Silver is one of the most effective modifiers of palladium [1]. Bimetallic Pd–Ag catalysts have been extensively used in industrially important gas-phase processes for the hydrogenation of acetylene traces contained in ethylene streams produced by petroleum cracking [2–6]. Recent research has also increasingly focused on the use of Pd–Ag-loaded catalysts for direct acetylene-to-ethylene production via selective hydrogenation [7–9]. Bimetallic particles with various compositions and structures serve as an active component in Pd–Ag catalysts. Due to the ability of palladium and silver to form systems distinguished by continuous mutual solubility of the components, the composition of the nanoalloys may vary widely [10]. Surface segregation of either component may lead to the generation of core–shells [11–13].

Alloying leads to changes in the electronic properties of metals. The 4d orbitals of metallic Pd are known to contain vacancies (a 4d shell contains 0.36 vacant electron holes per atom) due to the displacement of electrons from 4d to a vacant 5s sublevel [10]. The vacancies promote chemisorption of unsaturated compounds that act as electron donors, thus impairing selectivity due to the formation of ethane and oligomers [14]. Most researchers assume that silver compensates for the excess positive charge that appears on palladium atoms [10, 15]. The strength of ethylene adsorption on the surface of modified Pd decreases [16, 17]; hence, the ethylene readily desorbs and is not involved in further transformations. Moreover, introducing an second metal makes it possible to adjust the degree of isolation of Pd atoms on the surface of active particles and to generate sites with a specific geometry (the geometric effect) that hinders the adsorption of the target product [15, 18].

The state of the active component in Pd–Ag catalysts and, hence, their activity and selectivity are also affected by the support nature. It is important to choose a support with optimal textural and acid/base properties because these properties are responsible for the particle distribution on the surface, the probability of contact between the two metals, as well as the dispersion and electronic state of the palladium [19, 20].

Alumina, which has conventionally been used on an industrial scale as a support for Pd–Ag catalysts, has a disadvantages [2, 21]. The run and subsequent regeneration of alumina-supported catalysts may change the phase composition of the support, thus decreasing the accessibility of the active component. In particular, this may lead to the generation of strong Lewis acid sites, which will activate side oligomerization reactions and, ultimately, deactivate the catalyst. Carbon supports can serve as a promising alternative to oxide materials [19, 22]. Benavidez et al. [19] demonstrated that palladium supported on carbon black exhibits an appreciably higher selectivity in acetylene-to-ethylene hydrogenation than Al2O3- or MgO-supported catalysts, and they explain this advantage by the modified electronic properties of the metal. Carbon-supported Pd–Ag samples have been successfully tested by different research teams [22, 23]. However, research on Pd–Ag/C systems has been limited.

For the present study, we chose Sibunit as a support for Pd–Ag catalysts. Sibunit is a mesoporous carbon material distinguished by high ratings in properties such as thermal stability, mechanical strength, chemical purity, and developed specific surface area [24]. Our study was focused on interactions between palladium and silver in Pd–Ag/Sibunit catalysts that depend on the deposition sequence of metal precursors. The purpose was also to investigate the effects of the composition and structure of the active component on the catalytic performance in the hydrogenation of acetylene into ethylene.

EXPERIMENTAL

The mesoporous carbon material Sibunit (produced by a pilot plant of Center of New Chemical Technologies BIC, Omsk, Russia) was used as a catalyst support. Sibunit is a composite material based on globular and pyrolytic carbon [24, 25] with a specific surface area of 336 m2/g.

Pretreatment of the support consisted of a successive steps: washing with distilled water; grinding and screening to obtain 0.07–0.09 mm particles; oxidative treatment with a 5% nitric acid solution (to generate oxygen-containing functional groups on the support surface) [26]; and drying at 120°C for 2 h.

Pd–Ag/Sibunit samples and a Pd/Sibunit reference sample were synthesized by incipient wetness impregnation using aqueous solutions of palladium nitrate (prepared from PdCl2, AR grade, Krastsvetmet, Russia) and silver nitrate (AR grade, Krastsvetmet), followed by drying at 120°C for 2 h and reducing in H2 at 500°C for 3 h. To prepare a catalyst sample designated as Pd–Ag/Sibunit, the impregnating solution contained the salts of both metals, i.e. Pd(NO3)2 and AgNO3. Samples labeled Ag/Pd/Sibunit and Pd/Ag/Sibunit were prepared in two steps: initially, the first precursor solution—Pd(NO3)2 or AgNO3—was deposited, dried, and H2-reduced at 500°C for 3 h; then the second precursor solution was deposited, dried, and heat-treated under identical conditions. The finished samples contained 0.5 wt % Pd and 0.5 wt % Ag.

For X-ray diffraction analysis (XRD), a series of model samples were prepared by multiple impregnation. Each sample contained 7 wt % Pd and 7 wt % Ag because the XRD sensitivity was insufficient to analyze catalysts with lower metal concentrations. The analysis was carried out on a Bruker D8 Advance diffractometer in monochromated CuKα radiation in the 2θ range of 20°–90° with a step of 0.020° and counting time of 2 s/step (40 kV, 40 mA). The XRD patterns were processed using the ICDD PDF-2 (2006) database.

The catalysts were further examined by transmission electron microscopy (TEM) using a JEOL JEM 2100 microscope with an accelerating voltage of 200 kV and a resolution of 0.14 nm to characterize the morphology and check if the phase composition of the model catalysts agreed with that of the samples that contained 0.5 wt % of each component. The Pd–Ag/Sibunit, Pd/Ag/Sibunit, and Ag/Pd/Sibunit samples were ground in an agate mortar for 30 s. The powders were dispersed in ethanol using an UZDN-2T ultrasonic disintegrator, and the resultant suspension was deposited on nickel grids. The chemical composition of the samples was identified using an Oxford Instruments INCA-250 energy dispersive X-ray spectrometer equipped with a Si(Li) semiconductor detector (energy resolution 130 eV).

IR spectroscopy was used to characterize the surface functional composition both of the carbon support and the catalysts. The sample was ground in an agate mortar and sputter-deposited on a BaF2 plate as a uniform thin film in a 25-cm-high glass cylinder by sedimentation of fine particles. IR spectra were recorded on a Shimadzu IR Prestige-21 spectrometer in the range of 900–2000 cm–1 with a resolution of 4 cm–1. The spectra were obtained by averaging 50 scans.

The samples were exposed to a series of catalytic tests in the hydrogenation of acetylene into ethylene. The tests were carried out in a flow-type reactor at atmospheric pressure under heating from 25 to 95°C with step of 10°. A 0.0100 g catalyst sample was mixed with an inert diluent (SiO2) until a 1 cm3 layer was obtained. Prior to testing, the reactor was purged with nitrogen, after which a gas mixture of 4 vol % C2H2 in hydrogen was injected at a flow rate of 100 mL/min. The gas hourly space velocity (GHSV) was 600 000 mL g–1 h–1 in all test runs.

The reaction products were analyzed by GC on a Chromos GC-1000 chromatograph equipped with a 25 m×0.32 mm capillary column (60°C, SiO2 stationary phase, N2 carrier gas) and a flame ionization detector. The acetylene conversion (X, %), ethylene selectivity (SC2H4, %), and ethane selectivity (SC2H6, %) were evaluated by internal normalization [27]:

$$X = {{{[^{{\rm{feed}}}}{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{2}}}}-{\;^{{\rm{prod}}}}{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{2}}}}]} \over {^{{\rm{feed}}}{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{2}}}}}} \times 100,$$
$${S_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{4}}}} = {{{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{4}}}}} \over {{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{4}}}} + {N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{6}}}} + {N_{\rm{C}}}_{_{\rm{4}}{\rm{ + }}}}} \times 100,$$
$${S_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{6}}}} = {{{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{6}}}}} \over {{N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{4}}}} + {N_{\rm{C}}}_{_{\rm{2}}{{\rm{H}}_{\rm{6}}}} + {N_{\rm{C}}}_{_{\rm{4}}{\rm{ + }}}}} \times 100,$$

where N is the content of the compound indicated in the subscript (mol %); and feedNC2H2 and prodNC2H2 are the acetylene content in the reagent feed mixture and in the reaction product mixture, respectively (mol %).

The oligomer selectivity (SC4+, %) was calculated by the equation:

$${S_{{{\rm{C}}_{4 + }}}} = 100 - {S_{{{\rm{C}}_{\rm{2}}}{{\rm{H}}_{\rm{4}}}}} - {S_{{{\rm{C}}_{\rm{2}}}{{\rm{H}}_{\rm{6}}}}}.$$

Furthermore, the ethylene yield (YC2H4, %), i.e. the percentage of acetylene converted into ethylene, was evaluated as follows:

$${Y_{{{\rm{C}}_{\rm{2}}}{{\rm{H}}_{\rm{4}}}}} = X{{{S_{{{\rm{C}}_{\rm{2}}}{{\rm{H}}_{\rm{4}}}}}} \over {100}}.$$

The catalyst activity (A, mL[C2H2] gPd–1 s–1) was measured as the amount of acetylene converted over one second per 1 g Pd in the catalyst:

$$A = {{v \cdot X} \over {{m_{{\rm{Pd}}}}}} \times 100,$$

where v is the acetylene volume flow rate (mL/min); and mPd is the weight of Pd in the catalyst (g).

RESULTS AND DISCUSSION

XRD characterization. Figure 1 illustrates the XRD patterns of the support, as well as those of the Pd and Pd–Ag catalysts prepared with different precursor deposition sequences. For the model sample prepared by impregnating the support with a solution that simultaneously contained palladium and silver nitrates (Pd–Ag/Sibunit), two phases were identified: metallic silver and a Pd0.6Ag0.4 bimetallic compound (Table 1). The second phase was a substitutional alloy with a face-centered cubic (FCC) lattice type.

Fig. 1.
figure 1

XRD patterns of model catalyst samples: 7% Pd/Sibunit and 7% Pd–7% Ag/Sibunit.

Full size image
Table 1. XRD data for model catalyst samples: 7% Pd/Sibunit and 7% Pd–7% Ag/Sibunit
Full size table

The XRD patterns of the catalysts prepared by sequential deposition of the metal precursors (Ag/Pd/Sibunit and Pd/Ag/Sibunit) differ noticeably from those of Pd–Ag/Sibunit. The common feature of the XRD patterns of Ag/Pd/Sibunit and Pd/Ag/Sibunit is the presence of strong narrow reflections near 2θ of 38°, 44°, 64°, 77°, and 82°, assigned to the FCC faces (111), (200), (220), (311), and (222) of metallic silver (Fig. 1) [28]. The large extent of the coherent scattering region (CSR) of the Ag phase (about 24–36 nm) indicates that silver appeared as coarse particles (Table 1). In addition, to the right of the silver signals, the X-ray patterns of Ag/Pd/Sibunit and Pd/Ag/Sibunit contain weaker reflections associated with the formation of PdxAg(1–x) bimetallic particles.

TEM characterization. The TEM image of Pd–Ag/Sibunit (Fig. 2a) clearly shows, in agreement with our previous studies [8, 29], the presence of uniformly sized dispersed particles with an average diameter of 5.6 nm, mostly in the PdxAg(1–x) phase (d = 0.228 nm [30]). Furthermore, this sample contained some amounts of individual Pd particles (d = 0.225 nm, ICSD no. 41517) and Ag particles (d = 0.236 nm, ICSD no. 52257).

Fig. 2.
figure 2

TEM micrographs of 0.5% Pd–0.5% Ag/Sibunit catalysts prepared with different deposition sequences of Pd and Ag precursors: (a) Pd–Ag/Sibunit; (b) Ag/Pd/Sibunit; and (c) Pd/Ag/Sibunit.

Full size image

On the other hand, relevant EXAFS spectroscopy data (extended X-ray absorption fine-structure spectroscopy, a method that enables researchers to examine local atomic structures) show that palladium in a catalyst of this kind is predominantly surrounded by silver, and only a minority of metals exists as individual particles [8, 29]. In the case of Ag/Pd/Sibunit and Pd/Ag/Sibunit, particles varied widely in size. Along with 4–5 nm particles, large agglomerates up to 50–60 nm in diameter were also detected (Figs. 2b and 2c).

For a more detailed examination of Ag/Pd/Sibunit and Pd/Ag/Sibunit, a number of specific regions were mapped. The elemental mapping images for Ag/Pd/Sibunit (Fig. 3a) show a fairly uniform distribution of palladium on the support surface. In contrast, silver was mostly clustered within separate areas. Pd/Ag/Sibunit exhibited more uniform surface distributions both of palladium and silver (Fig. 3b) than Ag/Pd/Sibunit.

Fig. 3.
figure 3

Elemental mapping images: (a) Ag/Pd/Sibunit; and (b) Pd/Ag/Sibunit.

Full size image

The areas containing the largest particles (Fig. 4a) were further examined by local energy dispersive X-ray spectroscopy (EDX). These data are typical of monometallic particles (Table 2). Nonetheless, bearing in mind the presence of some silver uniformly distributed on the catalyst surface and the XRD data for the model systems, it would be fair to posit that a portion of silver entered into contact with palladium to form bimetallic particles and/or interfacial contacts between the two metals.

Fig. 4.
figure 4

TEM micrographs: (a) Ag/Pd/Sibunit; and (b) Pd/Ag/Sibunit.

Full size image
Table 2. Elemental compositions (wt %) of areas denoted as Spectrum 1–6 in TEM micrographs of Ag/Pd/Sibunit and Pd/Ag/Sibunit (see Fig. 4)
Full size table

The EDX data clearly show both the mono- and bimetallic natures of the largest agglomerates (Fig. 4b, Table 2). On the other hand, the XRD data indicate that finer-dispersed particles also represented both individual metals and PdxAg(1–x) bimetallic compounds.

The causes of the non-uniformity in size and composition of the particles formed in the catalysts synthesized by sequential deposition of precursors are worth closer consideration. It would be reasonable to assume that, as a result of the anchoring of the first metal (Pd or Ag) on the Sibunit by H2 treatment at 500°C, the O-containing functional groups (primarily carboxyl groups) located on the support surface were reduced. These groups are known to act as anchoring sites for active component precursors and, thus, suppress the reducibility of the carbon material [31–33]. The deposition of the second metal precursor on the surface free of oxygen-containing groups might lead both to the non-uniform distribution of the active component and to rapid reduction of a major portion of the second metal (Ag or Pd) precursor with larger particles being generated.

Sibunit, as a graphite-like carbon–carbon material, is known to possess electrochemical properties responsible for its redox interaction with metal precursors to form large metal particles [31, 32]. The TEM and elemental mapping images obtained in the present study indicate that the generation of large particles in Pd/Ag/Sibunit and Ag/Pd/Sibunit most likely followed a similar mechanism. The rapid reduction of the second metal precursor appeared to limit the chances of contact between the two metals, thus resulting in the formation both of alloys and of individual Pd and Ag particles.

Characterization of surface functional compositions of support and catalysts by IR spectroscopy. To confirm the hypothesis that the carbon support surface chemistry affects the generation of active particles in Pd–Ag/Sibunit, we examined the support as well as the H2-treated (at 500°C) Pd/Sibunit and Ag/Sibunit samples by IR spectroscopy.

The IR spectrum of the Sibunit support displays a strong peak near 1560 cm–1 attributed to C=C stretching vibrations in the aromatic rings of conjugated systems (Fig. 5). The broad peak at 1215 cm–1 corresponds to C–O stretching vibrations in lactones and phenol ethers. Between 1000 and 1200 cm–1, a number of peaks assigned to C–O vibrations in phenols and alcohols were observed. Along with these signals, the IR spectrum contains a weak peak near 1737 cm–1 (attributed to C=O stretching vibrations in carboxylic acids, ketones, and esters) and a peak near 1380 cm–1 (C–O stretching vibrations in carbonate ions).

Fig. 5.
figure 5

IR spectra of Sibunit support before/after treatment with 5% HNO3 solution, and of Pd/Sibunit and Ag/Sibunit catalysts after H2 reduction at 500°C.

Full size image

Treating the support with the 5% HNO3 solution led to the almost complete disappearance of C–O bonds in carbonate ions. On the other hand, this markedly increased the intensity of the peaks attributed to oxygen-containing moieties, such as the 1737 cm–1 peak corresponding to C=O stretching vibrations. In addition, a new weak peak, typical of C–OH bending vibrations in carboxylic acids, appeared at 916 cm–1. The appearance of oxygen-containing (e.g., carboxyl) groups on the support surface enhanced the Sibunit’s affinity for adsorptive interaction with the active component precursors, thus favoring uniform distribution of the precursor throughout the entire surface [33, 34].

The deposition of metal nitrate salts followed by reduction in H2 at 500°C resulted in the disappearance of the 916 cm–1 peak and a significant intensity drop of the 1737 cm–1 peak, indicating that the carboxyl groups were removed from the support surface. Furthermore, the lactone groups underwent partial degradation. Thus, it is safe to assume that, after the first metal was anchored, the depletion of the functional composition of the support surface caused a less uniform distribution of the second metal precursor and/or its partial reduction during the deposition and drying [31–33, 35].

Catalyst performance in acetylene hydrogenation. Figure 6 illustrates the catalytic test data for the samples in acetylene hydrogenation.

Fig. 6.
figure 6

Performance of Pd/Sibunit and Pd–Ag/Sibunit catalysts synthesized with different deposition sequences of Pd and Ag precursors in acetylene hydrogenation (GHSV = 600 000 mL g–1 h–1).

Full size image

The activity and selectivity of the catalysts display the following trend: Pd/Sibunit → Ag/Pd/Sibunit → Pd/Ag/Sibunit → Pd–Ag/Sibunit. This is the order in which the acetylene conversion decreased: the point XC2H2 = 50% on the conversion curve shifted from 30°C for Pd/Sibunit to 48°C for Pd–Ag/Sibunit (Fig. 6a). The catalytic activity measured at 35°C consistently declined from 964 mL(C2H2) gPd–1 s–1 for Pd/Sibunit to 805, 473, and 315 mL(C2H2) gPd–1 s–1 for Ag/Pd/Sibunit, Pd/Ag/Sibunit, and Pd–Ag/Sibunit, respectively. This can likely be explained by the increase in the number of less active bimetallic sites in the catalysts under study [15, 36]. The target product selectivity predictably increased in the same order (Fig. 6b): it grew from 63% for Pd/Sibunit to 68, 73, and 79% (at XC2H2 ≤ 70%, T = 25–40°C) for Ag/Pd/Sibunit, Pd/Ag/Sibunit, and Pd–Ag/Sibunit, respectively, primarily due to the decreasing content of the ethane byproduct (Fig. 6c).

The regularities found in the hydrogenation of acetylene are associated with the different types of particles formed in the Pd–Ag samples. The similarity in the properties of Ag/Pd/Sibunit and Pd/Sibunit can be explained by the fact that the metals in Ag/Pd/Sibunit primarily appeared as separate metallic phases. Only minor portions of Pd and Ag mutually interacted, thus causing the observed acetylene conversion drop along with the slight increase in selectivity. It is possible, the selectivity increase (as well as the activity drop) of Pd/Ag/Sibunit and, especially, of Pd–Ag/Sibunit was due to a large amount of PdxAg(1–x). These selective bimetallic particles adsorbed unsaturated hydrocarbons less strongly, thus ensuring rapid desorption of ethylene and preventing it from being completely hydrogenated into ethane [15, 16, 36]. In the case of Pd–Ag/Sibunit, its uniformly sized particles might provide additional improvement.

The maximum ethylene yield of the monometallic sample was 48% (Fig. 6e). The catalyst modification enhanced the ethylene yield up to 50, 64, and 71% for Ag/Pd/Sibunit, Pd/Ag/Sibunit, and Pd–Ag/Sibunit, respectively. It is also worth noting that, during the 10-h acetylene hydrogenation test at 85°C, Pd–Ag/Sibunit (the sample synthesized by co-impregnation with palladium and silver nitrate solutions) exhibited stable catalytic performance [37]. With this sample, high selectivity (73%) remained even at 97% acetylene conversion.

CONCLUSIONS

A series of bimetallic Pd–Ag catalysts were synthesized by depositing on a Sibunit mesoporous carbon support and tested in acetylene hydrogenation. For these catalysts, the effects of the deposition sequence of palladium and silver nitrate salts on the formation of the active component were investigated. Sequential deposition of precursors interleaved with a hydrogen treatment of the catalyst was shown to be less effective for the formation of highly selective PdxAg(1–x) bimetallic particles than using a solution that simultaneously contained precursors of both metals. The sequential synthesis preferentially produced particles non-uniform in size and composition because the anchoring of the first metal precursor by the H2 reduction led to the removal of oxygen-containing groups from the support surface (the O-groups are known to act as precursor anchoring sites and suppress the reducibility of the carbon material). The deposition of the second metal precursor on the reduced surface resulted in the formation of larger individual metal particles. When using the impregnating solution that contained both metal precursors, followed by H2 reduction, uniformly sized Pd0.6Ag0.4 bimetallic particles (dav = 5.6 nm) were predominantly formed, and this allowed ethylene to be produced by acetylene hydrogenation with a high yield of 71%.