Article pubs.acs.org/IECR
Hydrogenation of Dimethyl Oxalate over Copper-Based Catalysts: Acid−Base Properties and Reaction Paths Yanbo Song, Jian Zhang, Jing Lv, Yujun Zhao,* and Xinbin Ma Key Laboratory for Green Chemical Technology of Ministry of Education, Collaborative Innovation Center of Chemical Science and Engineering, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China S Supporting Information *
ABSTRACT: Hydrogenation of dimethyl oxalate (DMO) is a potentially important process in C1 chemistry, which produces ethylene glycol (EG) around 473 K and ethanol with higher alcohols (propanol, butanol, etc.) around 553 K. However, the detailed inter-relationship of formation paths for these products has not yet been discussed properly. In this study, we found that the formation paths of higher alcohols from DMO were inhibited around 473 K. On the basis of these results, a two-reactor system with different reaction temperatures was suggested to obtain less of the higher alcohols and more ethanol. Besides, it is found that a higher density of basic sites in the catalyst favors the formation of higher alcohols. An aluminum dopant was applied to decrease the basic sites in the catalysts accompanied by an increment of ethanol selectivity. When the aluminum-doped catalysts were introduced into the two-reactor system, a further improvement in ethanol selectivity was achieved.
1. INTRODUCTION With increasing attention to the energy crisis and environmental issues, more and more researchers focus on catalytic methods for the efficient and clean utilization of coal resources, causing a rapid global development of C1 chemistry.1 As one of the promising applications to C1 chemistry, the hydrogenation of dimethyl oxalate (DMO) drew much attention since the mass production of DMO was commercially realized by the synthesis method via syngas.2 Gong and his co-workers3 found that the distribution of its product varied with reaction temperature. Around 553 K, the major product of DMO hydrogenation is ethanol, which is a versatile feedstock for the synthesis of various products (e.g., chemicals, fuels, and polymers) and has also been commercially used as an additive or a potential substitute for gasoline.4 Accompanied with ethanol, higher alcohols (propanol, butanol, etc.) would be also produced during this hydrogenation process, diminishing the selectivity of ethanol.3 Therefore, to achieve better ethanol yield, it is meaningful to clarify the paths of higher alcohol generation and find a method to inhibit this process. As shown in eqs 1−8, selective catalytic hydrogenation of DMO involves several reactions. DMO reacted with hydrogen to methyl glycolate (MG), first, and then ethylene glycol (EG) generated from the hydrogenation of MG. At last ethanol would be obtained by the hydrogenation of EG (shown in eq 1−3). 1,2-Propanediol, 1,2-butanediol (C3−C4 diols), 2methoxyethanol (EGME), propyl alcohol (PrOH), and butyl alcohol (BuOH) (PrOH, and BuOH named C3−C4OH) as the side product would be produced through eq 4, 5, 6, 7, and 8, respectively. On the basis of the previous work, it was found that the products distribution of DMO hydrogenation varied greatly with reaction temperature upon copper-based catalysts.3 Around 473 K, the major product is EG (selectivity is 95%) with little MG (3%) and C3−C4 diols (2%).5,6 When the reaction temperature was raised to 553 K, the major product changed to ethanol and the major byproduct is C3−C4OH (eq © XXXX American Chemical Society
7 and 8). It has been supposed that C3−C4OH is probably produced via the Guerbet reaction on basic sites.7,8 However, the specific process of its formation from DMO is still unclear. CH3OOCCOOCH3 + 2H 2 → HOCH 2COOCH3 + CH3OH
(1)
HOCH 2COOCH3 + 2H 2 → HOCH 2CH 2OH + CH3OH (2)
HOCH 2CH 2OH + H 2 → C2H5OH + H 2O
(3)
HOCH 2CH 2OH + CH3OH → HOCH 2CH(OH)CH3 + H 2O
(4)
HOCH 2CH 2OH + C2H5OH → HOCH 2CH(OH)CH 2CH3 + H 2O
(5)
HOCH 2CH 2OH + CH3OH → HOC2H4OCH3 + H 2O (6)
HOCH 2CH 2OH + CH3OH + H 2 → C3H 7OH + 2H 2O (7)
HOCH 2CH 2OH + C2H5OH + H 2 → C4H 9OH + 2H 2O (8)
As shown in eqs 7 and 8, Guerbet reaction is the dimerization via the alcohol condensation of alcohol, like ethanol, propanol, and butanol. Veibel and Nielsen9 proposed that the mechanism of the Guerbet reaction is based mainly on three subsequent steps, including (a) dehydrogenation of Received: May 25, 2015 Revised: September 15, 2015 Accepted: September 18, 2015
A
DOI: 10.1021/acs.iecr.5b01928 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research
K. Before the analysis was conducted, the sample was degassed at 573 K for 3.5 h in vacuum. Pore size distribution was figured out by the method of Barrett−Joyner−Halenda (BJH) from the desorption branches of the adsorption isotherms. Mesoporous surface areas were calculated from the isotherms through the method of Brunauer−Emmett−Teller (BET). The actual copper content in the catalyst was determined by an inductively coupled plasma optical emission spectrometry (ICP-OES). The well-weighed sample was dissolved in the HF solution, neutralized with HBO3, and finally diluted with water to 50 mL. X-ray powder diffraction (XRD) measurement was carried out by employing a Rigaku C/max-2500 diffraction meter through the graphite filtered CuKa radiation (λ = 1.5406 Å) at room temperature. The Scherrer equation was used to test the particle size of the copper species in calcined and reductive catalysts. With the purpose of avoiding phase transformation from surface oxidation, catalysts were carefully collected under hydrogen atmosphere at room temperature and sealed into glass bottles before XRD analysis. The small diffraction data of calcination samples were recorded at an interval of 0.01° and a scanning speed of 4°/min from 2θ = 1° to 2θ = 10°. The wideangle data were obtained by the step of scanning with a rate of 12°/min from 2θ = 10° to 2θ = 90°. The specific surface area of metallic copper was measured by the dissociative N2O chemisorption and hydrogen pulse reduction described in the literature.18,19 Briefly, 100 mg of catalyst sample was reduced in 10% H2−Ar at 623 K for 4 h and cooled to 323 K. Then N2O was introduced to oxidize for 2 h, ensuring that surface Cu atoms were completely oxidized according to the reaction. The quantity of chemisorbed N2O was measured on a Micromeritics Autochem II 2910 equipped with a TCD by a hydrogen pulse chromatographic technique. The surface area of metallic copper was estimated from the total amount of hydrogen consumption with 1.46 × 1019 copper atoms per square meter.20,21 Acid and basic properties of catalysts were investigated by NH3 and CO2 temperature-programmed desorption (TPD) on a Micromeritics Autochem II 2910, equipped with a TCD. With regard to the basicity evaluation by CO2-TPD and NH3TPD, 100 mg of the catalyst sample in a quartz U tube was first reduced in 5% H2−Ar at 623 K for 4 h, and then cooled to 333 K. Besides, the gas switched to CO2 or NH3, which passed through the catalyst bed for 1 h. Weakly adsorbed CO2 or NH3 was removed by pure argon sweeping at room temperature until the baseline of TCD signals were stabilized. Under an atmosphere of pure argon, TPD was performed from room temperature to 1073 K at a heating rate of 10 K/min. 2.3. Catalytic Activity Test. As shown in Figure S2, two catalytic reaction systems were used in this study. One-step system consisted of a fixed-bed stainless steel reactor. Cu/ SiO2−xAl catalysts were placed in the constant temperature part of the reactor and silica wool was put both above and under the catalyst so as to fix it. The catalyst bed was packed with 0.5 g of catalyst particles (40−60 meshes), with an inner diameter of 8 mm and a height approximate of 33 mm. The catalyst was activated for 4 h in pure hydrogen atmosphere at 623 K under 2.5 MPa. After the temperature was cooled to 553 K, the reactant (20 wt % DMO in methanol solution) was injected into the reactor through a high-pressure pump (Lab Alliance Series II Pump). The system pressure of H2 was 2.5 MPa and the reaction temperature was 553 K. The amount of H2 flow was determined by the molar ratio of H2/DMO, which
alcohols to the corresponding aldehydes, (b) aldol condensation of the resulting aldehydes with the help of basic sites, as well as (c) hydrogenation of the unsaturated condensation products to produce the higher alcohols. The process of the Guerbet reaction is shown in Figure S1A and S1B. Takashi Tsuchida and co-workers10 found that hydroxyapatite catalyst with basic sites can conduct the Guerbet reaction. Carlini et al.11,12 reported a cross condensation between alcohols promoted by a bifunctional catalyst, containing both a dehydrogenating/hydrogenating metal species and a basic component, which facilitates aldol condensation. Copper catalysts with basic components were also reported in the Guerbet reaction.13 The reaction should occur on basic sites if C3−C4OH is generated via a Guerbet reaction. Therefore, the selectivity of C3−C4OH would be varied with different acid− base properties of catalysts and a method should be found to modify these properties, efficiently. As a common dopant for modulating the acid−base properties, aluminum has been applied in many catalysts. Several researchers found that the incorporation of alumina into many supports, such as MCM-4114 and SBA-315 as well as zeolites,16 can strengthen the acid sites associated with the presence of aluminum in the framework positions. Therefore, aluminum as dopant may be a choice to modify the acid and basic properties of the copper-based catalyst. Herein, a kind of aluminum modified Cu/SiO2 catalyst was prepared to ensure that the path leading to C3−C4OH production is the Guerbet reaction. We also obtained deep insight into the influence of aluminum doping on the structure and acid and base sites of the catalysts. Meanwhile a tworeactor system is proposed to provide guidance for achieving higher ethanol selectivity.
2. EXPERIMENT 2.1. Catalyst Preparation. Though the ammonia evaporation hydrothermal (AEH) method was better for the preparation of the Cu/SiO2 catalyst in our previous work,3,17 the commonly used ammonia evaporation (AE) method instead of AEH method was used in this study, with the purpose of distinguishing the major reasons for the side reactions more accurately. The catalyst preparation procedure is described as follows: Cu/SiO2−xAl catalysts: Cu/SiO2−xAl catalysts were prepared by an AE method, which was described briefly as follow: a certain amount of Cu(NO3)2·3H2O and Al(NO3)3·6H2O was dissolved in 100 mL of deionized water, then 25 wt % ammonia aqueous was added and stirred for 10 min. Subsequently, a required amount of 30 wt % silica sol was added to this solution and the mixture was stirred for 2 h with the initial pH of suspension adjusted to 11−12. All of these operations were conducted at room temperature. To deposit the copper and aluminum species on silica, the suspension was heated in a water bath from room temperature to 353 K to evaporate ammonia. The water temperature was kept at 353 K until the pH value of the suspension decreased to 6−7. The filtrate was washed with deionized water for three times and then dried at 353 K in vacuum atmosphere for 10 h. The resulting solid was calcined at 623 K for 4 h in air. The calcined sample was recorded as Cu/SiO2−xAl, where x stood for the weight percentage of alumina in the whole catalyst. 2.2. Catalyst Characterization. Textural properties of the catalyst were measured by the method of nitrogen adsorption using a Micromeritics TristarII 3000 analyzer instrument at 77 B
DOI: 10.1021/acs.iecr.5b01928 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research Table 1. Textural Parameters of Cu/SiO2 Catalysts
a
catalyst
Cu loadinga (%)
SiO2 Cu/SiO2 Cu/SiO2−0.5Al Cu/SiO2−1.0Al Cu/SiO2−1.5Al
19.6 18.9 19.3 19.0
Al loadinga (%)
specific surface area (m2/g)
pore volumeb (cm3/g)
pore sizec (nm)
copper dispersiond (%)
copper surface aread (m2/g)
0.5 1.0 1.5
102 386 355 344 240
0.30 1.38 1.34 1.40 1.37
11.9 11.5 11.2 14.9 20.1
10.3 10.5 9.5 7.6
12.7 12.9 11.9 10.3
Determined by ICP-OES. bAverage pore volume determined by BET. cMean pore diameter determined by BET. dDetermined by N2O titration.
Figure 1. (A) N2 adsorption−desorption isotherms of calcined Cu/SiO2−xAl samples; (B) pore-size distribution curves of calcined Cu/SiO2−xAl samples.
3. RESULTS 3.1. Textural Properties of Cu/SiO2−xAl. The physicochemical parameters of pure SiO2 and Cu/SiO2−xAl are listed in Table 1. It was found that pure SiO2 owned a specific surface area of 102 m2/g, and interestingly, catalysts loaded with copper species displayed larger specific surface areas. Cu/ SiO2 catalyst owned the highest specific surface area of 368 m2/ g, and catalysts doped with aluminum species displayed smaller ones. The formation of copper phyllosilicate would be responsible for the increase in the specific surface area of Cu/SiO2.22 The aluminum added into the catalysts usually weakens the interaction between Cu2+ and SiO2, leading to the formation of fewer copper phyllosilicate and more nano-CuO. As a result, the specific surface areas of catalysts decreased with the increase of aluminum dopant. The N2 adsorption−desorption isotherm is shown in Figure 1A. The four catalysts exhibited Langmuir type IV isotherms with similar hysteresis loop. Cu/SiO2, Cu/SiO2−0.5Al, and Cu/SiO2−1.0Al showed an H2-type hysteresis loop, indicating the lamellar structure of the three catalysts. Cu/SiO2−1.5Al owned an H1-type hysteresis loop, corresponding to a typical large-pore mesoporous material with a cylindrical channel. Moreover, the pore sizes of the catalysts calculated from the BJH desorption branch are listed in Table 1, and an increment of the pore size could be observed on the catalysts with higher aluminum contents. Pore size distributions of the calcined catalysts are shown in Figure 1B. Interestingly, Cu/SiO2 showed the highest peak intensity at about 3 nm and Cu/ SiO2−1.5Al catalysts dropped severely, suggesting that higher aluminum species in Cu/SiO2 catalysts would destroy the porous structure that might result in the small specific surface areas of the catalysts. During the catalyst preparation, Cu2+ could react with silica sol to form copper phyllosilicate with lamellar structure.3,6 When aluminum was added, the formation
was 200/1 and the weight liquid hourly space velocity (LHSV) of DMO was 2 h−1. The activity of Cu/SiO2−xAl catalysts was also tested when EG was used as reactant (20 wt % EG in methanol solution). Since 1 mol DMO corresponding to 1 mol EG, the molar flow velocity of DMO should be equal to that of EG. When LHSV of DMO is 2 h−1, the corresponding LHSV of EG should be 1 h−1 to keep the same molar flow velocity. Compared with a one-step system, the two-step system consisted of two reactors in series with two separate H2 flows. Cu/SiO2 catalyst was placed in the first reactor and Cu/SiO2− xAl catalyst was packed in the second one. Both of the catalysts were located in the constant temperature region of each reactor and silica wool was placed both above and below the catalyst so as to fix them. There was 0.1 g of Cu/SiO2 catalysts in the first reactor and 0.4 g of Cu/SiO2−xAl catalyst in the second one. The temperature for first reaction was 473 K and it was 553 K for the second one. All the catalysts were activated in pure hydrogen atmosphere at 623 K, similar to that of the one-step system. After both of the reactors reached the proposed temperature, the reactant (20 wt % DMO in methanol solution) was injected from the top of the first reactor. The gaseous products of the first reactor flowed directly into the second reactor through the pipe, which was used to connect the two reactors. In the first reactor, the molar ratio of H2/DMO was 80/1, and in the second one, it became 200/1 (H2 in the second reactor contained the rest of H2 from the first reactor and supplementary H2 from the top of the second reactor. 200/ 1 is the molar ratio of hydrogen that flew into the second reactor/the total amount of feeding DMO.) The pressure of two-step system was 2.5 MPa and LHSV of the whole system was 2 h−1 (LHSV = weight of DMO reactant hourly/(weight of the catalyst in both reactors)). To ensure repeatability, three to five separate GC samples were taken, the results were averaged for each experimental data point, and uncertainties were typically within 3%. C
DOI: 10.1021/acs.iecr.5b01928 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research
Figure 2. XRD patterns of the (A) calcined, (B) reduced Cu/SiO2−xAl catalysts: (a) Cu/SiO2, (b) Cu/SiO2−0.5Al, (c) Cu/SiO2−1.0Al, (d) Cu/ SiO2−1.5Al.
Figure 3. (A) NH3-TPD and (B) CO2-TPD of Cu/SiO2−xAl catalysts: (a) Cu/SiO2, (b) Cu/SiO2−0.5Al, (c) Cu/SiO2−1.0Al, (d) Cu/SiO2−1.5Al.
on Si−OH.23,24 Hidalgo et al.25 found that the peak desorbed at 500−600 K was always associated with the ammonia adsorbed on the acidic hydroxide group ≡Si−OH−Al, in accordance with our results. Meanwhile, the peak at high temperature may be ascribed to ammonia adsorbed on copper species.26 When the NH3-TPD profiles were integrated, a relatively linear dependence of the whole area of NH3-TPD peaks on the aluminum content could be illustrated, and the amount and density of acidic sites in catalysts are shown in Table 2. It was easy to find that the amount of acid sites increased with the aluminum concentration.
of copper phyllosilicate was inhibited and some Cu2+ became nano-CuO in the calcined catalyst. 3.2. XRD. XRD patterns for the calcined Cu/SiO2−xAl catalysts are shown in Figure 2A. The diffraction peak at 21.8° belonged to amorphous silica and the features at 35.4° as well as 38.7° corresponding to the crystal planes of monoclinic CuO phase (JCPDS 05-0661). The diffraction peak at 30.8°, 57.5°, and 63.4° could be ascribed to copper phyllosilicate (CuPS). With the amount of aluminum increasing, the characteristic peaks of copper phyllosilicate became weaker and finally disappeared. According to the XRD patterns, the particle of CuO in the Cu/SiO2, Cu/SiO2−0.5Al, and Cu/SiO2−1.0Al could not be detected, while aggregation of some CuO happened in Cu/SiO2−1.5Al. The XRD patterns of the reduced Cu/SiO2−xAl catalysts are shown in Figure 2B. Upon reduction at 623 K, the characteristic peaks of copper oxides and copper phyllosilicate disappeared. The characteristic peak at 36.6° for Cu2O(111) and the peak at 43.2° for Cu(111) became observable for the reduced catalysts. Small amounts of aluminum (