One Step Formation of Propene from Ethene or Ethanol through Metathesis on Nickel Ion-loaded Silica

Increased propene production is presently one of the most significant objectives in petroleum chemistry. Especially the one-step conversion of ethene to propene (ETP reaction, 3C2H4 → 2C3H6) is the most desired process. In our efforts, nickel ion-loaded mesoporous silica could turn a new type of ETP reaction into reality. The one-step conversion of ethene was 68% and the propene selectivity was 48% in a continuous gas-flow system at 673 K and atmospheric pressure. The reactivity of lower olefins and the dependences of the ETP reaction on the contact time and the partial pressure of ethene were consistent with a reaction mechanism involving dimerization of ethene to 1-butene, isomerization of 1-butene to 2-butene, and metathesis of 2-butene and ethene to yield propene. The reaction was then expanded to an ethanol-to-propene reaction on the same catalyst, in which two possible reaction routes are suggested to form ethene from ethanol. The catalysts were characterized mainly by EXAFS and TPR techniques. The local structures of the nickel species active for the ETP reaction were very similar to that of layered nickel silicate, while those on the inert catalysts were the same as that of NiO particles.


Introduction
The mainstay of petrochemical industries in the world is still ethene (C2 = ), while the need for propene (C3 = ) is rapidly increasing due to the increasing demand of polypropene, propene oxide, etc.
OPEN ACCESS [1,2]. This trend has led to the need for the conversion of C2 = to C3 = (ETP reaction) or of increased production of C3 = . Three kinds of measures are applied or suggested for this problem. First is the so-called mild-cracking: however, the increment of C3 = in this case is limited due to narrow range of applicable reaction conditions. Second is metathesis of C2 = and butenes (C4 = ) to form C3 = , for instance, the ABB Lummus process [3]. Its disadvantage is the necessity for equimolar amounts of C2 = and C4 = . Third is direct ETP conversion without any addition of other hydrocarbons. This would be the most desirable route, but no good catalyst for the reaction has been found so far. Supported molybdenum [4] and tungsten oxide [5] have been reported as possible catalysts, but their activity was so low as to be observed only in a closed recirculation system. On the other hand, various zeolites have been employed as catalysts for this reaction [6,7]. The reaction involves oligomerization/ polymerization of lower olefins, subsequent decomposition to yield C3 = or other species on the strong acid sites of the zeolites, and selective evolution of C3 = due to the shape-selectivity of zeolite pores. This process has the limitations of selectivity due to the shape selectivity and of catalyst lifetime owing to coke formation. The present objective is the selective formation of C3 = without the shape-selectivity.
The catalytic activity of Ni ion for the dimerization or oligomerization of olefins was found 50 years ago and has been widely studied [8]. In the case of heterogeneous catalysis, Ozaki et al. [9][10][11][12] reported the high catalytic activity of Ni/SiO 2 for the dimerization, though severe deactivation during the reaction prevented it from being applied in the practical process. They also found that acidic supports were effective for enhancement of the catalytic activity of nickel. A similar catalytic activity was also confirmed on various Ni-zeolites [13,14] or on Ni supported on MCM-41 [15] in a closed recirculation system. Since we had already found the acidic properties of silica MCM-41 [16][17][18][19][20][21][22][23], we tried the dimerization of C2 = to C4 = . During the study a subsequent reaction of the produced C4 = and unreacted C2 = to yield C3 = was uncovered. As a result we found that Ni ion-loaded mesoporous silica (Ni-MCM-41, abbreviated as Ni-M41) was highly active in the ETP transformation.
On the other hand the use of bio-ethanol (bEtOH) as an additive for automobile fuels has increased rapidly all over the World. This is one way of using renewable resources to suppress carbon dioxide emissions, while another challenge is the conversion of bEtOH into various olefins and their use for production of chemicals and polymers [1,2,. The latter would be very significant for the longterm fixation of carbon dioxide. Many efforts have therefore devoted to the development of systems for converting bEtOH to C2 = and other lower olefins. In particular conversion to C3 = is desirable due to the greater demand for C3 = derivatives [1,2].

Conversion of Ethene to Propene on Ni-M41 Catalysts
The reactions on Ni-M41 were examined as a function of reaction temperature. The dimerization of C2 = to C4 = mainly proceeded at 573 K. When 0.5 g of Ni-M41(Si/Ni=15) was used, the degrees of conversion of C2 = and the selectivity to C4 = reached 43 and 93%, respectively. The production ratio of 1-, trans-2-, and cis-2-butene was 0.5:1.0:0.3. At 673-723 K the major products were C3 = and C4 = .
The respective conversion levels were dependent on the partial pressure of C2 = and the contact time, as shown later. Hexenes, the product of C2 = trimerization, were observed at the wide temperature range but the yields were always less than 5%.
When silica gel was used as the support instead of M41 and nickel ion was loaded with the usual impregnation method, both the conversion level and the selectivity of C3 = were very poor. In addition, no C4 = was produced on M41 alone, indicating the necessity of nickel ion for the reaction. It follows that the coexistence of nickel ion and mesoporous structure of the support make the C3 = formation possible. The catalyst was continuously used at 673 K for 10 h to determine the possible deactivation. Small changes in the catalytic activity for the formation of C3 = were observed in the initial stage, but the activity became stable within 2 h and no deactivation was found during the 10 h experiment. The XRD patterns and the surface areas of Ni-M41 remained unchanged after the catalytic runs. Thus the stability of the present Ni-M41 catalysts under the present reaction conditions could be confirmed.
The correlations between the product distribution and the reaction conditions were then investigated. In the range P C 2 H 4 =10-50%, the conversion levels of C2 = and to C3 = and C4 = increased monotonously with increasing P C 2 H 4 . At P C 2 H 4 =49.7%, the respective conversions to C3, C4, and C6 olefins were 33, 29, and 6% on 0.3 g of Ni-M41 (20). The carbon balance was 99.8% in each experiment, which indicates almost no production of ''unknown products''. The degree of conversion to C3 = , 33%, appears to rather small but it should be noted that the concentration of unreacted C2 = was about 34% under the present conditions and the ratio of C3 = /C2 = =33/34 in carbon basis would be sufficiently great. Figure 1 shows the change in product distribution as a function of the weight of Ni-M41 employed, i.e., the contact time dependence of the reaction. Clearly, longer contact times resulted in greater conversion of C2 = and better selectivity for C3 = , while the selectivity for C4 = decreased and that of hexenes was almost constant. Propene is indeed the secondary product in the consecutive reaction of C2 = on Ni-M41. At 0.5 g of Ni-M41 (43), the degrees of C2 = conversion and C3 = selectivity were 55 and 54%, respectively. The reaction of C2 = and 1-butene was then studied to clarify the mechanism of the C3 = formation and the results are summarized in Figure 2A. One can recognize the selective formation of C3 = on Ni-M41 at the temperature of 623 K and higher. The increment in the amount of C4 = at 523-573 K is due to the dimerization of C2 = . The selective production of C3 = would indicate the progress of the metathesis reaction on this catalytic system. To confirm the reaction pathway in more detail, we examined two kinds of reactions. The first was the reaction of 1-hexene. When 1-hexene was introduced onto the Ni-M41 catalyst, methane, C2 = , C4 = , and pentenes were produced, besides C3 = , indicating the random scission of carbon-carbon bonds of 1-hexene. This indicates little possibility that C2 = and 1-butene first afford hexenes and the resulting hexenes homolytically decompose to give C3 = selectively in the experiments of Figure 2A. The second reaction examined was the retro-metathesis reaction. Namely the reaction of C3 = on Ni-M41 was investigated and shown to readily proceeded to yield equimolar C2 = and C4 = as shown in Figure 2B. The amounts of by-products were always small. It was further confirmed in separate experiments that the parent M41 was not active for the reaction of C2 = and C4 = . All of the results therefore strongly suggest the metathesis reaction on Ni-M41 and that the active center for the catalysis would be nickel ion.
Although at present we cannot preclude the possibility of a decomposition mechanism of higher olefins because other types of reaction mechanisms have been suggested on Cr [56] or Zr [57], we believe that the metathesis mechanism ( Figure 3) is the most plausible reaction mechanism for the C3 = formation on Ni-M41. That is, at first two C2 = molecules dimerize to give 1-butene on Ni, and the resulting 1-butene then isomerizes to 2-butene on the acid sites of M41, and finally the metathesis of the produced 2-butene with unreacted C2 = proceeds to form C3 = on Ni. The acidic properties of M41 silica were already been reported by us [16][17][18][19][20] and the other research groups [15,[21][22][23] and the isomerization of 1-butene to 2-C4 = , a typical acid-catalyzed reaction, was indeed confirmed on silica M41 [15].  (15). The codes C2-C6 mean ethene, propene, butenes, pentenes, and hexenes. In Figure 2B, the left vertical axis is the amount of propene and the right those of the products. As has been summarized by Grubbs [58,59] and Arpe [1], the metathesis reaction is one of the most important organic reactions. Despite world-wide study it is well known that the catalytically active species for the reaction are confined to Mo, W, Ru, and Re. The present results might suggest that nickel-ion loaded mesoporous silica is also active for the metathesis of C4 = and C2 = to yield C3 = in the gas-phase flow reaction. Mori et al. [60] suggested the possibility of metathesis on a Ni(0) complex in their discussion, while Baker et al. [61] concluded no progress of a metathesis reaction on Ni complexes. At the moment, no reports claim nickel ion as a catalytically active species for metathesis. It is noteworthy that the surface density of Ni is approximately 0.5 Ni/nm 2 in the case of Ni-M41 (20) on the assumption of the even distribution of nickel on the surface. The valence of nickel ion in the mesoporous silica were not studied here. There are two possibilities for the redox cycles of nickel species, Ni(I)-Ni(III) and Ni(0)-Ni(II). In Section 3.3 the TPR experiments will indicate the difficult reduction of nickel species loaded on M41 to Ni(0), which would be one important factor for generation of the catalytic activity. Therefore we speculate that the Ni(I)-Ni(III) system would be the possible redox cycle for the metathesis reaction. The stability of Ni(I) in the zeolites [62][63][64] and mesoporous materials [65,66] support the speculation that Ni(I) is an active center and a Ni(III) carbene is produced as an intermediate.  (30), or W (30) loaded M41 were all 5% or less at 673 K, and most of the products were "unknown products". It should be noted, however, that the gas-phase dimerization-isomerization-metathesis of C 2 = on tungsten catalysts was independently reported by Basset et al. [67] and the others [68]. The difference clearly results from the discrepancy of reaction conditions. Ru or Re loaded MCM-41 were prepared separately through the conventional impregnation method and employed as the catalyst for the present reaction at 673 K because of its high activity reported at lower temperatures, but no activity for the 3C 2 = → 2C 3 = reaction was observed in our experiments. This would be due to the lack of activity of Ru or Re for the dimerization of C 2 = and the difference of the reaction temperature applied. Clearly only nickel ion shows the unique activity for the ETP conversion in the gas-phase reaction at 673 K. The reason for the specific activity of nickel ion on MCM-41 would be a target of the future work.

Reaction of Ethanol on Ni-MCM-41
The influence of temperature on EtOH conversion over Ni-M41 is summarized in Figure 4. Many kinds of products were formed in addition to C2 = . Diethyl ether (DEE) was mainly obtained at around 523 K. DEE has been reported earlier as an intermediate compound in the dehydration, decomposing to yield EtOH and C2 = at higher temperatures [49,50]. The C2 = yield increased sharply at 573 K, and reached ca. 70% at 623 K or above. The C4 = yield reached a maximum at 623 K, while maxima in C3 = yield occurred at 673 and 723 K. Notably, AAD was formed at 573-723 K, although not in large amounts, which will be discussed later.
The stability of Ni-M41 was examined at 673 K. The catalytic activity did not change during 20 h of continuous time on stream. In addition, the carbon-based mass balances were always ca. 100%, within the experimental errors. The results demonstrate the stable catalytic activity of Ni-M41. However, there is the possibility that losses of catalytic activity could not be determined under these conditions because the catalytic activity of Ni-M41 was very high, as will be revealed in a following paragraph, and the conversion levels of EtOH were always ca. 100%. The yields of C2 = , C3 = , C4 = , and AAD were 67, 16, 5, and 7 %, respectively. The values should be compared with those of the reaction of C2 = on the same catalyst reported previously (see Section 3.1). At 673 K and P C2= =10 vol %, the C2 = conversion and selectivity to C3 = and C4 = were reported to be 42  Equation (7') is well-known as the Tishchenko reaction, and Equation (9) as the Fisher Esterification. The experimental results indicate the progress in Equation (7) instead of (7') on Ni-M41. It is already known that hydrolysis of ETA [Equation (8'), the reverse reaction of Equation (9)] gives acetic acid and EtOH, but Equation (8) is not popular. The reverse reaction of Equation (8), however, was already confirmed to proceed catalytically and was put into practical use by Showa Denko K.K., Japan [69]. To postulate Equation (8) is therefore legitimate. The sequence of reactions (6)-(7)-(8)-(9)-(8) would result in the formation of C2 = from AAD and EtOH through ETA and acetic acid as the intermediates.

Characterization of Nickel Species Loaded on the Mesoporous Silica
Three kinds of Ni-loaded M41 samples were prepared to clarify the state of the nickel ion. They were prepared by TIE, impregnation (IMP), equilibrium adsorption (EA) of [Ni(NH 3 ) x ] 2+ as shown later. The colors of the EA, TIE and IMP catalysts were pale ivory, pale ivory and pale blackish purple, respectively. The following results and discussion will be described on the premise of no essential difference in the pore structures among the M41 samples employed here.
The activity of the TIE catalyst for the ETP reaction was first compared with those of the IMP catalysts. Figure 7 shows the catalytic activities of Ni-M41, Ni/M41, and Ni/SiO 2 at 1 and 4 h after the beginning of the reaction. Only the Ni-M41 catalyst prepared by the TIE method showed high and stable activity for the ETP reaction, while the activity of Ni/M41 or Ni/SiO 2 was very low and decreased with the reaction time. To clarify the origin of the great difference between the activities of TIE-and IMP-catalysts, the catalysts were characterized by various methods. Surface areas of Ni-M41 and Ni/M41 calcined at 773 K were 856 and 822 m 2 /g, respectively. The values indicate little correlation between the surface area and the catalysis. The XRD measurements did not confirm any nickel-related crystalline phases on the TIE-catalysts, but showed the formation of NiO particles on the IMP catalysts.
More detailed characterization of the supported nickel species has been carried out by using the EXAFS and TPR techniques. Figure 8 shows radial structure functions (RSFs) of Ni-ion loaded catalysts and reference compounds. Most of the samples except the Ni foil gave two peaks at 0.15-0.16 and 0.26-0.28 nm though their respective intensities were depended on the samples. The latter peaks indicate the presence of Ni-Ni pairs. The conventional curve fitting analysis was applied to the spectra to determine the interatomic distance and the coordination numbers around the nickel atom and the results are summarized in Table 1. It should be noted in the table that the accuracy of coordination numbers estimated for the second coordination sphere has some uncertainty because the range of EXAFS spectra adopted here was limited to 120 nm −1 . We employed Ni foil, NiO, and two kinds of layered nickel silicates as the reference compounds. The reaction times after the beginning of the reaction are described in the parentheses. Catalysts: Ni-M41, Ni/M41, Ni/SiO 2 , Ni-NH 3 /M41, and Ni-NH 3 /SiO 2 . Catalyst 0.3 g, total flow rate 11 mL/min, 0.1 MPa, ethene 9.3% and water 1.4% (N 2 balance). Alkanes: methane and ethane.   Table 1 and Figure 8 reveal several important points from the comparison with the literature [70][71][72][73][74][75][76][77][78][79][80]. The distance and the coordination number of the first shells (oxygen backscatterer) on Ni-M41 indicate the presence of hexacoordinated Ni 2+ 6c ions [71] in the TIE sample. Yang et al. [72] reported that the nickel ion substituted for Si ion in the M41 framework has a tetrahedral coordination structure, which indicates a complete difference between the coordination states of the nickel ions in the present TIE sample and in the Ni-MCM-41 prepared by the sol-gel method. The distances of the second shell (Ni and Si backscatterer) of Ni-M41, 0.305 and 0.336 nm, were longer than those of NiO and Ni/M41, and similar to those of layered nickel silicates. The findings concerning the first and second shells strongly indicate that the nickel ion in the TIE sample has a layered nickel silicate-like structure. The typical layered nickel silicates consist of a NiO 6 layer sandwiched by one or two silica layers [70,73,74,81,82]. The EXAFS spectra of the two types of nickel silicates [ Table 1 and Figures 8h,i], however, were quite similar to each other, as has already been reported by several workers [70,71,75]. At the moment therefore we cannot determine the exact surface structure of nickel ion on the basis of the EXAFS spectra. In contrast, the Ni/M41 and Ni/SiO 2 catalysts gave spectra assignable to NiO species because the Ni-Ni distance was shorter than those of Ni-M41 and the layered-nickel silicates. It follows that the preparation methods have an essential effect for the appearance of the catalytic activity through the change in loading states of nickel ion on the supports. The role of layered-nickel silicates on M41 for the catalysis will be described in the later paragraphs in more detail.
The nickel species in the TIE catalyst gave a reduction peak at 931 K in the TPR experiments. This temperature was much higher than those of Ni/M41 (839 K), Ni/SiO 2 (746 K), and NiO alone (673 K). In TPR profiles of the nickel silicates, similar to previous works [74,76], the broad (i) 10 reduction peaks were observed on these samples. On the basis of many TPR experiments reported so far [72,79,83,84], we can summarize the TPR peak regions of nickel on silica as follows: Ni-oxide, the cationic form of nickel on the silica surface, and the nickel ion forming some surface composite compounds could be reduced at ca. 600-800, 800-900, and 900-1,000 K, respectively. The reduction temperature of Ni-M41 clearly falls within the region of the reduction of composite compounds. This further supports the above conclusion that the nickel ion in the TIE sample might form the layered nickel silicate like-structure on the surface. The amounts of H 2 consumed in the TPR experiments of the TIE sample and the layered silicates were almost equal to those of nickel ion contained in the respective samples. In contrast, the IMP catalysts gave a much higher ratio than unity. The composition of Ni oxide prepared by thermal decomposition of nickel carbonate at 773 K was reported to be NiO 1.13 and its color was black [85]. In addition, Ni-oxide prepared by the impregnation onto silica support was suggested to be most probable Ni 2 O 3 species [84]. The larger TPR peaks than those expected from H 2 /Ni = 1 and the pale blackish purple color of the present IMP catalysts both indicate the existence of the mixture of Ni 2 O 3 and NiO on the silica surface.
The surface layered nickel silicate is reported to be produced by loading of nickel ion as amine-complexes onto silica in a basic aqueous solution and then heating them in air at 623-1,073 K [71,72,76,[78][79][80]83,86,87]. Hadjiivanov et al. reported that the EA of [Ni(NH 3 ) x ] 2+ onto silica gel at pH 12.3 and the subsequent calcination at 623 K is effective for its preparation with ease [83]. We have here applied their method to prepare the samples containing the surface layered nickel silicate (the EA catalysts) to evaluate its role for the catalysis, in which the EA samples were finally calcined at 773 K. The XRD patterns of the Ni-NH 3 /M41 and Ni-NH 3 /SiO 2 catalysts did not show any diffraction peaks assignable to the layered nickel silicate, indicating the domain size of the surface layered nickel silicate was not large. The fine structure of nickel ion in the EA catalyst was studied by XAFS and the results are shown in Figure 8 and Table 1. It is clear that the spectra were very similar to those of the layered nickel silicate. The TPR profiles of the EA catalysts were separately measured. They have much resemblance to that of Ni-M41 though the reduction temperatures, 898 and 927 K, were somewhat lower than that of Ni-M41. All of the results clearly indicate the formation of the layered nickel silicate on the silica surface by the EA method, as has been reported by several authors.
The catalytic activity of the Ni-NH 3 /M41 and Ni-NH 3 /SiO 2 samples is shown in Figure 7. It was lower than that of the Ni-M41 sample (the TIE catalyst) while much greater than those of the IMP samples. Deactivation during the reaction was also observed with the EA samples, but the degrees were smaller than those of the IMP catalysts. More detailed investigation into the preparation conditions of the EA catalysts possibly leads to raising their catalytic activity to the same levels as that of the TIE catalyst. This estimation was indeed realized partly by Lehman et al. [88] All results presented here showed that the TIE method is the most effective to prepare the active nickel ion-loaded catalysts and the high catalytic activity would result from the effective formation of surface layered nickel silicate-like structure. It would be worth to note that we attempted to measure the dispersion of nickel metal on the TIE samples after the TPR experiments by the conventional CO adsorption [89,90] but we could not find any irreversible adsorption of CO. This means that the state of nickel metal on the TIE catalysts is entirely different from those on the conventional catalysts, which would be a target for the future study.

Conclusions
Our reports have for the first time claimed the gas-phase metathesis on nickel-containing catalysts at around 673 K. The specific characteristics of this finding are nickel, gas-phase, and high temperature. The reaction mechanism is suggested to be the dimerization of C2 = , the isomerization of the produced 1-C4 = , and the metathesis of C4 = and C2 = to yield C3 = . The reaction was then expanded to ethanol and we could also get C3 = from EtOH. Two reaction routes for the formation of C2 = from EtOH on Ni-M41 were revealed and proceeded in parallel. One is the dehydration route via DEE as intermediate. The other is a complicated route through AAD and ETA as intermediates. The reaction rate of the latter route is slower than that of the former, since the formation of AAD was observed in a wide range of SV values. The C2 = produced was converted to C3 = through dimerization, isomerization, and metathesis. The present results indicate that the formation of C3 = from C2 = or EtOH could be achieved by not using the shape selectivity well known in zeolite catalysis. The layered nickel-silicate like structure would be the active species for the new type of ETP reaction. More detailed investigation of the present system would develop a new horizon in gas-phase metathesis. and Tetsuo Suzuki of the NEDO research group for their helpful discussion. This work was financially supported by three Grants-in-Aids (JSPS, NEDO, and ALCA) from the Ministries, MEXT and METI of Japan.