Introduction

Fast growing population and unrestrained usage of natural resources are the major contributors for carbon dioxide emissions into the atmosphere 1. There are different methodologies to transform carbon dioxide (a cheap, recyclable and non-toxic carbon source) into useful chemical compounds 2. These include syngas (from CO2 and CH4) 3, methanol (from CO2 and H2) 4, urea (from ammonia and CO2) 5 among others. CO2 can also form linear carbonates, such as dimethyl (DMC) and diethyl carbonate (DEC), by its reaction with methanol 6 or ethanol, respectively 7. DEC is becoming increasingly important in the chemical industry; it is mainly used as an electrolyte in lithium-ion batteries 8 and as a fuel additive 9. In organic synthesis, DEC is used as a solvent, and as a carbonylation and alkylating agent due to the presence of CH3CH2O– and –CO– groups 10. The synthetic route from CO2 and ethanol to produce diethyl carbonate and water, avoids the problems caused by the use of toxic, corrosive and flammable gases in the standard synthesis routes, i.e., phosgene and hydrogen chloride (ethanolysis of phosgene 11, carbon monoxide (oxidative carbonylation of ethanol 12), or ethyl nitrite (carbonylation of ethyl nitrite 13), diethyl oxalate (decarbonylation of diethyl oxalate 14) and propylene carbonate 15. There is low DEC selectivity in some of these routes, e.g., propylene oxide, ethanol and carbon dioxide transesterification 16, ethanolysis of urea 17,18,19, and ethanol and dimethyl carbonate transesterification 20.

Several catalysts have been reported for the production of DEC starting from ethanol and CO2 (Table 1). The most common catalyst evaluated for DEC synthesis in the liquid phase are CH3CH2I–K2CO3 21, K2CO3 22, CeO2 23, CexZr1−xO2 24, CeO2 − SiO2 25 and ZrO2 26.

Table 1 Heterogeneous catalysts reported for DEC synthesis in the liquid phase.
Full size table

The catalytic activity for DEC formation is generally very low, because of the high thermodynamic stability and inert nature of carbon dioxide. Reports in the literature about DEC production in the liquid phase have focused on the effect of catalysts on activity and product distribution as a function of reaction pressure and temperature. An increase of catalytic activity (DEC yield) with reaction temperature was reported by Yoshida et al. 23 and Wang et al. 27 over CeO2 and CeZrO2 catalysts, respectively. Additionally, Prymak et al. 24 reported that the liquid phase reaction was favored at low temperature and high pressure. Those conditions caused rapid catalyst deactivation 28.

We previously reported the successful use of Cu, Ni and Cu–Ni supported on activated carbon 29 and ZrO2 30 as catalysts for the gas phase synthesis of DEC starting from ethanol and CO2 at moderate pressure (13 bar) and temperature (363 K). The Cu monometallic catalyst was three times more active (turnover frequency (TOF) = 5.1 h−1) than the monometallic Ni catalyst (TOF = 1.6 h−1). The catalytic activity increased further when both Cu and Ni were present in the catalyst, and the DEC rate increased with the Cu:Ni ratio. The TOF increased nearly 12 and 25 times for Cu:Ni molar ratios of 2:1 (53 h−1) and 3:1 (126 h−1), respectively. The enhanced catalytic activity over bimetallic catalysts was ascribed to the presence of Cu–Ni alloys on the surface of AC, which assignment was based on X-ray diffraction and TPR analysis.

As far as the authors know, there is no information about the effect of reaction parameters (temperature and pressure) on the gas-phase production of DEC. We report here the effect of temperature and pressure on the activity of Cu–Ni/AC catalyst. XPS analysis of Cu–Ni/AC evidence the presence of surface Cu–Ni alloys, presumably responsible for the increased activity of those catalysts. Additionally, a kinetic equation for the direct synthesis of DEC from ethanol and carbon dioxide is proposed.

Materials and methods

Catalyst preparation

Activated carbon AC supplied by Merck (90%, particle size < 100 µm) was used as support and the catalysts were synthesized by wetness impregnation. The metal precursors used were Cu(NO3)2·3H2O (supplied by Carlo Erba 99.5%) and Ni(NO3)2·6H2O (supplied by Merck 99%). A nominal 20% wt. loading of metal oxide (CuO + NiO) was used. Samples with different Cu:Ni molar ratios: 3:1, 2:1, 1:1, 1:2, and 1:3 were prepared. The metal precursors were dissolved in distilled water and mixed with the support for 24 h. The materials were then roto-evaporated for 3 h. The resulting samples were dried at 363 K for 12 h and calcined under a flow of 25 mL min−1 of N2 at 773 K for 3 h using a heating rate of 0.5 K min−1. The catalysts were finally reduced in a 5% H2/Ar mixture at 873 K.

Catalytic activity

Catalytic activity was evaluated using 1.0 g of catalyst in a cylindrical stainless steel fixed-bed reactor (internal diameter 0.7 cm). Ethanol was injected into the reactor by flowing a stream of He/CO2 at the appropriate flow rate through a stainless-steel bubbler containing liquid ethanol supplied by Merck (96%). The CH3CH2OH:CO2 molar ratio and partial pressures were regulated by saturating the stream with ethanol at different temperatures and by changing the He/CO2 ratio. A hot box system equipped with a K-type thermocouple was used to control the reaction temperature; additionally, a heating system was used to keep constant the outlet temperature at 420 K, avoiding possible condensation. After each catalytic test, argon gas was flowed (50 mL min−1) through the reactor for 24 h to clean the reaction system. Ethanol (m/z = 30), carbon dioxide (m/z = 22), DEC (m/z = 91) and diethyl ether (m/z = 74) were monitored online with a mass spectrometer (QMS Thermostar 200, Pfeiffer).

Kinetic tests were performed with different catalyst particle sizes and feed space times W/FETOH (50–130 g h mol−1) using 1.0 g of the best catalyst, a 2:1 molar ratio ethanol: carbon dioxide as a reactant mixture, and 13 bar total pressure. Temperature (383 ≤ T ≤ 463 K) and partial pressure of reactants (0.2 ≤ PETOH ≤ 0.34 bar, 0.32 ≤ PCO2 ≤ 1.31 bar) were varied in the catalytic tests. The temperature and pressure range were selected in accordance previous studies 31.

As the ethanol conversion was low throughout our study, we could assume differential operation for the experimental reactor, which allowed us to estimate the catalytic activity using Eqs. 1 and 2.

$$ r_{DEC} = \frac{{F_{DEC} }}{W} $$
(1)
$$ S_{DEC} = \frac{{F_{DEC} }}{{F_{DEC} + F_{DEE} }} \times 100 $$
(2)

where W = the catalyst weight (g), rDEC = rate of reaction (mol g−1cat min−1) of DEC, SDEC = DEC selectivity, FDEC = molar flow of DEC, FDEE = molar flow of diethyl ether.

Catalyst characterization

Chemical composition

Chemical composition of the catalysts was determined by atomic absorption spectroscopy in a Philips PU9200 spectrometer with a rhodium radiation source operating at 40 kV; 0.5 g of the catalyst were digested in HCl–HNO3 solution for the analysis.

X-ray diffraction (XRD)

X-ray diffraction (XRD) analysis was performed on a Phillips PW 1740 diffractometer. Diffraction patterns were recorded with Cu Kα radiation (20 mA, 40 kV) over a 5°–70° 2θ-range. The XRD patterns were compared with JCPDS files. The Scherrer Eq. (3), was used to calculate the crystallite size.

$$ D = \frac{K\lambda }{{\left( {\beta_{m} - \beta_{i} } \right){\text{Cos}} (\theta )}} $$
(3)

Here, D: size of the crystallite (nm), K: Scherrer constant, which is equal to 0.94 for spherical crystals, λ: radiation wavelength (0.154 nm), (βm − βi) is the broadening of the full width at half maximum of the main peak, βm: sample; βi: reference = 0.11, θ is the Bragg angle (degrees).

Temperature programmed reduction (TPR)

Temperature programmed reduction analysis of the samples was measured in a Micromeritics AutoChem II 2920. 50 mg of the sample were first heated from 278 to 523 K and kept at that temperature for 1 h in 70 mL min−1 He flow before cooling down to 313 K. The samples were then heated from 281 to 1073 K in 70 mL min−1 5% H2/Ar mixture flow.

Transmission electron microscopy (TEM)

A transmission electron microscope (JEOL 2010) operating at 200 kV and 124 μA was used to determine the morphology, size, and size distribution of metallic particles as it was reported by Tayyab et al. 32.

X-ray photoelectron spectroscopy XPS

The XPS measurements were collected with a spectrometer (Thermo VG Scientific Escalab 250) operated at 20 kV and 30 mA. All catalysts were reduced in flowing H2 at 593 K for 30 min before the XPS measurements. The binding energy (BE) of the C 1s peak was used as an internal standard. Relative concentrations of metals were estimated by calculating the integral of each XPS peaks using the cross sections 33.

Results and discussions

Catalytic activity

Figure 1 shows the catalytic activity (expressed as DEC reaction rate) during DEC synthesis at 383 K. The total Cu content varied from 0 to 100% and Cu:Ni molar ratio ranged from 1:1 to 3:1. The catalytic activity of monometallic samples was also evaluated.

Figure 1
figure 1

Catalytic activity of tested catalysts. Reaction conditions: 383 K, 13 bar, ethanol: CO2 feed molar ratio of 2:1, gas hourly space velocity of 988 h−1.

Full size image

There was no CO2 conversion when no catalyst was loaded neither when tests were carried out with activated carbon. The monometallic Cu catalyst exhibited higher catalytic activity (1.6 × 10−5 mol g−1cat min−1) than the Ni monometallic catalyst (0.7 × 10−5 mol g−1cat min−1). This result is according with reports by Bian et al. 34, for the formation of dimethyl carbonate. Deng et al. 35 used surface enhanced Raman spectroscopy and suggested that Cu can activate CO2 by forming negatively charged CO2δ− species, which can further convert to carbonate. The corresponding Raman bands were reported at 767 cm−1 (bending mode δ(OCO)) and at 1182 cm−1 (symmetric stretch mode). Figure 1 shows the DEC reaction rate as function of Cu:Ni molar ratio of our bimetallic catalysts. As a result of the Cu–Ni interaction the reaction rate for the bimetallic catalysts increased significantly when compared with the monometallic. The increase in the catalytic activity as a function of the Cu molar ratio agrees with data reported by Bian et al. 34 during linear carbonate synthesis in this composition range. In our case, the reaction rate increased about 2 and 2.5 times for Cu:Ni molar ratios of 2:1 and 3:1, respectively, when compared with the activity of the Cu/AC catalyst. The interaction between Cu and Ni also decreased the production of CO and improved the catalyst resistance to coke formation 36.

The selectivity to diethyl carbonate ranged between 82 and 88%, and it is similar to the values reported by Bian et al. 34 for the formation of dimethyl carbonate from methanol and carbon dioxide. The main side product was diethyl ether, that probably is produced by decomposition of DEC, rather than by dehydration of ethanol. This was verified by feeding either ethanol or diethyl ether to the reactor packed with 1.0 g of Cu:Ni/AC catalyst with a Cu:Ni molar ratio of 3:1, and operating at 13 bar and 383 K. The dehydration of ethanol has been reported to occur at 423 K, but in the presence of basic catalysts 17.

Effect of catalytic reaction conditions on DEC reaction rate

Effect of mass transfer limitations

The conditions in which there were no external or internal mass transfer resistances were identified. Figure 2 shows that DEC reaction rate remained constant when the flowrate was higher than 75 mL min−1, indicating the absence of external gas phase mass transfer resistance. Experiments with different catalyst particle sizes were performed to verify the absence of intraparticle mass transfer resistance. Figure 3 shows that the reaction is free of intraparticle mass transfer limitations for particle sizes below about 300 µm.

Figure 2
figure 2

Analysis of external mass transfer limitations. Reaction conditions: 383 K, total pressure: 13 bar, catalyst weight: 1.0 g, ethanol:CO2 feed molar ratio of 2:1.

Full size image
Figure 3
figure 3

Analysis of internal mass transfer limitations. Reaction conditions: 383 K, total pressure: 13 bar, catalyst weight: 1.0 g, ethanol: CO2 feed molar ratio of 2:1.

Full size image

Effect of reaction temperature

When the effect of temperature was tested, it was found that the DEC reaction rate increased with temperature from 343 K until it reached a maximum value at 383 K, decreasing continuously afterwards (Fig. 4). This effect suggests that an equilibrium restriction is involved, given the exothermic nature of the reaction 31. The presence of a maximum in rate in the 363–393 K range was reported by Orrego et al. 37, Chen et al. 38, and Bian et al. 34,39,40 for different alcohol–CO2 reactions.

Figure 4
figure 4

Reaction temperature effect. Reaction conditions: Total pressure: 13 bar, 1.0 g Cu:Ni-3:1, total flow: 75 mL min−1, ethanol: CO2 feed molar ratio of 2:1.

Full size image

The decrease of reaction rate at high temperature has been associated to blockage by CO2 adsorption 40, but that is not likely to be the case, as adsorption normally decreases when temperature increases. Furthermore, the decrease in DEC yield as a function of temperature was associated to a probable enhancement of catalyst deactivation at high temperatures and to an increased reactant mass throughput 41.

Pressure dependence of the reaction rate

The proposed kinetic equation and the estimated fitting parameters, as well as the corresponding standard errors are given in Table 2.

Table 2 Power law model and kinetic parameters for DEC formation at 13 bar total pressure and 383 K.
Full size table

The power law equation provides a good fit (r2 = 0.954) of the experimental results and appears to be an adequate alternative to represent pressure and temperature dependence of the rate (supplementary information). The proposed rate equation presented in this study is the first one proposed for this class of reactions, i.e., reaction in gas phase between CO2 and an aliphatic alcohol. The kinetic equations reported by other authors for DEC synthesis are reported in Table 3.

Table 3 Kinetic equations reported in literature for DEC synthesis.
Full size table

Kinetics studies for DEC synthesis from ethanol and carbon dioxide are scarce. Table 3 shows kinetic equations reported in open literature for DEC synthesis from ethanol and carbon dioxide. Décultot et al. 42 reported a kinetic equation for the reversible reaction using a model based on Langmuir–Hinshelwood mechanism for the reaction in liquid phase. Power law model was used for describing the kinetic of the reverse and forward reactions under multiphase conditions and used for the plant-wide process to produce diethyl carbonate from the direct reaction of CO2 with ethanol, using 2-cyanopyridine as an in situ dehydrating agent 42, and also reported by Giram et al. using CeO2 as catalyst 43. As the phases of the reaction influence the final kinetic equation, the reported rate equations cannot be compared.

Catalysts characterization

XRD results

Figure 5 presents the diffractograms of Cu/AC and Ni/AC with reflections at 2θ = 43.3° and 50.4° that correspond to the (111) and (200) planes of Cu0 face cubic centered cell (JCPDS 4-0836) and reflections at 2θ = 44.5° and 51.8° that correspond to 111 and 200 planes of Ni0 face cubic centered cell (JCPDS 4-0850). In Cu–Ni bimetallic samples an increase in Cu loading caused a linear shift of the (111) and (200) diffraction lines. A linear relationship between the lattice parameter and the Cu concentration agrees with Vegard law 44, and is a strong evidence of the formation of a Cu–Ni solid solution 45. The improved catalytic activity for DEC formation observed with the 3:1 and 2:1 catalyst appears to be related to the presence of such Cu–Ni alloy. Cu, Ni and Cu:Ni-3:1 mean particle size was estimated using Scherrer equation (Eq. 3). The particle size of the bimetallic Cu:Ni-3:1 sample was 22.2 nm, a value between those calculated for Cu (23.5 nm) and nickel (17.3 nm) monometallic samples.

Figure 5
figure 5

XRD of tested catalysts.

Full size image

Temperature programmed reduction

TPR profiles of the catalysts are presented in Fig. 6. Activated carbon has a single broad reduction signal above 873 K. The reduction patterns of the metal-containing samples have significant differences in the low to medium temperature range (473–673 K). The monometallic Cu catalyst exhibits one reduction peak centered at 524 K assigned to the reduction of bulk CuO 34. The monometallic Ni supported catalyst has two reduction signals; the low temperature signal has a maximum at 554 K with a shoulder at 515 K, and is assigned to the reduction of NiO species with different degrees of dispersion 46. The high temperature reduction peak has a maximum at 653 K and is attributed to the reduction of “free” NiO with activated carbon during the thermal treatment 47.

Figure 6
figure 6

TPR of support, Cu, Ni and bimetallic Cu:Ni at different molar ratios.

Full size image

In addition, the reduction temperatures of monometallic catalysts were higher than those for the unsupported monometallic CuO (486 K) and lower than those for NiO (631 K). The supported bimetallic catalysts with low Cu content had reduction patterns that combined those of the monometallic samples; however, the catalysts with the higher Cu content (Cu: Ni-2:1 and Cu:Ni-3:1) had a single reduction peak similar to the Cu sample. This clearly indicates that Ni reduction is strongly promoted by the interaction with Cu, and is another evidence of the formation of new phases between Cu and Ni, which was also confirmed by XRD. Our results agree with the report of Khromova et al. 45, who attributed the more facile reduction of CuO in the bimetallic Cu–Ni samples to the lower free energy of reduction for CuO (− 100.65 kJ mol−1 at 293 K) compared with NiO (− 12.31 kJ mol−1 at 293 K). The TPR profiles evidence a high synergy between Cu and Ni that may have an impact on DEC formation. All the samples show also a broad H2 consumption band at temperature higher than 773 K caused by the C–H2 reaction to form CH4 (observed also with pure activated carbon); however, the hydrogen activation does not affect the activity for DEC production, as the latter reaction occurs at much lower temperatures.

TEM analysis

Figure 7 shows the TEM images and particle-size of Cu:Ni-3:1 sample.

Figure 7
figure 7

TEM particle distribution of Cu:Ni-3:1. Light zones (support) and dark zones (metal particles).

Full size image

The TEM images show the morphological characteristics of the Cu:Ni-3:1 catalyst. In bimetallic samples, particles (dark zones) adopt spherical, rectangular, elliptical and irregular shapes as is reported by activated carbons 34, carbon nanotubes 40, nanowires 46, graphitic carbon 48 and ZnO–ZnS 49. Figure 7 shows a homogeneous distribution of metal particles over the carbon support, with some agglomerates observed in Cu–Ni sample. The particle size of the small nanocrystals of Cu–Ni alloy was 21 nm, in good agreement with XRD results. The diameter of Cu–Ni alloy is close to a simple addition of the diameters of Cu and Ni which suggests a possible insertion of Ni into Cu structure.

XPS analysis

The XPS spectra for Cu and Ni monometallic catalysts are presented in Figs. 8 and 9, respectively; results are compared with Cu:Ni-3:1 catalyst in both cases.

Figure 8
figure 8

Cu2p core-level spectra: (a) Cu catalyst and (b) Cu:Ni-3:1 catalyst.

Full size image
Figure 9
figure 9

Ni2p core-level spectra: (a) Cu catalyst and (b) Cu:Ni-2:1 catalyst.

Full size image

Figure 8a depicts the XPS spectra in the Cu 2p region for Cu and Fig. 8b Cu:Ni-3:1 bimetallic catalyst. Monometallic sample, Fig. 8a, shows two peaks at approximately 932.8 eV and 952.8 eV assigned to Cu2p3/2 and Cu2p1/2, respectively, indicating the zero-oxidation state of Cu after the reduction process, and a spin–orbit coupling energy of 20 eV and two strong satellite peaks at 944.2 eV and 962.6 eV. Experimental results indicate that copper is in the Cu0 or Cu1+ and Cu2+ states. The presence of Cu0 and Cu1+ was confirmed by a main peak in the 932.4–932.8 eV range associated with Cu2p3/2, while the presence of CuO was confirmed by high-intensity shake-up satellites at approximately 9–12 eV higher binding energy than the main Cu2p3/2 and Cu2p1/2 peaks. In bimetallic sample, Fig. 8b, binding energy shift was observed from 932.8 to 933.2 eV. The experimental results indicate that when Cu2+ is mixed with Ni, the monometallic catalyst becomes partially reduced, in agreement with TPR results.

The Ni monometallic catalyst (Fig. 9a) exhibits peaks at 854.1 eV (Ni 2 p3/2) and 872.6 eV (Ni 2 p1/2) which confirmed the presence of Ni metal and NiO, respectively. In Cu:Ni-3:1 sample (Fig. 9b) binding energy presents a shifting for Ni 2p3/2 from 854.1 to 856.2 eV, and from 932.8 to 933.2 eV for Cu 2p3/2. According to our experimental results, in the bimetallic sample part of Cu and Ni were in zero oxidation state. Additionally, a slight shifting in the binding energy of Cu and Ni indicated a change in the chemical environment of species related with formation of Cu–Ni alloy 50.

Conclusions

In this work, activated carbon supported Cu, Ni and Cu:Ni catalysts were prepared and tested for the gas phase formation of diethyl carbonate from ethanol and carbon dioxide. Based on the experimental results, it can be concluded that reaction rate of the bimetallic samples increases with the Cu content; Cu:Ni-3:1 molar ratio sample was the most active catalyst. The presence of Cu–Ni alloy was evidenced by XRD, TPR, and XPS analysis. The improved catalytic activity for the production of DEC over Cu:Ni bimetallic materials appears to be caused by the formation of Cu–Ni alloy nanoparticles with electronic and reducibility properties different to those of the monometallic samples. Experimental results showed that diethyl carbonate reaction rate reached a maximum at 383 K. A rate equation of second order with respect to ethanol and first order to respect carbon dioxide was proposed. This work presents a first kinetic equation to gas phase synthesis to dimethyl carbonate from CO2 and ethanol and it is expected that this work facilitates a better understanding of the mechanism of the direct synthesis of diethyl carbonate in gas phase, as previous kinetic reports are for the reaction in liquid phase.