7cy01890h-t6.tif(7)

S f is a parameter related to the kinetics, and was defined as the proportion of the rate of direct H2O2 formation to the sum of the rates of direct H2O2 formation and of direct H2O formation. The rate of H2O2 accumulation was determined using eqn (8), in which the H2O2 destruction rate was assumed to be proportional to [H2O2] and the catalyst concentration, [Cat], and kd is the rate constant.48,50Sf and kd include the terms of H2 and O2 partial pressures, pH2 and pO2, respectively. Sf and kd[Cat] were determined by applying a least-squares method to eqn (9) derived by integrating eqn (8). Here, image file: c7cy01890h-t7.tif and image file: c7cy01890h-t8.tif. The details of the calculations are described in the previous paper.48 In this paper, kd[Cat] will be used as it is as the parameter that shows the destruction activity. It should be noted that eqn (9) can be strictly applied only when both Sf and kd are constant. Otherwise the calculated values of Sf and kd would only be the apparent ones, which might show the average tendency.

d[H2O2]/dt = rH2Sfkd[H2O2][Cat](8)
image file: c7cy01890h-t9.tif(9)

2.4. Calculation of liquid phase H2 concentration

In a stationary state of the H2–O2 reaction in the semi-batch reactor, the reaction rate of H2, rH2, is equal to the net dissolution rate of H2 as shown in eqn (10), in which image file: c7cy01890h-t10.tif, kL, and a denote the concentration of H2 equilibrated to the H2 partial pressure in gas phase, the mass transfer coefficient of H2 through the gas–liquid boundary film, and the gas–liquid boundary surface area per unit volume, respectively. Because image file: c7cy01890h-t11.tif = pH2/HH2 (HH2: Henry's law constant of H2 in water), eqn (10) is transformed into eqn (11). We can know from eqn (11) the concentration of H2 in the solution, cH2, which decreases with increasing rH2. The kLa value (=1140 h−1) was determined in the previous paper and HH2 is 133 kPa L mmol−1 at 303 K in water.48
image file: c7cy01890h-t12.tif(10)
cH2 = pH2/HH2rH2/kLa(11)

3. Results and discussion

3.1. Effect of noble metal addition on the Pd-PVP catalyst

Approximately 0.5 atom% of Ru, Rh, Ir, Pt, or Au was added to Pd-PVP to examine the addition effect in the presence of H2SO4 and NaBr. The results are summarized in Table 1. Subscripts _init and _fin denote the initial and the final values of the variables. The reaction rates of H2, rH2, increased with time, resulting in rH2_fin > rH2_init except for the case of Ir, which will be discussed later. Ir and Pt enhanced the H2 reaction as well as the H2O2 formation, whereas Ru, Rh, and Au exhibited little effect. The rH2_init values of the Ir- and Pt-added catalysts were more than twice that of the original Pd-PVP catalyst. The final H2O2 concentration, [H2O2]UV, also doubled by Pt or Ir loading. Sf and kd[Cat] were successfully determined using eqn (9) with determination coefficients of R2 = 0.997–1.000. The selectivity of H2O2, (Se)UV and Sf, was somewhat decreased by loading of Pt or Ir. The H2O2 destruction rate parameter, kd[Cat], was affected little, resulting in the increase in the final H2O2 concentration.
Table 1 Effects of added noble metals on catalysis of Pd-PVPa
Added metal (atom%) p H2_init (kPa) p O2_init (kPa) r H2_init r H2_fin [H2O2]fin [H2O2]UV (Se)UV (—) S f (—) k d[Cat] (h−1)
(mmol L−1 h−1) (mmol L−1)
a Pd, 8.33 mg L−1; H2SO4, 0.01 N; NaBr, 0.001 N; 303 K, 1200 rpm, 2 h. Gas feed rates of H2, O2 and N2 were set to 10, 20 and 20 sccm, respectively. The subscripts _init and _fin are used to denote the initial and the final values, respectively.
None 17.1 36.0 21.9 29.1 27.5 27.7 0.54 0.86 0.53
Ru 1.12 17.3 36.4 20.4 30.1 24.3 24.8 0.48 0.76 0.58
Rh 0.50 16.5 36.2 25.1 30.3 28.4 27.8 0.50 0.77 0.47
Ir 0.58 15.7 31.6 70.4 70.0 56.8 54.7 0.38 0.55 0.35
Pt 0.55 16.1 33.0 52.1 64.9 58.0 54.6 0.46 0.75 0.50
Au 0.56 17.1 36.6 20.0 24.7 27.3 25.4 0.57 0.88 0.41

Fig. 3 shows the time courses of the H2–O2 reaction over the Pt- and Ir-loaded Pd-PVP together with that over the Pd-PVP. A significant change in rH2 with time was observed on the Pt-loaded catalyst, while little increase or constancy in rH2 on Pd-PVP alone or the Ir-loaded catalyst was observed, which will be discussed later. The effective H2O2 selectivity, Se, decreased with time due to the destruction of H2O2.

image file: c7cy01890h-f3.tif
Fig. 3 Time courses of H2 consumption and H2O2 formation for the H2–O2 reaction over non-, 0.55 atom% Pt-, and 0.58 atom% Ir-loaded Pd-PVP colloid catalysts in the presence of H+ and Br in water. The reaction conditions were the same as those in Table 1.

The catalytic activity of Pd-, Ru-, Rh-, Ir-, Pt-, and Au-PVP colloids was examined in separate experiments with and without the addition of H2SO4 and NaBr, and is summarized in Table 2. The apparent activity of single component catalysts was in the order Pt > Pd > Ir ≫ Ru, Rh, Au. The final H2O2 concentration was more or less improved by adding H+ and Br, indicating that the effect of the additives to suppress the side reactions commonly worked, in a similar manner as described in our previous papers for Pd catalysts.50,52 It is notable that Ir alone was apparently less active than Pd in Table 2, but Ir addition exhibited the largest acceleration effect in Table 1. It should be noted that rH2 was not decreased in the presence of H+ and Br in the cases other than Pd, indicating no prohibiting effects of HBr, which would afford a requirement for the enhancement effects of Pt and Ir.

Table 2 Effects of H2SO4 and NaBr on catalysis of metal-PVP colloidsa
Catalyst (mg-metal L−1) H2SO4 (N) NaBr (N) p H2_init (kPa) p O2_init (kPa) r H2 (mmol L−1 h−1) [H2O2]UV (mmol L−1)
Initial Final
a 303 K, 1200 rpm, 2 h b Probably more than 10 but rapidly deactivated.
Pd-PVP 8.33 0 0 9.2 38.2 56.1 55.3 0.12
0.01 0.001 17.1 36.0 21.9 29.1 27.7
Ru-PVP 8.36 0 0 19.8 36.7 1.7 1.4 0.12
0.01 0.001 19.7 36.8 2.5 1.5 0.64
Rh-PVP 8.36 0 0 19.9 36.8 1.0 0.7 0.06
0.01 0.001 19.4 36.7 4.9 3.6 0.90
Ir-PVP 8.20 0 0 19.2 37.3 6.0 4.1 0.12
0.01 0.001 16.7 37.7 19.3 10.0 0.53
Pt-PVP 1.04 0 0 17.5 38.1 46.6 39.5 0.47
0.01 0.001 12.7 38.9 41.8 42.3 5.30
Au-PVP 8.33 0 0 17.8 32.8 b 2.3 0.38
0.01 0.001 19.6 36.6 3.8 2.7 1.23

Before the detailed experiments, two significant confirmation tests were carried out. The first one was to confirm the amount of Pt remaining dissolved in the catalyst solution after H2 activation. In the confirmation experiment, 1 atom% of Pt(NH3)4(NO3)2 was added to the Pd-PVP solution of 83.4 mg-Pd L−1, and activated by H2 using the ordinary protocol, in which the respective amounts of Pd and Pt used were 10 and 20 times larger than those of the usual experiments. The resulting solution was filtrated with an ultrafiltration device described in the previous paper48 to remove the colloid particles. The amount of Pt remaining in the solution, analyzed by inductively coupled plasma mass spectrometry, was 18% of the addition amount, indicating deposition of most of Pt on the colloid particles. Secondly, 1 atom% of the Pt-PVP colloid was mixed into the Pd-PVP catalyst solution and the H2–O2 reaction was studied. The reaction rates were essentially the same as those on the Pd-PVP alone, demonstrating that the coexistence of the 1 atom% Pt particles and the Pd particles in a solution was not efficient to increase the activity and the present synergy effect could result from the direct interaction of Pd with Pt in one particle.

3.2. Effects of Pt addition on the kinetics of the Pd-PVP catalyst

Dependency on Pt/Pd ratio. The synergy effect of bimetallic Pd–Pt catalysts for the H2–O2 reaction was already reported at the level of about 5 atom% addition of Pt.36,38 By contrast, the current remarkable enhancement effects of Pt and Ir were observed at around 0.5 atom%, indicating the necessity of studying the new addition effects. In addition, the change in the catalytic activity with reaction time, observed in Fig. 3, is the important finding to be clarified. The first point was studied in this section and section 3.4 from changes in the catalysis with Pt/Pd and Ir/Pd atomic ratios in the presence of H2SO4 and NaBr. The second point was investigated in section 3.3.

The dependency on Pt/Pd ratio is summarized in Fig. 4, in which the reaction rate and the H2 concentration in the liquid phase are presented by the initial values, rH2_init and cH2_init. rH2_init, cH2_init, Sf, and kd[Cat] all showed respective continuous changes at Pt/Pd = 0–0.8 and 0.8–2.7 atom% and discontinuous changes around 0.8 atom%, the latter of which will be discussed later and the former will be first investigated in more detail. The rH2_init increased sharply by adding Pt and quadrupled with 1 atom% addition of Pt. Here, it is necessary to note that the values of rH2_init depended on the H2 concentration in water: that is, rH2_init should be corrected with cH2_init to compare the true activity. The values of rH2_init/cH2_init are plotted in Fig. 4c, which showed 3.7 and 9.2-fold improvement in the activity by addition of 0.55 and 1.09 atom% Pt to the Pd-PVP, respectively. The improvement in rH2_init/cH2_init became moderate with further increasing the Pt/Pd ratio and almost saturated around Pt/Pd > 1.5 atom%.

image file: c7cy01890h-f4.tif
Fig. 4 Influence of the Pt/Pd ratio on the kinetic parameters for the H2–O2 reaction over Pt-loaded Pd-PVP colloid catalysts in the presence of H+ and Br in water. Pd, 8.33 mg L−1; H2SO4, 0.01 N; NaBr, 0.001 N; pH2_init of 14–18 kPa; pO2_init of 34–36 kPa; 303 K, 1200 rpm, 2 h.

Although the values of Sf and kd[Cat], determined by applying data processing based on eqn (6),48 might be uncertain because of the rH2 variation during the reaction, they may be referred to as average values. Sf gradually decreased with Pt/Pd atomic ratio. The value of kd[Cat] was constant in part, but apparently showed distinct discontinuity and was lowered at higher Pt/Pd ratios in Fig. 4b, in contrast to rH2. The destruction of H2O2 is known to be mainly caused by hydrogenation,50,52 and therefore kd[Cat] is also corrected with rH2_init in Fig. 4c. kd[Cat]/cH2_init was roughly constant, although the data were somewhat scattered, indicating that the H2 activated on the Pt sites was not used for the hydrogenation of H2O2, as will be discussed later.

As shown in Fig. 3, a change in rH2 with the time course was observed over the Pt-loaded Pd-PVP catalysts. The final values of rH2 after 2 h reaction, rH2_fin, was plotted against rH2_init in Fig. 5, which revealed that there were two groups, rH2_init = rH2_fin (group A) and rH2_init < rH2_fin (group B). rH2 was constant during the reaction in group A, whereas it varied to a higher value in group B. The classification appeared to be dependent on the magnitude of rH2, or the H2 concentration in the liquid phase, cH2. Although considerable scattering of the plots is observed in Fig. 4 mainly due to experimental errors, it should be noted that the data at Pt/Pd < 0.82 atom% belong to group B and those at Pt/Pd > 0.82 to group A in Fig. 5. Two experiments at Pt/Pd = 0.82 atom% afforded two sets of results in Fig. 4. This would be due to the fact that they were experiments on the border separating groups A and B. That is, the initial structural change occurred or did not occur in the respective experiments, which might result from subtle experimental differences that are yet to be revealed.

image file: c7cy01890h-f5.tif
Fig. 5 Relationship between rH2 values at the initial and the final stages of the H2–O2 reactions summarized in Fig. 4.
Effect of H2 partial pressure. The possible dependency of the activity of the Pt-loaded Pd-PVP catalysts on the H2 concentration in the liquid phase, cH2, shown in Fig. 4 and 5, required study of the kinetic parameters as a function of H2 partial pressure. Fig. 6 summarizes the pH2 dependencies of the kinetic parameters when pH2_init was varied from 5 to 23 kPa. Clearly, two dependencies of the parameters were observed in the lower and higher ranges of pH2. Because rH2_init = rH2_fin in the former group and rH2_init < rH2_fin in the latter (not shown), the former group was assigned to group A and the latter to group B. Irrespective of considerable scattering of the plots, rH2_init/cH2_init was apparently constant in the both groups, and that in group A was higher than that in group B, indicating that the catalyst was more active in group A. The findings, rH2_init = rH2_fin for group A and rH2_init < rH2_fin for group B, led us to conclude that the catalyst transformation occurred at the beginning of the reaction in group A, and the transformation could occur more easily in the lower range of pH2. In the higher range of pH2, the transformation was slower, suggesting that plenty of the adsorbed hydrogen suppressed the transformation. Sf was independent of pH2 in both groups, although Sf for group B was higher than that for group A. kd[Cat]/cH2_init was roughly constant all throughout from group A to group B, indicating that kd[Cat] was proportional to cH2 and the H2O2 destruction activity was almost unchanged by the catalyst transformation. The H2O2 concentration accumulated for 2 h, [H2O2]UV increased with pH2, and group A showed a steeper slope as the overall result of rH2, Sf and kd[Cat]. Now we define stage I and stage II as the states of the catalyst before and after the transformation, respectively.
image file: c7cy01890h-f6.tif
Fig. 6 p H2 dependency of kinetic parameters of the H2–O2 reaction over Pt-loaded Pd-PVP colloid catalyst in the presence of H+ and Br in water. Pd, 8.33 mg L−1; Pt/Pd ratio of 0.55 atm%; H2SO4, 0.01 N; NaBr, 0.001 N; pO2_init of 30–42 kPa; 303 K, 1200 rpm; 2 h.

The H2O2 formation selectivity, Sf, over the Pt-loaded Pd-PVP catalyst is composed of the part over the Pd-PVP itself and that corresponding to the increment of rH2 by Pt addition. Because Sf decreased with increasing loaded Pt as shown in Fig. 4, the Sf originating from loaded Pt was lower than that from Pd-PVP. On the other hand, the Sf for group B was higher than that for group A, as shown in Fig. 6. Thus the order of Sf was Pd-PVP > stage I > stage II. The change in Sf will be discussed later.

A reasonable explanation for the difference between the states of stages I and II should be proposed to reveal the specific addition effect of a very small amount of Pt and its change. Changes in the particle sizes or in the Pt distribution among the particles might happen during the reaction. However, some reports showed high stability of Pd particle sizes under more severe reaction conditions in the liquid phase.54 Furthermore such changes would hardly explain the discontinuous pH2 dependency shown in Fig. 6. As shown later in more detail, Pt atoms would exist initially in a metastable state on the surface of a Pd particle and migrate to stabilized positions on the Pd particle by lattice rearrangement. Many bimetallic systems including Pt–Pd have recently been reported to be re-structured depending on the environmental conditions,46,47 supporting the mechanism.

3.3. Change in the catalysis of Pt-loaded Pd-PVP with reaction time

Influence of pretreatment after catalyst activation. In order to reveal the respective states of stages I and II, the catalyst solution was treated under various conditions after H2 activation of the catalyst, and then the H2–O2 reaction was carried out controlling pH2 by setting the H2 content in the feed gas, xH2, at 20 or 30 vol%. The results are summarized in Table 3. Without any treatment, the change in rH2 during the reaction was observed at xH2 = 30%, but not observed at xH2 = 20%, indicating that the catalyst transformation in the latter was finished immediately after the start of the reaction. Treatment of the catalyst under high H2 pressure (2 MPa) caused no change in rH2 at xH2 = 20%, which indicated that the initial adsorption of plenty of H did not prohibit the transformation of the catalyst. When the catalyst was treated in the absence of H2 at 303 K (the same as the reaction temperature), in the presence or absence of O2, and in the presence or absence of H2SO4 and NaBr, a change in rH2 was observed at xH2 = 30%. Clearly, these treatments did not cause any transformation of the catalyst. Treatment of the catalyst at 343 K for 1 h in N2 gave higher rH2 from the beginning of the reaction. The effectiveness of N2 treatment at higher temperature for the transformation of the catalyst would be due to promoted desorption of H2 and/or migration of the metal atoms to result in rearrangement. All the findings were in good agreement with the mobility of the Pt atoms accelerated under the actual reaction conditions owing to the lowered H2 concentration in the liquid phase and the reactions of the adsorbed species which was suggested in the previous section.
Table 3 Influence of catalyst treatment on the kinetic parameters of the H2–O2 reaction over a Pt-loaded Pd-PVP catalysta
Catalyst treatment after H2 reduction x H2 in feedb p H2_initial (kPa) Change in rH2c r H2 (mmol L−1 h−1) S f (—) k d[Cat] (h−1)
Atmosphere H+ and Br Temp. Time Initial Final
a Pd, 8.33 mg L−1; Pt, 0.083 mg-atom L−1; H2SO4, 0.01 N; NaBr, 0.001 N; 303 K, 1200 rpm, 2 h. b Total gas feed rate: 50 sccm; O2 feed rate: 20 sccm; N2 feed rate: balance. c A: rH2 was almost unchanged all throughout the reaction. B: Clear increment in rH2 was observed during the reaction. d Catalyst treatment was performed in an autoclave.
No treatment 20% 9.7 A 56.7 56.6 0.48 0.19
H2 (2MPa)d Present 303K 2h 20% 8.2 A 64.0 63.6 0.45 0.20
No treatment 30% 19.6 B 75.0 89.9 0.72 0.62
N2 Present 303K 1h 30% 19.3 B 76.9 97.1 0.66 0.80
O2+N2 (4:3) Present 303K 2h 30% 20.1 B 72.5 90.6 0.74 0.78
N2 Absent 303K 2h 30% 22.5 B 66.7 100.1 0.92 1.07
N2 Absent 343K 1h 30% 13.7 A 98.4 97.9 0.55 0.35

Change in kinetics with the reaction time. The time course of the H2–O2 reaction over the Pt-loaded Pd-PVP shown in Fig. 3 is drawn again in Fig. 7 together with the pH2, cH2 and rH2/cH2 data to discuss the activity change in more detail. rH2 as well as rH2/cH2 rose gradually at the first half (stage I), jumped up in the second half and became constant (stage II), suggesting that most of the catalyst particles suddenly changed in a short time. In the current experiments, each gas was supplied to the reactor at a constant rate. rH2 rose gradually at stage I (the reason will be discussed in the next paragraph), and correspondingly pH2 became lower. The rearrangement of the surface would take place statistically in this stage and be accelerated with decreasing pH2, because a lower pH2 was favorable for the transformation. When pH2 reached a critical level, a discontinuous change would proceed. The critical pH2 should be approximately 15 kPa (more exactly cH2 = about 60 μmol L−1), around which rH2 jumped up and pH2 dropped down at the same time. This corresponded well to the pH2 value distinguishing group A from group B in Fig. 6. The change in rH2/cH2 in Fig. 7 indicated that the catalytic activity doubled upon the discontinuous change. Although no distinct turning point is observed in the curve of [H2O2] in Fig. 7, one will be able to find that the linear [H2O2] increase was due to the discontinuous drop of the kd[Cat] value, as will be shown in Fig. 8.
image file: c7cy01890h-f7.tif
Fig. 7 Typical time course of the H2–O2 reaction with kinetic regime change over Pt-loaded Pd-PVP in the presence of H+ and Br. Pd, 8.33 mg L−1; Pt/Pd ratio of 0.55 atm%; H2SO4, 0.01 N; NaBr, 0.001 N; pH2 is shown in the figure; pO2 of 33–34 kPa; 303 K, 1200 rpm.

image file: c7cy01890h-f8.tif
Fig. 8 Kinetic analysis of the reaction shown in Fig. 7 based on eqn (2) (stage I) and (1) (stage II).

The gradual elevation of rH2 at stage I is similar to those frequently observed in H2–O2 reaction over the non-modified Pd-PVP catalyst, in which the increase in rH2 over Pd-PVP was attributed to the hydrogenation of H2O2.48 That is, H2, O2, and H2O2 were adsorbed on the catalyst at equilibrium, and the reactions of adsorbed H with adsorbed O2 and adsorbed H2O2 were rate determining. However, rH2 in the current study was almost constant at stage II independent of the H2O2 concentration, indicating little possibility that similar kinetics can be applied on Pt-loaded Pd-PVP. It was confirmed in our previous paper50 that the rate-determining step was the H2 activation, and rH2 was constant independent of the H2O2 concentration on Pd/C. It was also suggested that the proportion of highly unsaturated sites on Pd-PVP was much higher than on Pd/C.48 The structure of the Pd particles at stage II might be analogous to that of Pd/C.

The Pt atoms would initially occupy metastable positions such as adatom positions on relatively unsaturated surface sites like edges. They might easily migrate to slightly more stable positions with partial surface rearrangement. We assume that the Pt atoms then would move into the surface lattices to cause whole rearrangement of the surfaces to the substantially least unsaturated ones when pH2 was lower than the critical value. It is well known that restructuring of catalyst surfaces can be induced by adsorbates,53 which might be associated with the observed phenomena. Vigorous reactions on the Pt atoms would trigger the movement, which the adsorbed H atoms would prevent by stabilizing the Pt atoms at higher pH2. We cannot deny some involvement of other components like H+ or oxidation states of the Pt and Pd atoms in the change. Although we can only speculate the mechanism of the structural change right now, some dynamic theoretical calculations will clarify it in the future.

S f and kd[Cat] at stages I and II in Fig. 3 were analyzed as follows. Eqn (9) was applied to stage I, because it was confirmed in the previous paper that eqn (9) holds during the H2–O2 reaction over Pd-PVP in the presence of H+ and Br.48 On the other hand, eqn (8) was applied at stage II, since rH2 was constant. Good linear relationships were observed at both stages as shown in Fig. 8, and the determined values of Sf and kd[Cat] are shown in the figure. The value of Sf at stage I was larger than that at stage II. The value of kd[Cat] at stage I was also larger than that at stage II, but the values corrected with cH2 were at the same level. Incidentally, the values of the Sf and kd[Cat] determined by applying eqn (9) to the whole duration of the reaction were 0.75 and 0.50, respectively (see Table 1), reflecting more the results at stage I.

Discussion on changes in the kinetic parameters from stage I to stage II. The correlation between the kinetic parameters and the sates of catalysts in stages I and II is summarized as follows. 1) In stage I, in which Pt atoms would exist in a metastable state on Pd colloid particles, rH2 was higher, and Sf lower than those of the original Pd-PVP. Pt was not involved in the H2O2 destruction. 2) In stage II, in which Pt atoms would occupy more stable positions in the Pd lattice and the surface of the Pd particles would become more saturated, rH2 became much higher and Sf much lower. Pt was not involved in the H2O2 destruction either. In both cases, Pt accelerated the H2–O2 reaction, probably due to its high H2 activating ability. The higher reaction rates in stage II than those in stage I would lead to the higher activation ability in stage II.

If the role of Pt were only to supply activated H to the adsorbed O2 species on the Pd surface, the H2O2 selectivity, Sf, would be expected to be almost equivalent to those observed on Pd-PVP. The unexpected change in Sf was in fact recognized in Fig. 4. Firstly, it is possible to think that O2 species adsorbed on the Pt site were involved in the reaction by reacting with the activated H on the Pt sites or on the Pd surface. However, the change in kd[Cat] would make the following explanation more plausible. H2O2 is considered to adsorb on relatively more unsaturated sites in equilibrium, and to react with the adsorbed hydrogen in equilibrium over Pd-PVP.48 The density of the adsorbed H2O2 should be very low, because the density of the unsaturated sites was suppressed by the addition of H+ and Br.54 The H species activated on the Pt sites were very reactive, and the energy barrier for the reaction with the adsorbed O2 was so lowered that the H species would react instantly with abundantly adsorbed O2, which means little chance of the H species to react with the rarely adsorbed H2O2. H2O2 destruction would proceed via the H species adsorbed on the Pd surface. This consideration leads to another possibility for the reason of the lowered Sf. Hydroperoxy species on the Pd surface would react with adsorbed H species to form H2O2 or H2O + adsorbed O species, and the former would prevail on site B (less unsaturated sites)50,52 because the energy barrier for the former reaction would be lower. However, the difference in the energy barrier would become smaller with more active H species on the Pt site to cause the lowering of Sf. Adsorbed H species would be more active in stage II than in stage I resulting in the lower Sf.

Quantitative evaluation of promotion effect of Pt. It is difficult to analyze quantitatively the effect of Pt addition using the data in Fig. 4, due to the simultaneous change in cH2. rH2_fin would be more adequate than rH2_init to evaluate the effect of Pt addition, because the catalyst particles and their activity were stabilized at stage II. rH2_fin was thus plotted against the product of Pt concentration in the solution, [Pt], and the H2 concentration, cH2_fin, in Fig. 9 using the same experimental data. The plots for the H2–O2 reaction over Pt-PVP in the absence of Br (ref. 51) are also shown in the figure for comparison. A linear relationship was observed indicating that the increment of rH2 was proportional to both [Pt] and cH2. The slope, the activity of Pt, for Pt-loaded Pd-PVP was 10 times as high as that for Pt-PVP. Since the dispersion of Pt was ca. 50% in Pt-PVP, the Pt atoms on Pt-loaded Pd-PVP were estimated to be 5 times more active than the surface Pt atoms of Pt-PVP, although other factors may also be involved. Atomic dispersion of Pt on Pd would be important. It was reported that Pt–Pd alloys with a solid-solution structure, where Pt and Pd atoms were homogeneously mixed at the atomic level, adsorbed more H2 when Pt/Pd = 8/92 than when Pt/Pd = 21/79 or 50/50.47 The capability for adsorbing more H2 would be connected to the higher ability of the Pt atoms to activate H2 in the H2–O2 reaction.
image file: c7cy01890h-f9.tif
Fig. 9 [Pt]cH2vs. rH2 plot for the H2–O2 reaction over Pt-loaded Pd-PVP in the presence of H+ and Br, and over Pt-PVP in the absence of Br. The reaction data were taken from the experiments in Fig. 4 for Pt-loaded Pd-PVP, and from Fig. 3 in ref. 29 for Pt-PVP.

The oxidation states of Pt atoms are another possible factor. Our previous paper showed that the surface of Pt-PVP was covered by atomically and molecularly adsorbed oxygen under the reaction conditions, and thus the proportion of the reduced surface active for the reaction decreased.51 Pt atoms isolated in the Pd surface might be more resistant to oxidation.

3.4. Effects of Ir-addition on the catalysis of the Pd-PVP catalyst

The influence of the Ir-loading is summarized in Fig. 10. rH2 was elevated and Sf was lowered with Ir/Pd ratio. Ir/Pd dependency of rH2_init/cH2_init indicates that the activity enhancing effect of Ir was almost the same as that of Pt: kd[Cat]/cH2_init was almost constant, and its level was similar to that in the case of Pt. The results showed that Pt and Ir have similar addition effects to Pd, although the catalytic activity of Ir-PVP itself was very different from that of Pt-PVP as shown in Table 2. Although the reason for the remarkable distinction between the high activating effect of Ir and the moderate activity of Ir-PVP for the H2–O2 reaction is not evident yet, the discussion on the promotion effect of Pt would be also applied to the results of Ir. The lowest H2O2 productivity of Ir-PVP despite its moderate reaction rate in Table 2 indicated the easiness of the O–O cleavage. The H2 activating ability of Ir would be high, and O–O cleavage of the adsorbed O2 and/or OOH species would occur rapidly on the Ir surface. The latter reaction would increase the proportion of oxidized surface of Ir, losing the reduced surface necessary for H2 activation. A similar idea was already discussed for the Pt-PVP and Pd/C catalysts.50,51 Atomically dispersed Ir might be inert for the O–O cleavage.
image file: c7cy01890h-f10.tif
Fig. 10 Influence of the Ir/Pd ratio on the kinetic parameters for the H2–O2 reaction over Ir-loaded Pd-PVP colloid catalysts in the presence of H+ and Br in water. Pd, 8.33 mg L−1; H2SO4, 0.01 N; NaBr, 0.001 N; pH2_init of 12–18 kPa; pO2_init of 29–36 kPa; 303 K, 1200 rpm, 2 h.

On the other hand, changes in activity with the reaction time over Ir-loaded Pd-PVP catalysts were different from those over Pt-loaded Pd-PVP catalysts. As shown in Fig. 11, rH2 was substantially constant throughout the reaction when Ir was loaded, whereas without Ir it gradually rose with time as previously shown in Fig. 3. The behavior of rH2 over the Ir-loaded Pd-PVP catalysts was very similar to that in stage II of the Pt-loaded Pd-PVP. The structural change of the particles of Ir-loaded Pd-PVP would proceed in a similar manner to that of Pt-loaded Pd-PVP and take place so rapidly that stage II emerged from the beginning of the reaction, although the reason for the faster restructuring of Ir than Pt is undiscovered. The initial value of Se over Ir-loaded Pd-PVP was lower than that over the Pt-loaded one in Fig. 1 supported the structural change from the beginning of the reaction.

image file: c7cy01890h-f11.tif
Fig. 11 Time course of the H2–O2 reaction over Ir-loaded Pd-PVP colloid catalysts in the presence of H+ and Br in water. Pd, 8.33 mg L−1; H2SO4, 0.01 N; NaBr, 0.001 N; pH2_init of 12–18 kPa; pO2_init of 29–36 kPa; 303 K, 1200 rpm, 2 h.

3.5. Preliminary DFT studies

DFT studies were preliminarily performed to elucidate the great addition effects of Pt or Ir on the Pd-PVP colloid catalyst and the little effects of Ru, Rh and Au. Activation of hydrogen was focused on in the studies, because the concentration of H2, cH2, was directly related to the reaction rates. First-principles quantum mechanics calculations were carried out with supercell models using CASTEP,55 which employs the density functional theory plane-wave pseudopotential method. The GGA-PBE functional56 was used with ultrasoft pseudopotentials.57 The energy cutoff and Monkhorst–Pack mesh of k points were 400 eV and (4 × 4 × 1), respectively, for all the models. A vacuum space with a height of 20 Å was placed over the surface of each supercell model.

Adsorption energies of H were calculated using the models formed by replacing the central surface Pd atom of the Pd (111) 3 × 3 × 3 supercell with an M atom (M = Pt, Ir, Ru, Rh and Au) as shown in Fig. 12. These supercell slabs were geometrically optimized by fixing the third layer and under spin-polarized conditions applying the formal spin as the initial.

image file: c7cy01890h-f12.tif
Fig. 12 Supercell model for a M-substituted Pd(111) surface. Dark colored spheres represent Pd atoms and the light colored one a hetero atom M.

The energy level of a H atom located on the surface of a supercell slab was determined using eqn (12), in which ΔEH, E[Slab1], E[Slab0], and E[H2] represent the energy level of the H atom, the energies of the H-bearing slab, the mother slab, and H2, respectively. Here, calculation of [Slab1] and [Slab0] was performed in non-spin-polarized mode, because it was confirmed that the spin state should be ignored for determining ΔEH using eqn (12) in separate calculations.

ΔEH = E[Slab1] − E[Slab0] − E[H2]/2(12)

The adsorption energy profiles of H were determined right above the line segment connecting the center of the M atom and the point of contact of the two adjacent Pd atoms through the fcc hollow site. Points 0 to 8 were defined at equal intervals on the line segment, and H was vertically placed over each point by varying the height, h (0.1 Å intervals). Single-point energy calculation of E[Slab0] and E[Slab1] was carried out, and ΔEH was determined according to eqn (12) at each height. The adsorption energy of H at each point, ΔEad, was determined as the minimum value of ΔEH by applying a quadratic polynomial to the smallest 3 values of ΔEH.

Fig. 13 summarizes the results demonstrating that the energy levels of H adsorbed on top of M atoms (point 8) were distinctly dependent on the metal. The values of ΔEad on top of the M atom were in the order of Au > Pd > Rh > Ru > Pt > Ir, and their relative levels based on M = Pd were 7.02, 0.00, −5.13, −7.40, −10.27 and −15.23 kcal g-atom−1, respectively. Incidentally the relative energies at point 3 (fcc) and point 0 (bridge) for M = Pd based on the energy on top were −12.20 and −9.11 kcal g-atom−1. The relative energy level on top for Pt was comparable to those at fcc and the bridge of Pd, and that for Ir was lower than both. To use a geographical analogy by comparing the energy to the altitude, there are many narrow wells (fcc and hcp sites) and valleys (bridge sites) on the surface of a Pd colloid particle, and Pt or Ir atoms form wide ponds, which make us imagine that H2 would be efficiently activated there.

image file: c7cy01890h-f13.tif
Fig. 13 Energy profile of H adsorbed on a MPd(111)_plane supercell from the bridge site adjacent to M to the on-top site of M.

Furthermore, it should be noted that the energy level was relatively unchanged on the Pt and Ir atoms. Consequently, the adsorbed H atoms would be very mobile and the H–M bonds would be flexible there. The Pt and Ir atoms are expected to thus facilitate the reaction, because a H atom has to stand up on the metal surface to react with adsorbed O2 in the transition state.29 We already reported54 that H2O and H+ play important roles to mediate the reaction of the adsorbed H and the adsorbed O-containing species through hydrogen bonding. The high mobility and the flexibility of the H–M bond on Pt and Ir atoms would be advantageous also in this mechanism, because particular positional relations would be demanded in the transition state due to both the angles and the lengths of the hydrogen bonds.

The relative energy levels of adsorbed H on Ru and Rh were lower than that on Pd, although they were higher than those on Pt and Ir. The results do not explain distinctly why Ru and Rh showed almost no accelerating effects in Table 1. The reason might be that the energy levels were higher than those at fcc and bridge sites of Pd.

3.6. Simulation for evaluating the catalyst performances

It is important to evaluate the performances of the catalysts from a practical point of view, because the ultimate purpose of the study is to improve catalyst performance. In particular, increasing the concentration of H2O2 while obtaining a higher selectivity should be pursued first. Simulation of the reaction is one of the effective tools for evaluating the catalytic performances. Kinetics is useful not only for analyzing the reactions, but also for simulating the reactions by using the obtained kinetic parameters. We practiced simulation of the H2–O2 reaction over the Pd-PVP, Pt-loaded and Ir-loaded Pd-PVP catalysts to evaluate the catalyst performances.

The accumulation rate of H2O2 is expressed by eqn (8), and the time course of the H2O2 concentration follows eqn (13), which is obtained by integrating eqn (8).50 Because the consumption of H2 is given by rH2t provided that rH2 is constant during the reaction, the effective H2O2 selectivity Se is calculated using eqn (14). When t increases towards infinity, eqn (13) approaches the ultimate value [H2O2]t=∞, which is shown by eqn (15). [H2O2]t=∞ can be used as a convenient reference for evaluating the catalyst performance.

[H2O2] = {(rH2Sf)/(kd[Cat])}{1 − exp(–kd[Cat]t)}(13)
Se = [H2O2]/(rH2t)(14)
[H2O2]t=∞ = (rH2Sf)/(kd[Cat])(15)

By the way, reaction performance such as accumulation of H2O2 as well as H2O2 selectivity depends on the reaction conditions of course. The reaction conditions employed in our experiments, however, were not necessarily optimized to attain practical results, because the reaction system was designed to obtain appropriate kinetic data. In order to obtain better performance, conditions such as higher pressures and lower temperatures are generally employed. Whereas our reaction conditions were 0.1 MPa and 30 °C, Ishihara and his co-workers, for instance, employed conditions such as 1 MPa and 10 °C, and obtained the maximum H2O2 concentration such as around 1600 mmol L−1 using a Pd–Au colloid catalyst.26

We reported in our previous paper the influences of partial pressures pH2 and pO2 on the kinetic parameters over the Pd/C catalyst. Eqn (16) and (17), in which ks0, kd0 and Kx were constants, were attained for the reaction rate and the H2O2 degradation, whereas Sf was independent of pH2 and pO2.50 Assuming similar dependencies in the present catalyst systems, one can estimate the values of the kinetic parameters under any partial pressure using those obtained in the present experiments. It should be noted that the ratio kd/rH2 would be unaffected by pH2, whereas the ratio would be inversely proportional to pO2. Higher pO2 would suppress the degradation of H2O2 causing higher Se, although it would also suppress rH2 to some extent.

rH2 = ks0cH2[Cat]/{1 + Kx(pO2/cH2)}(16)
kd = kd0(cH2/pO2)/{1 + Kx(pO2/cH2)}(17)

We simulated the reaction time courses with respect to three catalyst systems, Pd-PVP, 1.64 atom% Pt-loaded Pd-PVP and 1.15 atom% Ir-loaded Pd-PVP, not only under ordinary pressure but also under 1 MPa. The conditions and the corresponding kinetic parameters as well as the values of [H2O2]t=∞ are shown in Table 4. Here, the individual values were determined by averaging the initial and the final values obtained in the experiment. Under 1 MPa, the values of pO2 were all set to 360 kPa, about 10 times higher than under ordinary pressure, and the values of cH2 to 100 μmol L−1, about the same value as in the case of Pd-PVP under ordinary pressure, in order to compare the results under the same conditions. The values of [H2O2]t=∞ under ordinary pressure were all lower than 300 mmol L−1 (about 1 wt%), although the values for Pt- and Ir- loaded Pd-PVP were considerably higher than that for the non-loaded Pd-PVP catalyst. On the other hand, the values under 1 MPa were significantly improved with the same order.

Table 4 Reaction conditions and kinetic parameters for time course simulation
A. Ordinary pressure (experimental conditions)
Loading (atom%) None Pt 1.64 Ir 1.15
p H2 (kPa) 16.36 14.75 17.12
c H2 (μmol L−1) 100.6 35.7 49.6
p O2 (kPa) 36.4 35.5 29.9
r H2 (mmol L−1 h−1) 25.5 85.7 90.2
S f (—) 0.858 0.424 0.460
k d[Cat] (h−1) 0.529 0.150 0.264
[H2O2]t=∞ (mmol L−1) 41 242 157

B. 1 MPa (cH2 and pO2 are given for simulation)
Loading (atom%) None Pt 1.64 Ir 1.15
c H2 (μmol L−1) 100 100 100
p O2 (kPa) 360 360 360
r H2 (mmol L−1 h−1) 12.6 50.9 48.0
S f (—) 0.858 0.424 0.460
k d[Cat] (h−1) 0.0265 0.0088 0.0117
[H2O2]t=∞ (mmol L−1) 409 2455 1891

Fig. 14 demonstrates the reaction time courses simulated using the kinetic parameters shown in Table 4, assuming that the catalysts were stable enough. One can easily understand that the values of [H2O2] approached those of [H2O2]t=∞ in relatively short times under ordinary pressure with rapid decreases in the values of Se. The decrease for the non-loaded Pd-PVP was particularly significant. Under 1 MPa, [H2O2] increased more effectively in all cases, and the values of [H2O2] reached about 1000 mmol L−1 at t = 60 h in the cases of the Pt- and Ir-loaded Pd-PVP catalysts. It should be noted that the values of Se at that time were 0.329 and 0.331, 78% and 72% of the initial values, respectively. In the case of the non-loaded Pd-PVP catalyst, the value of [H2O2] at t = 60 h was only 325 mmol L−1, and Se was more rapidly decreased from the initial value 0.860 to 0.430 by 50%. Practically the reaction time could be shortened by increasing the catalyst concentration. Even though the simulations were based on several assumptions such as the catalyst stabilities and extensibilities of the kinetics, the results in Fig. 14 suggest the promising performances of the Pt- and Ir-loaded Pd-PVP colloid catalysts. They should be experimentally examined in the future using an explosion-proof reactor, not available in our laboratory right now.

image file: c7cy01890h-f14.tif
Fig. 14 Time course simulation of H2–O2 reaction over non-loaded, 1.64 atom% Pt-loaded and 1.15 atom% Ir-loaded Pd-PVP catalysts under ordinary pressure and 1 MPa. Pd, 8.33 mg L−1; H2SO4, 0.01 N; NaBr, 0.001 N; 303 K, 1200 rpm. The kinetic parameters used are listed in Table 4.

4. Conclusions

The addition of 0.1–1.0 atom% of Pt and Ir brought about remarkable enhancement of the H2O2 formation rate of the Pd-PVP catalyst. Kinetic analyses of the modified catalysts indicated the following.

1) Only 0.5 atom% addition of Pt or Ir doubled the H2O2 formation rate of the Pd-PVP catalyst, while the addition of Ru, Rh, or Au affected little the activity in contrast to the effectiveness in alloy catalysts reported so far. Upon increasing the Pt/Pd or Ir/Pd ratio, the modified reaction rate, rH2/cH2, increased to 9 times the original value with the addition of 1 atom% Pt and became constant at 1.5 atom%. The H2O2 formation selectivity, Sf, decreased, while the H2O2 destruction rate parameter, kd[Cat]/cH2, was independent of the Pt/Pd or Ir/Pd ratio indicating that Pt or Ir was not involved in H2O2 destruction.

2) The kinetic parameters over Pt-loaded Pd-PVP varied with reaction time; rH2 increased and Sf and kd[Cat] decreased. Transformation of the catalyst from stage I to stage II was observed, in the former of which Pt atoms were located at coordinatively more unsaturated sites of the Pd particles and in the latter Pt atoms moved into the Pd lattice accompanied by rearrangement of the whole Pd conformation to a more stable structure. The transformation occurred more easily at lower pH2 and at higher temperature in the absence of H2.

3) The apparent activity per Pt atom of the Pt-loaded Pd-PVP in stage II was much higher than that of Pt-PVP, suggesting the peculiarity of atomically dispersed Pt atoms. The Pt atoms atomically dispersed in Pd would be more resistant to oxidation and adsorb more H2 with ease.

4) The catalysis of Ir-loaded Pd-PVP was almost similar to that in stage II on the Pt-loaded Pd-PVP catalyst. The rapid transformation from stage I to stage II was suggested at the initial stage of the H2–O2 reaction.

5) DFT studies were preliminarily performed to elucidate the great addition effects of Pt or Ir to the Pd-PVP colloid catalyst. The energy level of the adsorbed H atom on top of Pt was much lower than that on top of the surrounding surface Pd atoms forming an area, where H atoms would be efficiently adsorbed and their mobility would be higher than at the adsorbing sites on the Pd surface. The energy level for Ir was still lower. Those for Ru and Rh were higher than for Pt, though lower than for Pd.

6) The catalyst performances were evaluated by simulating the time courses of the H2–O2 reaction by using the obtained kinetic parameters for the non-loaded, 1.64 atom% Pt-loaded and 1.15 atom% Ir-loaded Pd-PVP catalysts. By elevating the reaction pressure (more correctly, pO2), Se was improved due to the decrease in kd[Cat]. H2O2 accumulation could reach 1000 mmol L−1 with an Se value of 0.33, under 1 MPa over the Pt- and Ir-loaded Pd-PVP, whereas the maximum accumulation (t = ∞) was only 400 mmol L−1 (Se = 0) over non-loaded Pd-PVP.

Conflicts of interest

There are no conflicts to declare.


The authors thank Tanaka Kikinzoku Kogyo K.K. for supplying the Pd-, Ru-, Rh-, Ir-, Pt- and Au-PVP colloid catalysts and providing helpful discussion. This work was financially supported by three Grant-in-Aids (JSPS, NEDO, and ALCA) from the Ministries MEXT and METI of Japan.


  1. Y. Yi, L. Wang, G. Li and H. Guo, Catal. Sci. Technol., 2016, 6, 1593–1610 CAS.
  2. J. H. Lunsford, J. Catal., 2003, 216, 455–460 CrossRef CAS.
  3. J. K. Edwards and G. J. Hutchings, Angew. Chem., Int. Ed., 2008, 47, 9192–9198 CrossRef CAS PubMed.
  4. C. Samanta, Appl. Catal., A, 2008, 350, 133–149 CrossRef CAS.
  5. R. Dittmeyer, J.-D. Grunwaldt and A. Pashkova, Catal. Today, 2015, 248, 149–159 CrossRef CAS.
  6. J. K. Edwards, S. J. Freakley, R. J. Lewis, J. C. Pritchard and G. J. Hutchings, Catal. Today, 2015, 248, 3–9 CrossRef CAS.
  7. K. Otsuka and I. Yamanaka, Electrochim. Acta, 1990, 35, 319–322 CrossRef CAS.
  8. I. Yamanaka, T. Onisawa, T. Hashimoto and T. Murayama, ChemSusChem, 2011, 4, 494–501 CrossRef CAS PubMed.
  9. T. Murayama, S. Tazawa, S. Takenaka and I. Yamanaka, Catal. Today, 2011, 164, 163–168 CrossRef CAS.
  10. T. Iwasaki, Y. Masuda, H. Ogihara and I. Yamanaka, Electrocatalysis, 2017 DOI:10.1007/s12678-017-0444-0.
  11. F. Thevenet, J. Couble, M. Brandhorst, J. L. Dubois, E. Puzenat, C. Guillard and D. Bianchi, Plasma Chem. Plasma Process., 2010, 30, 489–502 CrossRef CAS.
  12. M. Campos, W. Siriwatcharapiboon, R. J. Potter and S. L. Horswell, Catal. Today, 2013, 202, 135–143 CrossRef CAS.
  13. Y. Yi, J. Zhou, H. Guo, J. Zhao, J. Su, L. Wang, X. Wang and W. Gong, Angew. Chem., Int. Ed., 2013, 52, 8446–8449 CrossRef CAS PubMed.
  14. Y. Yi, L. Wang, Y. Yi, L. Wang, J. Yu, H. Guo, J. Zhang and C. Meng, AIChE J., 2017 DOI:10.1002/aic.15999.
  15. P. Landon, P. J. Collier, A. J. Papworth, C. J. Kiely and G. J. Hutchings, Chem. Commun., 2002, 2058–2059 RSC.
  16. J. K. Edwards, B. E. Solsona, P. Landon, A. F. Carley, A. Herzing, C. J. Kiely and G. J. Hutchings, J. Catal., 2005, 236, 69–79 CrossRef CAS.
  17. J. K. Edwards, A. Thomas, A. F. Carley, A. A. Herzing, C. J. Kiely and G. J. Hutchings, Green Chem., 2008, 10, 388–394 RSC.
  18. J. K. Edwards, N. E. Ntainjua, A. F. Carley, A. A. Herzing, C. J. Kiely and G. J. Hutchings, Angew. Chem., Int. Ed., 2009, 48, 8512–8515 CrossRef CAS PubMed.
  19. T. Ishihara, Y. Hata, Y. Nomura, K. Kaneko and H. Matsumoto, Chem. Lett., 2007, 36, 878–879 CrossRef CAS.
  20. Y. Nomura, T. Ishihara, Y. Hata, K. Kitawaki, K. Kaneko and H. Matsumoto, ChemSusChem, 2008, 1, 619–621 CrossRef CAS PubMed.
  21. M. Piccinini, E. Ntainjua, J. K. Edwards, A. F. Carley, J. A. Moulijn and G. J. Hutchings, Phys. Chem. Chem. Phys., 2010, 12, 2488–2492 RSC.
  22. S. J. Freakley, M. Piccinini, J. K. Edwards, E. N. Ntainjua, J. A. Moulijn and G. J. Hutchings, ACS Catal., 2013, 3, 487–501 CrossRef CAS.
  23. J. Li and K. Yoshizawa, Catal. Today, 2015, 248, 142–148 CrossRef CAS.
  24. D. Gudarzi, W. Ratchananusorn, I. Turunen, M. Heinonen and T. Salmi, Catal. Today, 2015, 248, 58–68 CrossRef CAS.
  25. F. Menegazzo, M. Manzoli, M. Signoretto, F. Pinna and G. Strukul, Catal. Today, 2015, 248, 18–27 CrossRef CAS.
  26. T. Ishihara, R. Nakashima, Y. Ooishi, H. Hagiwara, M. Matsuka and S. Ida, Catal. Today, 2015, 248, 35–39 CrossRef CAS.
  27. A. Villa, S. Freakley, M. Schiavoni, J. Edwards, C. Hammond, G. Veith, W. Wang, D. Wang, L. Prati, N. Dimitratos and G. Hutchings, Catal. Sci. Technol., 2016, 6, 694–697 CAS.
  28. S. Yook, H. Kwon, Y. Kim, W. Choi and M. Choi, ACS Sustainable Chem. Eng., 2017, 5, 1208–1216 CrossRef CAS.
  29. A. Staykov, T. Kamachi, T. Ishihara and K. Yoshizawa, J. Phys. Chem. C, 2008, 112, 19501–19505 CAS.
  30. H. C. Ham, G. S. Hwang, J. Han, A. W. Nam and T. H. Lim, J. Phys. Chem. C, 2009, 113, 12943–12945 CAS.
  31. R. Todorovic and R. J. Meyer, Catal. Today, 2011, 160, 242–248 CrossRef CAS.
  32. J. Li, A. Staykov, T. Ishihara and K. Yoshizawa, J. Phys. Chem. C, 2011, 115, 7392–7398 CAS.
  33. J. Li, T. Ishihara and K. Yoshizawa, J. Phys. Chem. C, 2011, 115, 25359–25367 CAS.
  34. H. Xu, D. Cheng and Y. Gao, ACS Catal., 2017, 7, 2164–2170 CrossRef CAS.
  35. P. Tian, L. Ouyang, X. Xu, C. Ao, X. Xu, R. Si, X. Shen, M. Lin, J. Xu and Y.-F. Han, J. Catal., 2017, 349, 30–40 CrossRef CAS.
  36. Q. Liu, J. C. Bauer, R. E. Schaak and J. H. Lunsford, Appl. Catal., A, 2008, 339, 130–136 CrossRef CAS.
  37. G. Bernardotto, F. Menegazzo, F. Pinna, M. Signoretto, G. Cruciani and G. Strukul, Appl. Catal., A, 2009, 358, 129–135 CrossRef CAS.
  38. J. Xu, L. Ouyang, G.-J. Da, Q.-Q. Song, X.-J. Yang and Y.-F. Han, J. Catal., 2012, 285, 74–82 CrossRef CAS.
  39. S. Sterchele, P. Biasi, P. Centomo, P. Canton, S. Campestrini, T. Salmi and M. Zecca, Appl. Catal., A, 2013, 468, 160–174 CrossRef CAS.
  40. E. N. Ntainjua, S. J. Freakley and G. J. Hutchings, Top. Catal., 2012, 55, 718–722 CrossRef CAS.
  41. S. Freakley, Q. He, J. Harrhy, L. Lu, D. Crole, D. Morgan, E. Ntainjua, J. Edwards, A. Carley, A. Borisevich, C. Kiely and G. Hutchings, Science, 2016, 351, 965–968 CrossRef CAS PubMed.
  42. S. Maity and M. Eswaramoorthy, J. Mater. Chem. A, 2016, 4, 3233–3237 CAS.
  43. P. Tian, X. Xu, C. Ao, D. Ding, W. Li, R. Si, W. Tu, J. Xu and Y.-F. Han, ChemSusChem, 2017, 10, 3342–3346 CrossRef CAS PubMed.
  44. G. J. Hutcings and C. J. Kiely, Acc. Chem. Res., 2013, 46, 1759–1772 CrossRef PubMed.
  45. S. R. Bharadwaj, A. S. Kerkar, S. N. Tripathi and S. R. Dharwadkar, J. Less-Common Met., 1991, 169, 167–172 CrossRef.
  46. H. Kobayashi, M. Yamauchi, R. Ikeda and H. Kitagawa, Chem. Commun., 2009, 4806–4808 RSC.
  47. H. Kobayashi, M. Yamauchi, H. Kitagawa, Y. Kubota, K. Kato and M. Takata, J. Am. Chem. Soc., 2010, 132, 5576–5577 CrossRef CAS PubMed.
  48. T. Deguchi, H. Yamano and M. Iwamoto, J. Catal., 2012, 287, 55–61 CrossRef CAS.
  49. T. Deguchi and M. Iwamoto, Ind. Eng. Chem. Res., 2011, 50, 4351–4358 CrossRef CAS.
  50. T. Deguchi, H. Yamano and M. Iwamoto, Catal. Today, 2015, 248, 80–90 CrossRef CAS.
  51. T. Deguchi, S. Takenouchi, H. Yamano and M. Iwamoto, Catal. Sci. Technol., 2016, 6, 4232–4242 CAS.
  52. T. Deguchi and M. Iwamoto, J. Catal., 2011, 280, 239–246 CrossRef CAS.
  53. G. A. Somorjai, J. Mol. Catal. A: Chem., 1996, 107, 39–53 CrossRef CAS.
  54. T. Deguchi and M. Iwamoto, J. Phys. Chem. C, 2013, 117, 18540–18548 CAS.
  55. S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M. J. Probert, K. Refson and M. C. Payne, Z. Kristallogr., 2005, 220, 567–570 CAS.
  56. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS PubMed.
  57. D. Vanderbilt, Phys. Rev. B, 1990, 41, 7892–7895 CrossRef.


Electronic supplementary information (ESI) available. See DOI: 10.1039/c7cy01890h
Present affiliation: Tokyo Ohka Kogyo Co., Ltd., 1590 Tabata, Samukawa-machi, Koza-gun, Kanagawa 253-0114, Japan.
§ Present affiliation: Tanaka Kikinzoku Kogyo K.K., 2-14 Nagatoro, Hiratsuka, Kanagawa 254-0021, Japan.
Present affiliation: Research Institute for Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan.

This journal is © The Royal Society of Chemistry 2018