Aqueous biphasic catalysis: Ruhrchemie/Rhône-Poulenc oxo process

Transcription

1 Applied Catalysis A: General 221 (2001) Aqueous biphasic catalysis: Ruhrchemie/Rhône-Poulenc oxo process Christian W. Kohlpaintner a,, Richard W. Fischer a, Boy Cornils b,1 a Celanese Chemicals Europe GmbH Werk Ruhrchemie, D Oberhausen, Germany b Celanese Chemicals Europe GmbH Werk Ruhrchemie, D Hofheim, Germany Abstract The use of water-soluble catalysts represents a significant advance in homogeneous catalysis in general and in the manufacture of n-butyraldehyde in special. Although catalyzed homogeneously, the technique of an immobilized catalyst in mobile phase offers all advantages of a heterogeneous process. Thus, the combination of homogeneous catalysis with heterogeneous catalyst handling yields the most straightforward and soundest oxo process with superior economics Elsevier Science B.V. All rights reserved. Keywords: Water-soluble catalysts; Hydroformylation; Oxo process 1. Introduction With production figures over 6 MM tons per annum, the oxo synthesis (hydroformylation, hydrocarbonylation) belongs to the important organochemical syntheses. It is the most relevant homogeneously catalyzed transformation to bulk and fine chemicals [1]. The bulk production of n-butyraldehyde by addition of the constituents of syngas (CO and H 2 )to propylene is commercially still dominant [2], (Fig. 1). So far, 60 years of oxo syntheses experienced at least five quantum leaps of development ((1) Diaden process with heterogeneous cobalt catalysts; (2) high pressure process with homogeneous Co catalyst; (3) introduction of Rh as the central atom of complex catalysts and (4) of ligand-modified Rh or Co catalysts; (5) the two-phase catalysis [3,4]). As for all homogeneously catalyzed reactions the problem of separating Corresponding author. address: ckohlpainter@celanese.com (C.W. Kohlpaintner). 1 Fax: the homogeneously dissolved catalyst according to definition is immanently disadvantageous. The difficult separation between the catalyst and the reaction products after reaction could only be overcome by expensive recycle processes [3,12]. The variation of the application phase of hydroformylation by biphasic catalysis [5] offered the chance to separate catalyst and reaction products just by decantation. In aqueous, homogeneous two-phase catalysis, the active catalyst for the reaction is (and remains) dissolved in water, so that the reactants and reaction products, which are ideally organic and relatively non-polar, can be separated off after the reaction is complete by simply separating the second phase from the catalyst solution, thus making it easy to recirculate the latter. This technique, makes it possible to combine not only the positive influences of water as a solvent [6 8] but also the inherent advantages of homogeneous catalysis. The resultant basic flowscheme is as simple as possible (Fig. 2). A major aspect of the new technology (Ruhrchemie/ Rhône-Poulenc [RCH/RP] process; commercially X/01/$ see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S X(01)

2 220 C.W. Kohlpaintner et al. / Applied Catalysis A: General 221 (2001) Fig. 1. used for hydroformylation since 1984 [9 13]) is the tailoring of organometallic complexes as catalysts, as is nowadays becoming increasingly important both for industrial application and for new reactions and new products. The unusual development sequence (large-scale industrial application prior to detailed scientific interest of the academia, [13]) was politely referred to by Joó [14] who is together with Kuntz [15] one of the pioneers of aqueous-phase catalysis as induction period, although it was more like a serious case of idea-transfer inhibition. Several compilations review the state of the art [16,17], more or less seen from the academia. Details of the development of RCH/RPs oxo process have also been released [13,18]. In the new oxo process, much was completely different from before, including (a) the catalyst, for the first time applied in water soluble form, in satisfying quality and inexpensevely; (b) the process procedure; (c) the reactor with special mixing of the two-phases; (d) the particular energy balance, as a result of which the process becomes a net steam supplier; and (e) the completely new, simple circulation of the aqueous catalyst solution [11 13]. Various water-soluble ligands have extensively been tested but only TPPTS (tri(m-sulfonyl)triphenylphosphine, vulgo triphenylphosphine, trisulfonated) has proven economically as far as handling and the ratio of price and performance is concerned [12,18]. The new generation oxo process thus becomes essentially a process with the following characteristics: (1) high selectivity, producing virtually exclusive Fig. 2. Flow-scheme of RCH/RP process.

3 C.W. Kohlpaintner et al. / Applied Catalysis A: General 221 (2001) aldehydes with up to 98% linearity; (2) simplicity in terms of apparatus and process operation; (3) net steam supplier, making excellent utilization of the heat of reaction by means of the heat-transfer medium n-butyraldehyde; (4) very simple recycling of the homogeneous oxo catalyst by immobilization using the mobile support water; (5) excellent economics owing to minimal losses of the catalyst metal rhodium; (6) low purity demands on the reactants; (7) excellent process potential from safety and environmental points of view [19]. The new oxo process has been realized at different locations, using propylene as well as butenes as starting olefins. The total capacity is over 600,000 tons per year, which accounts for roughly 10% of the world capacity. The progress of the last years has been described elsewhere [9 13]. 2. State of the art 2.1. Status of Celanese s plants Today, the RCH/RP process is sucessfully operated at two locations in the world. In the early 1980s, the process was developed and commercialized at the Oberhausen site of the former Ruhrchemie AG (now Celanese AG). The first unit started-up in In 1988 and in 1999 two additional trains went on stream. The actual combined capacity of all three units at Oberhausen amounts to over 500,000 tons of C 4 -products, thus representing one of the largest oxo installation at one location. The n/iso-c 4 -aldehydes produced are converted either to the corresponding C 4 -alcohols or after aldolization and hydrogenation to 2-ethylhexanol (2-EH). With an annual volume of 360,000 tons of 2-EH the Oberhausen facility serves mainly the Asian and European markets. In 1996, the RCH/RP process was licensed to Hanwha Corp. in South-Korea. With a total capacity of 120,000 tons the unit started-up sucessfully in The aldehydes are dedicated mainly for the manufacturing of butanols. In the early 1990s developmental work was started to apply the RCH/RP process also to the conversion of higher alkenes than C 4. It was shown that C 4 -olefines or mixtures thereof could easily be converted to C 5 -aldehydes. The overall economics looked attractive and the selling-markets for the target products were picking up as expected. Consequently, a 12,000 tons unit went on stream at the Oberhausen site in 1995, to convert raffinate-2 (a mixture of butene-1 and butene-2) into C 5 -aldehydes which are predominantly converted to the corresponding alcohols and acids. All units are running well without any major problems. The great tolerance of the RCH/RP process with respect to the purity of alkene feedstocks is unsurpassed and adds great flexibility to the plants. The process can be operated for long periods of time at steady and stable conditions thus, allowing a further reduction in workforce even beyond the already low levels realized a couple of years ago. In Table 1 typical process conditions and the composition of the oxo crude product are shown as 15 years averages (Table 1) Recent developments In spite of the outstanding advantages of the modified Rh-TPP based low pressure hydroformylation process (high conversion and selectivity, low pressure operations), the high cost rhodium inventory, the energy balance, and especially the Rh recycle can be counted as downsides of the successor of the old cobalt processes. The technical implementation of a two-phase process, operating with highly selective water-soluble Rh catalysts marks a turning-point in applied homogenous catalysis since A crucial point of development was the manufactute of TPPTS. Sulfonation of the less water soluble triphenylphosphine followed, after neutralisation, by ph-controlled selective extraction and re-extraction procedures, yields an aqueous solution of the tri-sodium salt of tri(m-sulfonyl)triphenylphosphine in good yields. Until today, the ligand synthesis is object of permanent improvement and investigation. Described by a quite simple reaction scheme, the sulfonation technology and the consecutive work up procedure requires significant know how [20]. More than 15 years of experience, with the operating system, demonstrate that it is of utmost importance to reduce by-product formation during the ligand synthesis. The elimination of specific impurities bear a big potential for further economic savings preparing a next step on the way to be a low cost producer, compared to the peers. To minimize catalyst deactivation [21,22] (formation of inactive Rh species, ligand decomposition, e.g. P C bond cleavage by direct oxidative

4 222 C.W. Kohlpaintner et al. / Applied Catalysis A: General 221 (2001) Table 1 Conditions and results: 15 year average of RCH/RPs process Unit Typical value Variance n-butyraldehyde (%) iso-butyraldehyde (%) n-butyralcohol (%) iso-butyralcohol (%) <0.1 <0.1 Butyl formates (%) Traces Traces Heavy ends (%) n/iso ratio Selectivity toward C 4 products (%) >99.5 >99 Selectivity toward C 4 aldehydes (%) Temperature ( C) Total pressure (MPa) CO/H 2 ratio Aqueous/organic phase ratio Heat recovery a (%) >99 >99 Conversion (%) Propylene quality (% propene) a Minus losses by radiation. insertion of the Rh metal or formation of PDSPP [= propyldi(sulfophenyl)phosphine] acting as strong electron donor reducing the amount of active Rh catalyst) it turned out to be beneficial to control the P(III)/Rh ratio very carefully. In addition, it proven that an intermittendly or an almost continuous addition of fresh ligand solution increases the catalyst lifetime (catalyst activity and thereof resulting productivity) considerably [23]. A pseudo-continuous ligand addition mainly compensates the formation of phosphine oxides formed by traces of oxygen which are brought into the system by the feedstock and the syngas. A special technique with specific chemical additives makes it possible to improve the catalyst performance again after a significant drop in activity. Doing so, it is possible to increase the catalyst lifetime considerably. Besides TPPTS a number of other sulfonated phosphine ligands were investigated and tested on a pilot plant scale. Among them are systems which are derived from biphenyl [BISBIS = sulfonated bis(diphenylphosphine)biphenyl, varying grade of sulfonation] or binaphthyl structures (BINAS = sulfonated NAPHOS, sulfonation grade between six and eight). It turned out that Rh/BINAS is the most active and selective water-soluble hydroformylation catalyst [12,24 26]. Its reactivity is up to 10 times higher than TPPTS, even low P(III)/Rh ratios give n/iso selectivities of 98/2 (TPPTS: exceeding 96/4). Concerning the outstanding performance data, BI- NAS would be clearly the ligand of choice. However, compared to the rather simple TPPTS, such complicated systems result in higher manufacturing costs. This disadvantage could be compensated by the lower amount of ligand needed. But BINAS shows higher decomposition rates than TPPTS, up to now the overriding reason not to change the established system. Further development is going on to overcome this problem. As a consequence up to now TPPTS is the best choice for technical application in water based two-phase catalyst systems. The TPPTS ligand easily carries all of the used Rh into the water phase by forming Rh/TPPTS complexes analogous to the complexes known from the simple Rh/TPP system (Fig. 3). Due to the sufficient strong coordination of TPPTS to the Rh metal center almost no Rh (<10 9 g Rh per kg n-butyraldehyde, [12]) leaching into the organic phase is observed a critical success factor and performance prerequisite with respect to the desired economics of the process. Even with the low solubility of propylene or butene in water, the reactivity of the used catalyst system is high enough to provide a very efficient hydroformylation rate [13]. To increase the solubility of propylene in the water phase, plant tests were performed with polyethylene glycols with

5 C.W. Kohlpaintner et al. / Applied Catalysis A: General 221 (2001) Fig. 3. various chain lengths, acting as an auxiliary agent for better propylene solubility. Good reaction rates (rate increase by more than 10%) were achieved, demonstrating a simple tool to increase productivity or to switch to milder conditions than already applied. In summary, the major advantages of the TPPTS system are the simple, mild catalyst recycle, the comparable low Rh inventory (rhodium savings), the high n/iso ratio of >20 and the low energy consumption of the whole process as well as the improved utilisation of propylene and syngas. The water based process is normally operated at a ph-value between 5 and 6 to provide reaction conditions which suppress unwanted side reactions of the aldehydes. A careful ph-control is beneficial for various operation parameters of the protic reaction media: catalyst reactivity and selectivity are directly influenced by and thus dependent on the H 3 O + concentration present in the aqueous phase Short or long chained alkenes as feedstocks With the RCH/RP process it is possible to hydroformylate propene up to pentene with satisfying space time yields (STY). Currently, the process is solely used for the conversion of propene and butene. Compared to propylene, butenes show a lower solubility in water, analogous to their higher homologs [13]. Due to this property the observed hydroformylation reaction rate of butene mixtures can vary due to source, process, concentration, and quality and is significant lower than for propylene. However, already a slight increase in the catalyst concentration is sufficient to achieve satisfying STYS for the manufacture of valeric aldehydes. As a very selective catalyst, the Rh/TPPTS system does not catalyze the hydroformylation of butene-2 providing a simple method for the sole conversion of the terminal alkenes. Thus, the butanes and the not converted butene-2 are separated from the reaction products in a stripping column. Advantageously, the RCH/RP technology can be applied to the hydroformylation of C 3 - as well as C 4 -alkenes successively using the same production unit, only minor changes (e.g., reaction temperature, partial pressures) are necessary to switch between the different feedstock streams [27,28]. This, however, is not the case for higher olefins. Nevertheless, the latter correspond to approximately 25% of the world wide oxo business [1]. A biphasic hydroformylation process would solve the problem of product/catalyst separation in a very elegant manner: for C 10 olefins and higher, the crude aldehyde products can not be separated by distillation due to thermal instability at the required temperatures. There are numerous attempts to overcome the problem of low reactivity of higher olefins using two-phase catalyst and reaction-systems; summarized briefly as: (1) use of amphiphilic water-soluble ligands (influence of olefin solubility or increase of catalyst concentration at the phase interface area), (2) use of co-solvents (alcohols, detergents to improve interface exchange of the

6 224 C.W. Kohlpaintner et al. / Applied Catalysis A: General 221 (2001) feedstock and the reaction products), (3) temperature controlled switch of the catalyst system from the aqueous phase to the organic phase, (4) supported aqueous phase catalysis or the use of immobilized rhodium (unmodified) on a polymeric water-soluble support [29]. The addition of poly(ethylene-glycol) and poly- (ethylene-glycol)derivatives (with medium molecular masses from 400 to 1000) to aqueous two-phase systems yields a significant increase in alkene conversion. Thus, for pentene-1, hexene-1 or octene-1, respectively, the olefin conversion increases to 113% (C 5 ), 182% (C 6 ) and ca. 790% (C 8 ) by applying 12, 16, or 35 wt% of PEG-400. In the case of octene-1 however, the n/iso ratio drops from 95/5 to 83/17, in the other cases these ratios are affected only (changes of ca. 1 to 2%). Importantly, using PEG derivatives, Rh leaching can be kept on a relatively low level ( ppm), turning out as a significant advantage compared to reaction media using methanol or ethylene-glycol as auxiliary agents. The method proved technically feasible. Besides TPPTS and the addition of auxiliary agents, water soluble sulfonated chelating ligands like BI- NAS or BISBIS show excellent performance in the two-phase hydroformylation of higher alkenes. After the addition of traces of tetradecylammoinum-binas (e.g wt.%) the conversion is increased up to 84% without loosing the outstanding n/iso ratio from 99/1. In general, the raw materials for higher oxo-alcohols are based on mixtures of internal olefins with usually low content of terminal isomers. Thus, the next step in development should be a catalyst capable of isomerization and consecutive hydroformylation as well as catalyst systems which can also convert branched alkenes Economics The highly sophisticated RCH/RP process, which together with Union Carbide s Mark IV process represents the state-of-the-art in today s oxo technology, is also reflected in the variable and fixed costs per ton of product. Two major factors attribute for the reduction in the overall manufacturing costs: a highly efficient energy recovery system and minimum capital requirements for an RCH/RP oxo unit. Both are results of the greatly simplified process design which is only possible by applying a water-soluble catalyst. The catalyst/product separation is achieved by a simple phase separation. Therefore, the exothermal reaction heat of the hydroformylation reaction can easily be recovered by a heat exchanger inside the reactor (high energy efficiency) and, secondly, no further distillation steps and columns are required to separate the catalyst from product (with the consequence of less capital requirements). Both factors together with the reduced fixed costs makes the RCH/RP process roughly 8 10% cheaper in manufacturing costs compared to other Rh based processes applying a homogeneous phosphine-modified catalyst [11]. Higher syntheses costs for the ligand TPPTS are included in the overall process economics (variable costs) thus compensating the potentially lower catalyst costs of a Rh-phosphine catalyst. 3. Conclusions Homogeneous catalysis was never the prototype of a heat-and-beat technology. With the described opportunity of quickly and easily separating the catalyst from the reaction products the homogeneous aqueous biphasic process will open the scene for developments other than hydroformylations or hydrogenations. This applies particularly to the possibility of tailoring the catalysts by varying the central atom, the ligand, or even the phase to be used. Thus, this variant of homogeneously catalyzed syntheses will lead to an even more sophisticated field of chemistry. References [1] C.D. Frohning, C.W. Kohlpaintner, in: B. Cornils, W.A. Herrmann (Eds.), Applied Homogeneous Catalysis with Organometallic Compounds, VCH, Weinheim, Germany, 1996, p. 61 (Chapter ). [2] W. Gick, Celanese AG, private communication. [3] J. Falbe (Ed.), New Syntheses with Carbon Monoxide, Springer, Berlin, 1980, (Chapter 1). [4] B. Cornils, J. Mol. Catal. A 143 (1999) 1. [5] J. Manassen, in: F. Basolo, R.L. Burwell (Eds.), Catalysis Progress in Research, Plenum Press, London, 1973, p [6] A. Lubineau, J. Augé, in: Topics in Current Chemistry, Vol. 206, p. 1. [7] A. Lubineau, J. Augé, in: B. Cornils, W.A. Herrmann (Eds.), Aqueous-Phase Organometallic Catalysis, Wiley, Weinheim, 1998, p. 17 (Chapter 2).

7 C.W. Kohlpaintner et al. / Applied Catalysis A: General 221 (2001) [8] Grieco, Organic Synthesis in Water, Blackie, London, [9] B. Cornils, J. Falbe, in: Proceedings of the 4th International Symposium on Homogeneous Catalysis, Leningrad, September 1984, p [10] H. Bach, W. Gick, E. Wiebus, B. Cornils, in: Proceedings of the First IUPAC Symposium on Organic Chemistry, Jerusalem, 1986, p. 295 (Abstracts). [11] E. Wiebus, B. Cornils, Chem.-Ing.-Technol. 66 (1994) 916. [12] B. Cornils, E. Wiebus, CHEMTECH 25 (1995) 33. [13] B. Cornils, Org. Proc. Res. Dev. 2 (1998) 121. [14] F. Joó, A. Kathó, J. Mol. Catal. A 116 (1997) 3. [15] E.G. Kuntz (then Rhône-Poulenc), FR Patent 2,349,562; 2,366,237; 2,733,516 (1976). [16] P. Kalck, F. Monteil, Adv. Organomet. Chem. 34 (1992) 219. [17] W.A. Herrmann, C.W. Kohlpaintner, Angew. Chem. Int. Ed. Engl. 32 (1993) [18] B. Cornils, E.G. Kuntz, J. Organomet. Chem. 502 (1995) 177. [19] B. Cornils, E. Wiebus, Recl. Trav. Chim. Pays-Bas. 115 (1996) 211. [20] R. Gärtner, B. Cornils, H. Springeer, P. Lappe, AG Ruhrchemie, DE Patent 3,235,030 (1982). [21] J.A. Kulpe, Dissertation, Technische Universität München, [22] C.W. Kohlpaintner, Dissertation, Technische Universität München, [23] C.D. Frohning, C.W. Kohlpaintner, in: B. Cornils, W.A. Herrmann (Eds.), Aqueous-Phase Organometallic Catalysis, Wiley, Weinheim, 1998, p [24] W.A. Herrmann, H. Bahrmann, W. Konkol, J. Mol. Catal. 73 (1992) 191. [25] W.A. Hermann, C.W. Kohlpaintner, R.B. Manetsberger, H. Bahrmann, H. Kottmann, J. Mol. Catal. 97 (1995) 65. [26] H. Bahrmann, H.W. Bach, C.D. Frohning, H.J. Kleiner, P. Lappe, D. Peters, D. Regnat, W.A. Herrmann, J. Mol. Catal. 116 (1997) 49. [27] C.D. Frohning, C.W. Kohlpaintner, in: B. Cornils, W.A. Herrmann (Eds.), Aqueous-Phase Organometallic Catalysis, Wiley, Weinheim, 1998, p [28] H. Bahrmann, C.D. Frohning, P. Heymanns, H. Kalbfell, P. Lappe, D. Peters, E. Wiebus, J. Mol. Catal. 116 (1997) 35. [29] H. Bahrmann, S. Bogdanovic, in: B. Cornils, W.A. Herrmann (Eds.), Aqueous-Phase Organometallic Catalysis, Wiley, Weinheim, 1998, p. 307.