The transformation of biomass and its derivatives to commodity chemicals is an attractive means of addressing the anticipated shortage of fossil fuel sources [1, 2]. As a by-product in biodiesel processing, glycerol is regarded as a biomass derivative and thus a renewable chemical building block [3, 4]. Tremendous efforts have been directed toward the conversion of glycerol to valuable chemicals through reduction processes such as dehydration [5, 6, 7, 8] and hydrogenolysis [9, 10, 11, 12, 13, 14, 15, 16, 17], while hydrogenolysis to propanediols is one of the most promising approaches. As shown in Scheme 1, the hydrogenolysis of glycerol produces 1,3-propanediol (1,3-PD), 1,2-propanediol (1,2-PD), 1-propanol (1-PO) and 2-propanol (2-PO), of which 1,3-PD is the most valuable. Generally, efficient catalysts for the production of 1,3-PD from glycerol are composed of a highly reducible noble metal (Pt, Ir, or Rh) together with a species containing an oxophilic metal such as Re or W [10, 11, 12, 13, 14, 15, 16, 17], including Ir-ReOx/SiO2, Pt-Re/C and Pt/WO3/AlOOH. Recently, our group proposed an Ir-Re alloy catalyst supported on KIT-6 (an ordered mesoporous silica with a cubic arrangement of interconnected pores) as a potential glycerol hydrogenolysis catalyst, and determined that this material exhibits the fastest yet 1,3-PD formation rate (25.6 mol·molIr−1·h−1) at 63.3% glycerol conversion [18].
The hydrogenolysis of glycerol is a structure-sensitive reaction [12, 19]. The catalytic performance relies heavily on the structure of the bimetallic catalyst [12, 13], which in turn is related to the catalyst preparation parameters [13, 14, 20, 21, 22, 23, 24]. As an example, variations in both the reduction and calcination temperatures can change the bimetallic Pt-Cu interaction and the extent of surface coverage by copper, thus affecting the catalytic performance of Pt-Cu catalysts during nitrate reduction [24]. Daniel and co-workers found that higher reduction temperatures result in improved atomic mixing of Pt and Re and better catalytic performance of a Pt-Re/C catalyst during the hydrogenolysis of glycerol [13]. Recently, we proposed that the bimetallic Ir-Re structure and thus its catalytic performance could be influenced by applying different thermal treatment procedures [18]. Producing the Ir-Re/KIT-6 catalyst without a calcination pretreatment renders the Re species fully reducible and thus the Re couples with Ir to form an Ir-Re alloy following reduction at 500 °C. On the contrary, the combination of calcination and subsequent reduction results in the formation of an Ir-ReOx structure [10, 17]. In our previous work, we proposed a possible bifunctional mechanism for the Ir-Re alloy catalyst involving both metallic Ir and acidic (Ir-)Re-OH sites. The higher surface acidity of the Ir-Re alloy catalyst was seen as possibly being responsible for its higher activity. Although analogous bifunctional mechanisms have been proposed in the cases of bimetallic Ir-Re, Rh-Re and Pt-Re catalysts during C-O scission reactions based on studies involving density functional theory (DFT) simulations [12, 23, 27, 33], to the best of our knowledge, the role of the surface acidity of a bimetallic Ir-Re catalyst and the possible bifunctional mechanism have not yet been fully demonstrated based on experimental results.
In the present work, a series of bimetallic Ir-Re/KIT-6 catalysts was prepared by a direct reduction method. The effects of the reduction temperature on the nature of the bimetallic Ir-Re catalysts and on their catalytic performance were studied as a means of examining the structure-activity relationship. This study therefore provides insights into the role of surface acidity during glycerol hydrogenolysis over bimetallic Ir-Re catalysts.
KIT-6 was synthesized using P123 (Aldrich, MW = 5800) as the structure-directing agent and TEOS (Shanghai Lingfeng Chemical Reagent Co., Ltd., 99.0%) as the silica source, according to a previously published procedure [30]. In a typical preparation, 4.0 g P123 was dissolved in a mixture of 144 g distilled water and 7.9 g HCl (35 wt%) and mixed at 35 °C for 3 h. After complete dissolution of the P123, 4.0 g n-BuOH was added and the mixture was stirred for 1 h. Subsequently, 8.6 g TEOS was added dropwise to the solution. The mixture was then stirred at 35 °C for 24 h before hydrothermal treatment at 110 °C in a polypropylene bottle for a further 24 h. The resulting precipitate was collected by filtration and then calcined at 550 °C in air for 5 h.
Bimetallic Ir-Re catalysts supported on KIT-6 were prepared by a consecutive incipient wetness impregnation method using an aqueous solution of H2IrCl6 (Sigma-Aldrich, 99.9%) and NH4ReO4 (Sigma-Aldrich, 99.0%) [18]. In a typical procedure, Ir/KIT-6 was first prepared by the incipient wetness impregnation of KIT-6 with the required amount of an aqueous solution of H2IrCl6, aged overnight and then dried at 120 °C for 12 h. The Ir-Re/KIT-6 was then obtained by impregnating the dried Ir/KIT-6 with an aqueous solution of NH4ReO4. The impregnated samples were aged overnight and then dried at 120 °C for 12 h. Subsequently, the samples were reduced at 400, 500, 600 or 700 °C in pure H2 (70 mL/min) for 3 h, then passivated in 1% O2/Ar for 20 min at room temperature. The exact loadings of Ir and Re were 4.1 and 3.8 wt%, respectively, as determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES) analysis.
Reactions were carried out in a 100-mL stainless steel autoclave (Parr Instruments) equipped with an electronic temperature controller and a magnetic stirrer. In a typical procedure, 20 g of an aqueous 20 wt% glycerol solution and 0.15 g catalyst were loaded into the autoclave. The reactor was purged three times with 2 MPa H2, then heated to 120 °C, pressurized to 8 MPa and stirred at 500 r/min. The samples were analyzed using ultra performance liquid chromatography (UPLC, Waters 2414) together with a refractive index (RI) detector and a C18 AQ column. The carbon balance in all tests was found to be greater than 95%. Conversion, selectivity and specific reaction rates were calculated as follows:
${\rm{Conversion}}\;(\% ) = \;\frac{{({\rm{mol}}\;{\rm{of}}\;{\rm{glycerol}}\;{\rm{converted}})}}{{({\rm{mol}}\;{\rm{of}}\;{\rm{glycerol}}\;{\rm{charged}})}}\; \times 100$ ${\rm{Selectivity}}\;(\% ) = \;$ $\frac{{({\rm{mol}}\;{\rm{of}}\;{\rm{product}}) \times ({\rm{number}}\;{\rm{of}}\;{\rm{carbon}}\;{\rm{atoms}}\;{\rm{in}}\;{\rm{the}}\;{\rm{product}})}}{{({\rm{Sum}}\;{\rm{of}}\;{\rm{carbon - based}}\;{\rm{mol}}\;{\rm{for}}\;{\rm{all}}\;{\rm{liquid}}\;{\rm{products}})}}\;$ ${\rm{Specific}}\;{\rm{rate (mol}} \cdot {\rm{mo}}{{\rm{l}}_{{\rm{Ir}}}}^{ - 1} \cdot {{\rm{h}}^{ - 1}}{\rm{)}} = \;\frac{{({\rm{mol}}\;{\rm{of}}\;{\rm{glycerol}}\;{\rm{converted}})}}{{({\rm{mol}}\;{\rm{of}}\;{\rm{Ir}}\;{\rm{charged}}) \times \;({\rm{reaction}}\;{\rm{time}})}}\;$
Metal loadings were determined by ICP-AES (Vanan 710). N2 adsorption-desorption was performed using a Micromeritics ASAP 2010C apparatus at −196 °C after out-gassing the samples at 300 °C and 133.3 Pa for 6 h. X-ray diffraction (XRD) was performed with a Rigaku D/Max 2550VB/PC diffractometer using Cu Kα (λ = 0.15406 nm) radiation. Transmission electron microscope (TEM) characterization of the reduced Ir-Re/KIT-6 catalysts was conducted using a JEOL JEM 2100 with an accelerating voltage of 200 kV and a point resolution of 0.18 nm. Mean particle sizes and particle size distributions were determined by measuring more than 200 randomly selected metal particles.
Temperature-programmed reduction (TPR) data were acquired using a Micromeritics AutoChem 2920 together with a U-shaped quartz reactor equipped with a thermal conductivity detector, under 5% H2/Ar (30 mL/min). The sample quantities ranged from 40 to 70 mg and the temperature was increased from ambient to 800 °C at 10 °C/min. CO chemisorption was carried out at 45 °C. Prior to each measurement, the sample was reduced in situ at 500 °C for 1 h under H2/Ar (40 mL/min), then purged with Ar at 530 °C for 30 min (40 mL/min) and finally cooled to 45 °C. The number of surface Ir sites was taken as equal to the irreversible CO uptake based on the CO diffuse reflectance infrared Fourier transform (DRIFT) spectra.
a Calculated using the BET method (SBET).
b Pore volume (Vp), calculated using the BJH method.
c Pore size (dp), calculated using the BJH method.
d Dispersion (CO/Ir), estimated from CO chemisorption.
e Estimated from NH3-TPD.
f Estimated from deconvolution of NH3-TPD profiles.
DRIFT spectra were recorded on a Nicolet 6700 FTIR spectrophotometer in the diffuse reflectance mode, using an in situ cell equipped with a sample cup, heater and ZnSe windows. Pre-reduced or passivated samples were placed into the sample cup and reduced in situ at 350 °C under a H2 flow for 1 h. Each sample was then cooled to room temperature, after which 2% CO/Ar was introduced until the CO coverage reached its saturation level. Spectra were collected at room temperature after purging with Ar.
Temperature-programmed desorption of ammonia (NH3-TPD) was performed using a Micromeritics AutoChem 2920 to assess the acidic properties of the Ir-Re/KIT-6 catalysts. All samples were pre-reduced under a hydrogen flow at a corresponding temperature for 3 h and subsequently passivated under 1% O2/Ar for 20 min (30 mL/min) at room temperature. The reduced Ir-Re/KIT-6 samples were used for NH3-TPD measurements, while the total amount of acid was also determined by measuring the NH3 uptake of the specimen. Prior to measurements, the sample was degassed in a He stream (20 mL/min) at 500 °C for 1 h and then cooled to 100 °C in He. The sample was subsequently exposed to a mixed gas flow of 10% NH3 and 90% He (50 mL/min) to allow for sufficient adsorption of NH3 and then exposed to a He flow for 30 min. Finally, the temperature was raised to 500 °C at 5 °C/min under a He flow (50 mL/min). The acid amounts were estimated from the calibrated peak areas of the NH3-TPD profiles based on the relationship between the peak area and the amount of NH3 desorbed.
N2 adsorption-desorption was performed to investigate possible changes in the textural properties of the Ir-Re/KIT-6 catalysts upon applying different thermal treatments. The resulting data (Table 1 and Fig. 1) indicate that the introduction of metal species results in a decrease in both the surface area and the pore volume, most likely because of the localization of metal particles inside the mesoporous channels [18]. Nevertheless, these data show that changes in the reduction temperature have only a minimal effect on the textural properties of the catalysts. All catalysts exhibit mesoporous characteristics with a similar mean pore size of approximately 7 nm.
Fig. 2 shows the TPR profiles of Ir/KIT-6, Re/KIT-6, and Ir-Re/KIT-6 without prior calcination treatments, as a means of assessing their reducibilities. Re/KIT-6 displays two hydrogen consumption peaks at 292 and 432 °C. Because the position of the lower temperature peak is exactly the same as that observed for a physical mixture of NH4ReO4 and KIT-6, this peak can be assigned to the reduction of physisorbed ReO4− species. The hydrogen consumption peak appearing in the Re/KIT-6 data at 432 °C could be due to weakly chemisorbed ReO4− species on the oxide support, possibly through Re-oxygen interactions resulting from the high oxophilicity of Re [27]. The TPR profile of Ir/KIT-6 shows three main hydrogen consumption peaks, possibly owing to the reduction of different ionic iridium species ([IrCl6]2−, [IrCl5]− and IrCl4) formed via ligand exchange [34], thus providing evidence for different types of iridium-support interactions [31].The presence of ionic iridium species is further demonstrated by the yellow appearance of the Ir/KIT-6 because IrO2 is black. These observations are in accordance with our previous report that the metal precursors are present in the form of ions on the hydrated support layer following impregnation and drying [18].
In the case of Ir-Re/KIT-6, two main hydrogen consumption peaks are observed at 160 and 227 °C. The peak at the higher temperature likely can be attributed to the reduction of chemisorbed Re species promoted by Ir. Compared with the Re/KIT-6 data, the hydrogen consumption peaks generated by bimetallic Ir-Re/KIT-6 are obviously shifted to lower temperatures. This result indicates close contact between Ir and Re and shows that Ir can promote the reduction of Re species in bimetallic Ir-Re/KIT-6 catalysts due to the hydrogen spillover effect. Furthermore, the metallic states were estimated based on the corresponding quantities of hydrogen consumption. It is evident that both Ir and Re species in monometallic catalysts can be reduced to metallic states. The formation of Re metal from Re/KIT-6-R500 and NH4ReO4 during TPR was further evidenced by XRD data (Fig. 3(a)). The full reduction of NH4ReO4 also indicates that the ReO4− species is reduced to the metallic state in one step rather than two [32]. In the case of Ir-Re/KIT-6, the hydrogen consumption amount equals that required to fully reduce both metal species. Based on these results, direct reduction of Ir-Re/KIT-6 catalysts above 250 °C can fully reduce both metal species and thereby generate an Ir-Re alloy.
Fig. 4 presents the TEM images and particle size distributions of Ir-Re/KIT-6 catalysts. It should be noted here that the reduction temperature was raised from room temperature to each target temperature at a very slow ramp rate of 1 °C/min to inhibit the possible aggregation of metal particles. It can be seen that metal particles are well dispersed on the supports, with mean particle sizes of 2.5 to 2.8 nm. Furthermore, all the Ir-Re/KIT-6 catalysts show very weak XRD peaks (Fig. 3b), similar to those previously reported for Ir-Re alloys [18, 25]. Since the crystal size detection limit of XRD is ca. 3 nm, this phenomenon could be ascribed to the high degree of metal dispersion and/or the possible presence of an amorphous phase [25, 26].
Fig. 5 shows the STEM-EDX elemental analysis data for randomly selected metal particles, and indicates the co-existence of both Ir and Re in a single metal particle. This result provides evidence for direct contact between Ir and Re. As suggested by H2-TPR results, the Ir and Re species undergo close interaction and can be completely reduced to metallic states. On this basis, it can be proposed that the Ir-Re alloy structure forms during reduction at 400-700 °C.
Fig. 6 summarizes the CO-DRIFT spectra of the Ir-Re/KIT-6 catalysts. In a previous study [18], red-shifts of the CO band as well as changes in the band shape were proven to be effective and sensitive indicators of the formation of Ir-Re alloy on Ir-Re/KIT-6 catalysts. As can be seen, all the Ir-Re/KIT-6 catalysts exhibit a similar CO adsorption band at approximately 2055 cm−1, a position that is exactly intermediate between the locations of 2068 and 2040 cm−1 that are assignable to Ir/KIT-6 and Re/KIT-6, respectively. Comparing the red-shift of the CO bands to the Ir/KIT-6 spectra suggests the increased back-donation of electrons from Ir to CO molecules as a result of the electron enrichment of Ir alloying with Re [29]. In addition, the bridged CO band at approximately 1800 cm−1 is absent for the Ir-Re/KIT-6 catalysts but is observed for the Ir/KIT-6, indicating that continuous surface Ir ensembles are separated by Re atoms. These results suggest that the Ir-Re/KIT-6 catalysts reduced at 400-700 °C are in the form of Ir-Re alloys.
Furthermore, it can be seen that the band intensities of these catalysts are very similar, implying that they possess similar CO adsorption capabilities. The Ir dispersions were further determined by CO chemisorption (Table 1). It should be noted that, as indicated by the CO-DRIFT spectra, the Re/KIT-6 shows negligible CO chemisorption compared to the Ir/KIT-6. This is consistent with previous observations of Pt-Re catalysts [12]. As such, the Ir dispersion was considered to be correlated to the quantity of chemisorbed CO. In keeping with the CO-DRIFT results, all the Ir-Re/KIT-6 catalysts exhibit similar levels of Ir dispersion, and these levels are significantly lower than that of the Ir/KIT-6 with a similar particle size (i.e., 2.3 nm). The lower CO chemisorption capability of Ir-Re/ KIT-6 compared with Ir/KIT-6 can be explained by (i) the changes in the electronic properties of Ir resulting from alloying with Re and (ii) the dilution effect of the Re.
The acidity of bimetallic M (Pt, Rh, Ru, or Ir)-Re catalysts can be assessed by NH3-TPD [12, 17, 23, 27, 36], and therefore this technique was employed to detect changes in the surface acidity of Ir-Re/KIT-6 catalysts with different reduction temperatures. As can be seen from Fig. 7, the NH3-TPD profiles of all Ir-Re/KIT-6 catalysts exhibit two peaks centered at approximately 185 and 260 °C, indicating their similar acid strengths. It should be noted that the siliceous KIT-6 support typically shows negligible surface acidity [28]. The acidity in these materials likely originates from Re-OH sites formed via the interaction between oxophilic Re and water molecules with the assistance of reducible metals such as Pt, Rh and Ir [12, 17, 23, 27]. The resulting Re-OH groups can act as Brönsted acid sites because of their ability to donate protons, due to the high oxophilicity of Re [23, 27, 35]. Moreover, the Re-OH interacts strongly with the Ir (i.e., Ir-Re-OH), increasing the electron deficiency of the Re (as indicated by CO-DRIFTS), and hence generating higher acid strengths. Therefore, the observed different strengths of acid sites could be due to varying extents of Ir-Re interaction. In this regard, as reported previously, Re species interacting weakly with Ir produce weaker acid strengths as evidence by the NH3-desorption peak centered at 175 °C in the case of the Ir-ReOx catalysts [18]. Furthermore, the deconvolution results show that increasing the reduction temperature leads to a higher percentage of strong acid sites (Table 1), possibly due to the stronger Ir-Re interaction at higher temperatures.
As shown in Table 1, the acid amount of Ir-Re/KIT-6 evidently increases with reduction temperature and reaches a maximum at 600 °C. The opposite trend was observed in the case of a Rh-Re/C catalyst by Chia et al. [23], such that the surface acidity of Rh-Re/C decreased from 64 to 38 μmol·gcat−1 with increases in the reduction temperature from 250 to 450 °C. These differing trends could be due to the different reduction temperatures employed and the nature of the metals. We propose that the higher number of acidic (Ir-)Re-OH sites in our work resulted from more uniform mixing of Ir and Re during reduction at high temperatures (Scheme 2), such that bimetallic Ir-Re/KIT-6 catalysts with more efficient Ir-Re interactions were obtained. Daniel et al. [13] has also reported that Pt-Re/C reduced at 700 °C shows better atomic mixing of Pt and Re than when reduced at 450 °C. In addition, the slight decrease in the acid amount in the case of the catalyst reduced at 700 °C could be the result of its larger particle size (2.8 vs. 2.5 nm).
Glycerol hydrogenolysis trials with Ir-Re/KIT-6 catalysts reduced at various temperatures were conducted using conditions consisting of 8 MPa H2, 120 °C and 20 g of an aqueous glycerol solution (20 wt%), with the results summarized in Table 2. It is evident that the activity (expressed as a specific rate) increases with the reduction temperature from 400 to 600 °C and then remains almost unchanged when increasing the temperature to 700 °C. A bifunctional mechanism for glycerol hydrogenolysis has been proposed for bimetallic M(Pt, Rh, Ir)-Re catalysts that possess both metallic M sites and acidic (M-)Re-OH sites [12, 17, 23, 27]. It has been suggested that the acidic (Ir-)Re-OH sites and the metallic Ir sites are responsible for glycerol adsorption and hydrogen activation during the hydrogenolysis reaction, respectively [10, 18]. Therefore, the catalytic performance of Ir-Re/KIT-6 during glycerol hydrogenolysis is related to both the Ir dispersion and the acidic properties of the material.
As noted, the reduction temperature has a negligible effect on the number of surface Ir sites. In addition, the acid strength is not obviously changed by the reduction temperature. Having excluded the effects of Ir dispersion and acid strength on the catalytic performance, the acid amount can be directly correlated with the reactivity (expressed as a specific rate), as shown in Fig. 8. The activity is observed to increase linearly with the acid amount. This linear relationship strongly indicates the direct influence of surface acidity on activity, and provides experimental evidence for the bifunctional mechanism of Ir-Re alloy catalysts.
Because the reaction order with respect to glycerol over Ir-Re/KIT-6 is zero [18], the adsorption of glycerol on acidic (Ir-)Re-OH sites must be rather strong [17, 18]. This means that the acidic (Ir-)Re-OH sites are involved in the rate-determining step and that acid density plays a vital role in determining the activity of Ir-Re catalysts. These findings provide fundamental knowledge concerning the role of Re in the Ir-Re alloy and suggest measures that could be taken to improve the performance of Ir-Re catalysts, such as increasing the Re content or enhancing the Ir-Re alloying extent.
The hydrogenolysis of glycerol over bimetallic Ir-Re catalysts generates the primary hydrogenolysis products 1,3-PD and 1,2-PD and the secondary hydrogenolysis products 1-PO and 2-PO [10]. As shown in Table 1, with increasing glycerol conversion, the selectivity for 1,3-PD decreases with a concomitantly increased selectivity for 1-PO. However, the selectivity for 1,2-PD does not undergo a significant decrease. The difference in the behaviors of 1,3-PD and 1,2-PD implies changes in the microstructures of the Ir-Re catalysts that might be related to the synergy between Ir and Re-OH groups and their changing acid strengths. The effect of acidity on product distribution therefore requires further investigation.
In this work, we investigated the effects of reduction temperature on the properties of bimetallic Ir-Re/KIT-6 catalysts prepared by a direct reduction method. The results showed that the bimetallic Ir-Re structures of catalysts reduced at 400-700 °C are in the form of Ir-Re alloys that exhibit similar mean particle sizes, Ir dispersions and acid strengths but varied acid densities. It has been found that increasing the reduction temperature from 400 to 600 °C leads to a higher number of surface acid sites on the Ir-Re/KIT-6 catalyst due to the more efficient Ir-Re interaction, resulting in increased activity during glycerol hydrogenolysis and decreased selectivity for 1,3-propanediol. However, the activity of an Ir-Re/KIT-6 catalyst reduced at 700 °C was observed to slightly decrease, likely as a result of the growth of the metal particles. Furthermore, a linear relationship between activity during glycerol hydrogenolysis and acidic (Ir-)Re-OH density was established. This fact strongly suggests the vital role of acidity in glycerol hydrogenolysis over Ir-Re alloys. The present study provides fundamental knowledge that should assist in optimizing bimetallic Ir-Re catalysts and designing novel bimetallic catalysts for glycerol hydrogenolysis.
2015, Vol. 36 Issue (10): 1750-1758
PDF (1143 KB)
2015, Vol. 36 Issue (10): 1750-1758
PDF (1143 KB)