Qing-Cheng Zhanga,
Kun-Peng Chenga,
Li-Xiong Wen*ab,
Kai Guob and
Jian-Feng Chenab
aState Key Laboratory of Organic-Inorganic Composites, Beijing University of Chemical Technology, Beijing 100029, China. E-mail: wenlx@mail.buct.edu.cn; Fax: +86-10-64434784; Tel: +86-10-64443614
bResearch Center of the Ministry of Education for High Gravity Engineering and Technology, Beijing University of Chemical Technology, Beijing 100029, China
First published on 24th March 2016
CuO/ZnO/Al2O3 catalyst precursors were precipitated in a novel micro-impinging stream reactor (MISR) and a traditional stirred tank reactor (STR), respectively, followed by a period time of aging in the mother liquid. Being different from the simultaneous precipitating and aging of catalyst precursors within the same STR reactor, these two processes occurred in two separate containers in the MISR route, hence providing a more uniform and steady environment for both the precipitating and aging processes on top of the higher micromixing efficiency and better process control of the MISR. Therefore, substantial changes in the phase compositions and microstructures of the catalyst precursors were obtained with MISR, which resulted in smaller and more homogeneous catalyst particles with a larger BET surface area and specific copper surface area, better Cu/Zn dispersion as well as higher catalytic activity when compared to those prepared in the STR. The aging process also played an important role in catalyst preparation and it could be controlled more easily and precisely in the MISR route to form a more desirable phase structure, morphology and eventually more superior catalytic performance in methanol synthesis for the final catalysts.
The precipitation of CuO/ZnO/Al2O3 catalyst precursors is often followed by an aging step. During the aging process, the precipitate remains in contact with its mother liquid and several reactions may take place at the same time, which can strongly affect the properties and consequently the catalytic performance of the final catalyst.8,21,22 Firstly, the initially formed precipitates will slowly transform by partial dissolution/re-precipitation reactions, and a recrystallization will occur at the same time accompanying with a change in color from blue to bluish green and, as expected, a change in chemical composition, particle size and morphology.23,24 Secondly, a mixture of intermixed hydrocarbonates such as rosasite (Cu,Zn)2(OH)2CO3 and aurichalcite (Cu,Zn)5(OH)6(CO3)2 will be formed through exchange reactions between the preformed phases.10,25 Here, the efficient incorporation of Zn into malachite forming rosasite (zincian malachite) is regarded as the key to CuO/ZnO catalysts preparation, since a significant amount of atomically distributed Zn in rosasite will lead to an effective stabilization of the Cu phase and thus highly active CuO/ZnO catalysts will be obtained after thermal decomposition.10,26 Therefore, the aging of fresh precipitates is also a crucial step for the synthesis of homogeneous CuO/ZnO/Al2O3 catalysts, and hence a uniform and steady environment will play a key role for the aging process as well as the quality of the final products. However, the aging environment in STR-operation mode is barely uniform or stable not only because of the simultaneous precipitating and aging processes but also the poor micromixing performance in STR. All precipitated particles get accumulated in STR for the whole precipitation process (it usually lasts ∼20 min or even longer in the drop-adding method),13,19,20,27 and the aging of precursors has already started before the precipitation is completed, which can lead to an inhomogeneous product distribution, as the initially formed precursors may differ significantly from the phase formed at the end of the precipitation.
A micro-impinging stream reactor (MISR) was built and applied to prepare CuO/ZnO/Al2O3 catalysts for methanol synthesis in our previous studies.28,29 The MISR is constructed with two steel capillaries connected to a commercial T-junction, but no tube is connected to the T-junction outlet, which provides a big-sized channel to lead the reacted precipitates into sample collectors for the next step. The two streams impinge on each other directly inside the T-junction to create a quick and constant supersaturation level and therefore a more homogeneous nucleation environment for the quick precipitating process, which may have been completed within the T-junction chamber. Therefore, the following aging process will have a uniform and steady environment. MISR also offers a better process control on reactant concentration, pH, volumetric flow rates and etc., and the residence time distribution for both of precipitating and aging processes is sharp, hence a better product quality of CuO/ZnO/Al2O3 catalysts will be achieved. In addition, unlike most of micro-structured devices such as microchannel reactor,30 microfluidic reactor31 and micromixer,32 which are difficult to be fabricated due to their complicated configuration or easy to have severe blocking problems during the precipitation process due to the tiny channels inside the reactors, MISR can be easily constructed and has negligible blocking problem because of its bigger geometric dimensions and faster stream flows as compared to conventional microreactors.
In previous work, CuO/ZnO/Al2O3 catalysts have been prepared in MISR with better microstructures and properties as well as higher catalytic activity in methanol synthesis than those prepared in STR,29 as a result of the intensified micromixing within the MISR reactor for the precipitating process as well as a more uniform and steady environment for the subsequent aging process. In this work, we combined fast precipitation and freeze-drying to investigate the phase transitions during the first seconds of the precipitation, and then provided a detailed study on the effects of post-precipitation processes on the properties and activity of CuO/ZnO/Al2O3 catalysts and their precursors produced in MISR and traditional STR, respectively.
The co-precipitation was also performed through a traditional route by simultaneously dropping two aqueous solutions containing metal nitrates and sodium carbonate, respectively, into a glass beaker filled with a small amount of deionized water under vigorous stirring at 80 °C. The metal nitrates solution was added at a rate of 5 mL min−1, while the adding rate of the Na2CO3 solution was controlled to maintain the desired pH (7.0) in the mother liquid. After complete addition of the solutions, the suspension was further aged under continuous stirring at 80 °C for 0, 10, 30, 60 and 120 min, respectively. The post-processes were as same as those in the MISR route.
In addition, in order to investigate the phase transitions occurred during the first seconds of the precipitation, a part of non-aged precipitates were filtrated immediately after complete precipitation and no washing procedure was applied so that to minimize possible effects on the solids. The precipitates were then dried overnight in the freeze-dryer for subsequent analyses.
Fourier transform infrared spectra (FT-IR) were measured in the range of 4000–400 cm−1 using the KBr disc technique on a PE2000 FT-IR spectrometer.
The particle morphologies were observed with a JEOL 3010 transmission electron microscopy (TEM) operated at 300 kV. The samples were ultrasonically dispersed in high purity ethanol and then deposited onto ultrathin carbon-coated nickel grids.
The temperature-programmed reduction (TPR) experiments were performed with an Autochem II 2920 multifunctional adsorption instrument. The samples (50 mg) were firstly purged with Ar for 0.5 h to remove physically absorbed water and then reduced in 10 vol% H2/Ar mixture gas (30 mL min−1) at a heating rate of 10 °C min−1 up to 600 °C. The consumption of H2 was analyzed by gas chromatograph (SP-2100A, China) equipped with a thermal conductivity detector (TCD).
X-ray photoelectron spectra (XPS) were conducted with an ESCALAB 250 spectrometer using Al-Kα radiation. The binding energies were calculated with respect to C 1s peak at 284.8 eV.
The specific surface area (SBET) of the calcined catalysts was calculated by the BET method from nitrogen adsorption–desorption isotherms obtained with an ASAP 2010 surface area analyzer (Micromeritics Instrument Corporation, USA). All samples were pretreated under vacuum at 200 °C for 2 h.
The exposed copper surface area (SCu) was determined by N2O reactive frontal chromatography (N2O-RFC) and carried out with a Micromeritics AutoChem 2920 instrument through the method proposed by Gao et al.33 The catalysts (0.1 g) were first reduced with 10 vol% H2/Ar mixture (30 mL min−1) at 350 °C for 2 h, followed by purging with Ar for 30 min and cooling to 65 °C. The reduced catalysts were then exposed to N2O (30 mL min−1) for 1 h to oxidize surface copper atoms to Cu2O. After that the samples were flushed with Ar to remove the N2O and cooled to room temperature. Finally, the Cu2O at the surface was reduced back to metallic Cu at 350 °C in the H2/Ar flow (30 mL min−1). The SCu was calculated from the amount of consumed H2 during the reduction steps, assuming a molar stoichiometry of N2O/Cus = 2 and a value of 1.4 × 1019 Cu atoms per m2 as the copper atoms density.34
The conversion of CO is defined as:
![]() | (1) |
The selectivity to methanol is defined as:
![]() | (2) |
Fig. 2a shows the XRD patterns of precursors prepared in MISR and aged for different length of time. It was found that, after washing and oven-drying processes, the diffraction peaks belonging to Na2Zn3(CO3)4 were mostly disappeared and the complete removal of crystalline NaNO3 was achieved as the diffraction peak at 2θ = 29.0° was not observed anymore, which was very strong for the samples without washing as shown in Fig. 1. Therefore, a structural rearrangement of the system took place when the fresh precipitate was exposed to water and it was unlikely that any Na2Zn3(CO3)4 would have remained in the sample after being vigorously stirred and washed in STR-route, followed by drying at 110 °C overnight.36–39 When the precipitates were intensively stirred in the mother liquid for a certain aging time (10, 30, 60 or 120 min), Na2Zn3(CO3)4 was completely converted into zinc-enriched hydrocarbonate phases. Meanwhile, the crystallization of malachite Cu2(OH)2CO3 and rosasite from amorphous copper-containing species would start as well, accompanying with a color change from blue to bluish green. It was found that the characteristic peaks of malachite (JCPDS no. 41-1390) and rosasite (JCPDS no. 18-1095) at 2θ = 14.8°, 17.6°, 24.2° and 31.9° appeared after 30 min aging and increased in intensities with time, which is in good accordance with the studies of Farahani et al.13 who proposed that further increase in aging time after the appearance of color change would cause a progressive crystalline growth as a result of Ostwald ripening. In addition, the Zn–Al hydrotalcite Zn6Al2(OH)16CO3·4H2O (JCPDS no. 38-0486) represented by the diffraction peaks at 2θ = 11.5° and 23.3° was observed for precursors aged for 120 min. Therefore, aging of the freshly precipitated solids would eventually result in a mixture of hydrocarbonate phases, which could strongly affect the microstructures and consequently the catalytic performance of the final catalyst.
![]() | ||
Fig. 2 XRD patterns of the catalyst precursors aged for different length of time: (a), MISR-prepared; (b), STR-prepared; (c), the position of 20![]() |
The XRD patterns of precursors prepared in STR aged for different length of time are shown in Fig. 2b. No noticeable variation was found in the XRD patterns of the hydrocarbonate phases through the washing treatment except for the removal of crystalline NaNO3, and then STR samples followed the same transitional tracks in aging process as those prepared in MISR. In addition, a slight shift of the malachite diffraction peak near 2θ = 32° (20 peak) to higher angles was observed when the aging time was increased from 30 min to 60 min (Fig. 2c), which was correlated with the substitution chemistry of malachite as Zn2+ is incorporated into malachite structure forming rosasite.10,41 Therefore, the exchange reactions between the preformed phases took place during the aging of precursors, which would enhance the Cu/Zn dispersion of final catalysts. However, the 20
peak shifted back to lower angles when the precursor was aged for 120 min, which was due to the formation of Zn–Al hydrotalcite as shown in the XRD patterns.
Compared to the MISR-route, all precursors prepared in STR exhibited much sharper copper-enriched hydrocarbonates (malachite and rosasite) diffraction peaks indicating larger crystallites according to Scherrer equation. In addition, a considerable shift of 20 peak towards higher angles (2θ = 32.1°) was observed for the precursor prepared in MISR as compared to STR samples, which suggested a highest degree of Cu-substitution by Zn in rosasite phase should be achieved in MISR-route after 60 min aging of precursor, and thus CuO/ZnO/Al2O3 catalyst of fine dispersion was obtained. Such differences in the compositions and microstructures of precursors were traced back to the different pathways of solid formation for two precipitation routes. As discussed earlier, in the drop-adding stirred reactor, particle growth takes place when a droplet of precipitating agent is added into the reactor in which precipitates have already existed.42 On the other hand, the intense impinging in MISR creates a fast and uniform micromixing as well as a high supersaturation level, which favors nucleation rather than crystal growth, thus results in smaller and more uniform particles in the precipitation.43,44 In addition, the aging mechanism has shown that no binary precipitates can be directly obtained,9 however, MISR can be used to form a pseudo-homogeneous precipitate with the best possible distribution of Cu- and Zn-intermediates due to its homogeneous nucleation environment, which will further accelerate the exchange reactions between the preformed phases and, in turn, favor the Cu/Zn dispersion of final catalysts.
A comparison of the FT-IR spectra for precursors prepared at different conditions (different routes and aging times) is shown in Fig. 3. All samples displayed a weak absorption band around 1630 cm−1 ascribed to the bending vibration of water. For the MISR-prepared samples (Fig. 3a), there was a small peak at 1385 cm−1 representing the asymmetric N–O vibration in the non-aged precursor, which was due to the existence of a small amount of gerhardtite Cu2(OH)3NO3 generated by the reaction between Cu(OH)2 and NO3−. However, an unambiguous discrimination of poorly crystallized samples on the basis of XRD patterns is difficult due to their high similarity in crystal structures, thus gerhardtite was not detected in XRD analysis in Fig. 2. Gerhardtite was believed to exert adverse effect for production of active catalyst by forming larger copper particles (i.e. sintering effect).11 Fortunately, gerhardtite would then slowly react upon aging with hydroxide and carbonate to form amorphous georgeite Cu(OH)2CO3 (the chemical formula of georgeite was the same as that of malachite),21 thus the asymmetric N–O vibration disappeared when the precipitate was aged for 10 min. The strong absorption bands of the amorphous precursors (0, 10 min aging) around 1480 and 837 cm−1 were quite similar to those of georgeite reported by Shen et al.20,27 In addition, the characteristic bands in the region of 1520–1390 cm−1 were attributed to asymmetric C–O stretching vibration, whereas the bands around 835 cm−1 were ascribed to out-of-plane OCO bending mode.41 Increasing the aging time from 10 min to 30 min forced the asymmetric C–O stretching vibration around 1480 cm−1 shift to 1499 cm−1, which was mainly ascribed to the recrystallization of amorphous georgeite, either directly crystallized to malachite or transformed into rosasite.20,27,35 All crystallized samples displayed the spectra with only small differences in position and intensity, hence the characterization of a phase mixture would be complicated for the carbonate bands originating from the malachite, rosasite and aurichalcite that overlapped significantly. However, previous studies45,46 have suggested that the incorporation of Zn2+ into the malachite lattice forming rosasite would result in a shift of the asymmetric C–O stretching vibration (around 1500 cm−1) to higher wavenumber. Compared to other precursors, the precursor aged for 60 min showed the higher wavenumber (1506 cm−1) of the asymmetric C–O stretching vibration; therefore, a more homogeneous microstructure of the CuO/ZnO/Al2O3 catalyst might be obtained in the MISR with 60 min aging of precursor.
![]() | ||
Fig. 3 IR spectra of the catalyst precursors aged for different length of time: (a), MISR-prepared; (b), STR-prepared. |
For the samples prepared in STR (Fig. 3b), it was found that the non-aged precursor displayed a very small peak at 1385 cm−1 representing the asymmetric N–O vibration; however, its intensity was much lower than the MISR-prepared sample (Fig. 3a), which was due to the gradual transformation of gerhardtite to georgeite during the precipitation process in STR as already discussed. Therefore, it confirmed that several reactions took place in the long residence time before the complete precipitation in STR. The small differences of IR spectra in positions and intensities (Fig. 3a and b) indicated that the two precipitation routes generated precipitates and precursors with different chemical compositions and phase structures, which was mainly due to the different micromixing performances, nucleation environments and residence time distributions of the STR and MISR, respectively.
![]() | ||
Fig. 4 XRD patterns of the CuO/ZnO/Al2O3 catalysts aged for different length of time: (a), MISR-prepared; (b), STR-prepared. |
TPR profiles were given in Fig. 5 to investigate the reduction patterns of copper species for the CuO/ZnO/Al2O3 catalysts prepared at different conditions. Samples prepared in both routes exhibited a broad peak without any shoulders in the range of 200–300 °C, which was obtained from the reduction of CuO. Under the experimental conditions in this work, ZnO and Al2O3 would not be reduced.47 For the MISR-prepared catalysts (Fig. 5a), the non-aged catalyst showed a reduction peak at 265 °C, and a shift of reduction temperature towards 238 °C was observed when the aging time was increased to 10 min, which was correlated to smaller CuO crystallites in the prepared catalyst as demonstrated by the weak CuO diffraction peaks in XRD patterns (Fig. 4a). In addition to the particle size effect, the shift of reduction peak to lower temperature might also be caused by the better copper dispersion of the mixed oxides.48,49 Therefore, the shift of reduction peak to lower temperature might be mainly owing to the higher zinc content into the malachite structure forming rosasite when aging time was prolonged (from 10 min to 60 min) as already discussed in XRD patterns (Fig. 2a) and IR analysis (Fig. 3a), thus favoring the copper dispersion and finally improving the reducibility of CuO/ZnO/Al2O3 catalysts. In addition, compared with the narrow reduction peaks for the CuO/ZnO/Al2O3 catalysts obtained from MISR route, the TPR profiles of the catalysts prepared in STR (Fig. 5b) were broader and shifted to higher reduction temperature, suggesting much larger CuO particles with a much wider size distribution were obtained in STR. Previous studies concerning the reducibility of copper-based catalysts for methanol synthesis and methanol steam reforming have revealed that a high catalytic activity is associated with good reducibility of catalysts;50,51 therefore, higher catalytic performance would be expected for the MISR-prepared catalysts.
![]() | ||
Fig. 5 TPR profiles of the CuO/ZnO/Al2O3 catalysts aged for different length of time: (a), MISR-prepared; (b), STR-prepared. |
Fig. 6 illustrates the typical XPS spectra of Cu 2p3/2 levels of the CuO/ZnO/Al2O3 catalysts prepared at different conditions. The binding energy (BE) range for the main Cu 2p3/2 peak was around 933.5 eV, generally with characteristic satellite peaks between 940 and 945 eV due to the electron shake-up process.52 Various studies53–55 have shown that the positions of these peaks depend on the chemical composition and crystalline structure of the catalyst, and particularly on the near environment of Cu2+. The binding energy of Cu 2p3/2 for pure CuO was 934.0 eV, which was higher than those of ternary CuO/ZnO/Al2O3 catalysts, indicating that CuO and other metal oxides were not simply physically mixed in the CuO/ZnO/Al2O3 catalysts but an interaction between them was present. This is because that the electronegativity of Cu (1.90) is stronger than that of Zn (1.65), which induced the migration of electron cloud to Cu and finally resulted in a shift of Cu 2p3/2 to lower BE.53 In addition, it was found that the MISR-prepared catalyst that was aged for 60 min displayed lower BE compared to other catalysts, which was mainly attributed to the enhanced interaction between CuO and ZnO because of their fine particle size and better Cu/Zn dispersion.
![]() | ||
Fig. 6 XPS spectra of the CuO/ZnO/Al2O3 catalysts aged for different length of time: (a)–(e), MISR-prepared, aging t = 0, 10, 30, 60, 120 min, respectively; (f), STR-prepared, aging t = 60 min. |
TEM images of the CuO/ZnO/Al2O3 catalysts prepared at different conditions and after methanol synthesis are shown in Fig. 7. The samples showed very similar morphology and comprised agglomerates of round and oval particles. TEM analysis of MISR-prepared catalysts as a function of aging time revealed a decrease of primary particle size from ∼10 nm (non-aged) to ∼6 nm (10 min aging) (Fig. 7a and b), which was well consistent with the XRD analysis in Fig. 4a. It was also found that, after aging for 60 min, the catalyst produced in MISR contained mostly round particles with a mean size of 8–10 nm (Fig. 7c), whereas slightly larger and more irregular particles were obtained in STR route (Fig. 7d). This observation agreed well with the XRD analysis (Fig. 4) and TPR results (Fig. 5), which indicated that larger and uninformed catalyst particles were obtained in STR due to its poor mixing and mass transfer performance.17 In addition, the TEM images showed that significant aggregation of catalyst particles took place after methanol synthesis reaction for both MISR- and STR-prepared catalysts (Fig. 7e and f), but with a more severe aggregation for the STR-prepared catalyst. Such aggregation might be caused by the highly exothermic effect of the methanol synthesis reaction and would result in less active sites and a decrease of catalytic activity during the synthesis process.56–58
Reactor | Aging time (min) | SBET (m2 g−1) | SCu (m2 g−1) | XCO (mol%) | SMeOH (C mol%) | YMeoH (mol%) |
---|---|---|---|---|---|---|
a Reaction conditions: T = 250 °C, V(CO)![]() ![]() ![]() ![]() |
||||||
MISR | 0 | 54.1 | 7.5 | 12.8 | 97.7 | 12.4 |
MISR | 10 | 141.6 | 16.9 | 25.9 | 97.9 | 25.3 |
MISR | 30 | 103.3 | 18.5 | 30.6 | 98.9 | 30.2 |
MISR | 60 | 103.4 | 22.4 | 33.7 | 99.3 | 33.5 |
MISR | 120 | 77.0 | 16.1 | 28.4 | 98.6 | 28.0 |
STR | 60 | 64.7 | 13.4 | 27.2 | 98.8 | 26.8 |
The catalytic performances in methanol synthesis via CO hydrogenation for the CuO/ZnO/Al2O3 catalysts prepared at different conditions are also shown in Table 1. It demonstrated that the catalytic activity (CO conversion, XCO) would be enhanced greatly with increasing SCu for the MISR-prepared samples, which was mostly enhanced with increasing aging time until it reached a proper length of time (60 min). But the positive influence of SCu on the CH3OH selectivity (SMeOH) was much slighter. Therefore, the non-aged MISR-prepared catalyst had the lowest XCO and SMeOH, while the highest XCO and SMeOH were achieved by the MISR-prepared catalyst with 60 min aging-treatment. Such results agreed well with the previous observations that the aging process with proper time would lead to more homogeneous microstructure of the precursors and final CuO/ZnO/Al2O3 catalyst, and the well dispersed CuO could enhance the synergy reaction with ZnO crystallites and hence improve the catalytic activity of the catalyst. Table 1 also showed that the MISR-prepared catalyst with 60 min aging-treatment had significantly higher SBET, SCu and XCO than those of the STR-prepared catalyst with 60 min aging-treatment, which agreed well with its larger particle size and poorer Cu/Zn dispersion for the STR products and clearly demonstrated the advantages of the MISR-route over the STR-route for catalyst preparation.
It is widely accepted that the methanol synthesis via CO hydrogenation is a structure-sensitive catalytic reaction. The different catalytic performances of these catalysts were associated with the combining effects of their different phase structures, crystallite sizes, Cu/Zn dispersion, exposed copper surface area and specific surface area as already discussed in previous sections. In addition, defects on the surface and/or in the bulk structure, such as microstrain, impurities, and structural disorder were also suggested to be responsible for the catalytic performances of the copper-based catalysts, which needs further investigation.59–62 Therefore, both precipitating and aging processes will play key roles for the catalytic performance of the final CuO/ZnO/Al2O3 catalysts.
This journal is © The Royal Society of Chemistry 2016 |