摘要:

Mono- and polymetallic lanthanide-containing functional assemblies a field between tradition and novelty Claude Piguet*a and Jean-Claude G. Bünzlib a Department of Inorganic Analytical and Applied Chemistry University of Geneva 30 Quai E. Ansermet CH-1211 Geneva 4 Switzerland. E-mail Claude.Piguet@chiam.unige.ch b Institute of Inorganic and Analytical Chemistry University of Lausanne BCH 1402 CH-1015 Lausanne Switzerland. E-mail Jean-Claude.Bunzli@icma.unil.ch Received 19th May 1999 The variable and versatile co-ordination behaviour of lanthanide metal ions LnIII limits their selective introduction into organised molecular or supramolecular architectures. The design of lanthanide-based devices is thus a special challenge since their specific electronic magnetic or spectroscopic properties result from a precise control of the co-ordination sphere around the metal ions.The lock-andkey principle associated with the preorganisation of rigid macropolycylic multidentate ligands tailored for one particular LnIII only partially fulfils these structural requirements. The development of less constrained macrocyclic ligands or macrocycles bearing pendant arms allows a smooth transition toward flexible (predisposed) receptors leading to the application of the induced fit principle in lanthanide co-ordination chemistry. According to this concept programmed secondary non-covalent interstrand interactions (p-stacking hydrogen bonds electrostatic re- Claude Piguet was born in 1961. He studied chemistry at the University of Geneva and obtained his PhD thesis with felicitations in 1989 in the field of biomimetic copper–dioxygen complexes.He pursued his formation as a postdoctoral fellow in the group of Prof. J.-M. Lehn (University of Strasbourg 1989–1990) and then collaborated as Maître-assistant with Prof. A. F. Williams (University of Geneva 1990–1994) and Prof. J.-C.G. Bünzli (University of Lausanne 1995). In 1995 he received the Werner Medal of the New Swiss Chemical Society for his research in the field of supramolecular chemistry of dand f-block metal ions. Recipient of the Werner grant for the project ‘Toward Organized Luminescent Materials’ and Lecturer at the University of Geneva (1995–1998) he was nominated full Professor in Inorganic Chemistry in 1999.His research interests and topics include the design of controlled self-assembled supramolecular complexes with d-block and fblock metal ions the development of functional lanthanide probes and sensors the preparation of lanthanide-containing metallomesogens and the application of paramagnetic NMR for characterizing co-ordination complexes in solution. Claude Piguet This journal is © The Royal Society of Chemistry 1999 1 Mono- versus polymetallic lanthanide-containing functional devices The lanthanides correspond to the first period of the f-block elements starting at lanthanum (Z = 57) and ending at lutetium (Z = 71). As a result of the low energy of the 4f orbitals one 5d and two 6s electrons are easily removed leading to a complete series of stable LnIII cations possessing characteristic 4fn open- Jean-Claude Bünzli was born in 1944.He earned a degree in chemical engineering in 1968 and a PhD in 1971 (Swiss Federal Institute of Technology Lausanne) for his work on the kinetic behaviour of Nb and Ta pentachloride adducts. He spent two years at the University of British Columbia as a teaching postdoctoral fellow (photoelectron spectroscopy) and one year at the Swiss Federal Institute of Technology in Zürich (physical organic chemistry). He was appointed assistant-professor at the University of Lausanne in 1974 and started a research program on the spectrochemical properties of f-elements. He was promoted as a full Professor of Inorganic and Analytical Chemistry in 1980.He acted as the elected Dean of the Faculty of Sciences (1990–1991) and as one of the elected Vice-Rectors of the University (1991–1995). He held positions of invited professor at the Université Louis Pasteur Strasbourg in 1996 and at the Science University of Tokyo in 1998. He is a member of the Finnish Academy of Sciences and Letters and of the editorial boards of Spectroscopy Letters and Acta Chemica Scandinavica. His present research focuses on supramolecular chemistry of lanthanide ions with a special emphasis on the design of luminescent devices. Jean-Claude G. Bünzli 347 Chem. Soc. Rev. 1999 28 347–358 pulsion) assist the complexation process leading to an ultrafine tuning of the metallic co-ordination sites.These two complementary approaches are discussed and evaluated for the design of organised mono- di- and polymetallic lanthanide complexes together with the consideration of new semirigid multidentate podands which combine both aspects. shell configurations (n = 0 for La ? n = 14 for Lu). These 4f orbitals have less radial extension than the filled 5s2 and 5p6 orbitals and are thus shielded from external perturbations1 and the 4f electrons are little involved in covalent interactions upon the formation of chemical bonds leading to a poor stereochemical control in lanthanide-containing edifices. Consequently the LnIII ions display large and variable co-ordination numbers (CN = 8–12)2 which are difficult to predict because the hard LnIII ion will complete its co-ordination sphere by binding small molecules or anions (water chloride hydroxide etc.) if the number of available sites offered by the host is too low.On the other hand steric constraints strongly influence the coordination sphere so that a given multidentate receptor may impose a particular co-ordination number around the metal. Finally when one realises that the fine tuning of the fascinating electronic spectroscopic and magnetic LnIII properties required for the design of functional molecular building blocks with potential uses in biology medicine and materials sciences results from a precise structural control of the metal ion sites (accessibility geometry symmetry type and number of the donor atoms nature of the different ions in polymetallic assemblies),3 it is obvious to conclude that the selective incorporation of LnIII ions into highly organised architectures is the limiting step for designing programmed functional devices.Most of the systems of practical use today rely on monometallic lanthanide complexes with optimised structural and electronic properties. Some homodi- and homotrimetallic complexes have been tested too with their designs based on the principles prevailing for monometallic species. Prospects for the development of such systems are good as far as chemical properties are concerned for instance hydrolytic catalysts for biological systems4 or porphyrinogen complexes for the fixation and reduction of dinitrogen.5 On the other hand physico-chemical properties of homodimetallic species are often not much enhanced compared to those of the monometallic systems and the development of heteropolymetallic architectures in which a judicious combination of different 4f ions imbedded at specific locations would offer new perspectives for programming advanced functional devices such as (i) micelles or polymers in which the luminescent ion is sensitised by light-harvesting ligands organised around a nonluminescent ion (ii) vectorial devices for energy and/or electron migration (iii) systems acting both as luminescent probes and contrast agents (iv) precursors for doped materials in which the lanthanide ions have to be inserted at given distances.Such heteropolymetallic complexes create an additional task for the chemist since they require the recognition of specific lanthanide ions by the different compartments of a polytopic receptor.This represents a real challenge because as far as co-ordination chemistry is concerned the only difference between LnIII ions is a small monotonous contraction of the ionic radius with increasing atomic numbers. In going from LaIII to LuIII the total relative contraction amounts to only ca. 16% a figure that can be compared to the difference in ionic radius observed between Na+ and K+ (26%) and careful attention has therefore to be given to the design of polytopic receptors for preparing pure heteropolymetallic lanthanide-containing devices. In this paper we review the various methods adopted so far to master the chemical environment of the lanthanide ions in monometallic systems and we explore ways to produce polytopic receptors for heteropolymetallic 4f–4f edifices based on the experience gained with homodimetallic 4f–4f and heterodimetallic 4f–3d complexes.2 Underlying concepts for the control of the LnIII co-ordination sphere Detailed thermodynamic investigations of lanthanide complexation processes in water point to a remarkable compensa- Chem. Soc. Rev. 1999 28 347–358 348 tion effect responsible for the emergence of the well-known electrostatic trend i.e. a monotonous increase of the formation constants of the complexes with the decreasing size of LnIII.6 Two successive steps can be written for the net complexation reaction of a LnIII ion with a ligand L corresponding to dehydration [eqn.(1)] followed by the combination of the desolvated partners [eqn. (2)]. [Ln(H2O)n]3+ + [L(H2O)p]x2 " [Ln(H2O)m]3+ + [L(H2O)q]x2 + (n 2 m + p 2 q) H2O (1) (2) [Ln(H2O)m]3+ + [L(H2O)q]x2 " [LnL(H2O)m + q](32x)+ Opposite enthalpic and entropic contributions are expected for each reaction but the compensation effect assumes that the free energy for the dehydration process [eqn. (1)] is negligible at room temperature because DH1 Å TDS1. The global free energy of the complexation process is thus dominated by the enthalpy-driven combination step (DGglobal = DG1 + DG2 Å DG2) leading to increased formation constants with increasing charge density on the cation. Although this simple approach is often used as a guideline for interpreting thermodynamic data for lanthanide complexes the introduction of steric constraints preorganisation and/or chelate effects within sophisticated receptors may alter the expected electrostatic trend because specific intramolecular interactions and solvation effects are not considered in this simple model.Some selectivity ( = deviation from the electrostatic trend) for the complexation of LnIII has been evidenced for polycarboxylates and related acyclic polydentate receptors but the rationalisation of these observations is difficult and the associated structural control is limited.6 On the other hand the recently emerging applications of lanthanide-containing systems in chemistry biology and medicine require a high degree of structural and electronic control as demonstrated by the following three examples.(a) Luminescent sensors and molecular light-converters can take advantage of the specific properties associated with 4fn open-shell configurations of LnIII ions (narrow emission lines in the visible or IR range long-lived excited states) if the following requirements are fulfilled in the final devices (i) a precise structural and geometrical arrangement of the donor atoms around LnIII (ii) a good protection of the metallic site from solvent molecules possessing high-frequency C–H N–H or O–H vibrations which deactivate the metal-centred excited states (iii) sufficient thermodynamic stability and kinetic inertness and (iv) suitable ligand- and metal-centred excited levels for efficient intramolecular energy transfers.7 The design of a receptor satisfying these criteria is difficult for monometallic complexes and becomes a stimulating challenge for extended heteropolymetallic assemblies working as directional light-converting devices or logic gates (Fig.1a).8 (b) Similar considerations apply to contrast agents used in magnetic resonance imaging (MRI Fig. 1b) and based on GdIIIcontaining edifices in view of the 8S7/2 electronic configuration of this paramagnetic ion. The accessibility to the metal ion has to be strictly controlled since water molecules must enter the first co-ordination sphere to maximise inner-sphere relaxivity but no de-complexation is tolerated because of the high toxicity of the aquo-ion.9 Moreover the exchange rate between coordinated and bulk water should be optimised to ensure an efficient transfer of the paramagnetic information.The development of successful contrast agents again relies on the careful engineering of organised architectures as recently exemplified by the design of a calcium-sensitive magnetic resonance imaging contrast agent based on a dimetallic GdIII supramolecular edifice.10 (c) The catalytic sequence-specific cleavage of phosphodiester bridges in RNA and DNA is required in antisense technology used for the development of new therapies. Recent developments have shown that co-ordination of water mole-Fig. 1 Schematic representation of lanthanide-containing functional devices working as a) a UV–vis light converter b) a MRI contrast agent and c) a catalyst for the hydrolysis of the phosphodiester bond.cules and phosphate residues to LnIII respectively assists the deprotonation of the nucleophile and activates the substrate providing the needed catalytic effect which has been first demonstrated with monometallic macrocyclic LnIII complexes conjugated to oligonucleotides (Fig. 1c).11 Interestingly impressive accelerations have been observed for homodimetallic lanthanide complexes in which the metal ions activate simultaneously the nucleophile (water) and the electrophile (phosphate residue).4 This brief survey reveals that a considerable gap exists between the molecular organisation provided by usual multidentate ligands bound to LnIII and the precise structural and electronic control required for programming lanthanide-containing functional devices.For more than two decades the lockand-key principle12 combined with preorganisation of rigid receptors has been extensively investigated for solving this problem but the recent consideration of more flexible systems has led to the (re)discovery of the induced fit principle associated with the predisposition of the receptor.13 From a thermodynamic point of view the preorganisation of the receptor according to the lock-and-key concept aims at minimising the entropy cost of the complexation process while maximising the enthalpy-favourable ligand–metal interactions and related work on the encapsulation of LnIII ions by rigid macrocyclic and macropolycyclic ligands is presented below.This approach is critically compared with the parallel development of less rigid receptors (i) macrocycles with grafted pendant arms and (ii) multidentate ligands and podands which tend to overcome the increased entropy cost of the assembly process through optimisation of convergent non-covalent interactions according to the induced fit concept. 3 Structural thermodynamic and electronic control in monometallic lanthanide complexes 3.1 The lock-and-key principle The principle has deep roots in biological processes in which molecules can only be chemically active if they are attached onto a receptor. According to Emil Fischer (1894) the fixation must be selective which implies a strict steric match between the receptor and the host hence the “lock-and-key” image to illustrate the requirements of molecular recognition.These notions (fixation molecular recognition) together with the concept of co-ordination brought forward by Werner at the turn of the 20th century are the basis of supramolecular chemistry.14 The simplest example of molecular recognition is the selective complexation of a spherical entity for instance a spherical metal ion (NaI CaII AlIII CuI LnIII etc.) and to meet this challenge several types of monocyclic (coronands) and polycyclic (cryptands) receptors have been developed. The receptor contains the necessary chemical information stored in its structure and the relevant properties for the recognition process are its form its size and the number and nature of the anchor sites (donor atoms).The characteristics of the latter include their position in the receptor their electronic properties (charge density polarisability ability to engage in van der Waals interactions) and their chemical reactivity upon coupling with the host (acid–base or redox processes). Following the initial work on alkaline and alkaline earth cations crown ethers were chosen to test size-discriminating effects along the LnIII series (Scheme 1).15 In propylene carbonate the maximum stability of the coronates is gradually shifted from 18-crown-6 for the lighter LnIII ions to 15-crown-5 Chem. Soc. Rev. 1999 28 347–358 349 Scheme 1 for the heavier ones. The entropy-driven macrocyclic effect (the difference in stability between the coronate and the complex with the open-chain analogue) is usually at a maximum for the best adjusted macrocycle.The discriminating effect is nevertheless small with DlogK = 1–2 between two consecutive macrocycles as shown in Fig. 2 mainly because the cavity size cannot be tuned finely enough.16 Adding one -(CH2)2-O group enlarges the mean cavity diameter by more than 0.5–0.7 Å,14 a difference 2–3 times as large as the contraction of the ionic radius along the entire LnIII series! Moreover the macrocycles are not completely rigid and they adapt their conformation when the ionic radius decreases. Combined with the electrostatic trend this leads to almost identical stability constants for the LuIII coronates with 15-crown-5 and 18-crown-6 ethers.If the macrocycle is too small sandwich 1+2 complexes form and differentiation between the LnIII ions is now determined by steric repulsion of the ligands when the Ln–O distance becomes shorter complexes with lighter LnIII ions are more stable but again DlogKLa–Lu is only of the order of 1–2. Better discrimination can be achieved by rigidifying the ligand upon fusion with benzyl rings (DlogKLa–Lu = 2.6 for dibenzo- 18-crown-6) at the cost of the overall stability which drops by almost 4 orders of magnitude for LaIII. On the other hand adding more flexibility into the ring by replacing two ether functions in 18-crown-6 with two amine groups results in an increased stability of over 8 orders of magnitude while any size-discriminating effect is lost.Moving to macrobicyclic receptors such as cryptands does not improve the situation since the addition of a -(CH2)2-O group in one of the side chains also enlarges the cavity too much to get a fine tuning adapted to the small size difference between two consecutive LnIII ions. As a matter of fact there is almost no gain in stability over the diazapolyoxocycloalkanes and the stability remains almost constant along the LnIII series or has the tendency to be governed by electrostatic effects depending on Chem. Soc. Rev. 1999 28 347–358 350 Fig. 2 Macrocyclic and size-discriminating effect in LnIII 1+1 complexes with crown ethers (-) and with the corresponding open-chain analogues (5) (adapted from ref.16). the solvent. In addition small particles such as water fluoride or even an oxygen atom from a perchlorate group can slide between the arms and interact with the metal ion. However interesting bicyclic receptors have been designed which provide high stability and good energy transfer for the sensitization of the EuIII and TbIII ions. One example is the tris(bipyridine) cryptand (bpy.bpy.bpy) used in homogeneous immunoassays.17 In parallel to this work and to circumvent some of the limitations encountered with rigid monocyclic and bicyclic receptors ionisable macrocycles have been developed bearing pendant arms fitted most frequently with carboxylic or phosphorous-containing acid functions. The presence of the latter leads to the formation of strong complexes owing to a hard ion–ion interaction between LnIII and the carboxylate anions.The ligands are intermediate between preorganised and predisposed entities since the complexed pendant arms usually have a different conformation than in the free ligand and since they can adapt to the varying size of the LnIII ions. Several platforms have been used to graft these arms crown ethers diazapolyoxocycloalkanes cyclen and other polyazacycloalkanes and more recently calixarenes. Although significant sizeselective effects are usually not observed within the LnIII series these ionisable macrocycles proved to be good chelating agents for the selective complexation of lanthanides over alkali and alkaline earth cations.18 A general trend is that the stability can be slightly tuned by adjusting the ring size and the number of donor atoms of the platform as well as the number of arms to form a suitable cage structure for the LnIII ion.Among all the ligands tested DOTA (1,4,7,10-tetrakiscarboxymethyl- 1,4,7,10-tetraazacyclododecane) proved to be one of the best sequestering agent for LnIII ions with logK in the range 22–26 and non-cyclic ligand behaviour i.e. the stability of the complexes increases with the increase in charge density. Thanks to its high stability the complex [Gd(DOTA)]2 has been developed as an efficient contrast agent for magnetic resonance imaging. 3.2 The induced fit principle The ever-increasing demand for improved selectivity in the complexation of lanthanide metal ions has led to the consideration of the induced fit principle which uses flexible receptors in order to optimise the interactions with the metal ions which act as template agents.The increased entropic cost of the complexation process compared with preorganised systems is overcome by the programming of stabilising intramolecular non-covalent interstrand interactions which can be modulated by judicious design of the receptor. The coding of suitable structural and electronic information in a receptor which will be expressed in the final complex and optimised for one particular metal ion is not an easy task especially for 4f ions.19 To the best of our knowledge Grenthe described the first lanthanide complexes [Ln(L1-2H)3]32 exhibiting secondary intramolecular interactions affecting the thermodynamic trend along the lanthanide series according to the induced fit concept (Scheme 2).20 The cumulative formation constants log(b3) for [Ln(L1- 2H)3]32 in water display the classical electrostatic trend for large LnIII (Ln = La–Tb) reach a plateau around Dy–Er which corresponds to a weak peak of selectivity for these ions and then decrease for the smaller LnIII (Ln = Tm–Lu; Fig.3). In the crystal structures of [Ln(L1-2H)3]32 the de-protonated dipicolinate ligands are wrapped around LnIII to give triple-helical complexes in which the metal ion is nine co-ordinate in a pseudo-tricapped trigonal prismatic site. This D3-symmetrical structure is maintained in solution along the complete lanthanide series and the decrease of log(b3) for the heavy LnIII (Ln = Tm–Lu) cannot be attributed to different co-ordination numbers 351 Scheme 2 in the final complexes.Steric constraints were invoked in the original paper,20 but a careful consideration of the molecular structure of [Ln(L1-2H)3]32 shows that the building of the metal ion cavity brings the negatively charged carboxylate side arms of one ligand very close to those belonging to the neighbouring strands (Fig. 3a). For large LnIII ions the contact distance remains sufficient to produce minor electrostatic repulsion (minimum average O···O distance = 3.46 Å for [La(L1- 2H)3]32) but the contraction of the LnIII radius induces an increased repulsion between the carboxylate groups (average O···O distance = 3.26 Å for [Lu(L1-2H)3]32) which eventually destabilises the complexes with the heavy LnIII ions.Although these secondary interactions remain limited and b3 decreases by a factor less than 10 between [Er(L1-2H)3]32 and [Lu(L1- 2H)3]32 these preliminary observations suggest that the induced fit principle can be applied for the selective recognition of LnIII ions. In order to substantiate this interpretation a series of neutral symmetrical tridentate ligands containing a central pyridine ring connected to variable side arms L2–5 has been investigated for the complexation of LnIII ions.21–23 The diester derivative L2 is structurally similar to L1 but it reacts with 4f ions to give thermodynamically unstable and kinetically labile cationic complexes [Ln(L2)3]3+ which exist as an intricate mixture of conformers in dynamic equilibria in solution.21 Formation constants in acetonitrile [log(b3)] are small compared to those measured for [Ln(L1-2H)3]32 in water (a more competing solvent!) and follow the typical electrostatic trend (Fig.4). A related behaviour is observed for 2,2A+6A,2B-terpyridine (L5) which provides unstable and labile triple-helical complexes [Ln(L5)3]3+ in acetonitrile,22 but no reliable stability constants have been determined for these complexes. Improving the affinity of the neutral side arms for LnIII is the key feature for preparing stable cationic nine co-ordinate triple-helical com- Chem. Soc. Rev. 1999 28 347–358 Fig. 3 a) Formation and structure of the triple-helical complexes [Ln(L1- 2H)3]32 and b) cumulative formation constants log(b3) for [Ln(L1-2H)3]32 in water (I = 0.5 M 298 K) given versus the inverse of the ionic radii of nine-coordinate LnIII.Fig. 4 Cumulative formation constants log(b3) of [Ln(L2)3]3+ (5) [Ln(L3)3]3+ (:) and [Ln(L4)3]3+ (-) in acetonitrile (298 K) given versus the inverse of the ionic radii of nine-coordinate LnIII. plexes with defined structural properties. Three parallel approaches have been explored. Firstly Thummel and co-workers have introduced rigid ethylene bridges in L6 which force the three connected pyridine units to adopt a cisoid conformation preorganised for their meridional trico-ordination to LnIII in the triple-helical complexes [Ln(L6)3]3+.22 NMR data point to a remarkable increase in stability and an improved resistance toward water hydrolysis but no size-discriminating effects for specific lanthanide metal ions have been reported.Secondly Piguet and co-workers have increased the electronic density on the co-ordinating oxygen atom of the side arms by replacing ester functions with carboxamide groups in L3. The affinity of the side arms for the hard lanthanide metal ions is significantly improved leading to the formation of inert D3-symmetrical complexes [Ln(L3)3]3+ which are stabilised by a factor of 3]3+.21 A detailed study of their 105–106 compared to [Ln(L2) solution structure demonstrates the existence of two different isostructural series controlled by secondary steric constraints between the terminal diethylamino groups which are brought Chem.Soc. Rev. 1999 28 347–358 352 close together by the wrapping of the strands about LnIII. For the large lanthanide metal ions (Ln = Ce–Tb) the central pyridine is weakly bound to LnIII leading to dynamically averaged D3hsymmetrical complexes [Ln(L3)3]3+ in solution. For the small lanthanides (Ln = Er–Yb) a compact rigid and inert D3- symmetrical triple-helical structure is obtained. This structural segregation between large and small LnIII is encouraging but the incriminated constraints are too weak to affect significantly the thermodynamic data which display the typical electrostatic trend within experimental error for the complete lanthanide series (Fig. 4). Thirdly the extension of the aromatic planes on going from pyridine in L5 to benzimidazole side arms in L4 produces triple-helical complexes [Ln(L4)3]3+ in which the benzimidazole rings of neighbouring strands are closely packed along the C3 axis.Strong intramolecular interstrand p-stacking interactions are evidenced with large LnIII leading to an optimal stabilisation around GdIII. A further contraction of the cavity with smaller LnIII results in a strong destabilisation of the triplehelical structure because of the repulsive van der Waals interactions occurring between interpenetrating electronic shells.23 Thermodynamic studies in acetonitrile support this interpretation and show that the third cumulative formation constants [log(b3)] of [Ln(L4)3]3+ decrease by a factor of 103 between GdIII and LuIII when the relative contraction of the ionic radii amounts only to 7%! (Fig.4). Since the origin of the peak selectivity is firmly established specific structural variations of the receptors give predictable modulations in the final complexes. The attachment of bulky groups to the benzimidazole side arms in L4 (X = 3,5-dimethoxybenzyl) prevents a regular wrapping of the three strands thus removing the close packing between benzimidazole rings and leading to a complete loss of selectivity and an impressive decrease in stability.23 On the other hand the introduction of the strong 4-(diethylamino)- phenyl donor group in the 4-position of the pyridine ring in L7 strengthens the N(pyridine)–LnIII bond and shifts the wrapped strands in the final triple-helical complexes [Ln(L7)3]3+ leading to a new peak of selectivity centred around TbIII.23 This fine and predictable tuning of structural and thermodynamic properties in monometallic complexes justifies the efforts focused on the development of predisposed ligands for programming organised lanthanide-containing architectures but the simultaneous implementation of predetermined functions (e.g.light-conversion energy transfer paramagnetic relaxation) requires better control and adjustment of electronic properties. In order to improve the structural and electronic properties in the final complexes two different side arms have been connected to the central pyridine units leading to unsymmetrical tridentate binding units.However the coordination of three unsymmetrical strands to a spherical LnIII provides a mixture of two isomers depending on the relative orientations of the bound chelating units. According to a statistical distribution we expect the formation of 25% of the facial isomer (HHH)-[LnL3] (C3-symmetry) and 75% of the meridional isomer (HHT)-[LnL3] (C1-symmetry) but only the facial isomer possesses the structural and electronic characteristics compatible with the development of advanced functional devices (high symmetry and vectorization). The quantitative preparation of the desired facial complex may be achieved through the connection of three unsymmetrical strands to a semi-flexible covalent tripod. A very elegant approach has been described by Orvig and co-workers who have connected hydrophilic sulfonated bidentate aminophenolate strands to various constrained tripods in L8–10.24 An intramolecular network of non-covalent N–H···O hydrogen bonds predisposes the deprotonated podand [L8-3H]32 for its complexation to LnIII.Stable 1+1 neutral podates [Ln(L8-3H)(H2O)6] are readily formed in which the ligand acts as a tridentate donor toward LnIII via the phenolate groups (Fig. 5a). A second equivalent of podand [L8-3H]32 displaces five water molecules to give the 1+2 podates [Ln(L8-3H)2(H2O)]32 whose surprisingly large Fig. 5 a) Formation of the lanthanide podate [Ln(L8-3H)2(H2O)]32 controlled by intramolecular hydrogen bonds.24 b) Encapsulation of LnIII by the constrained hexadentate podates L9–10 24 and c) encapsulation of LnIII by a predisposed nonadentate podand [L12+H]+.25 stability [log(K2) > log(K1)] results from the considerable contribution of the second step (K2) to the translational entropy.Moreover the observed electrostatic trend for the cumulative formation constants of [Ln(L8-3H)2(H2O)]32 [log(b2) = log(K1) + log(K2)] shows an unprecedented selectivity for the small LnIII ions [Dlog(b2) = log(b Lu 2 ) 2log(b La 2 ) = 5.5] whose origin is enthalpic and probably associated with the tightening of the intramolecular hydrogen bond network. The replacement of terminal hydrophilic phenolate rings in L8 with lipophilic aromatic groups in L11 produces similar 1+2 complexes [Ln(L11)2]3+ in which the metal is six co-ordinate by three negatively charged phosphinato donor groups of each podand.The cumulative formation constants log(b2) are significantly reduced because of the weaker affinity of the binding groups for LnIII but the unusual order of the successive stability constants log(K2) > log(K1) is reinforced because of the specific formation of a stabilising equatorial hydrophobic belt made of six closely packed phenyl units in the 1:2 podates [Ln(L11)2]3+.24 In both cases secondary intramolecular noncovalent interactions (hydrogen bonds in [Ln(L8-3H)2(H2O)]32 and dispersion forces in [Ln(L11)2]3+) control the structural and thermodynamic properties of the final complexes. A contraction of the covalent tripod in L9–10 hinders the intramolecular hydrogen bond network and reaction with LnIII produces classical encapsulated complexes [Ln(Li-6H)(H2O)3]32 (i = 9 10) in which the chelating units are bidentate; the co-ordination sphere being completed by three water molecules (Fig.5b).24 The logical extension toward tridentate chelating units connected to a covalent tripod to give nonadentate podands is strongly limited by the severe structural requirements resulting from the helical wrapping of three bent tridentate chelating units around a spherical metal ion. The partial co-ordination of potentially nonadentate podands to LnIII and/or the formation of interpenetrated dimers are generally observed except for the protonated podand [L12+H]+ which produces regular nine coordinate C3-symmetrical podates [Ln(L12+H)]4+ in the solid state and in solution (Fig.5c).25 Prior to complexation [L12+H]+ exists as a mixture of two conformers which are stabilised by an intramolecular trifurcated (respectively bifurcated) N–H···(ONC)n hydrogen bond involving the protonated apical nitrogen atom of the tripod and the proximal carbonyl groups. The resulting preorganisation of the receptor overcomes the large electrostatic repulsion between [L12+H]+ and Ln3+ leading to stable and rigid podates [Ln(L12+H)]4+. Compared with the non-clipped triple-helical analogues [Ln(L3)3]3+ the podates are stabilised by entropic effects and the protection of the metal ion is improved; two crucial points for the development of luminescent lanthanide-containing devices.However 353 Chem. Soc. Rev. 1999 28 347–358 [L12+H]+ and L3 do not exhibit significant size-discriminating effects along the lanthanide series and further secondary interactions must be encoded to improve structural control in nine co-ordinate lanthanide podates if such an effect has to be implemented. An alternative approach uses facial octahedral d-block complexes as non-covalent helical tripods for controlling the facial co-ordination of three unsymmetrical tridentate binding units about LnIII.26 The segmental ligand L13 has been designed for this purpose and possesses a bidentate domain analogous to 2,2A-bipyridine coded for the recognition of octahedral d-block ions connected to an unsymmetrical tridentate binding unit coded for f-block ions.A complete characterisation of the energy hypersurface of the thermodynamic assembly process between L13 MII and LnIII (M = Fe Co Zn; Ln = La–Lu) predicts the quantitative formation of the non-covalent podates [LnM(L13)3]5+ for stoichiometric conditions and a total ligand concentration larger than 5 3 1025 M. This is indeed observed and the structures of the final self-assembled complexes [LnM(L13)3]5+ have been characterised both in solution and in the solid state. The d-block ions are facially pseudo-octahedrally co-ordinated by the three bidentate binding units thus forming a non-covalent tripod which arranges the tridentate binding units for their facial complexation to nine-coordinate LnIII ions (Fig. 6).26 The spectroscopically inactive ZnII ion in Fig.6 Self-assembly of the non-covalent podates [LnM(L13)3]5+. The structure on the right corresponds to the X-ray crystal structure of [EuZn(L13)3]5+. [LnZn(L13)3]5+ acts as a structural organiser and has no effect on the electronic and spectroscopic properties associated with LnIII. In [EuZn(L13)3]5+ a strong red luminescence is detected upon irradiation with UV light which can be optimised by minor structural changes in [EuZn(L14-H)3]2+. The latter complex is strongly emissive and stable in water which opens up new perspectives for the development of self-assembled luminescent stains in aqueous media. The introduction of d-block ions with predictable electronic and kinetic properties into the non-covalent tripod is the starting point for the preparation of lanthanide-containing molecular devices with predetermined properties.CoII and FeII give structurally similar non-covalent podates [LnCo(L13)3]5+ and [LnFe(L13)3]5+ which display different functions. The lightyellow complexes [LnCo(L13)3]5+ act as templates for the selective preparation of kinetically inert CoIII podates [LnCo(L13)3]6+ which can be decomplexed to give the chiral inert preorganised nonadentate receptor FAC-[Co(L13)3]3+. The dark purple complexes [LnFe(L13)3]5+ work as tunable thermal switches which simultaneously change optical (dark purple Ô orange) and magnetic (diamagnetic Ô paramagnetic) properties. 26 These self-assembled complexes can be described as lanthanide-containing anchors whose structures solution behaviour and electronic properties can be modulated by the choice of the non-covalent tripod a first step toward the development of vectorial heterodimetallic f–f assemblies.Chem. Soc. Rev. 1999 28 347–358 354 4 Toward functional polymetallic f–f supramolecular complexes Considering the difficulties inherent in the preparation of organised monometallic lanthanide complexes it is not surprising that only limited research work has been focused on the preparation of polymetallic lanthanide-containing molecular or supramolecular complexes. Homodimetallic f–f complexes based on symmetrical bis-compartmental ligands represent simple cases which have been tested for improving the efficiency of paramagnetic MRI contrast agents27 and luminescent probes,7,8 but the new properties associated with the weak intramolecular intermetallic f–f interactions occurring in these complexes introduce essentially limiting factors for their use as functional devices (i) relaxivity of water molecules bound to GdIII is limited at high magnetic field by the increased electronspin relaxation induced by dipolar coupling,27 (ii) paramagnetism is modified by weak antiferromagnteic coupling28,29 and (iii) emission properties are affected by intermetallic energy transfers energy migration and/or emission/re-absorption processes.7,8 The related heterodimetallic f–f complexes have been much less investigated because their preparation requires the recognition of each metal ion by a specific site and we are aware of a single report describing the quantitative formation and complete characterisation of a pure heterodimetallic complex [(L15-3H)(Yb(C3H6O))(La(NO3)2)]+ in the solid state.28 In order to overcome this limitation statistical mixtures of complexes containing homo- and heteropairs have been synthesised and spectroscopically characterised but the interpretation of the data and their potential applications are obviously limited.8 On the contrary a large amount of heterodimetallic d–f complexes have been synthesised by taking advantage of the pronounced stereochemical preferences of soft 3d-block ions (FeII CoII NiII CuII and ZnII) which strongly contrast with those of hard LnIII ions.Compartmental ligands (macrocycles or podands) possessing different binding units adapted for the recognition of one particular type of metal ions have been systematically developed leading to isolated d–f heteropairs in the solid state and in solution.30 Magnetic interactions between d- and f-block metal ions is a current theme of interest but less attention has been focused on intramolecular dÔf energy transfers.30 Many efforts dedicated to the preparation of polymetallic lanthanide complexes involve macrocyclic receptors and the lock-and-key principle but recent advances in metallosupramolecular chemistry have led to the design of self-assembled lanthanide complexes according to the induced fit concept.Both aspects are briefly considered in the two following sections. 4.1 The lock-and-key principle In order to make use of the lock-and-key principle for embedding two identical 4f ions in a single receptor host molecules have to be designed which are comprised of two compartments each able to strongly bind one metal ion (Scheme 3).Basically two approaches can be used. The first relies on two macrocycles linked together. One of the few examples is the recent design of a calcium-sensitive magnetic resonance contrast agent based on a homodimetallic GdIII complex. A selective complexing agent for CaII 1,2-bis(o-aminophenoxy)- ethane-N,N,NA,NA-tetraacetic acid has been modified by grafting two GdIII receptors (Scheme 3). In the absence of calcium the iminodiacetates of the calcium-binding compartment interact with GdIII hindering water coordination onto the paramagnetic ion.In the presence of calcium the iminodiacetates bind this ion thereby allowing water to interact more efficiently with GdIII and the relaxivity of the polymetallic assembly increases by 80%.10 Another category of potential hosts would be molecules in which two macrocycles are held at a suitable Scheme 3 distance by two bridging units to form a cylindrical macrotricyclic receptor similar to the hosts used to selectively complex organic amines.14 Attempts to use this concept in our laboratories failed because oligomeric species were formed. To design heterodimetallic edifices similar host molecules could be engineered bearing macrocycles with different cavity sizes and/or pendant arms. To the best of our knowledge this synthetic way has not been attempted probably in view of the difficulty encountered for the fine tuning of each compartment to the requirements of a specific 4f ion and also in view of the work required for synthesizing such a non-symmetric receptor.The second approach makes use of a macrocycle with a large cavity and possessing two compartments each coded for the coordination of one 4f ion. Several homodimetallic 4f–4f edifices have been isolated mainly with macrocyclic Schiff bases30 (Scheme 3) and with calixarenes29 (Scheme 1). Macrocyclic Schiff bases are derived from the condensation of a substituted 2,6-diformylphenol with an appropriate diamine and LnIII ions often play the role of template agents. The homodimetallic [Gd2(NO3)4(La-2H)]·H2O complex displays a propeller conformation with pseudo D2 symmetry; the metal ions are ten coordinate being bound to four oxygen atoms from two bidentate nitrate ions to two oxygen and two nitrogen atoms from the macrocycle and to the two phenolate functions acting as bridging groups.The metal–metal distance is short 3.97 Å but not unusual for this type of compound.30 Both homo- and heterodimetallic 4f–4f complexes have been isolated with Lb but the hetero compounds probably consist of a mixture of homo- and heterodimetallic complexes. The ligand displays some size-discriminating effect when equimolar quantities of LaIII and SmIII or DyIII are reacted the isolated compounds contain respectively 20% or 90% LaIII. Moreover directional intramolecular TbIII?EuIII energy transfer occurs in the heterodimetallic material while the sensitization of the metal ion luminescence by energy transfer from the triplet state of the ligand is reasonably good.30 The interest in the field seems now to turn on iminophenolate cryptates obtained by template synthesis in which the metal–metal distance is particularly short 3.45 Å.8 Calixarenes result from the condensation of substituted phenols and formaldehyde and they offer a rich chemistry with almost unlimited possibilities since functions or functional arms can be grafted both on their upper and lower (phenolic) rims.Calix[8]arenes (H8L) have a cavity large enough to accommodate two LnIII ions and in the presence of a sufficiently strong base (triethylamine) they react in DMF or DMSO with 4f ions to give neutral dimetallic [Ln2(H2L)(solv)5]·xsolv entities.In the case of p-tert-butylcalix[ 8]arene (b-H8L) and p-nitrocalix[8]arene (n-H8L) available crystal structures show the LnIII ions being eight coordinate two deprotonated phenol groups and one solvent molecule acting as bridging ligands. The overall symmetry is C2 with the symmetry axis perpendicular to the intermetallic Eu–Eu axis. A striking feature is the two-bladed propeller conformation adopted by the ligand upon coordination which requires a substantial rearrangement energy. Indeed the free ligand has a somewhat flat and “ondulated” conformation which is retained in the 1+1 complexes with CaII or LnIII ions. In the dimetallic complexes the ligand can be described as two calix[4]arenes with cone conformation and placed side by side in a “transoid” configuration.The large conformational change sustained by the ligand upon the transformation of the 1+1 complexes into the 1+2 complexes is reflected in the half-lives of this reaction step 5500 s while the reaction leading to the 1+1 complex occurs in the ms range. From this point of view it is clear that calix[8]arenes cannot not be viewed as strictly preorganised macrocycles but rather as flexible receptors able to adapt somewhat to the changing size of the host metal ions. A slight size-discriminating effect has been reported for b- H2L62 which has a preference for cations in the middle of the lanthanide series but this effect is too small to be used for selective complexation.The dimetallic complexes with calix- [8]arenes present interesting functionalities in that the energy transfer from the ligand onto the trivalent LnIII ions can be easily tuned by simple changes in the para substituent of the phenol groups. For instance Fig. 7 clearly points to b-H2L62 being a good sensitizer for TbIII (the detection limit in DMF upon luminescence monitoring is < 10210 M) but transferring very poorly to EuIII while n-H2L62 behaves as a good EuIII sensitizer but the energy of its 3pp* state is lower than the 355 Chem. Soc. Rev. 1999 28 347–358 Fig. 7 Luminescence spectra showing how the sensitisation of EuIII and TbIII ions can be tuned by changing the substituent in the para position of the phenol groups of calix[8]arenes.energy of the excited 5D4(TbIII) level preventing any ligand-tometal energy transfer. The close proximity of the two ions in [Ln2(H2L)(DMF)5] about 3.6–3.9 Å allows magnetic interaction to take place between the two ions. At T < 20 K the magnetic susceptibility of the GdIII dimetallic edifice with b- H2L62 deviates from the Curie law and the temperature dependence can be fitted with a simple model taking into account two weakly anti-ferromagnetically coupled S = 7/2 ions with J = 20.063 cm21.29 Given the extensive efforts made presently to graft arms with various functional groups on calixarenes both on the lower and upper rims there is no doubt that some of the resulting dimetallic edifices will be of considerable interest in several applied fields.4.2 The induced fit principle Until now the only pure and fully characterised heterodimetallic f–f complex [(L15-3H)(Yb(C3H6O))(La(NO3)2)]+ results form the incorporation of two different lanthanide metal ions into a flexible nonadentate podand according to the induced fit principle. YbIII is seven co-ordinate by the podand (four nitrogen atoms of the tripod and three oxygen atoms of the bridging phenolate groups) and occupies the upper part of the cavity of the podate while LaIII is co-ordinated by the three bridging phenolates and the terminal methoxy groups.28 However no reliable solution study has been reported and we cannot conclude that the final heterodimetallic podate is the thermodynamic product of a strict self-assembly process because kinetic factors and/or selective crystallisation of one particular species within a mixture of equilibrating complexes cannot be excluded.In order to preliminarily explore the mechanism of directional intramolecular Ln1 ? Ln2 energy transfers the symmetrical bis-tridentate ligands L16–18 have been synthesized.31 Selfassembly with LnIII provides homodimetallic triple-stranded helicates [Ln2(Li)3]6+ (i = 16,17) and [Ln2(L18-2H)3] in which the nine co-ordinate metal ions are facially co-ordinated by three wrapped tridentate chelating units and separated by 8.8–9.1 Å (Fig. 8). The electronic and thermodynamic characteristics can be tuned by a judicious choice of the terminal donor group of the tridentate binding units.[Eu2(L16)3]6+ is luminescent upon UV irradiation but energy back-transfer processes Chem. Soc. Rev. 1999 28 347–358 356 Scheme 4 Fig. 8 Schematic representation of the heterodimetallic triple-stranded helicate [TbEu(L16)3]6+ working as a directional light-converter. dis = involving ligand-centred excited states quench the emission of the analogous Tb-complex. This drawback is removed for [Ln2(L17)3]6+ and [Ln2(L18-2H)3] (Ln = Eu Tb) leading to efficient UV–vis light-converting devices in acetonitrile and water respectively.31 Under stoichiometric conditions the assembly processes are highly selective and the reaction of L16 (3 equiv.) with an equimolar mixture of Ln1(iii) and Ln2(iii) (Ln1 ­ Ln2 and Ln = La Eu Tb Lu) produces only three complexes two homodimetallic helicates [(Ln1)2(L16)3]6+ and [(Ln2)2(L16)3]6+ and the heterodimetallic analogue [(Ln1)(Ln2)(L16)3]6+.31 Systematic deviations from the statistical distribution {[(Ln1)2(L16)3]6+ (25%) [(Ln2)2(L16)3]6+ (25%) and [(Ln1)(Ln2)(L16)3]6+ (50%)} are observed when Ln1 and Ln2 have different sizes and at least one LnIII is smaller than GdIII.These results imply a destabilisation of the heterodimetallic complex produced by subtle interstrand interactions relevant to the induced fit concept which favours the formation of triple-stranded helicates possessing two identical metallic cavities. The speciation of the thermodynamic mixture for the EuIII/TbIII pair leads to a disproportionation constant K 0.94 [eqn.(3)] which can be compared to the expected statistical value of 0.25. 2 [TbEu(L16)3]6+ [| [(Eu)2(L16)3]6+ + [(Tb)2(L16)3]6+ (3) The extra destabilisation of the heterodimetallic complex amounts to 1.6 kJ mol21 compared to a complex possessing two independent metallic sites which demonstrates the sensitivity of flexible self-assembled architectures to minor variations in the size of the metal ions (REu NC=9 = 1.120 Å RTb NC=9 = 1.107 Å). Evidence for an efficient intramolecular Tb ? Eu energy transfer has been found (h = 76%) which provides a directional light-converting supramolecular device if pure heterodimetallic complexes can be prepared (Fig. 8). Similarly the podand L19 reacts with LnIII to give the side-by-side homodimetallic complexes [Ln2(L19-3H)2] in which each LnIII is eight-co-ordinate in a pseudo-square antiprism; one phenolate of each ligand strand bridges the two metal ions which are held 3.918 Å apart.32 This structure is maintained in solution and the use of statistical Tb/Eu mixtures exhibits an efficient intramolecular directional Tb?Eu energy transfer in the heterodimetallic complex [TbEu(L19-3H)2] but no speciation has been determined which prevents the evaluation of a possible thermodynamic control of the assembly process.A few extensions toward organised polymetallic lanthanide complexes have been reported for aesthetically appealing solid state networks using lanthanide as connectors between divergent bidentate ligands,33 but their electronic spectroscopic or magnetic properties have not been considered.Recently monometallic lanthanide building blocks have been tentatively introduced into liquid crystals displaying calamitic mesophases and macroscopic structural ordering.34 Binnemans and coworkers have demonstrated that the intermolecular interactions responsible for the emergence and stability of the mesophases depend on the size of the lanthanide metal ions encapsulated in the cavity of the lipophilic receptors. The resulting control of the fusion and isotropization processes suggests that non-covalent intermolecular interactions may also obey the induced fit principle leading to lanthanide-containing organised macroscopic materials with predetermined properties. 5 Summary and outlook This short overview compares two fundamental principles in metallosupramolecular chemistry the lock-and-key and the induced fit which are apparently antagonistic in their essences and definitions.As far as lanthanide co-ordination complexes are concerned the lock-and-key approach is difficult to apply in a strict sense since the high level of rigidity associated with a satisfying preorganisation of the receptor is not compatible with the ultra-fine tuning required for the recognition of LnIII ions and their subsequent selective incorporation into specific coordination sites. The selectivity or size-discriminating effect observed for rigid macrocyclic ligands remains modest along the lanthanide series in contrast with the remarkable selectivity established for spherical alkali and alkaline earth ions.More flexible macrocyclic receptors lead to improved stability and the associated easy control of the arrangement and orientation of the donor atoms in macrocycles is a considerable advantage for programming lanthanide complexes which is counter-balanced by the loss of selectivity. The induced fit principle explores the reverse approach in which the flexibility of acyclic ligands is used to build a cavity around the metal ions whose size and geometry is optimised by secondary interstrand interactions. The thermodynamic and structural properties of the resulting lanthanide complexes can be finely tuned but the relative orientation of unsymmetrical binding units around LnIII is limited.This represents a severe drawback for the simultaneous tuning of electronic and spectroscopic properties. In this context semi-rigid podands combine some of the advantages of the two approaches. The orientation and the geometric arrangement of the binding units co-ordinated to LnIII in the final complexes is ensured by the tripod (covalent or noncovalent) while the intrinsic flexibility of the dangling side arms allows the programming of non-covalent secondary interactions (p-staking hydrogen bonding steric repulsion). It is thus not so surprising that the only well characterised heterodimetallic f,fA complex is a podate,28 but only a few examples of well-defined lanthanide podates have been reported because of the difficulty to fulfil the large co-ordination number requirement of LnIII ions with flexible side arms.The development of nonadentate podands containing three tridentate binding units is a logical solution to this problem although structural restrictions induced in the tripod generally prevent the formation of organised and predictable assemblies. Nevertheless pioneer work in that field24–26 suggests that the association of a carefully designed tripod with specific dangling chelating units may overcome these limitations leading to vectorial lanthanide-containing anchors acting as fundamental building blocks for the design of functional polymetallic lanthanide-containing devices. 6 Acknowledgments This work is supported through grants from the Swiss National Science Foundation and the Werner Fondation.Chem. Soc. Rev. 1999 28 347–358 7 References 1 S. Cotton Lanthanides and Actinides McMillan Education London 1991. 2 J.-C. G. Bünzli in Rare Earths ed. R. Saez-Puche and P. Caro Editorial Complutense Madrid 1998 pp. 223–259. 3 C. Piguet and J.-C. G. Bünzli Chimia 1998 52 579. 4 A. Tsubouchi and T. C. Bruice J. Am. Chem. Soc. 1994 116 11614; K. G. Ragunathan and H.-J. Schneider Angew. Chem. Int. Ed. Engl. 1996 35 1219. 5 E. Campazzi E. Solari C. Floriani and R. Scopelliti Chem. Commun. 1998 2603. Chem. 1995 34 1756 ; G. R. Choppin Lanthanide Probes in Life 6 J. Huskens J. A. Peters H. van Bekkum and G. R. Choppin Inorg. Chemical and Earth Sciences ed. J.-C. G. Bünzli and G. R. Choppin Elsevier 1989.7 N. Sabbatini M. Guardigli and J.-M. Lehn Coord. Chem. Rev. 1993 123 201; J.-C. G. Bünzli Lanthanide Probes in Life Chemical and Earth Sciences ed. J.-C. G. Bünzli and G. R. Choppin Elsevier 1989. 8 J.-C. G. Bünzli P. Froidevaux and C. Piguet New J. Chem. 1995 19 661; F. Avecilla A. de Blas R. Bastida D. E. Fenton J. Mahia A. Macias C. Platas A. Rodriguez and T. Rodriguez-Blas Chem. 9 S. Aime M. Botta M. Fasano and E. Terreno Chem. Soc. Rev. 1998 Commun. 1999 125. 27 19. 10 W.-H. Li S. E. Fraser and T. J. Maede J. Am. Chem. Soc. 1999 121 1413. 11 A. Roigk R. Hettich and H.-J. Schneider Inorg. Chem. 1998 37 751; S. J. Oh Y.-S. Choi S. Hwangbo S. C. Bae J. K. Ku and J. W. Park Chem. Commun. 1998 2189. 12 F. W. Lichtenthaler Angew.Chem. Int. Ed. Engl. 1994 33 2364. 13 D. E. Koshland Angew. Chem. Int. Ed. Engl. 1994 33 2375; J. Rowan D. G. Hamilton P. A. Brady and J. K. M. Sanders J. Am. Chem. Soc. 1997 119 2578. 14 J.-M. Lehn Supramolecular Chemistry Concepts and Perspectives VCH Weinheim 1995. 15 J.-C. G. Bünzli in Handbook on the Physics and Chemistry of Rare Earths ed. K. A. Gschneidner Jr. and L. Eyrings Elsevier Science Amsterdam 1987 Vol. 9 Ch. 60. 16 J.-C. G. Bünzli and F. Pilloud Inorg. Chem. 1989 28 2638. 17 G. Mathis in Rare Earths ed. R. Saez-Puche and P. Caro Editorial Complutense Madrid 1998 pp. 285–297. 18 F. Arnaud-Neu Chem. Soc. Rev. 1994 235. 19 C. Piguet G. Bernardinelli and G. Hopfgartner Chem. Rev. 1997 97 2005. 20 I. Grenthe J. Am. Chem. Soc.1961 83 360; J. M. Harrowfield Y. Kim B. W. Skelton and A. H. White Aust. J. Chem. 1995 48 807. 21 F. Renaud C. Piguet G. Bernardinelli J.-C. G. Bünzli and G. Hopfgartner Chem. Eur. J. 1997 3 1660. 22 C. Mallet R. P. Thummel and C. Hery Inorg. Chim. Acta 1993 210 223. 23 S. Petoud J.-C. G. Bünzli F. Renaud C. Piguet K. J. Schenk and G. Hopfgartner Inorg. Chem. 1997 36 5750. 24 M. P. Lowe P. Caravan S. J. Rettig and C. Orvig Inorg. Chem. 1998 37 1637 . 25 F. Renaud C. Piguet G. Bernardinelli G. Hopfgartner and J.-C. G. Bünzli Chem. Commun. 1999 457. 26 S. Rigault C. Piguet G. Bernardinelli and G. Hopfgartner Angew. Chem. Int. Ed. Engl. 1998 37 169; C. Edder C. Piguet J.-C. G. Bünzli 27 E. Toth L. Helm A. E. Merbach R. Hedinger K. Hegetschweiler and and G. Hopfgartner J. Chem. Soc. Dalton Trans. 1997 4657. A. Janossy Inorg. Chem. 1998 37 4104. 28 J.-P. Costes F. Dahan A. Dupuis S. Lagrave and J.-P. Laurent Inorg. Chem. 1998 37 153. 357 29 J.-C. G. Bünzli F. Ihringer and F. Besançon in Calixarenes Molecules for Separation ACS Symposium Series ed. G. Lumetta A. Gopalan and R. D. Rogers American Chemical Society Washington DC in the press. 32 R. C. Howell K. V. N. Spence I. A. Kahwa and D. J. Williams J. Chem. Soc. Dalton Trans. 1998 2727. 33 D. M. L. Goodgame S. Menzer A. M. Smith and D. J. Williams Chem. Commun. 1997 339. 34 K. Binnemans R. Van Deun D. W. Bruce and Y. G. Galyametdinov Chem. Phys. Lett. 1999 300 509; H. Nozary C. Piguet P. Tissot G. Bernardinelli J.-C. G. Bünzli R. Deschenaux and D. Guillon J. Am. Chem. Soc. 1998 120 12274. 30 J.-P. Costes F. Dahan A. Dupuis and J.-P. Laurent Chem. Eur. J. 1998 4 1616; P. Guerriero S. Tamburini and P. A. Vigato Coord. Chem. Rev. 1995 139 17. 31 N. Martin J.-C. G. Bünzli V. McKee C. Piguet and G. Hopfgartner Inorg. Chem. 1998 37 577; M. Elhabiri R. Scopelliti J.-C. G. Bünzli and C. Piguet Chem. Commun. 1998 2347. Review 8/04240C Chem. Soc. Rev. 1999 28 347–358 358

ISSN:0306-0012    

DOI:10.1039/a804240c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Regio- and diastereoselective rearrangement of cyclopentane-1,3-diyl radical cations generated by electron transfer Waldemar Adam* and Thomas Heidenfelder Institute of Organic Chemistry University of Würzburg Am Hubland D-97074 Würzburg Germany. E-mail adam@chemie.uni-wuerzburg.de. Fax +49 9318884756. Internet www.organik.chemie.uni-wuerzburg.de Received 17th May 1999 Cyclopentane-1,3-diyl radical cations are readily available from azoalkanes or bicyclo[2.2.0]pentanes (housanes) by electron transfer. These short-lived intermediates may stabilize by numerous chemical processes predominantly by 1,2-migration to form the corresponding cyclopentene derivatives after back electron transfer. This review focuses on the regio- and diastereoselectivities in the rearrangement of the strained cyclopentane-1,3-diyl radical cations and explores the mechanism of this novel 1,2-migration.This highly selective transformation offers attractive opportunities for the convenient and efficient synthesis of complex polycyclic structures. 1 Introduction Radical cations are mechanistically versatile intermediates in numerous chemical transformations (Scheme 1) which include isomerization rearrangement cycloreversion fragmentation Diels–Alder cycloadditions nucleophilic capture and dimerization. 1 For the cyclopentane-1,3-diyl radical cations generated from azoalkanes or bicyclo[2.2.0]pentanes (housanes) 1,2-migration represents the dominant reaction path to yield the corresponding cyclopentene derivatives (path A) after back electron transfer (BET).2–5 Nevertheless on appropriate substitution other stabilization processes may be triggered to compete with the 1,2-shift.Thus urazol annelation significantly Waldemar Adam born in 1937 in the Ukraine was raised in Germany and received his education in the United States (BSc 1958 University of Illinois; PhD 1961 MIT with F. D. Greene). He started his academic career in 1961 at the University of Puerto Rico (Rio Riedras) where he was promoted to full Professor in 1970. In 1980 he was appointed to the Chair of Organic Chemistry at the University of Würzburg. He has received numerous prizes and is the (co)author of about 780 scientific publications. Waldemar Adam This journal is © The Royal Society of Chemistry 1999 alters the reactivity of the intermediary radical cations through the hetereoatom substitution.Instead of the 1,2-alkyl shift observed for the carbocyclic analogs these nitrogen-substituted radical cations prefer to deprotonate (path B).6 Furthermore such urazol-annelated 1,3-radical cations are efficiently trapped by methanol to afford hemiaminal products (path C).6 The cyclobutene-annelated cyclopentane-1,3-diyl radical cation is intramolecularly trapped by the juxtaposed double bond to afford bicycloheptadiene and norbornadiene derivatives (path D).7 In the case of norbornadiene annelation fragmentation to cyclopentadienes competes with the 1,2-shift (path E).8 The chemical reactivity of cyclopropane radical cations the parent structures for the more strained cyclopentane-1,3-diyl radical cations has been extensively explored in recent years.9–17 The rich array of chemical transformations for cyclopropane radical cations are particularly well manifested for the strained cyclopentane-1,3-diyl derivatives (Scheme 1) which are the subject of the present review.These short-lived intermediates are readily generated from azoalkanes and housanes by photochemical and chemical electron transfer of which the latter process is more convenient. Their existence is established by means of spectral methods most definitively through EPR spectroscopy and chemical trapping. Of the chemical transformations in Scheme 1 we shall focus on the rearrangement process (path A) since it is the most abundant reaction for which the regio- and diastereoselectivities are now well understood and which illustrate the most characteristic transformation of these unusual transient species.Thomas Heidenfelder born in 1970 in Germany commenced his chemistry studies in 1989 at the University of Würzburg and joined Professor Adam’s group in 1995 (Diplom 1995 Doctorate 1998). His doctoral work was concerned with the regioand diastereoselectivity of the rearrangement of cyclopentane-1,3-diyl radical cations. Thomas Heidenfelder 359 Chem. Soc. Rev. 1999 28 359–365 Scheme 1 2 Generation of radical cations A vide variety of chemical and physical methods which range from chemical electron transfer (CET) electrochemical oxidation g radiolysis electron-impact ionization and photoinduced electron transfer (PET) have been used for the generation of radical cations.For mechanistic studies in solution the desired radical cations are most conveniently produced chemically18 or photochemically.19 In the case of photosensitized electron transfer a set of suitable sensitizers has become available20 for the selective generation of radical cations through appropriate matching of the energies of the excited states to allow exothermic electron transfer between the electron-accepting sensitizer and electrondonating substrate. Alternatively chemical electron transfer with persistent and isolable radical cation salts may be used e.g. the well-known trisarylaminium salts.21,22 Such aminium salts serve as one-electron oxidants whose oxidation potential is conveniently tunable by the type degree and pattern of substitution of the three aryl rings.An extensive comparative study of the PET and CET processes has revealed that the CET mode is particularly advantageous for the oxidation of azoalkanes and housanes. Specifically the CET oxidations proceed catalytically in a clean manner to afford the rearranged olefins in high yields (Scheme 2) while in the PET mode side reactions with the reduced photooxidant lead to by-products and low mass balances. Even when the aminium salt possesses a lower oxidation potential than the azoalkanes or housanes complete conversion of the substrate may be achieved due to the driving force that originates from the irreversible exothermic rearrangement step after the endothermic electron transfer.The catalytic cycle is completed by BET from the amine Ar3N to the cyclopentene radical cation to form the cyclopentene product and regenerate the aminium salt Ar3N·+. For most of the applications presented herein the CET mode has been utilized to generate the radical cations except in the EPR-spectral studies. Since these must be conducted under matrix isolation at cryogenic temperatures neither CET nor PET is feasible and g radiolysis constitutes the most expedient method. Chem. Soc. Rev. 1999 28 359–365 360 3 Evidence for transient radical cations Besides EPR spectroscopy under matrix-isolation conditions23,24 and pulse-radiolysis25,26 studies also chemical trapping7,27 proved helpful in detecting and characterizing the transient radical cations derived from azoalkanes and housanes.Thereby valuable mechanistic insight has been gained into the chemical behavior of these short-lived reactive intermediates. Scheme 2 Note that the radical cations as drawn are structurally not to be construed to constitute closed cyclopropane-type radical cations; they are planar open 1,3-diyl radical cations with the unpaired electron located either on the H- or the R-bearing terminus and the two possible limiting structures are represented for convenience and economy by means of the dashed line. EPR spectroscopy under matrix isolation is the most direct and structurally informative method to detect radical cations.For example the cyclopentane-1,3-diyl radical cation 1·+ derived from the corresponding housane by g radiolysis at 80–90 K was shown to possess a puckered conformation.23 On warm-up to ca. 105 K the relatively labile radical cation 1·+ rearranged diastereoselectively to the olefin radical cation 2·+ by a 1,2-shift of the axial hydrogen. For the spectral characterization of the phenyl-substituted radical cations 3·+ and 4·+ EPR was inconclusive because of unresolved hyperfine structure. However time-resolved optical absorption spectroscopy allowed the detection and characterization of such transients generated from the corresponding housanes or azoalkanes by pulse radiolysis.25,26 Unfortunately for the cyclopentane-annelated case only the corresponding 1,2-radical cation species 3·+ was observed because the initially formed 1,3-radical cation was too short-lived to be detected at the time resolution of the pulse radiolysis (ca.1 ms). Nevertheless for the corresponding urazole-annelated derivative the initial 1,3-radical cation 4·+ was detected.25,26 Chemical trapping provides indirect evidence through product analysis for the detection and characterization of the cyclopentane-1,3-diyl radical cations in solution. While for cyclopentane-1,3-diyl diradicals trapping experiments with dioxygen or aminoxyls as scavengers serve as a valuable mechanistic tool,28–30 such trapping studies are not feasible for the corresponding 1,3-diyl radical cations because the rates are too slow.However numerous studies on intermolecular and intramolecular nucleophilic trapping of 1,2-radical cations through reactions with alcohol,31,32 amine,31 and nitrile31 functionalities are known. 4 Rearrangement of cyclopentane-1,3-diyl radical cations 4.1 Regioselectivity2–5 As already mentioned cyclopentane-1,3-diyl radical cations derived from diazabicyclo[2.2.1]heptene (DBH) derivatives or housanes exhibit a high propensity to rearrange by a 1,2-shift to the corresponding 1,2-radical cations which after back electron transfer yield substituted cyclopentenes. For an unsymmetrical substrate regioisomeric products are expected due to the relative stabilization of the cation and radical sites in the oxidized 1,3-diyl species.Indeed the one-electron oxidation of the unsymmetrical methyl-substituted housane 6a yielded exclusively 3-methylcyclopentene (8a) whereas its phenyl analog 6b gave mainly 1-phenylcyclopentene (7b) as rearrangement product (Scheme 3).4,24 Moreover the distinctly different product distributions in the PET reactions of the azoalkanes 5 and the corresponding housanes 6 provide strong evidence for the involvement of diazenyl radical cations in the denitrogenation of the oxidized azoalkanes 5. This was already documented by the matrix EPR studies in the radiolytic oxidation of housanes and azoalkanes. Thus like the diazenyl diradicals the corresponding radical cations 5·+ expel N2 with a concomitant 1,2-hydrogen shift through backside attack on the remaining C–N bond.24,33 Recent ab initio calculations furnished a detailed mechanistic trajectory for the 1,2-shift in the methyl-substituted cyclopentanediyl radical cation 6·+ (Fig.1).34 Thus the 1,3-radical cation 6·+ possesses a stable puckered conformation due to population of a bonding orbital by the unpaired electron. The puckered species requires ca. 3 kcal mol21 of activation to form the twisted conformer which lies in an energy well of ca. 2.6 kcal mol21 due to methyl stabilization of the localized positive Scheme 3 Note that in this Scheme the product distributions in the PET reactions of the azoalkanes 5a,b (the first two rows) and the housanes 6a,b (last two rows) are shown. Fig. 1 QCISD//MP2-631G* reaction coordinate for the 1,2-H shift of the bridgehead methyl-substituted housane (energies in kcal mol21); n stands for endo and x for exo.charge. Subsequently the twisted conformer needs about 3.4 kcal mol21 for the Wagner–Meerwein-type hydrogen shift to the methyl-substituted cation site. The migration to the unsubstituted radical site lies with ca. 10 kcal mol21 much higher. The reason for this substantial difference (ca. 6 kcal mol21) in energy barrier of these two modes of hydrogen migration is due to the fact that for the shift to the cationic center a favorable two-electron/two-orbital interaction applies while the shift to the radical site requires an unfavorable threeelectron/ two-orbital interaction. As a consequence 1,2-hydrogen migrations in cations are facile but in radicals are reluctant.35,36 The low energy barrier of ca.3 kcal mol21 for the hydrogen shift is corroborated by the EPR findings in that the 1,3-radical cation 6·+ does not persist even at 77 K.24 The alternative 1,2-shift through a puckered transition-state structure (not shown in Fig. 1) requires a prohibitive activation barrier of ca. 40 kcal mol21.34 361 Chem. Soc. Rev. 1999 28 359–365 To assess the relative stabilization of the cation and radical sites in the cyclopentane-1,3-diyl radical cations the readily accessible cyclopentane-annelated housanes 9 were prepared and the regioselectivity of the CET-induced 1,2-methyl rearrangement determined.2,3 The CET reactions that afford the olefinic products 10 and 11 are displayed in Scheme 4.A Scheme 4 complete reversal in the regioselectivity of the 1,2-shift was observed for the CH3 CH2OH CH2OCH3 CH2F and C6H4-p- Me derivatives the regioisomer 10 is preferred (migration to the X-substituted terminus) but for the CH2CN CHO COCH3 C6H4-p-Cl C6H4-p-CO2CH3 and C6H4-p-CN cases the regioisomer 11 is favored (migration to the Ph-substituted terminus). 2,3 These regioselectivities reflect a profound electronic effect of the X substituent at the migration terminus in the intermediary cyclopentane-1,3-diyl radical cations. The relevant orbital interaction for the 1,2-methyl shift engages the LUMO(s*) of the cyclopentane-1,3-diyl radical cations 9·+ and the HOMO(s) of the migrating C–Me s bond. The required relative LUMO(s*) energies may be assembled in a qualitative manner through the interaction of the orbitals for the fragments R–CMe2 and Ph–CMe2 in the 1,3-radical cation.This is shown in Fig. 2 for two extreme cases namely complete methyl migration to the X or to the Ph terminus. Fig. 2 Schematic orbital-interaction diagram of the radical fragments. The relative ordering of the orbital fragments is given by the corresponding eSOMO orbital energies which are readily accessible through AM1 calculations. For convenience the De quantity is defined for which positive values (De > 0) apply SOMO of when the eSOMO of the X-substituted fragment lies above the cumyl one (the phenyl substituent is taken as reference point) while negative values (De < 0) are observed when the e the X-substituted fragment lies below the cumyl one.As a consequence for De > 0 the LUMO in the radical cation 9·+ will be in energy more similar to the X-substituted fragment and also carry the larger coefficient at this site (Fig. 2). Hence the 1,2-shift will take place preferentially to the X site to yield the regioisomer 10. In contrast for De < 0 the LUMO of the Chem. Soc. Rev. 1999 28 359–365 362 1,3-radical cation 9·+ lies closer in energy to the cumyl fragment and the phenyl terminus (Fig. 2) bears the larger coefficient such that the regioisomer 11 is preferred. Indeed a plot of the logarithm of the regioisomeric ratios [ln(10/11)] versus the semiempirical orbital energy differences (De) displays an excellent linear correlation (r2 = 0.989).This linear correlation substantiates that electronic and not steric effects of the X substituent control the observed regioselectivities in the radical cations 9·+.2,3 4.2 Stereochemical memory effect of the migrating group A remarkable stereochemical memory effect was disclosed in the PET chemistry of the stereolabeled syn- and anti-5-methylbicyclo[ 2.1.0]pentanes syn/anti-12 the anti stereoisomer furnished only 1-methylcyclopentene as the rearrangement product while the syn one afforded predominantly 3-methylcyclopentene (Scheme 5).24 On the basis of this diastereoselectivity combined with EPR spectroscopy24 and ab initio calculations,34 the mechanism in Scheme 5 was deduced. Scheme 5 Oxidation of the housanes anti,syn-12 affords the persistent puckered 1,3-radical cations anti,syn-12·+(p) as confirmed by EPR spectroscopy under matrix isolation.Breakage of the oneelectron bond generates the twisted conformers anti,syn- 12·+(t) which on Wagner–Meerwein rearrangement leads to the respective olefin radical cations 13·+ and 14·+ and final BET produces the corresponding cyclopentenes. The ring-flip in the syn-12·+(t) isomer competes with a 1,2-shift; presumably methyl migration leads to the thermodynamically less stable dialkylated olefin radical cation 13·+ whereas hydrogen migration for the anti-12·+(t) isomer results in the more stable trialkylated one 14·+. Also in the CET reaction of the deuterium-labeled housanes syn/anti-15 the initial syn/anti deuterium distribution was conserved quantitatively within the experimental error (ca.3%) in the rearrangement to the corresponding cyclopentene products 2-D and 3-D (Scheme 6).4 Scheme 6 These labeling results confirm the already mentioned stereochemical memory effect also for the more persistent disubstituted 1,3-diyl radical cations syn/anti-15·+. Evidently analogously to the twisted conformation of the radical cation syn/anti-11·+(t) the original syn substituent acquires a pseudoaxial orientation in almost perfect coplanar alignment with the 2p orbital at the bridgehead position while the pseudoequatorial substituent is located essentially parallel to the nodal plane of the 2p orbitals. Clearly migration of the pseudo-axial substituent is favored and this stereoelectronic control accounts for the diastereoselective migration in the CET-induced rearrangement of the deuterium-labeled housanes in Scheme 6.In contrast for the methyl-stereolabeled syn/anti-16 housanes (Scheme 7),4 only a 1,2-hydrogen shift had occurred for both the anti and the syn isomer. AM1 calculations reveal that the resulting tetraalkylated olefin radical cations (H migration) 17·+ and 18·+ possess ca. 12 kcal mol21 less energy than the corresponding trialkylated ones (Me migration). Presumably a common planar radical cation intermediate is involved in this rearrangement in which hydrogen migration results in the thermodynamically favored product. Scheme 7 4.3 Diastereoselectivity Cyclopentane annelation as in the bicyclo[3.3.0]octane skeleton provides an inherent stereochemical label to assess the diastereoselectivity of the rearrangement process in cyclopentane-1,3-diyl radical cations (Scheme 8).2–5 The product data show that the diastereoselectivity in the CET-induced Scheme 8 rearrangement of these cyclopentane-annelated housanes 19 depends on the type of substitution at the bridgehead positions.Thus also some 20(endo-Me) product is observed when the rearrangement terminus bears alkyl groups i.e. CH2OH CH2OMe CH2F or CH2CN whereas exclusively the exo product is obtained when this site carries an aryl substituent. The formation of both the exo-Me and endo-Me diastereomers 20 as olefinic products suggests a planar radical cation geometry in the rearrangement step.If a twisted conformation analogous to 12·+(t) were to be involved only the 20(exo-Me) diastereomer should have been formed. The preference for the exo-Me diastereomer of the regioisomer 20 results from the larger steric interaction with the annelated cyclopentane ring in the TS-A·+ transition state during the transposition of the endo-methyl group. Also steric effects are responsible for why aryl substitution at the rearrangement terminus suppresses endo-methyl product completely for the regioisomer 21 (Scheme 8) as displayed in the transition-state structure TS-B. Inspection of molecular models reveals that severe repulsive interactions between the aryl group and the annelated cyclopentane ring in the radical cation oblige the phenyl ring to align conformationally in a skewed orientation with respect to the planar cyclopentane-1,3-diyl ring.As a consequence the endo-methyl group is sterically blocked by the skew aryl substituent and exclusively the 21(exo-Me) diastereomer is produced (Scheme 8). Thus the diastereoselectivity of the 1,2-shift is controlled by steric factors in the intermediary planar 1,3-radical cations TS-A·+ derived from the cyclopentane-annelated housanes 19.2,3 4.4 Synthetic potential The high degree of stereoselectivity and the control of regioselectivity through appropriate bridgehead substitution in 363 Chem. Soc. Rev. 1999 28 359–365 the housane offer an opportunity to employ the rearrangement of cyclopentane-annelated cyclopentane-1,3-diyl radical cations (path A in Scheme 1) for the synthesis of complex ring systems e.g.the wide-spread diquinanes (Scheme 8).37 In particular the CET methodology with trisarylaminium salts constitutes an effective method for this purpose. On one hand back electron transfer to the cyclopentane-1,3-diyl radical cations is minimized and on the other hand this reaction may be run on a preparative scale.37 Moreover when the methylene bridge of the housane is spiro-substituted this oxidative rearrangement allows the preparation of polycyclic structures through stereocontrolled ring expansion. For instance the triquinane-related olefin 23 in Scheme 9 was obtained in excellent yield upon oxidation of the spiro-substituted tricyclooctane 22 with catalytic amounts of TBA·+ in which the quarternary center is perfectly diastereo- and regioselectively introduced.Surely this novel synthetic methodology provides an efficent and convenient access to unusual cyclopentanoid derivatives.37 Scheme 9 Scheme 10 Scheme 11 5 Perspectives Of the various transformations for the distonic cyclopentane- 1,3-diyl radical cations displayed in Scheme 1 the product studies establish that the rearrangement by a 1,2-shift (path A) is the prominent stabilization channel. The driving force Chem. Soc. Rev. 1999 28 359–365 364 originates from the energy-favored formation of the proximate cyclopentene-1,2-diyl radical cation (olefin radical cation) ab initio computations disclose a relatively low (ca. 3 kcal mol21) activation barrier for the 1,2-H migration.It is thus not surprising that low-temperature (ca. 80 K) matrix isolation is essential for direct EPR-spectral detection of these elusive species while unequivocal trapping is reserved for a few specialized cases e.g. intermolecularly by methanol for the nitrogen-substituted derivative (path C) or intramolecularly for the cyclobutene-annelated system (path D) in Scheme 1. To date very few other transformations are known to compete with the facile rearrangement process (path A) in Scheme 1 and then only under special circumstances. One of these is fragmentation (path E) which is detailed in Scheme 10 for the norbornene-annelated derivative; in fact cycloreversion into the two cyclopentadienes dominates over the diastereoselective 1,2-methyl migration.8,27 The mechanistic reasons for this reactivity preference are yet to be elucidated but presumably the rigidity of the annelated bicyclic ring encumbers the proper conformational alignment for the methyl shift (cf.the twisted conformation in Fig. 1) such that the annelating s bonds are cleaved preferentially. Of course considerable impetus for fragmentation derives from the formation of the diphenylsubstituted cyclopentadiene radical cation stabilized through conjugation. A dramatic example constitutes the deprotonation (path B Scheme 1) of the diaza-substituted 1,3-radical cation 24·+ which offers exclusively the bisolefin 25 (Scheme 11) without even traces of rearrangement product.6,25 It was shown that the monoradical 25(H)· intervenes derived from the 1,3-radical cation 24·+ on proton loss.Subsequently 25(H)· is oxidized by a second equivalent of Ar3N·+ to yield the corresponding cation 25(H)+ and a second deprotonation affords finally the bisolefin 25. Evidently the two nitrogen atoms in the urazole-bridged radical cation stabilize it by conjugation while the juxtaposed carbonyl group facilitates proton loss such that 1,2-methyl migration is completely suppressed.6,25 The profound differences in the chemical behavior of cyclopentane- norbornene- and urazole-annelated 1,3-radical cations illustrate that on appropriate substitution other stabilization processes may be triggered to compete with the 1,2-shift. Now that we understand quite well the regio- and diastereocontrolled rearrangement process (path A) a future goal of the research should be to uncover competitive chemical reactivity of cyclopentane-1,3-diyl radical cations and elucidate the structural and electronic features that steer it.For instance a worthwhile task would be the exploration of methylene-bridge substitution (the intervening carbon center between the radical sites) and how it dictates reactivity modes. Poorly migrating groups should enhance alternative reaction channels of the cyclopentane-1,3-diyl radical cations as preferred stabilization process (Scheme 1). Little if anything is known about this mechanistic query even for carbocations and the readily accessible housane-derived 1,3-radical cations offer challenging opportunities.6 Acknowledgement Our work in this area was generously supported by the Fonds der Chemischen Industrie and the Volkswagen-Stiftung. The former doctoral students Markus Dörr Herbert Walter Jürgen Sendelbach Coskun Sahin Thomas Kammel Fumio Kita and Lluis Blancafort (chronological order) deserve special praise and appreciation for their diligence imagination motivation and perseverance; their contributions are cited in the references. 7 References 1 For a recent review see M. Schmittel and A. Burghart Angew. Chem. Int. Ed. Engl. 1997 36 2551. 2 W. Adam V.-I. Handmann F. Kita and T. Heidenfelder J. Am. Chem. Soc. 1998 120 831. 3 W. Adam and T. Heidenfelder J. Am. Chem. Soc. 1998 120 11858. 4 W. Adam A.Corma M. A. Miranda M.-J. Sabater-Picot and C. Sahin J. Am. Chem. Soc. 1996 118 2380. 5 W. Adam and C. Sahin Tetrahedron Lett. 1994 35 9027. 6 W. Adam and T. Kammel J. Org. Chem. 1996 61 3172. 7 W. Adam T. Heidenfelder and C. Sahin J. Am. Chem. Soc. 1995 117 9693. 8 W. Adam and J. Sendelbach J. Org. Chem. 1993 58 5310. 9 G. Boche and H. M. Walborsky in Updates from the Chemistry of Functional Groups Cyclopropane-derived Reactive Intermediates ed. S. Patai and Z. Rappaport John Wiley and Sons Chichester 1990 ch. 5 pp. 207–236. 10 H. D. Roth Top. Curr. Chem. 1992 163 131. 11 T. Herbertz and H. D. Roth J. Am. Chem. Soc. 1998 120 11904. 12 J. P. Dinnocenzo and D. A. Conlon J. Am. Chem. Soc. 1988 110 2324. 13 P. G. Gassmann and B. A. Hay J.Am. Chem. Soc. 1985 107 4075. 14 T. Miyashi Y. Takahashi H. Ohaku H. Ikeda and S.-I. Morishima Pure Appl. Chem. 1991 63 223. 15 S. Nishida M. Murakami H. Oda T. Tsuji T. Mizuno M. Matsubara and N. Kikai J. Org. Chem. 1989 54 3859. 16 J. P. Dinnocenzo T. R. Simpson H. Zuilhof W. P. Todd and T. Heinrich J. Am. Chem. Soc. 1997 119 987. 17 P. Du D. A. Hrovat and W. T. Borden J. Am. Chem. Soc. 1988 110 3405. 18 N. G. Connelly and W. E. Geiger Chem. Rev. 1996 96 877. 19 G. J. Kavarnos Fundamentals of Photoinduced Electron Transfer VCH New York 1993. 20 G. J. Kavarnos Top. Curr. Chem. 1990 156 21. 21 W. Schmidt and E. Steckhan Chem. Ber. 1980 113 577. 22 L. Eberson and B. Olofsson Acta Chem. Scand. 1991 45 316. 23 W. Adam H. Walter G.-F. Chen and F.Williams J. Am. Chem. Soc. 1992 114 3007. 24 W. Adam C. Sahin J. Sendelbach H. Walter G.-F. Chen and F. Williams J. Am. Chem. Soc. 1994 116 2576. 25 W. Adam T. Kammel and S. Steenken Angew. Chem. Int. Ed. Engl. 1996 35 543. 26 W. Adam T. Kammel M. Toubartz and S. Steenken J. Am. Chem. Soc. 1997 119 10673. 27 W. Adam and J. Sendelbach J. Org. Chem. 1993 58 5316. 28 W. Adam and M. Dörr J. Am. Chem. Soc. 1987 109 1570. 29 W. Adam and S. E. Bottle Tetrahedron Lett. 1991 32 1405. 30 W. Adam S. E. Bottle R. Finzel T. Kammel E.-M. Peters K. Peters H. G. von Schnering and L. Walz J. Org. Chem. 1992 57 982. 31 P. J. Kropp in Organic Photochemistry ed. A. Padwa Marcel Dekker New York 1979 vol. 4 pp. 1–142. 32 S. Dai J. T. Wang and F. Williams J. Chem. Soc. Perkin Trans. 1 1989 1063. 33 W. Adam U. Denninger R. Finzel F. Kita H. Platsch H. Walter and G. Zang J. Am. Chem. Soc. 1992 114 5027. 34 W. Adam L. Blancafort and M. A. Robb J. Am. Chem. Soc. in press. 35 J. W. Wilz in Free Radicals ed. J. K. Kochi John Wiley and Sons New York 1973 vol. I ch. 8 pp. 333–501. 36 A. L. J. Beckwith and K. U. Ingold in Rearrangements in Ground and Excited States Academic Press New York 1980 vol. 1 ch. 4 pp. 161–310. 37 W. Adam T. Heidenfelder and C. Sahin Synthesis 1995 1163. Review 9/03931G 365 Chem. Soc. Rev. 1999 28 359–365

ISSN:0306-0012    

DOI:10.1039/a903931g

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Compounds containing a planar-tetracoordinate carbon atom as analogues of planar methane Walter Siebert and Anuradha Gunale Anorganisch-Chemisches Institut der Universität Heidelberg Im Neuenheimer Feld 270 D-69120 Heidelberg Germany. E-mail ci5@ix.urz.uni-heidelberg.de; Fax +49(0)6221/54-5609 Received 25th February 1999 Over the past years the number of examples of compounds containing a planar-tetracoordinate carbon atom has increased. However the presence of a carbon atom with a 360° sum of angles does not imply that the species is a derivative of planar methane; there must be an appropriate electronic stabilization. In the case of complexes 21a and 21b the central carbon atom is indeed stabilized by s-donors and pacceptors as required for planar methane.1 Introduction One and a quarter centuries ago van’t Hoff and LeBel1 convinced the chemical community of the tetrahedral arrangement of the four hydrogen atoms around carbon in methane. According to calculations2 planar CH4 (A X = XA = H) has two two-center two-electron (2c-2e) C–H bonds (X = H) one three-center two-electron (3c-2e) bond (HCH; XA = H) and a doubly occupied pp orbital. To obtain the planar species an enthalpy of 150 kcal mol21 needs to be overcome which well exceeds the bond dissociation energy of the C–H bond. Therefore achieving planar methane is unlikely. It should however be possible to stabilize a planar-tetracoordinate carbon (ptC) by incorporating substituents functioning as strong s-donors and p-acceptors which supply the lacking s electron density and remove the energetically unfavorable p-electron density.This idea was first introduced and backed up with extended Hückel calculations by Hoffmann et al.,2 who Walter Siebert studied chemistry at Philipps-Universit�at Marburg Germany where he obtained his MSc degree in 1964 and earned his PhD with Max Schmidt in 1965. Having completed his Habilitation in the Chemistry Department at Julius- Maximilians-Universit�at W�urzburg in 1971 he became Associate Professor at his Alma Mater Philipps-Universit�at in 1973. After rejecting a chair at the FU Berlin he moved to a full professorship at Ruprecht- Karls-Universit�at Heidelberg in 1980. His research interests focus on boron heterocycles as ligands in sandwich oligodecker and polydecker complexes on cluster chemistry and on compounds containing a planar-tetracoordinate carbon atom.Walter Siebert This journal is © The Royal Society of Chemistry 1999 suggested that silyl or boryl groups might stabilize a ptC. Schleyer Pople et al.3 extended this concept to a variety of electropositive substituents expected to be capable of stabilizing a ptC in A (X = XA = SiH3 BH2 Li). Their MO studies indicate that in 1,1-dilithiocyclopropane the lithium-substituted carbon atom favors a planar-tetracoordinate environment over a tetrahedral arrangement by 7 kcal mol21. Over the past twenty years there have been a number of compounds reported having planar-tetracoordinate carbon atoms and this area was reviewed recently by Erker et al.4 The majority of examples belong to the class of p-stabilized carbon atoms with ptC as part of the p-system of an arene as depicted in B and of an olefin as in C.In B the metal centers M and MA are forced into the plane of the arene by interactions with the ortho substituents R. There is one example (see 3) in which the metals M = MA are directly bonded to the ortho carbon atoms of the ptC in benzene and bridged by a ligand L. The ptC in the olefinic compounds C is connected to the metal centers through a 3c-2e MCMA bond. In addition M binds through a 2c-2e bond to the other carbon atom of the double bond and the metals are also bridged by a ligand L. Thus two metal centers with chelating ability seem to be required for stabilization of a ptC in B and C.It is apparent that compounds with a ptC incorporated in a p-system do not need the stabilization pattern predicted2 for compounds of type A (X = XA = SiH3 BH2) as sufficient pelectron density is delocalized onto the arene or olefin moiety. In other words B and C are not stabilized as type A because there is no lone pair present in B and C. Anuradha Gunale born in India (1968) completed her BA in chemistry at Northwestern University with Duward Shriver and her MSc degree at Indiana University with George Christou. In 1994 she began to work in the group of Walter Siebert at the University of Heidelberg Germany on compounds containing a planar-tetracoordinate carbon atom. She finished her PhD thesis in 1997.Anuradha Gunale 367 Chem. Soc. Rev. 1999 28 367–371 A second class of compounds with a ptC that should follow the structural and electronic features is schematically depicted in A D and E.5 The latter two represent the syn and anti isomers of a singlet carbene stabilized by two s-donor/pacceptor groups X = XA (e.g. BR2) and two transition metal complex fragments M MA. In D and E the metal centers could be symmetrically or asymmetrically bonded to the ptC. Asymmetric bonding would indicate that one center has better donor and weaker acceptor properties than the other resulting in an electronic push/pull effect. Another approach for the realization of a ptC is its incorporation into a hydrocarbon framework F in which solely the steric forces are responsible for a ptC conformation.As examples of such compounds fenestranes and more recently alkaplanes F have been investigated.6,7 Octaplane has been calculated to have an unusually low ionization energy (ca. 5 eV) and a ptC in the radical cation.7 The bonding in octaplane and its cation is similar to that in A in the neutral molecule the HOMO is the doubly occupied pp-like orbital and the sp2- hybridized ptC is involved in two 2c-2e C–C bonds and one 3c- 2e CCC bond. The pp orbital exhibits a slight deviation of the radial symmetry which indicates that it is not a lone pair with pure p character but is involved to a small extent in bonding with the neighboring C atoms. In contrast the singly occupied MO (SOMO) of the radical cation shows no deviation.2 Planar-tetracoordinated carbon incorporated in a p system Complexed arene Cotton et al.8 reported the crystal structure of 1 in which four 1,3-dimethoxy-substituted phenyl groups coordinate to the triply bonded V2 unit. Two of the phenyl rings are bonded through the ipso C atom and one MeO substituent to the V2 unit whereas the other two phenyl groups coordinate through both MeO groups. A 3c-2e VCV interaction of the ipso-C atom completes the planar tetracoordination first recognized by Keese et al.9 The related 1,3-dimethoxy-2-lithiobenzene 2 also seems to have an ipso ptC. However this compound proved to be a tetramer in the crystal,10 in which two dimeric units of 2 are Chem. Soc. Rev. 1999 28 367–371 368 located on top of each other and rotated with respect to each other by 90°.This indicates that the ipso-C atoms as well as the Li centers are actually pentacoordinated. The ipso-C atom in dimeric phenyllithium(tmeda) is tetrahedrally coordinated.11 However calculations by Schleyer et al. indicated a ptC for phenyllithium and cyclopropenyllithium.12 In the substituted benzene-1,3-biszirconium complex 3 reported by Buchwald et al.13 the two Cp2Zr moieties are bridged by a methyl group and both metals are in a bonding 3c- 2e interaction with the carbon atom in the 2-position which is a ptC stabilized by the aromatic p-system. Poumbga Bénard and Hyla-Kryspin14 studied compounds 1 and 3 by extended Hückel ab initio Hartree–Fock and CI calculations.It was pointed out that the ptC is part of the psystem and bears a minus charge (s4p1 configuration). Donation of s-electron density into empty d orbitals with metal–metal bonding character occurs. Here again the mode of stabilization is not that in A. In the naphthalene derivatives 4 and 5 the Ti–C9 distance in 4 [2.442(7) Å)]15 and the B–C9 distance in 5 [2.045(5) Å)]16 are shorter than the sum of the van-der-Waand it may seem that a ptC is present (especially in the case of 4). However on the basis of orbital symmetry there is no bonding interaction between Ti or B and the bridging carbon atom C9. Complexed allene and analogues Chisholm et al.17 reported the ditungsten allene complex 6 in which the central carbon atom of the allene has the coordination characteristics of a ptC.The bonding of the allene to the W2 unit is unique (see formula drawing of 6).17 In the V-shaped C3 ligand three 2p orbitals form bonding nonbonding and antibonding combinations and the central carbon atom has an empty p orbital orthogonal to the three 2p combinations. The allene functions as a 4e donor and one of the W2 p bonds interacts with the empty 2p orbital of the ptC. The complexes 7 obtained from W2(OR)6 and the allene analogues18 XNCNY (X = Y = N–R X = O Y = N–Ph) have been described in which the central atom is a planartetracoordinate carbon. Related to 6 and 7 are the complexes 8 and 9. In the allenyldipalladium complex 819 the Pd centers with formally d8 configuration are directly connected through a 2c-2e Pd–Pd bond.The ptC is bonded to both Pd centers [Pd–C 2.361(2), 2.431(3) Å] and the C3 unit is almost linear (173.2°). In 920 the ptC is nearly symmetrically bonded to the Pd atoms [Pd–C 2.280(7) 2.331(9) Å] and the CS2 ligand is in the plane defined by the Pd atoms and the bridging phosphorous atom. Complexed olefins and analogues The essential features of olefin complexes 10–1221 are that the ptC is part of the CNC double bond and that one of its sp2 hybrid orbitals is involved in a 3c-2e bond with both metal centers. The bonding in complexes 10 and 11 may be compared with the familiar electronic situation in B2H6 where two 3c-2e bonds hold the two BH3 molecules together. Here the metal centers are electronically connected through the two 3c-2e bonds M1CM2 and M1XM2.Complexes 10 and 12 are different in that only one 3c-2e bond is present in 12 (M1CAl) and the bridging chlorine atom is involved in two 2c-2e bonds. Erker et al.4,21 have reported more than 50 examples of ptC compounds of types 10–12 which all belong to class C. In these compounds there is no unfavorable p-lone pair as the carbon atom is incorporated (vide supra) into an olefinic p-bond. Therefore examples of class C cannot be stabilized according to the Hoffmann model2 for planar methane (A) and its derivatives (D E). In compounds of the type 10–12 the sp2 carbon atom is involved with one of its sp2 hybrid orbitals in a 3c-2e bonding to two metals (M1 M2). Gleiter Hyla-Kryspin et al.22 have studied complexes of type 11 by ab initio and extended Hückel methods and found that the stabilization of the ptCs in the d0 complexes depends on the presence of an in-plane acceptor orbital at M1 and that the delocalization of the p-electron density of the ptC does not play any role.The structural alternatives for the compounds 10–12 are the classical heterocyclic structures with the planar-tetracoordinate environment at carbon being represented as trigonal planar. Compounds of the types 10 and 11 have been studied with respect to their isomerization. It was found that their conventional structures 13 and 14 are > 30 kcal mol21 and ca. 12–14 kcal mol21 less stable than 10 and 11 respectively. 3 Derivatives of planar-tetracoordinate methane Although the all-boron substituted compounds C[B(OR)2]4 23 have been known for quite some time the methyl derivative (R = CH3) has only recently been structurally characterized.In the case of C(BCl2)4 the possibility of planar tetracoordination was considered from the beginning. Schleyer Pople et al.3a have carried out MO calculations on the structures and stabilities of the geometric isomers of tetraborylmethane 15 and on the spiro-cyclic compound 16 which indicate that the latter could be a candidate for a stabilized ptC; the tetrahedral environment is preferred by only 6 kcal mol21.3a We have tried to prepare compound 17 however with no success. An example of a compound with a ptC is the methane dication CH4 2+ which has been observed in the gas phase.24 According to a comparison of the measured and calculated ionization energies an “anti-van’t Hoff–LeBel” isomer of CH4 2+ is formed when an electron is removed from the tetrahedral methane monocation.25 High-level ab initio calculations by Wong and Radom26 show that the structure of CH4 2+ of lowest energy has C2v symmetry (and not D4h as for squareplanar methane with four H–C–H bonds of 90º) with two short and two long C–H bonds.26 This dication is therefore a complex between CH2 2+ and molecular hydrogen.There has only been a limited number of methane derivatives containing transition metals. C[AuP(cyclo-C6H11)3]4 from Schmidbaur et al.27 has been known for quite some time but it was not structurally characterized.Although this compound was assumed to have a tetrahedral geometry—a prediction that is substantiated by the presence of the bulky phosphane ligands which might prevent the angle of 109° between the Au centers from decreasing towards 90°—the strong Lewis basicity28 may indicate a planar structure in which in accordance with the Hoffmann model the filled p orbital perpendicular to the molecular plane is available for attack by an electrophile. In analogy to the methane dication oxidation of tetragoldmethane may lead to a metal-containing derivative of CH4 2+. The complex cations [(AuPPh3)4(m4-CR)]n+ (R = H n = 1;28 R = Me n = 1;29 R = S(O)Me2 n = 030) have been isolated and characterized. They have a distorted square-pyramidal structure in which the central carbon atom is substituted with four basal gold atoms and the apical substituent R and can therefore be viewed as donor–acceptor complexes5 between a nucleophile (H2 CH32) and [C(AuPPh3)4]2+.In compound 18 synthesized by Marks et al.,31 the presence of a ptC for the bridging methylene group has been proposed in the transition state of a dynamic process. This is supported by extended Hückel calculations and would imply that this is the first planar dimetallamethane derivative to be investigated but a reversible metal–CH2 bond cleavage is also a possible mechanism. A carbidotetrarhenium cluster32 could be regarded 369 Chem. Soc. Rev. 1999 28 367–371 as an example of a tetrametallamethane. However in the complex anion [{I(OC)3Re}C{Re(CO)4}3]2 (19) the carbon atom is at the center of a tetrahedrally distorted square of rhenium atoms.The folding in the Re4 ring of 42° is explained by repulsive interactions between ligands on adjacent metals. We have investigated the complexation chemistry of the nonclassic boriranylideneboranes 20.33–35 Upon treatment with two equivalents of [Co(C5H5)(C2H4)2] 20a b react with cleavage of the ring C–C bond and migration of the aryl substituent R from one boron atom to the other to provide 21a and 21b. The boriranylideneboranes have been transformed into chain structures which can be considered as the complexstabilized diborylcarbenes 21a b. Crystals of 21a that were suitable for an X-ray structure analysis were obtained. Microcrystals of the mesityl derivative 21b could also be grown and studied with an image plate (IPDS Stoe) although they were poor in quality.The refinement of the weak-intensity data yielded a structure similar to that of 21a but did not allow the discussion of any structural details. The crystal structure of 21a shows a planar tetracoordinated carbon atom (sum of angles 359.9°) coordinated by four electropositive centers and with short B–C bond lengths (Fig. 1). However this Fig. 1 Ortep representation of the central framework of 21a. Selected bond lengths and angles C4–B3 1.483(10) B3–C2 1.474(9) C2–B1 1.522(10) B1–C5 1.601(10) Co1–C5 2.180(7) Co1–B1 2.276(6) Co1–C2 2.009(6) Co1–B3 2.081(8) Co2–C4 2.114(6) Co2–B3 1.973(7) Co2–C2 1.887(6); B1–C2–B3 150.3(6) B1–C2–Co1 78.9(4) B1–C2–Co2 139.0(5) B3–C2– Co1 71.5(4) B3–C2–Co2 70.7(4) C2–B3–C4 138.0(6); (see also ref.35). coordination geometry is not sufficient for a compound to be a derivative of planar methane. The bonding situation must also be in agreement with the description presented by Hoffmann et al.2 Hyla-Kryspin Gleiter et al.5 have studied the electronic stabilization of this unusual configuration by extended Hückel and ab initio SCF calculations. In the simplified system in which all substituents were replaced by hydrogen atoms the complex 21c was built up from the cobalt dimer fragment Chem. Soc. Rev. 1999 28 367–371 370 [Co(C5H5)]2 and the bridging diborylcarbene ligand H2CNB– C–BH2 (22c). It was determined that the electronic structure of 22c is similar to that of planar methane.The HOMO is mainly localized on the ptC and is almost purely 2pp in character. The low-lying LUMO is an out-of-phase combination of the inplane 2p orbitals and it has the correct symmetry to accept selectron density from the CoCp dimer fragment. This indicates that the features of the electronic situation for the free diborylcarbene 22c are similar to those of planar methane. However the energetics are already a bit better according to a natural population analysis the electron density at the ptC in free diborylcarbene 22c is s2.978p1.501 as compared to s2p2 in planar methane.2 In other words with respect to planar methane the energetically unfavorable p-electron density has been decreased and there is additional electron density for bonding in the s plane.Upon complexation an additional shift in electron density takes place and the final electronic configuration at the ptC in 21c is s3.944p1.356. The in-plane s-electron density increases from 2.978 to 3.944e and the out-of-plane p-electron density decreases from 1.501 to 1.356e. Thus in 21c the overall stabilization manner of the ptC corresponds to that predicted by Hoffman et al.2 as required for realizing this unusual geometry 21c is a derivative of planar methane. Since the natural charge on the ptC increases from 20.482 in 22c to 21.310 in 21c a net shift of electron density takes place from the Co(C5H5) units to the bridging diborylcarbene ligand upon complexation. In total the two cobalt atoms are donors with respect to the ptC but their in-plane and out-of-plane interactions have different character.According to an analysis of the occupancies of the in-plane and out-of-plane natural atomic orbitals Co1 is a strong s-donor (decrease from 5.007 to 4.157 upon complexation) and a strong p-acceptor (increase from 2.968 to 3.886) whereas Co2 is best described as a weak s-acceptor (increase from 5.007 to 5.206) and a moderate pdonor (decrease from 2.970 to 2.340). Therefore s-electron density is transferred from Co1 through the ptC atom to Co2 whereas p-electron density moves in the opposite direction in the plane perpendicular to that of the four substituents on the ptC. This push/pull interaction is responsible for the stabilization of the ptC.4 Conclusions Over the past years the number of examples of compounds containing a planar-tetracoordinate carbon atom has increased. The confirmation of the presence of this now not so unusual geometry has for the most part come from crystal structure analyses. However the presence of a carbon atom with a 360º sum of angles does not imply that the species is a derivative of planar methane. Theoretical calculations are required to determine the nature of the interactions in the system. In most cases investigated the mode of stabilization is different from that predicted for planar CH4. Very often the delocalization of pelectron density does not play a role in stabilization of the geometry as sufficient density is already distributed into the psystem of the arene or olefin (as with 3 6 and 10–13).In the case of 2 which has often been listed as a compound containing a ptC the carbon atom is in reality coordinated by five atoms and not four. Another point of interest is that there are few purely organic species that come into question (for example the octaplanes) and most examples involve transition metal complexes. Therefore the interaction of the orbitals of carbon with those of metals must be particularly well suited to stabilizing a ptC. Compounds 21a and 21b are the first analogues of planar methane to be structurally characterized and investigated. It is hoped that further examples can be found and trends recog-nized for the factors required to obtain this still very rare species.5 References 1 J. H. van’t Hoff Arch. Neerl. Sci. Exactes Nat. 1874 445; J. A. LeBel Bull. Soc. Chim. Fr. 1874 22 337. 2 R. Hoffmann R. W. Alder and C. F. Wilcox Jr. J. Am. Chem. Soc. 1970 92 4992; R. Hoffmann Pure Appl. Chem. 1971 28 181. 3 (a) J. B. Collins J. D. Dill E. D. Jemmis Y. Apeloig P. von R. Schleyer R. Seeger and J. A. Pople J. Am. Chem. Soc. 1976 98 5419; (b) M.-B. Krogh-Jespersen J. Chandrasekhar E.-U. Würthwein J. B. Collins and P. von R. Schleyer ibid. 1980 102 2263. 4 D. Röttger and G. Erker Angew. Chem. 1997 109 840; Angew. Chem. Int. Ed. Engl. 1997 36 812 and references therein. 5 I. Hyla-Kryspin R. Gleiter M.-M. Rohmer J. Devemy A. Gunale H. Pritzkow and W. Siebert Chem. Eur. J. 1997 3 294.6 W. Luef and R. Keese Adv. Strain Org. Chem. 1993 3 229; W. C. Agosta in The Chemistry of Alkanes and Cycloalkanes eds. S. Patai and Z. Rappoport Wiley New York 1992 p. 927. 7 J. E. Lyons D. R. Rasmussen M. P. McGrath R. H. Nobes and L. Radom Angew. Chem. 1994 106 1722; Angew. Chem. Int. Ed. Engl. 1994 33 1667. 8 F. A. Cotton and M. Millar J. Am. Chem. Soc. 1977 99 7886; F. A. Cotton G. E. Lewis and G. N. Mott Inorg. Chem. 1983 22 560. 9 R. Keese A. Pfenninger and A. Roesle Helv. Chim. Acta 1979 62 326. 10 H. Dietrich W. Mahdi and W. Storck J. Organomet. Chem. 1988 349 1; S. Harder J. Boersma L. Brandsma A. van Heteren J. A. Kanters W. Bauer and P. von R. Schleyer J. Am. Chem. Soc. 1988 110 7802. 11 D. Thoennes and E. Weiss Chem. Ber. 1978 111 3157.12 A. Streitwieser S. M. Bachrach A. Dorigo and P. von R. Schleyer in Lithium Chemistry eds. A.-M. Sapse and P. von R. Schleyer Wiley New York 1995 p. 1; K. Sorger P. von R. Schleyer and D. Stalke J. Am. Chem. Soc. 1996 118 1086. 13 S. L. Buchwald E. A. Lucas and W. M. Davis J. Am. Chem. Soc. 1989 111 397. 14 C. M. Poumbga M. Bénard and I. Hyla-Kryspin J. Am. Chem. Soc. 1994 116 8259. 15 M. A. G. M. Tinga G. Schat O. S. Akkerman F. Bickelhaupt W. J. J. Smeets and A. L. Spek Chem. Ber. 1994 127 1851. 16 A. Hergel H. Pritzkow and W. Siebert Angew. Chem. 1994 106 1342; Angew. Chem. Int. Ed. Engl. 1994 33 1247. 17 R. H. Cayton S. T. Chacon M. H. Chisholm M. J. Hampden-Smith J. C. Huffman K. Folting P. D. Ellis and B. A. Huggins Angew.Chem. 1989 101 1547; Angew. Chem. Int. Ed. Engl. 1989 28 1523; S. T. Chacon M. H. Chisholm K. Folting J. C. Huffman and M. J. Hampden- Smith Organometallics 1991 10 3722. 18 F. A. Cotton and E. S. Shamshoum J. Am. Chem. Soc. 1985 107 4662; F. A. Cotton and E. S. Shamshoum Polyhedron 1985 4 1727; F. A. Cotton W. Schwotzer and E. S. Shamshoum Organometallics 1985 4 461. 19 S. Ogoshi K. Tsutsumi M. Ooi and H. Kurosawa J. Am. Chem. Soc. 1995 117 10415. 20 P. Leoni M. Pasquali G. Pier A. Albinati P. S. Pregosin and H. Rüegger Organometallics 1995 14 3143. 21 (a) G. Erker Comments Inorg. Chem. 1992 13 111; (b) M. Albrecht G. Erker and C. Krüger Synlett 1993 441; (c) M. Albrecht G. Erker M. Nolte and C. Krüger J. Organomet. Chem. 1992 427 C21; (d) G.Erker M. Albrecht C. Krüger S. Werner P. Binger and F. Langhauser Organometallics 1992 11 3517; (e) P. Binger F. Sandmeyer C. Krüger J. Kuhnigk R. Goddard and G. Erker Angew. Chem. 1994 106 213; Angew. Chem. Int. Ed. Engl. 1994 33 197; (f) P. Binger F. Sandmeyer C. Krüger and G. Erker Tetrahedron 1995 51 4277. 22 R. Gleiter I. Hyla-Kryspin S.-Q. Niu and G. Erker Angew. Chem. 1993 105 753; Angew. Chem. Int. Ed. Engl. 1993 32 754. 23 R. B. Castle and D. S. Matteson J. Am. Chem. Soc. 1968 90 2194; R. B. Castle and D. S. Matteson J. Organometal. Chem. 1969 20 19. 24 T. Ast C. J. Porter C. J. Proctor and J. H. Beynon Chem. Phys. Lett. 1981 78 439; C. J. Porter C. J. Proctor T. Ast P. D. Bolton and J. H. Beynon Org. Mass Spectrom. 1981 16 512.25 K. Lammertsma P. von R. Schleyer and H. Schwarz Angew. Chem. 1989 101 1313; Angew. Chem. Int. Ed. Engl. 1989 28 1321. 26 M. W. Wong and L. Radom J. Am. Chem. Soc. 1989 111 1155. 27 H. Schmidbaur and O. Steigelmann Z. Naturforsch. Teil B 1992 47 1721. 28 H. Schmidbaur F. P. Gabbai A. Schier and J. Riede Organometallics 1995 14 4969. 29 O. Steigelmann P. Bissinger and H. Schmidbaur Z. Naturforsch. Teil B 1993 48 72. 30 J. Vincente M. T. Chicote R. Guerrero and P. G. Jones J. Am. Chem. Soc. 1996 118 699. 31 G. M. Smith M. Sabat and T. J. Marks J. Am. Chem. Soc. 1987 109 1854. 32 T. Beringhelli G. Ciani G. D’Alfonso A. Sironi and M. Freni J. Chem. Soc. Chem. Commun. 1985 978. 33 A. Gunale H. Pritzkow W. Siebert D. Steiner and A. Berndt Angew. Chem. 1995 107 1194; Angew. Chem. Int. Ed. Engl. 1995 34 1111. 34 A. Gunale H. Pritzkow W. Siebert D. Steiner A. Berndt I. Hyla- Kryspin and R. Gleiter in Advances in Boron Chemistry ed. W. Siebert The Royal Society of Chemistry Cambridge 1997 pp. 350-353. 35 A. Gunale D. Steiner D. Schweikart H. Pritzkow A. Berndt and W. Siebert Chem. Eur. J. 1998 4 44. Review 8/01225C 371 Chem. Soc. Rev. 1999 28 367–371

ISSN:0306-0012    

DOI:10.1039/a801225c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Chemical aspects of the toxicity of inhaled mineral dusts B. Fubinia and C. Otero Are�anb a Dipartimento di Chimica Inorganica Chimica Fisica e Chimica dei Materiali Università di Torino Via Pietro Giuria 7 10125 Torino Italy b Departamento de Química Universidad de las Islas Baleares 07071 Palma de Mallorca Spain Received 4th May 1999 Some humans are often exposed to airborne mineral dusts at the workplace or in daily life. When inhaled some kinds of mineral dusts can trigger a pathological response of the respiratory system. Silicosis (from silica dust) and asbestosis (from asbestos fibres) are the most commonly known diseases originating from inhaled mineral dusts; other examples are bronchogenic carcinoma and mesothelioma. Detailed knowledge of the chemical (and physical) factors underlying mineral dust toxicity is much needed in order to evaluate the relative risks from exposure to different kinds of materials both natural and synthetic.These pathogenic factors have been reviewed with a focus on the surface chemistry of mineral particles and interface phenomena. To facilitate understanding an outline of the anatomy of the respiratory system and of the etiology of the main diseases involved is also given. 1 Introduction Dating from antiquity lung complaints related to inhaling mineral dusts are among the oldest occupational diseases known to mankind. Indeed Pliny the Younger (61–113 AD) had already referred to the typical sickness of asbestos workers and during the Middle Ages both Paracelsus (1493–1541) and Agricola (1494?-1555) extensively wrote about the miner’s Bice Fubini was educated at the University of Torino (Italy) where she is currently Associate Professor of General and Inorganic Chemistry.She spent one yeat at the School of Chemistry of the University of Bath (UK) carrying out research on solid state and surface chemistry. In the past ten years she has developed studies on the chemical basis of the toxicity of solid materials which is presently her mainresearch interest. She has been invited to several consensus workshops in this field and besides original scientific papers she has authored several reviews on the toxicity of mineral dusts including contributions to IARC (International Agency for Research on Cancer) and to ECVAM (European Centre for the Validation of Alternative Methods).Bice Fubini This journal is © The Royal Society of Chemistry 1999 disease. By the turn of the century the first epidemiological studies on the health hazards associated with airborne mineral dusts mainly silica and asbestos began to appear and the medical community was rendered increasingly aware of the pathogenicity of these materials. This led to severe law regulations which now apply in most western countries. For instance asbestos is today one of the most strictly regulated materials and in the United States the EPA (Environmental Protection Agency) has required all schools and other public buildings to be inspected and analysed for any friable materials containing asbestos.Similar regulations apply in most western countries. Besides silica and asbestos many other kinds of airborne mineral dusts (particulates) both natural and synthetic are today recognised as being potential health hazards carbonaceous dust glasswool and rockwool are only a few examples. These materials can be present in both the occupational and the general environment and the resulting effects on health must be related to the quantity of particles inhaled as a whole taking into account that different kinds of particulates may act not only additively but also in a synergistic (or perhaps sometimes inhibiting) way. Diseases (mainly of the lung) related to inhaled mineral dusts and fibres are usually progressive and cumulative and some of them have latencies of around 15 to 20 years.For these reasons recent epidemiological studies have tracked groups of workers mainly miners over long time periods. These studies have shed some light on the etiology of the diseases mainly relating the Carlos Otero Are�an was educated at the University of Madrid (Complutense) and obtained PhD degrees in Chemistry from the Universities of Bath (UK) and Madrid. He has carried out postdoctoral research at the French CNRS (Orl�eans) and at the University of Oxford (ICL). Currently he is Professor of Inorganic Chemistry at the Universidad de las Islas Baleares (Spain) where his main research interests are in several aspects of solid state and surface chemistry including the interaction between inorganic solids and biological molecules.Carlos Otero Are�an 373 Chem. Soc. Rev. 1999 28 373–381 exposure to a particular kind of particulate to the corresponding illness. However the mechanisms leading to pathogenicity of mineral dusts and the underlying chemistry involved are still poorly understood. And yet knowledge of the chemical basis of particulate toxicity is much needed in order to evaluate the relative risks from exposure to different kinds of materials. This is so much the case at a time when popular concern has grown about the hazards of both natural materials (e.g. asbestos) and synthetic replacements such as fibrous alumina or glasswool. The available literature on the chemical aspects of the toxicity of mineral dusts is somewhat fragmentary and conflicting.1,2 Comparative analysis of the results is complicated by the fact that many experiments (in vivo and in vitro) have been performed under non-equivalent conditions and quite often the allegedly pathogenic materials have not been properly characterized. The long latency periods (already referred to) of several pathological conditions also hinder the evaluation of experimental results. However short-term animal studies (over a period of days to weeks) following exposure to mineral dusts shed some light on the initial stages of the pathological response,1 including inflammation pre-fibroplastic and preneoplastic changes. The increasing use of specific techniques2,3 (FTIR spectroscopy microcalorimetry EDAX (Energy Dispersive X-ray Analysis) atomic force microscopy etc.) for surface characterization is also contributing to our understanding of the pathogenicity of mineral dusts.Altogether some unifying ideas are beginning to emerge and we aim at reviewing them. For the benefit of the non-specialized reader we first give an account of the anatomy of the lungs the fate of particles that enter the respiratory system and the main health effects of inhaled particles. 2 Structure and function of the respiratory system an outline The main function of the lung is to act as a gas exchange system. Venous blood is pumped into the lungs and circulated through the pulmonary capillaries which surround millions of aircontaining alveoli.These alveoli with a total surface of about 80 m2 constitute the gas exchange area of the lungs. During circulation of venous blood through the pulmonary capillaries surrounding the alveoli diffusion gradients favour exchange of oxygen and carbon dioxide between the inspired air and the blood; venous blood becomes thus oxygenated and CO2 is expelled. The upper respiratory tract begins at the nose and mouth and continues through the larynx. After passing through the larynx inspired air enters the trachea which divides into two main bronchi each one leading into a lung. In the lung the bronchus repeatedly divides into smaller and smaller air ducts. The narrowest and most numerous are the bronchioles which have a lumen of about 0.5 mm. The terminal bronchioles end up forming alveolar ducts (0.2 mm in diameter) which open into the alveoli.This architecture and the final dimension of air passages have major implications for deposition and clearance of particulates entering the lung in inspired air. Particles having an aerodynamic size favouring the sedimentation or interception (mainly at bifurcations) have little chance of reaching the alveoli and usually are removed by the mucociliary escalator (see later). Another important poinr in mind is that the air in the lung is basically saturated with water vapour as it becomes moistened on passing through the nose bronchi and smaller air ducts to the alveoli. Therefore inhaled particles necessarily become fully exposed to water vapour before reaching the distal parts of the lung.To preserve the structural and functional integrity of the alveoli specialized cells secrete a pulmonary surfactant rich in phospholipids. These surfactant lipids spread as a monolayer at Chem. Soc. Rev. 1999 28 373–381 374 the air–water interface lining the alveoli and prevent alveolar collapse (at low lung volumes) by decreasing the surface tension of the curved air–liquid interface. The pulmonary surfactant layer may also be involved in non-ciliary transport of inhaled particles from alveoli to bronchioles thus facilitating mucociliary transport and clearance. Besides lipids the lung surfactant contains among other chemical species antioxidants (ascorbate and glutathione) and proteins (mainly immunoglobulins) which can interact specifically with inhaled minerals.Lining the inner surface of the air ducts from the upper respiratory tract down to the terminal bronchioles there is a structure composed of ciliated cells mucus-secreting cells and glands; this mucociliary system (or escalator) plays a major role in removing solid particles from the lung. The cilia act as beaters which continuously transport mucus (and particles deposited on it) upward towards the trachea and larynx where the mucus is then expectorated or swallowed. This is the main mechanism whereby particles deposited on the surface of air passages are removed. In addition particles can also pass into the lymphatic system and be removed. The lymphatic system is a network of thin-walled vessels found throughout all parts of the body (except the central nervous system) including the lung and pleura (see later).Along their course small vessels merge to form larger lymphatic channels. The lymphatic system has the main purpose of removing excess interstitial fluid (the lymph) which is finally drained into the subclavian veins at the root of the neck. For this purpose the lymphatic vessels have valves like those of veins which prevent back flow. The lymph nodes filter off bacteria (a mechanism against infection) and foreign particles and contribute leukocytes (a kind of white blood cell) and antibodies. Foreign particles present in the lung can be engulfed by macrophages which are migratory cells capable of transporting ingested materials through the lymphatic circulation towards the lymph nodes.Although primarily found at these nodes the macrophages (or phagocytes) form part of a larger organisation known as the reticuloendothelial system which includes similar cells in other body tissues. Those present in the lung contribute to the clearance of inhaled particulates. However upon ingestion of particulates macrophages can also liberate fluids contributing to potentially dangerous free radical generation. The lung is covered with two layers of a thin membrane the pleura. The outer layer (parietal pleura) lines the inside of the thorax the inner (visceral pleura) is directly attached to the outside of the lung and is rich in lymphatic drainage channels. When considering the fate of inhaled particulates it is important to realize that the lung alveoli are adjacent to the visceral pleura.Hence particles reaching the distal structures of the lung can pass into the pleura in addition to those that move by way of the lymphatic channels. We shall now summarise the possible fate of inhaled mineral particles and the most frequent pathological conditions they can originate. The interested reader is directed to comprehensive medical texts4 for greater details on the respiratory system. 3 The fate of mineral particles that enter the respiratory system To a large extent the behaviour of particulates entering the respiratory system depends on their size and shape which determine relevant aerodynamic characteristics.The anatomic structure of the respiratory system prevents large particles from reaching the distal parts of the lung; they tend to be intercepted along the passages of the nose or at branching points of bronchi. When particles show a more or less spherical shape only those having a diameter smaller than 5 mm are likely to reach the alveoli; larger ones would be deposited primarily in the upper air passages and removed by the mucociliary escalator. Fibres represent a more complex case since their aerodynamic behaviour depends on length and diameter. Most airborne mineral fibres have a diameter of a few tenths of a mm while length can go from a few mm to several hundreds. Although fibres as long as 200 mm have been found in pathological studies of lung tissue the vast majority5 are shorter than 50 mm.Besides affecting the rate of clearance from the lung the relative dimensions of mineral fibres could also have some differential effect on pathogenicity. It has been suggested6 that fine long fibres (diameter smaller than 0.25 mm and length greater than 8 mm) are more carcinogenic than short and thick fibres. However this hypothesis is not free from controversy,7 and chemical factors are becoming ever more evident. Airborne mineral particles which have reached the lung alveoli can undergo several different processes. First they can simply remain free in the alveolus (some of them would be eventually removed via the mucociliary escalator). Secondly they can be partially or completely dissolved over a period of time.Particles of limestone marble or dolomite (composed of CaCO3 containing variable amounts of other metal carbonates) are likely to have this fate since lysosomal fluids released mainly by alveolar macrophages have an acidic pH and can thus dissolve metal carbonates. Similarly chrysotile (due to its higher magnesium content) is progressively leached by alveolar fluids while amphibole asbestos is more insoluble. This different behaviour could be correlated with the lower toxicity of chrysotile. A third possible fate of inhaled mineral particles reaching the alveoli is to be ingested by macrophages. Alveolar macrophages have a life span averaging 50 days although it can be shortened following fibre ingestion. When they die ingested particles are discharged and can be reingested by other macrophages.This cycle can be repeated indefinitely. Macrophages have a diameter of about 10 mm and are capable of engulfing particles smaller than 5 mm. Larger fibres cannot be completely engulfed but they can pierce the plasma membrane of macrophages giving rise to the discharge of lysosomal fluids and enzymes. A fourth fate involves migration of mineral particles either naked or inside macrophages across the alveolar membrane and into the interstitial lung tissue. They can either remain there or reach the lymphatic system. Particles entering the lymphatic ducts tend to be filtered off at lymph nodes where they may stay indefinitely. However some of them pass (by way of the lymph) into the blood circulation and can thus reach other organs of the body.Finally inhaled mineral particles can pass through the alveolar membrane and into the pleura although the mechanism of this translocation does not seem to be clear. 4 Main diseases related to inhaled mineral dusts 4.1 Pneumoconiosis Pneumoconiosis has been defined4b as a non-neoplastic reaction of the lungs to inhaled minerals and the resultant alteration in their structure excluding asthma bronchitis and emphysema. Pneumoconioses can have several origins and they are termed accordingly; those occurring more often are silicosis asbestosis and coal worker’s pneumoconiosis. Inhaled silica dust can give rise to silicosis. Silica is widespread in nature; it occurs as a component of both the soil and many rocks apart from forming quartz and several other mineral species.Exposure to finely divided silica occurs in a variety of jobs; high-risk workers include miners sandblasters and people involved in drilling works. When the intensity and duration of exposure are high silica particles accumulate in alveoli from where they spread into the lung and lymphoid tissue. Crystalline silica dust causes an intense cellular reaction with local proliferation of macrophages lymphocytes and fibroblasts. The characteristic histological lession in the lung is a silicotic nodule the granuloma. When silicotic nodules are discrete the condition is termed simple silicosis and is usually observed to develop only after ten or more years of exposure to silica dust.When the disease progresses silicotic nodules tend to grow and coalesce into conglomerates leading to massive fibrosis a condition termed conglomerate silicosis. Lung function is reasonably preserved in patients having simple silicosis but after the onset of conglomeration lung function can become severely impaired. Shortness of breath (dyspnea) chronic cough and bronchitis are common symptoms. Advanced conglomerate silicosis can lead to a severe condition of the cardiorespiratory system. Asbestosis is the form of lung fibrosis that results from inhaling asbestos fibres; it is frequently associated with pleural calcification (plaques). Exposure to asbestos can occur in many occupations; high-risk groups include asbestos miners and millers asbestos textile workers shipyard workers and labourers in various construction and insulation trades.As in the case of silicosis asbestosis is a disease which develops slowly over the years. The onset of substantial lung fibrosis needs exposure to a relatively high air concentration of asbestos fibres for at least 10 years. However once initiated fibrosis can progress even without further exposure. In contrast to the discrete nodular fibrosis observed in silicosis the fibrogenic reaction typical of asbestosis usually spreads out along the supporting structures of the lung. The mechanism of asbestos-induced fibrogenesis is not clear. The sequence of adverse health effects starts with alveolitis (inflammatory process at the alveoli) and is followed by a generalised fibrogenic response which leads to destruction of the alveolar architecture and adjacent vascular structures.8 Long fibres seem to be more active than short ones.The reason for this size-dependent response seems to be that long fibres cannot be taken entirely by macrophages; as a result the pierced macrophage leaks and discharges fibrogenic factors. Note also that because of their size long fibres cannot be cleared from the alveoli as efficiently as smaller mineral particles. Diffuse fibrosis generated by inhaled asbestos is often difficult to distinguish from other forms of lung fibrosis having a different etiology (e.g. idiopathic pulmonary fibrosis). A unique (and diagnostic) feature of asbestosis is the presence in histological sections of the affected lung of asbestos bodies.9 Asbestos (or ferruginous) bodies is the name given to fibres which become coated with a thick layer of iron oxyhydroxide mixed with mucopolysaccharides and proteins.Long fibres tend to form asbestos bodies. Heavy exposure to carbonaceous dust can lead to a condition known as coal miner’s pneumoconiosis. The main occupations concerned are coal mining and processing but the mining and processing of graphite and the fabrication of carbon black and carbon electrodes can also pose health hazards. Chimney sweeps also constitute a risk group. Unlike silica or asbestos pure coal or graphite dusts do not cause an intense cellular reaction in the lung and they are less fibrogenic.Inhaled carbonaceous dust gives rise to localized nodules (in the area of alveoli and alveolar ducts) which are 1 to 2 mm in diameter and contain macrophages laden with carbon dust. Ingestion of carbon particles by macrophages does not prematurely kill them but heavy and prolonged exposure to airborne carbonaceous dust overloads the cleansing mechanisms of the lung. At the stage of localized nodules lung function is nearly normal and the resulting condition is less severe than simple silicosis. In some patients however conglomerate lesions develop with an outcome similar to that found in conglomerate silicosis. It should also be noted that coal 375 Chem. Soc. Rev. 1999 28 373–381 in mines is often associated with silica. Therefore a coal miner can develop silicosis.4.2 Cancer Exposure to respirable mineral dusts can be a risk factor for developing several forms of cancer. Among them bronchogenic carcinoma and malignant mesothelioma are the most common forms. However not all particulate minerals are carcinogenic. Table 1 lists the mineral carcinogens recognized by the International Agency for Research on Cancer (IARC). Table 1 Carcinogens (according to the IARC) which often occur as mineral dusts. (*) Possibly carcinogenic Frequent occupational exposure Material Asbestos Asbestos mining asbestos textile industry insulation trades Metal plating zinc ore mining and processing Graphite processing carbon black fabrication Tanning of leather Mining Metal plating Metal plating Mining rock drilling sandblasting Cadmium Carbon black (*) Ceramic fibres (*) Chromium(VI) compounds Coal tar pitches Erionite Glasswool (*) Lead compounds (*) Nickel compounds Nickel metallic (*) Silica crystalline Rockwool (*) Talc containing asbestiform fibres Bronchogenic carcinoma (lung cancer) arises from the epithelial lining of the lung airways.It usually has a latent period of at least 10 to 20 years between exposure to the carcinogen and clinical manifestation.1 There is some evidence10 that lung cancer is frequently associated with diffuse pulmonary fibrosis (asbestosis) rather than with other forms of pneumoconiosis. Diffuse malignant mesothelioma is a cancer of the mesoderm developing either at the pleura or the peritoneum (the lining membrane of the abdomen).Epidemiological studies10 have shown an excess risk of developing mesothelioma among workers exposed to asbestos or to erionite (see below). It is also known6,7 that asbestos fibres can induce mesothelial tumours in experimental animals. 4.3 Other adverse health effects of inhaled mineral dusts Other lung conditions attributable to inhaled mineral dusts include chronic cough chronic bronchitis emphysema and thickening (calcification) of the pleura. Bronchitis (and alveolitis) results from inflammatory responses triggered by particle deposition in the respiratory tract. Inflammation is a defensive reaction to the irritation arising either from a direct effect of the inhaled dust or as a consequence of the activity of lymphocytes and macrophages.Alveolar macrophages can generate oxidizing free radicals during phagocytosis and this activity can cause local tissue damage and inflammation. Impaired lung function brought about by fibrosis can cause over-inflation of the air spaces thus leading to pulmonary emphysema. The condition is aggravated when conglomerate lesions occur. Finally chronic infection (mainly due to anaerobic bacteria) occurs occasionally at the stage of conglomerate silicosis or diffuse fibrosis. There is also an increased risk of developing pulmonary tuberculosis. There are reports in the literature11 linking exposure to silica (or asbestos) with a variety of immunoregulatory disorders and autoimmune diseases.The best example of this link is the Chem. Soc. Rev. 1999 28 373–381 376 occasional coexistence of silicosis and rheumatoid arthritis (Caplan’s syndrome). Other autoimmune diseases which have been implicated in the pathogenesis of silicosis (and to a lesser extent asbestosis) include systemic sclerosis lupus and chronic renal disease. The mechanisms of synergy between exposure to mineral dusts and autoimmune conditions are poorly understood. However there may be a link between proliferation and destruction of macrophages (triggered by the presence of mineral dusts in the lung and lymphatic system) and immunoregulatory disorders. 5 Main materials of concern The main inorganic materials of concern in relation to diseases caused by inhaled dusts can be divided into two categories natural minerals and synthetic mineral fibres.Among natural minerals silica asbestos and erionite are those of main concern; coal can be cited in a second place. 2) is a widespread mineral which can occur in a Silica (SiO large number of polymorphic forms (more than 20 have been described). The most common form of silica is quartz cristobalite can be cited in second place. Quartz is a major constituent of a number of rocks such as granite and sandstone. It also occurs alone either in a highly crystalline form or forming poorly crystalline minerals (often referred to as being cryptocrystalline). Synthetic silicas include silica gel and pyrogenic silica. Silica can also have a biogenic origin.Among biogenic silicas diatomaceous earth (or kieselguhr) is mined in a large scale and used mainly in filtration plants and as an insulation material. Calcination of kieselguhr yields cristobalite. All silica minerals both natural and synthetic are formed by a framework of corner sharing SiO4 tetrahedra. Different polymorphs arise from slightly different arrangements of these tetrahedra. The surface of the mineral contains mainly silanol (Si–OH) and siloxane (Si–O–Si) groups. In freshly crushed silica some highly reactive chemical species can also be present (see Section 6.1). 3Fe2 3+Si8O22(OH)2 respectively. Asbestos minerals are fibrous silicates which can be divided into two groups serpentines and amphiboles. Chrysotile is the main representative of the serpentine group it has the chemical composition Mg3Si2O5(OH)4.Amosite and crocidolite are the best known representatives of the amphibole group their chemical formulae are (Mg Fe2+)7Si8O22(OH)2 and Na2(Mg Fe2+) Basically the crystal structure of all the minerals in the serpentine group can be thought of as being formed by a double layer consisting of a tetrahedral (silicate) sheet of composition (Si2O5)n 2 n2 in which three of the oxygen atoms in each SiO4 tetrahedron are shared by adjacent tetrahedra and an octahedral (brucite) sheet of composition [Mg3O2(OH)4]n 2 n+ formed by edge-sharing MgO2(OH)4 octahedra (iron can substitute for magnesium in this layer). The two sheets are bonded together forming a double layer in which the apical oxygens of the (Si2O5)n 2 n2 sheet are shared with the brucite layer as shown in Fig.1a. There is a slight misfit between the tetrahedral and octahedral sheets of the double layer which causes curling (like rolling up a carpet) to form concentric cylinders. Curling takes place keeping the silicate layer inside and the brucite layer on the outside of the curve (Fig. 1b). The average diameter of the cylinder thus formed (a chrysotile fibril) is about 25 nm. The crystal structure of the amphiboles (Fig. 1c) can be described in terms of a basic structural unit formed by a doubletetrahedral chain (corner linked SiO4 tetrahedra) of composition (Si4O11)n 6 n2. These silicate double-chains share oxygen atoms with alternate layers of edge-sharing MO6 octahedra where M stands for a variety of cations mostly Mg2+ Ca2+ Fe2+ or Fe3+.Variations in the stacking of the tetrahedral and octahedral Fig. 1 (a) The double layer of the serpentine-type structure. (b) Schematic representation of chrysotile asbestos fibres. (c) The amphibole-type structure. (d) Typical morphology of amphibole fibres. layers give rise to slightly different structural types. The amphiboles exhibit prismatic cleavage. When they are crushed or milled they fracture along the octahedral layers giving rise to acicular fragments having the typical morphology shown in Fig. 1d. Individual fibres are usually about 0.2 mm in diameter and they tend to aggregate into bundles. Serpentine and amphibole asbestos are the most typical fibrous minerals.However some other silicates can present a fibrous habit. Zeolites (which are three-dimensional aluminosilicates) should be mentioned in this context since one of them (erionite) is known to cause mesothelioma when inhaled. Erionite is one of the few fibrous zeolites; its typical composition can be described as (K Na)5Ca2[(AlO2)9- (SiO2)27]·27H2O. It should be borne in mind however that the Na+ K+ and Ca2+ ions can be replaced by other cations. Synthetic inorganic fibres are produced either to replace fibrous minerals or for more specific applications. Main uses of these synthetic materials are thermal and acoustic insulation fireproofing fibre optics and the manufacturing of fibre reinforced composites. Typical compositions of synthetic inorganic fibres are summarized in Table 2.Table 2 Typical composition (weight%) of synthetic inorganic fibres Ceramic fibre Slagwool Rockwool Component Glasswool 2O3 Al2O3 B2O3 CaO FeO/Fe K2O MgO Na 2 0–96 0–14 0–0.7 0–0.8 0–0.1 0–0.5 0–0.2 — 0–53.9 0–1.6 0–8 0–92 11.8–12.5 — 37.5–40.0 0.9–1.0 0.3–0.4 — 0.2–1.5 — 40.6–41.0 0.4–0.5 — — 6.5–13.4 — 10.8–30.3 1.5–12.4 1.0–1.6 — 2.3–2.5 — 45.5–52.9 0.5–2.0 — — 2.4–14.5 3.5–8.5 5.5–22.0 — 0.5–3.5 3.0–5.5 0.5–16.0 0.0–59.0 34.0–73.0 0.0–8.0 — 0.0–4.0 2O PbO SiO TiO2 Y2O3 ZrO2 Fibrous glass includes glasswool and special-purpose fibre glass (e.g.optical fibre). It is an amorphous material made of fused silica to which other glass-forming oxides are added in variable concentration. Mineral wool includes rockwool made from magma rock and slagwool derived from molten slag produced mainly in iron and steel metallurgical processes. Refractory ceramic fibres can be made from clays and other aluminosilicates or from several metal oxides such as SiO2 Al2O3 or ZrO2. Whiskers constitute a less common type of refractory fibres; they are made mainly from metal carbides or boron nitride and frequently they are single crystal materials. To a large extent the possible pathological effects of synthetic mineral fibres in relation to respiratory diseases are as yet unknown. Some epidemiological studies12 suggest that excess mortality from at least some types of synthetic mineral fibres is not high.However because of limited knowledge it would be prudent to treat these materials with similar precautions to those recommended for asbestos. 6 Chemical factors affecting pathogenicity 6.1 General background The pathogenic effects of inhaled mineral dusts can be related to both physical and chemical factors. As already considered in Sections 2 and 3 the size and shape of inhaled particulates affect the rates at which deposition and clearance take place. Large airborne particles tend to be deposited in the air ducts of the respiratory system before reaching the terminal bronchioles and alveoli and are efficiently removed by ciliated cells.Macrophages can ingest small particles and short fibres thus helping to transport them out of the lung. Fibres of intermediate size thin enough to reach the distal parts of the lung but too long to be entirely engulfed by macrophages may be a major culprit in causing lung disease. Indeed this notion known as the Stanton hypothesis,6,13 dominated early studies on the toxicity of inhaled mineral fibres. More recently however attention is being focused on the chemical aspects of particulate pathogenicity since it has been recognized that morphological factors alone are unable to account for the differential toxicity of mineral dusts having different chemical composition and structure.14 When dealing with solids which are (mostly) insoluble in biological fluids the chemical aspects of their toxicity have to be related mainly to surface properties.It is often considered that the chemical reactivity of a solid material is determined by its chemical composition and crystal structure. However it should be realized that at the surface of a Chem. Soc. Rev. 1999 28 373–381 377 (crystalline) solid the three-dimensional periodicity of the crystal lattice is lost. This fact affects electrostatic potentials and local electron distribution giving rise to electronic states (surface states) which confer to the surface characteristic chemical and physical properties. Surface states can also arise from (or can be modified by) surface relaxation or reconstruction and also from structural imperfections and adsorbed impurities (or surface segregation of chemical species from the bulk).The presence of surface states affects both the electrical properties of the surface and the chemical reactivity by modifying the affinity of the surface for electrons. When dealing with ionic solids the presence at their surface of exposed cations and anions is of primary importance regarding chemical reactivity and interaction with biological molecules and living cells. Exposed coordinatively unsaturated cations act as electron acceptors (Lewis acid sites in the broad sense) and provide interaction points for electron donors. Similarly surface exposed anions can interact with electron acceptors or dipolar molecules. At the solid–liquid interface (between solid particles and biological fluids) the ionic composition of the surface layer can be altered by preferential transfer of cations or anions into the aqueous liquid phase.This phenomenon alters the electrical neutrality of the interface and the solid surface becomes electrically charged. The corresponding electrokinetic potential thus generated usually termed the zeta potential can reach hundreds of millivolts and can critically affect the interaction of mineral particles with living cells. Indeed Light and Wei15 have found a strong (positive) correlation between the zeta potential of asbestos fibres and their hemolytic activity (ability to rupture the erythrocyte membrane). The idealized surface of covalent solids (such as graphite or quartz) has a homopolar character with no charge separation.However as a result of mechanical fracture several reactive species can be generated which represent localized surface states. As an example let us consider silica. 4 tetrahedra share corners In all silica polymorphs SiO forming a three-dimensional network of Si–O–Si bonds which have mainly a covalent character. When these bonds are broken as a result of mechanical fracture two different situations can occur as depicted in Fig. 2. Homolytic cleavage gives rise to Fig. 2 Homolytic (A) and heterolytic (B) cleavage of Si–O–Si bonds in silica. free radical •Si and Si–O• species often referred to as dangling bonds. Heterolytic cleavage generates electrically charged surface species (Si+ and SiO2).Both types of surface states are highly reactive and they are likely to be the main cause of the pathogenic effects induced by inhaled silica dust.14e Another factor to consider is that charge separation at the silica surface results in increased hydrophilicity of the mineral. This hydrophilicity can affect interaction with biological molecules. 6.2 Hydrophilicity and hydrophobicity Solids presenting a heteropolar surface are hydrophilic because exposed surface ions interact strongly with the electrical dipole of water molecules. Homopolar surfaces tend to be hydrophobic. Hydrophilic mineral particles show the strongest interaction with biological molecules which are mainly polar or Chem. Soc. Rev. 1999 28 373–381 378 even contain charged groups.They can therefore act as electron donors or acceptors. Extremely hydrophobic surfaces can also strongly adsorb proteins by a mechanism known as the hydrophobic interaction which involves a favourable entropic term arising from the release of water molecules forming the hydration sphere of the protein. However such a high degree of hydrophobicity does not occur among mineral dusts. Hydrophilic surfaces favour protein adsorption and denaturation. They also favour cell-surface adhesion which can lead to injury.2,16 The outer cell membrane (the plasma membrane) is composed of amphiphilic phospholipids proteins and steroids which form a thin bilayer. This bilayer is structured so as to have the hydrophobic cores of the biological molecules in each monolayer pointed inward while the hydrophilic head groups are pointed outward towards the surrounding (aqueous) space and the internal cytoplasm.These hydrophilic head groups which in vivo tend to be negatively charged may interact strongly with surface exposed cations of mineral particles (usually displacing adsorbed water molecules). Such an interaction can adversely affect membrane structure and dynamics leading to membranolysis. As an example Fig. 3 shows the hemolytic activity of a cristobalite powder as a function of hydrophilicity. When cristobalite was calcined at increasing temperature in order to Fig. 3 Erythrocyte hemolysis (white circles) and water vapour adsorption (black circles) on cristobalite powder preheated at increasing temperature.render it less hydrophilic the hemolytic potential was also found to decrease.17 Note that the degree of hydrophilicity was tested by measuring the extent of water vapour adsorption at an equilibrium pressure of 5 Torr. Although this experiment has to be interpreted with some caution because heat treatment of cristobalite could also reduce the concentration of surface free radicals it is tempting to think that the relative hydrophobicity of carbon dust (as compared to silica or asbestos) could be one factor determining its less intense cellular reaction in the lung (Section 4.1). 6.3 Mineral induced free radical generation Many intracellular processes including the mitochondrial respiratory chain reduce molecular oxygen to the superoxide radical (O· 22) or to hydrogen peroxide.In a low concentration these chemical species are only moderately reactive towards biological molecules and their potential adverse effects are minimized by the antioxidant defense mechanisms of the cell which include the action of the enzymes catalase superoxide dismutase and glutathione peroxidase. However in the presence of transition metal ions the radical species O· 22 and also free oxygen and H2O2 can generate the hydroxyl radical (•OH). This free radical is a highly reactive species capable of causing among other deleterious effects DNA damage protein oxidation and lipid peroxidation. Iron is the most ubiquitous transition metal in both natural minerals and synthetic mineral fibres.It can occur either as a major component (for example in many asbestos minerals) or as an impurity. Exposed iron ions at the surface of mineral dusts can generate •OH by the following (iron-catalysed) Haber– Weiss cycle Reductant + Fe(iii) ? oxidized reductant + Fe(ii) (1) Fe(ii) + O2 ? Fe(iii) + O• 22 (2) O• 22 + 2H+ + e2 ? H2O2 (3) 2O2 ? Fe(iii) + OH2 + •OH Fe(ii) + H (4) Eqn. (4) is known as the Fenton reaction. Several metabolites can act as the reductant species in eqn. (1); ascorbate cysteine and glutathione are a few examples. Generation of hydroxyl radicals by the Haber–Weiss cycle needs iron only in catalytic (trace) amounts and the turnover of free radicals can overload the antioxidant defence mechanisms of living cells.In natural minerals (e.g. the amphibole group minerals amosite and crocidolite) and in some man-made fibres iron can occur either as Fe(iii) or as Fe(ii) and some controversy has arisen as to which oxidation state can be more dangerous. 14,18–20 Note however that the oxidation state of iron in the bulk mineral could be of little concern for in vivo effects. It should be borne in mind that (i) because of exposure to (humid) air iron containing particulates would show Fe(iii) ions at their surface regardless of the oxidation state of the metal in the bulk material and (ii) once inhaled mineral dusts come into contact with reductant chemical species (e.g. ascorbate superoxide ions etc.) capable of converting Fe(iii) into Fe(ii). Iron-catalysed free radical generation is known to be an important factor enhancing acute lung inflammation and it also appears to be a major carcinogenic factor.1 Hobson et al.21 have shown that •OH radicals enhance asbestos fibre uptake by bronchial epithelial cells and many pathologists agree that bronchogenic carcinoma (which originates at the epithelial cells of the lung airways) frequently occurs in association with exposure to iron containing asbestos.22 Similarly malignant mesotheliomas have often been associated with a history of occupational exposure to amphibole asbestos.1 The role of free radicals in asbestos-induced diseases has been recently reviewed.14,23 Kane1 has discussed in detail the possible mechanisms of mineral fibre carcinogenesis.Generation of free radicals (mainly •OH) constitutes one such mechanism.Free radicals are known to induce DNA damage which according to Kane,1 can lead to carcinogenesis by any one of the following routes (i) alterations produced in oncogenes and growth factors (ii) alterations in tumour suppressor genes and (iii) alterations in growth regulatory genes. The Haber–Weiss cycle is not the only mechanism involved in the generation of free radicals by inhaled mineral dusts. They are also produced as a consequence of macrophage activity as stated in Section 4.3. Fig. 4 depicts a summarising scheme and a micrograph showing macrophage morphology. It should be noted that administration of antioxidants such as superoxide dismutase or catalase was found to result in amelioration of cell injury caused by asbestos in experimental animals.7 This fact lends further support to the hypothesis linking free radical release with asbestos-induced pathogenesis.6.4 Adsorption of endogenous chemical species Inhaled mineral dusts have the potential to adsorb (and concentrate on their surface) endogenous molecules and metal ions which may influence pathogenicity. The extent to which adsorption takes place depends on such factors as specific surface area of the mineral dust chemical composition hydrophilicity or hydrophobicity and zeta potential. Endogenous iron is one of the metals known to concentrate on the surface of inhaled dusts. This accumulation of iron ions can boost free radical generation and consequent pathogenicity; Fig.4 (a) Scheme of processes relating to phagocytosis and free radical generation by mineral particulates. (b) Macrophage phagocytizing chrysotile asbestos fibres (courtesy of Dr D. B. Warheit). it can also result in depressed resistance to infection. Thus Ghio et al.24 suggested that the increased incidence of tuberculosis observed among silicate workers may be explained by accumulation of iron complexed by inhaled dust particles and made available to dormant mycobacteria. Endogenous iron deposited on inhaled mineral dusts can have several origins; the hemolytic activity of the mineral particles,15 and the iron store proteins ferritin and hemosiderin are among the possible sources. Ferritin in particular was often found to be present in the coating of ferruginous bodies,9 and it could constitute a local source of iron ions capable of supporting a Haber–Weiss cycle leading to radical damage to DNA and cell organelles.In a recent in vitro study,25,26 ferritin was found to adsorb strongly on the asbestos fibres amosite and crocidolite. When traces of ascorbic acid were added these mineral fibres containing adsorbed ferritin were found to cause significant radical damage to DNA. These results are in agreement with the increased DNA damage found for amosite-core asbestos bodies when compared to the effect produced by the naked fibre.27 379 Chem. Soc. Rev. 1999 28 373–381 Macromolecules from the pulmonary surfactant can also be adsorbed onto inhaled mineral particles. Among these macromolecules immunoglobulins are of some concern.They constitute a major component of the lung surfactant (cf. Section 2) and can trigger a pathogenic response when adsorbed on mineral particles. In vitro cell assays,28 using alveolar macrophages from guinea pigs have shown that immunoglobulin G (IgG) is strongly adsorbed on asbestos fibres conferring to them antigenic determinants which enhance attack by macrophages with release of (potentially dangerous) superoxide anions. As the authors conclude these in vitro studies may have significance for the interaction of asbestos with the alveolar macrophage in vivo. The human IgG subclasses have isoelectric points ranging from 7.2 to 8.6 so that they tend to be positively charged at physiological pH.Minerals such as crocidolite which show a negative zeta potential15 in a physiological buffer are expected to be major adsorbents of IgG. Contrary to the effect produced by IgG adsorption the pathogenic response to inhaled particles present at the alveoli can be mitigated when phospholipids (from the pulmonary surfactant) become adsorbed onto the mineral particle.29 6.5 Adsorption of exogenous substances Airborne particulates can adsorb vapours from ambient air some of which e.g. polyaromatic hydrocarbons are highly toxic. The solid particle may act as a carrier of these compounds into the lung where they can damage cells and tissues. Equilibria at interfaces regulate some of these processes since the lung surfactant may dissolve adsorbed substances.The increased risk of lung cancer among smokers exposed to silica and asbestos dust (as compared to non-smokers) has been related to adsorption of polyaromatics (from cigarette smoke) on the mineral particulates.30 Cigarette smoke was also found to potentiate the damaging action of rockwool on isolated DNA.31 Nitric oxide is among the atmospheric contaminants which can be adsorbed onto airborne particulates. Its presence on inhaled mineral dusts can upset metabolic regulation of this biologically significant molecule. It can also result in increased free radical generation.32 6.6 Mechanical fracture and grinding Many environmental dusts originate as a consequence of mechanical fracture and milling of bulk materials. In these cases the final state of the surface depends considerably upon the way the fracture occurred and the composition of the atmosphere during grinding.Crystalline silica dusts in respirable size are usually generated during processes (mining drilling etc.) whereby large crystals become fractured. Since mechanical fracture does not usually follow crystal planes the particles thus generated are very irregular and show sharp edges and acute spikes. In a dry atmosphere mechanical cleavage of Si–O bonds can give rise to very reactive surface species (Fig. 2). When water vapour is present surface SiOH species tend to be formed instead of dangling bonds.14e In experimental animals freshly ground silicas have shown a higher degree of toxicity than aged ones.33 The cause of acute silicosis which can occur among workers involved in sandblasting drilling or grinding has to be sought in the peculiar properties of freshly cleaved crystals.14e 6.7 Synergic effects Mixed dusts often contain several components which may interact in a synergistic (or sometimes inhibiting way) with each Chem.Soc. Rev. 1999 28 373–381 380 other.2 Examples of synergic interaction are the enhanced toxicity of freshly fractured quartz containing traces of iron,33 and that of metallic cobalt mixed with tungsten carbide.34 In the latter case it was shown that WC/Co composite particles display a much higher cytotoxicity than either component alone and this fact correlates with in vitro physicochemical studies showing that the composite material can generate free radicals in contact with (aerated) water.However in similar conditions tungsten carbide generates no free radicals and cobalt shows little free radical activity.2,35 Moreover epidemiological and clinical studies suggest that association of cobalt to tungsten carbide is the determining factor causing lung disease when the composite dust is inhaled. This particular pathology has been termed hard metal disease.34 Examples of inhibiting effects are the decreased carcinogenicity of quartz dust when it comes into intimate contact with coal dust or with some metals such as zinc tungsten or gold. Finally it should be mentioned that simple adherence between different airborne particulates can modify the rates at which they are removed from the lung or translocated into tissues adjacent to the lung alveoli.These effects are particularly relevant when dealing with association between fibrous and non-fibrous particulates.1 6.8 Biopersistence Biopersistence can be defined as the retention in the lung over time of inhaled mineral dusts. Factors affecting biopersistence include particle size and shape (which affect the rate of clearance from the lung) chemical composition surface area and structural parameters (which affect the rate of dissolution). Changes in any of these parameters may alter the toxicity of mineral dusts. Biopersistence is also dependent on the site and rate of deposition; a large increase in the rate of deposition in the alveoli can overwhelm macrophage clearance mechanisms.Regarding the rate of dissolution the relationship between in vivo solubility and chemical composition has been demonstrated36 for a number of different mineral fibres for which solubility was found to follow the order glass fiber > rockwool > ceramic fibre > chrysotile > amphibole asbestos. Significantly their toxic potential appears to increase in the same order as decreasing solubility. The concept of biopersistence has gained importance in the last decade; clearly it should be taken into account when considering the toxicity of particulates. However it is not easy to assess (in humans) the relative biopersistence of different mineral dusts because of the many processes involved. For more details the interested reader may consult the recent topical issue of Environmental Health Perspectives.36 7 Conclusions It should be clear that toxicity of inhaled mineral dusts stems from a combination of (interrelated) factors which greatly complicates assessment of the role played by each individual process.However our current knowledge about toxicity of inhaled particulates is being advanced by increasing awareness of the main physical and chemical parameters which determine adverse health effects. Among physical factors particle size and shape determine the rate of deposition of airborne particulates and that of clearance from the lung. Surface roughness may mediate inflammatory processes and to some extent chemical behaviour and dissolution rate.The main chemical factors determining particulate pathogenicity have been reviewed. They can all be related to processes occurring at the interface between the mineral particle and the pulmonary tissue. Therefore surface chemical composition (which may differ in several ways from that of the bulk material) and active surface states are the key chemical determinants of biological response. It should be emphasized that a single chemical (or physical) factor is not likely to be the only pathogenic determinant for any kind of particulate. However studies (both in vivo and in vitro) carefully designed to analyse the effect of each individual factor are very valuable and they represent most of the current research work in the field of chemical toxicity of inhaled mineral dusts.Such studies should pave the path for further research concerning the interplay of the many factors affecting particulate toxicity which have already been identified. 8 References 1 A. B. Kane in Mechanisms of Fibre Carcinogenesis Ed. A. B. Kane P. Boffetta R. Saracci and J. D. Wilbourn IARC Scientific Publication No. 140 Lyon 1996 p. 11 and references therein. 2 B. Fubini A. E. Aust R. E. Bolton P. J. A. Borm J. Bruch G. Ciapetti K. Donaldson Z. Elias J. Gold M. C. Jaurand A. B. Kane D. Lison and H. Muhle ATLA 1998 26 579. 3 B. T. Mossman Toxicol. Pathol. 1999 27 180. 4 See for example (a) C. Nagaishi Functional Anatomy and Histology of the Lung University Park Press Baltimore 1972; (b) W. R. Parkes Occupational Lung Disorders Butterworths London 1982; (c) R.Rhoades and R. Pflanzer Human Physiology Saunders London 1996. 5 V. Timbrell Ann. N. Y. Acad. Sci. 1965 132 255. 6 M. F. Stanton M. Layard A. Tegeris E. Miller M. May E. Morgan and A. Smith J. Natl. Cancer Inst. 1981 67 965. 7 L. A. Goodglick and A. B. Kane Cancer Res. 1990 50 5133. 8 B. T. Mossman and A. Churg Am. J. Respir. Crit. Care Med. 1998 157 1666. 9 A. M. Churg and M. L. Warnock Am. J. Pathol. 1981 102 447. 10 J. C. Wagner J. C. Gilson G. Berry and V. Timbrell Br. Med. Bull. 1971 27 71. 11 K. Steenland and D. F. Goldsmith Am. J. Ind. Med. 1995 28 603; M. Klockars P. Koskela E. Jarvinen P. Kolari and A. Rossi Br. Med. J. 1987 294 997; R. N. Jones M. Turner-Warwick M. Zisking and H.Weill Am. Rev. Respir. Dis. 1976 113 393. 12 P. E. Enterline G. M. Marsh V. Henderson and C. Callahan Ann. Occup. Hyg. 1987 31 625. 13 A. Morgan R. J. Talbot and A. Holmes Br. J. Ind. Med. 1978 35 146. 14 (a) D. W. Kamp P. Graceffa W. A. Pryor and S. A. Weitzman Free Radical Biol. Med. 1992 12 293; (b) B. Fubini in Fiber Toxicology Ed. D. B. Warheit Academic Press London 1993 p. 229; (c) J. A. Hardy and A. E. Aust Chem. Rev. 1995 95 97; (d) B. Fubini Environ. Health Persp. 1997 105 1013; (e) B. Fubini in The Surface Properties of Silicas Ed. A. P. Legrand Wiley Chichester 1998. 15 W. G. Light and E. T. Wei Nature 1977 265 537. 16 K. Donaldson B. G. Miller E. Sara J. Slight and C. Brown Int. J. Exp. Pathol. 1993 74 243. 17 D.H. Hemenway M. P. Absher B. Fubini and V. Bolis Arch. Environ. Health 1993 48 343. 18 H. Pezerat R. Zalma J. Guignard and M. C. Jaurand in Non Occupational Exposure to Mineral Fibres IARC Scientific Publication No. 90 Ed. J. Bignon J. Peto and R. Saracci) IARC Lyon 1997. 19 A. Nejjari J. Fournier H. Pezerat and P. Leanderson Br. J. Ind. Med. 1993 50 501. 20 P. S. Gilmour P. H. Beswick D. M. Brown and K. Donaldson Carcinogenesis 1995 16 2973. 21 J. Hobson J. L. Wright and A. Churg FASEB J. 1990 4 3135. 22 P. T. Cagle in Pathology of the Lung Ed. W. M. Thurlbeck and A. M. Churg Thieme Medical Publishers New York 1995 pp. 437–573. 23 B. Fubini in Mechanisms of Fibre Carcinogenesis Ed. A. B. Kane P. Boffetta R. Saracci and J. D. Wilbourn IARC Scientific Publication No. 140 IARC Lyon 1996. 24 A. J. Ghio T. P. Kennedy R. M. Schapira A. L. Crumbliss and J. R. Hoidal Lancet 1990 336 967. 25 B. Fubini F. Barceló and C. Otero Areán J. Toxicol. Environ. Health 1997 52 343. 26 C. Otero Areán F. Barceló and B. Fubini Res. Chem. Intermed. 1999 25 177. 27 L. G. Lund M. G. Williams R. F. Dodson and A. E. Aust Occup. Environ. Med. 1994 51 200. 28 R. K. Scheule and A. Holian Am. J. Respir. Cell. Mol. Biol. 1990 2 441. 29 X. Liu M. J. Keane J. C. Harrison E. V. Cilento T. Ong and W. E. Wallace Toxicol. Lett. 1998 96 77. 30 I. J. Selikoff E. C. Hammond and J. Churg J. Am. Med. Assoc. 1968 204 106. 31 P. Leanderson V. Lagesson and C. Tagesson Environ. Health Persp. 1997 105 1037. 32 C. C. Chao S. H. Park and A. E. Aust Arch. Biochem. Biophys. 1996 326 152. 33 V. Castranova V. Vallyathan D. M. Ramsey J. L. McLaurin D. Pack S. Leonard M. W. Barger J. Y. C. Ma N. S. Dalal and A. Teass Environ. Health Persp. 1997 105 1319 and references therein. 34 D. Lison P. Carbonelle L. Mollo R. Lauwerys and B. Fubini Chem. Res. Toxicol. 1995 8 600 and references therein. 35 G. Zanetti and B. Fubini J. Mater. Chem. 1997 7 1647. 36 J. Bignon R. Saracci and J. C. Touray (Eds.) Environ. Health Persp. 1994 102 Supplement 5. Review 8/05639K 381 Chem. Soc. Rev. 1999 28 373–381

ISSN:0306-0012    

DOI:10.1039/a805639k

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Chiral non-racemic N-cyanomethyloxazolidines the pivotal system of the CN(R,S) method Henri-Philippe Husson and Jacques Royer Laboratoire de Chimie Thérapeutique UMR 8638 du CNRS associée à l’Université René Descartes 4 avenue de l’Observatoire 75270 Paris cedex 06 France Received 5th January 1999 The a-cyanomethyloxazolidine ring system has been integrated into piperidine and pyrrolidine structures providing chiral non-racemic building blocks. Reaction conditions have been determined for regio- and stereoselective substitution of the ring atoms. This review highlights the applications of the methodology for the diastereoselective synthesis of both natural and unnatural derivatives containing either the piperidine or the pyrrolidine ring as a substructure.Introduction The piperidine and pyrrolidine rings are an integral feature in the structure of a large number of alkaloids and synthetic products of biological interest. In the past ten years there has been an intense interest in the enantioselective synthesis of such heterocyclic compounds. Most of the synthetic efforts have involved a plethora of approaches specific for each target molecule whereas the development of general methodologies in which preformed chiral non-racemic building blocks are used for the construction of a wide diversity of simple or complex structures remained a challenging task. Fifteen years ago we felt that for the design of a new and general strategy one should be able to substitute each position of the heterocyclic rings.Simple and conjugated iminium and Professor H.-P. Husson was born in 1939. He received his Pharmacy degree in 1961 from Paris University and his PhD in Chemistry from the same University. He was a post-doctoral fellow with Professor E. Wenkert at Indiana University Bloomington. He spent the first part of his career in the Institut de Chimie des Substances Naturelles (CNRS) at Gif-sur-Yvette in Professor Potier’s Group as a Director of Research. In 1987 he was appointed Full Professor and Chairman of the Department of Medicinal Chemistry of the Faculty of Pharmacy in Paris (Université René Descartes). The main research interests of Professor Husson are the chemistry of natural products and the development of new chemical methods for enantioselective synthesis of biologically active molecules.Henri-Philippe Husson This journal is © The Royal Society of Chemistry 1999 corresponding enamine systems might be potentially valuable for this purpose. The idea was to take advantage of the reactivity of a-aminonitriles of type A bearing a chiral appendage which was supposed to induce the formation of new chiral centers (Scheme 1). Departure of CN2 gives rise to the prochiral Scheme 1 iminium ion B which on axial addition of a nucleophile R2 under stereoelectronic control leads in a stereospecific manner to C. Deprotonation of A affords the corresponding anion which can trap an electrophile reagent to give D. Departure of CN2 from D affords the prochiral iminium E which reacts with H2 as a nucleophile according to the same control as for its analogue B.Thus according to their mode of formation compounds F and Jacques Royer was born in 1947. He studied Chemistry at the University of Paris-XI (Orsay) where he received his PhD in the laboratory of Professor Michel Vilkas. After working as a postdoctoral fellow at Delalande Research Center (a French pharmaceutical company now Sanofi-Synth�elabo) he moved to the Institut de Chimie des Substances Naturelles (CNRS in Gifsur-Yvette) in the group of Professor H.-P. Husson. Since 1990 he has been Director of Research (CNRS). His main research interest is concerned with the asymmetric synthesis of nitrogen containing compounds alkaloids amino acids and amino alcohols.Jacques Royer 383 Chem. Soc. Rev. 1999 28 383–394 C are diastereomers since the newly created stereogenic centers have the opposite absolute configuration. Elimination of the chiral appendage would give enantiomeric amines. As far as pyrrolidine and piperidine rings are concerned it was necessary to obtain a multifunctional system allowing the substitution at the a,aA and b,bA positions of the nitrogen atom. For this purpose we designed the N-cyanomethyloxazolidine system where the nitrogen atom is an integral part of both aaminonitrile and a-aminoether functions which represents a synthetic equivalent of various synthons as illustrated in Scheme 2. The opening of the oxazolidine ring can generate Scheme 2 another iminium and regiospecific substitutions could therefore be envisaged.Condensation of a dialdehyde and a b-aminoalcohol in the presence of KCN furnished the N-cyanomethyloxazolidine system. The latter was first integrated into a piperidine structure to produce stable synthetic equivalents of piperidine synthons. Thus (2)-5-cyano-3-phenylhexahydro-5H-[1,3]oxazolo[ 3,2-a]pyridine 1 was easily prepared on the kilogram scale by treating glutaraldehyde with (R)-(2)-phenylglycinol and KCN (Scheme 3).1,2 Scheme 3 The choice of the chiral b-aminoalcohol was crucial since it is not a recoverable chiral auxiliary but a source of nitrogen and Scheme 4 Chem. Soc. Rev. 1999 28 383–394 384 chirality. Thus the chiral moiety must be easily cleavable as a nitrogen protective group.Phenylglycinol commercially available in both enantiomeric pure forms proved suitable because of the benzylic position of the amino group allowing facile hydrogenolysis. For this reason phenylglycinol was definitely more interesting than the norephedrines which were previously very popular in asymmetric synthesis. The simple structure of coniine the poisonous hemlock alkaloid was an attractive target for checking the validity and efficiency of our strategy and hence the absolute configuration of the first stereogenic center formed at the C-2 position of the piperidine ring. The synthetic scheme for the preparation of (S)-(+)-coniine (4) involved alkylation of 5-cyano-3-phenylhexahydro-5H- [1,3]oxazolo[3,2-a]pyridine 1 through LDA deprotonation followed by treatment with propyl bromide to give 2 in very high yield (Scheme 4).Reductive decyanation of 2 using NaBH4 in CH3OH furnished amino alcohol 3 with entire diastereoselectivity. Finally debenzylation allowed to obtain natural (S)-(+)-coniine (4).1 Conversely the propyl chain could be introduced at the C-2 position of 1 in the opposite (R) configuration with PrMgBr after prior complexation of the cyano group with silver salts. Compound 5 was obtained in low yield (25%) but with complete C-2 stereoselectivity. It will be shown below that this reaction occurred in much better yields (see Section 2.1.3) with several more complex structures. Reductive opening of the oxazolidine ring (NaBH4 CH3OH) afforded 6 which is in diastereomeric relationship with 3.Debenzylation of 6 led to enantiomeric (R)-(2)-coniine (4). 1 The high diastereoselectivity observed for the reduction or alkylation of the iminium salt generated from the aminonitrile function could be explained by an axial attack onto the preferred conformer of the iminium. This iminium intermediate formed upon addition of a nucleophile Nu2 (H2 or R2) could exist as the two conformers I1 or I2 (Fig. 1). The I1 conformer would be preferred since the bicyclic structure is much less constrained despite the A1,2 allylic strain present in such a conformer. Addition of the nucleophile reagent should occur from the axial direction (upper face) to give a chair transition state under the stereoelectronic control.The enantiodivergent synthesis of (R)- or (S)-coniine illustrates the mainspring of the CN(R,S) method which allows by CN elimination remarkable stereocontrol for the formation of the new R or S chiral centers. This result led us to name this method in recognition of the “Centre National de la Recherche Scientifique” which supported this work. With respect to the noteworthy stereoselectivity of the formation of the first chiral center at the C-2 potion of the Scheme 5 Fig. 1 piperidine ring it appeared promising to envisage further stereocontrolled functionalization. Indeed successive substitutions of three other carbon centers i.e. C-3 C-5 and C-6 proved possible if taking advantage of the capacity of the Ncyanomethyloxazolidine system to react as masked or potential iminium salts (or their corresponding enamines).Elimination of the chiral appendage permitted the generation of a secondary amine which could be incorporated into a new heterocycle. Finally the cyano group could be reduced hydrolyzed or involved in a spirocylization reaction. A similar synthetic strategy could be conducted for the pyrrolidine series using 5-cyano-3-phenylhexahydropyrrolo[ 2,1-b][1,3]oxazole 40 (see Section 2.2.1). However in this case the exploitation of the enamine reactivity was not possible due to the facile aromatization to a pyrrole. 2 Potential and masked iminium reactivity The use of potential and masked iminium salt reactivity of 1 was proved to be valuable in the synthesis of both enantiomers of coniine 4 (Scheme 4).A clear indication of this dual behaviour has been shown in various and more complex examples. 2.1 a-Substituted pyrrolidines and piperidines included in complex ring systems The access to indolizidines quinolizidines benzoquinolizidines and the unprecedented skeleton of tetraponerine alkaloids was possible using the reactivity of the a-aminonitrile and the nucleophilicity of the nitrogen atom after elimination of the chiral appendage. 2.1.1 Indolizidine3 and quinolizidine.4 The preparation of hydroxylated indolizidines was imagined following a scheme where the cyclization step is an intramolecular opening of an epoxide by the piperidine nitrogen. It was first shown that when the anion of 1 was reacted with an aldehyde a b-amino alcohol could be obtained in a diastereoselective manner.The synthesis of b-conhydrine 85 is an example of the usefulness of this reaction (Scheme 5). In a similar manner treatment of the anion of 1 with crotonaldehyde followed by NaBH4 reduction gave 9 (Scheme 6) as the major threo derivative. The latter was submitted to stereoselective epoxidation of the double bound. The major epoxide 10 after N-debenzylation was cyclized to the indolizidine skeleton and furnished 123 after O-deprotection. Scheme 6 An efficient synthesis of unnatural dihydroxyquinolizidine 16 from 1 was reported by McIntosh.4 Alkylation of the anion of 1 with iodobutanal ethylene ketal (Scheme 7) permitted the Scheme 7 preparation of enantiomerically pure (2S)-piperidine derivative 13 which was transformed to allylic alcohol 14 by classical reactions.Sharpless epoxidation of 14 in the presence of (+)-diethyl tartrate (DET) led to epoxide 15 which was cyclized to 16 upon treatment with sodium naphthalenide. 2.1.2 Benzoquinolizidine.6† A great deal of interest in azaanalogues of podophyllotoxin has developed on account of the antitumor activity of semi-synthetic derivatives of natural products. Building block 1 was the precursor for two syntheses of benzoquinolizidine analogues of podophyllotoxin. One of them will be described in detail (Scheme 8). Piperidine 17 was obtained from alkylation of the anion of 1 with piperonyl bromide followed by reduction and debenzylation. Upon treatment with 3,4,5-trimethoxybenzoyl chloride amide 18 could be obtained in good yield.Subsequent cyclization via a Bischler–Napieralski reaction and reduction of the iminium gave predominantly (10:1) the cis isomer 19. Once again the major isomer resulted from a stereoelectronically controlled reduction. † Benzoquinolizidine is 1,3,4,6,11,11a-hexahydro-2H-benzo[b]quinolizine. Chem. Soc. Rev. 1999 28 383–394 385 Scheme 8 2.1.3 Tetraponerines.7 The concept of the preparation of the a-cyanomethyloxazolidine and the rationale of the CN(R,S) method were both used in the asymmetric synthesis of the unprecedented skeleton of tetraponerines derived from either a piperidine or a pyrrolidine ring. Eight alkaloids T-1 to T-8 have been extracted from the defensive secretion of Tetraponera sp.a New-Guinea ant. All of them were prepared via the CN(R,S) method from the piperidine 1 or pyrrolidine 40 building blocks allowing the determination of the absolute configuration of the natural enantiomers. As an example the synthesis of epimeric alkaloids T-7 and T-8 will be described herein. The cyano aminal 22 appeared to be a suitable intermediate in Scheme 9 their synthesis (Scheme 9). This key intermediate was thought to be formed by a cross-condensation of two aminoaldehyde equivalents 21 and aminobutyraldehyde acetal. According to the established procedure the ketal 21 was prepared in 3 steps from 1 with complete stereocontrol. Acidic hydrolysis of 21 followed by condensation with commercially available aminobutyraldehyde acetal in the presence of KCN at slightly acidic pH gave the cyanoaminal 22 in high yield and as Chem.Soc. Rev. 1999 28 383–394 386 a sole isomer. This condensation parallels the preparation of 1 where a bis-aldehyde was condensed with an amino alcohol in the presence of CN2. As expected the cyanoaminal 22 had the configuration depicted in Scheme 9 which is in accordance with a thermodynamically controlled process. Introduction of the pentyl side chain onto cyanoaminal 22 could be controlled according to the CN(R,S) strategy. Axial addition of the Grignard reagent gave tetraponerine T-7 with (S) configuration in 97% de. On the other hand electrophilic substitution via LDA deprotonation followed by hydride axial attack furnished tetraponerine T-8 with (R) configuration and complete diastereoselectivity.2.1.4 Formation of a quaternary center. Alkylated aminonitriles such as 23 still possess two sites of potential iminium salt reactivity allowing the addition of nucleophiles. It was found that tert-butyldimethylsilyltrifluoromethane sulfonate promoted a clean elimination of CN2 despite possible competitive complexation of the silyl reagent with the oxazolidine oxygen atom. The probably more stable substituted iminium is formed since only the cyano group is antiperiplanar to the nitrogen lone pair. Thus sequential treatment of 23 with TBDMSOTf followed by a Grignard reagent led to the formation of oxazolidine 24 bearing a quaternary center at C-2 in highly diastereoselective fashion8 (Scheme 10).The diaster- Scheme 10 eoselectivity could be explained by the mechanism discussed above for the asymmetric synthesis of (+)- and (2)-coniine (Fig. 1). The diastereoselective formation of a quaternary center a to the nitrogen appeared as a valuable synthetic pathway it will be used in the synthesis of euphococcinine which will be described further below (see Section 2.2.2.3). 2.2 a,aA-Disubstituted pyrrolidines and piperidines 2.2.1 Simple derivatives. Simple a,aA-disubstituted pyrrolidines and piperidines can currently be synthesized by numerous routes of which some allow the control of optical activity. However the main problem is the control of the cis or trans relative relationship of the substituents. The first stereogenic center being created as R or S as desired at the C-2 position of the piperidine ring the diastereoselective formation of cis or trans 2,6-disubstituted derivatives could be envisaged.This was illustrated in the syntheses of cis and trans piperidine alkaloids. 26 by treatement with TMSCN which gave aminonitrile 29 in good yield (Scheme 12). Then alkylation of 29 led to a C-6 alkylated derivative which furnished the oxazolidine 30 by CN2 elimination and alcohol deprotection. Reductive (NaBH opening of the oxazolidine gave the trans derivative (trans/cis 70:30). During the reduction step the hydride ion enters on the same face of the iminium I6 (Fig. 2) as the Grignard reagent did in the previous synthesis now giving rise to the trans dialkylated product.After debenzylation pure (2S,6S)-(+)-solenopsin-A (31) 9 could be easily isolated. Conditions were found to selectively remove the cyano group of alkylated aminonitriles without opening of the oxazolidine ring. Treatment of 25 (Scheme 11) with AgBF4 followed by reduction with Zn(BH4)2 at 278 °C led to compound 26. The reduction was totally stereoselective and a 2S configuration was obtained. Addition of propyl Grignard reagent to the oxazolidine ring of 26 gave cis-dialkylpiperidine 27 (after separation of the 8:2 cis/trans mixture) from which (2S,6R)- (+)-dihydropinidine (28) was obtained after simple hydrogenolysis. By reversing the order of alkyl substitution i.e. propyl and then methyl optically pure (2)-dihydropinidine was easily prepared in a similar fashion.1 Once again it was possible to prepare both enantiomers starting from a common intermediate.The major formation of the cis isomer is explained by an axial 4. This 3 which exhibits a strong A1,2 strain attack (stereoelectronic control) onto the iminium I conformer is preferred to I (Fig. 2). A wide range of 2,6-disubstituted piperidines was obtained through the established reactivity of Grignard reagents with the hexahydro-5H[1,3]oxazolo[3,2-a]pyridine system. Interestingly some new results have been obtained concerning the diastereoselectivity of this reaction. Recently Higashiyama et al.10 described the reaction of C-2 substituted hexahydro- 5H[1,3]oxazolo[3,2-a]pyridine 26 and 34. The 2-methyl derivative 26 (Scheme 13) previously prepared by the CN(R,S) method afforded on reaction with vinyl or ethynyl magnesium bromide a majority of trans isomers 33.This behaviour is in contrast with the reactivity of simple alkyl Grignards that gave poor selectivity. The C-2 diastereomer 34 on the contrary gave exclusively 2,6 cis products. In this manner a totally diastereoselective synthesis of (2)-pinidine 36 has been achieved. The stereochemical outcome of the latter reaction has been explained by a strong A1,2 allylic strain as depicted in Fig. 2 for iminium I3. However the mechanism of the reaction affording only the trans isomer remains uncertain. Solenopsin-A a fire ant venom was an interesting target since it belongs to the trans-2,6-dialkylated piperidine series generally more difficult to synthesize.In order to obtain such derivatives we took advantage of the mechanism of formation of the cis compound in the previous synthesis (Fig. 2). Indeed a cyano group was introduced at the C-6 position of oxazolidine Scheme 11 Fig. 2 Scheme 12 4) 387 Chem. Soc. Rev. 1999 28 383–394 Scheme 13 Scheme 14 Simple 1,3-oxazolidines derived from phenylglycinol have served as starting materials for the construction of phenyloxazolipiperidines which underwent well established regio- and stereospecific chemistry. For their synthesis of (2)-desoxoprosopinine Agami et al.11 have elaborated a diastereoselective preparation through N-Boc oxazolidine derivatives of the 5-cyano-3-phenylhexahydro-5H-[1,3]oxazolo- [3,2-a]pyridine 37.The latter was submitted to CN(R,S) manipulations to create the last two stereogenic centers with excellent selectivity (Scheme 14). A similar synthetic strategy could be conducted for the pyrrolidine series using 5-cyano-3-phenylhexahydropyrrolo[ 2,1-b][1,3]oxaxole 40 which is analogous to 1. The preparation of 4012 (Scheme 15) parallels the preparation of 1 Scheme 15 using dimethoxytetrahydrofuran as a stable equivalent of succinaldehyde. Compound 40 was isolated as a mixture of two epimers at C-2 which did not influence the diastereoselectivity of the alkylation reactions at all as will be shown below. The anion of 5-cyano-3-phenylhexahydropyrrolo[2,1- b][1,3]oxazole 40 was alkylated to give aminonitrile 41 as a 1:1 mixture of diastereomers (Scheme 16).Nevertheless the reductive decyanation produced oxazolidine 42 in 60% yield as a unique compound. The decyanation was achieved by an alternative method using Li in liquid ammonia. Subsequent reaction with a Grignard reagent in ether led to the trans dialkyl derivative 43 being obtained as the major isomer (trans/cis 7:3). In the pyrrolidine series the trans stereoselectivity could be explained by the addition of the Grignard reagent on the less hindered face of the iminium salt formed by the opening of the oxazolidine ring. The asymmetric synthesis of (2S,5S)- (+)-trans-2-ethyl-5-heptylpyrrolidine 44 a component of the ant Solenopsis punctaticeps has been achieved following this strategy (Scheme 16).13 Chem. Soc. Rev. 1999 28 383–394 388 Scheme 16 By just replacing KCN with benzotriazole in the condensation of (S)-(+)-phenyglycinol with dialdehydes Katritsky14 obtained compounds 45 and 46 which are very similar to the aminonitrile series (Scheme 17). Interestingly in the contrast to Scheme 17 the latter series the piperidine derivative 46 was obtained as a mixture of diastereomers while the pyrrolidine homologue 45 was isolated as a single more stable trans isomer. The benzotriazole moiety was easily substituted by different nucleophiles in a highly diastereoselective manner for both the piperidine and the pyrrolidine series. This approach is limited to this type of reaction since compounds 45 and 46 only acted as iminium equivalents.Nevertheless monoalkyl and cis-dialkyl substituted piperidines and pyrrolidines were obtained using the same methodology as for the aminonitrile series. 2.2.2 Bicyclic systems. The synthesis of a,aA-disubstituted pyrrolidines and piperidines was extended to bicyclic systems Scheme 18 Scheme 19 following the possibility of cyclization of the a-side chain onto the nitrogen atom (octahydroindolizines octahydro-2H-quinolizines etc.) or at the aA position (tropane and related structures). 2.2.2.1 Indolizidines.‡ This family of alkaloids is very large and it has been shown with tens of isolated products that some of its members occur as trace amounts in natural sources. In some cases their structures have been determined only by racemic syntheses.Thus chiral approaches are needed for the determination of the absolute configuration. (2)-Monomorine-I (49) is an ant trail pheromone from Monomorium pharaonis. Only the relative configuration of this indolizidine alkaloid was known until the first asymmetric synthesis15 of the levorotatory enantiomer was achieved in only four steps starting from (2)-1. Introduction of an alkyl chain bearing a protected ketone onto aminonitrile 1 gave oxazolidine 47 (Scheme 18). Treatment of 47 with a methyl Grignard reagent in ether led to the major cis derivative 48 which was easily separated from the diastereomeric mixture (cis/trans 4:1) by means of chromatography and subjected to hydrogenation in an acidic medium. These one-pot conditions allowed the cleavage of the ketal hydrogenolysis of the chiral appendage ‡ Indolizidines are octahydroindolizines.and reduction of the cyclized iminium with the expected cis stereochemistry. (3S,5R,8aR)-(2)-Monomorine 49 was obtained after chromatographic separation from the epimeric material. This synthesis permitted determination of the absolute configuration of natural (+)-monomorine-I as 3R,5S,8aS. Obviously the use of (+)-1 (available from (S)-(+)-phenylglycinol) as starting material or reversal of the order of introduction of substituents would have led to the natural enantiomer. (2)-Gephyrotoxin-223AB extracted from the skin of neotropical frogs exhibits an indolizidine structure 53 analogous to that of monomorine but with the opposite configuration at C-3.It was anticipated to introduce the butyl side chain at this position via an a-aminonitrile function. The preparation of 51 with a cis configuration (Scheme 18) was performed through the same sequence as for the previous synthesis. This compound was sequentially debenzylated in neutral conditions and treated with KCN in the acidic medium to give the aminonitrile 52 in excellent yield. Alkylation of 52 with butylmagnesium bromide in ether at low temperature gave (3R,5R,8aR)-(2)-gephyrotoxin-223AB (53)16 as the major isomer. Once again the addition of the Grignard reagent onto the iminium intermediate took place on the same face as the addition of hydrogen in the synthesis of monomorine-I giving rise to the expected R configuration at C-3.389 Chem. Soc. Rev. 1999 28 383–394 Scheme 20 Scheme 21 Using a derived strategy Grierson17 reported an interesting preparation of indolizidine alkaloids (Scheme 19). Aminonitrile 54 was prepared through the CN(R,S) strategy in good yield and transformed to 56 which was obtained as a unique isomer. An elegant and original ring contraction was elaborated to furnish the indolizidine skeleton. Compound 56 underwent a crucial ring opening with diethylcyanophosphonate to give 57. Finally the anion generated a to the nitrogen atom of 57 by decyanation under dissolving metal conditions led to an unprecedented ring cyclization giving diastereomeric indolizidines 58 and 59. This approach has been carried out only in the racemic series and deserves to be more widely exploited.2.2.2.2 Pyrrolizidine.§ The synthesis of simple alkyl pyrrolizidines is quite similar to those of monomorine-I or gephyrotoxin-223AB. The preparation of the ant venom (+)-xenovenine 63a18 was achieved using the same pathway (Scheme 20). After alkylation and complete diastereoselective decyanation oxazolidine 61 was alkylated with a Grignard reagent to give the trans pyrrolidine 62a (trans/cis 77:23). Reductive cyclization gave 63a in high yield. The absolute configuration of synthetic 63a was deduced from the synthetic scheme but was in disagreement with data reported in the literature. This discrep- § Pyrrolizidine is hexahydro-1H-pyrrolizine. Chem. Soc. Rev. 1999 28 383–394 390 ancy was eventually lifted by another research group which confirmed our results.2.2.2.3 Tropanes and homotropanes. To achieve the formation of indolizidine pyrrolizidine and related skeletons an aside chain was cyclized onto the nitrogen atom of the piperidine or pyrrolidine ring. It was anticipated that this cyclization could also occur onto the potential iminium salt of the oxazolidine function to furnish bicylic derivatives of the tropane and homotropane series. Euphococcinine 67 which possesses a homotropane skeleton has been extracted from the plant Euphorbia atoto as well as from the defense secretion of the ladybugs Cryptolaenus montrouzieri and Epilachna varisvetis. The synthesis of the bicyclic system was planned through the intramolecular Mannich reaction of 65 which could be obtained by nucleophilic alkylation of aminonitrile 64 (Scheme 21).The anion of 1 was treated with 3-bromo-2-methoxyprop-1-ene to give 64. The alkylated compound was then reacted with TBDMSOTf to promote the elimination of the nitrile followed by a methyl Grignard addition and PPTS treatment to give the bicyclic product 66. The final steps consisted of hydrolysis of the ketal function and hydrogenolysis of the chiral appendage which were achieved in a one-pot reaction to give (2)-euphococcinine 67.19 The elaboration and control of the R absolute configuration at the C-2 quaternary center was obtained by the Scheme 22 entirely diastereoselective addition of the Grignard reagent onto the iminium derived from aminonitrile 64.The tropane alkaloid ferruginine 72 was isolated from the arboreal species Darlingiana ferruginea and D. darlingiana. Interesting nicotinic agonist activity was reported for this type of compound. It was anticipated that starting from 40 a Mannich reaction according to a 6-endo-trig cyclization would lead to the bicyclic system of ferruginine. In the published syntheses of this alkaloid introduction of the double bond in the latter steps appeared as a tedious task. This problem was overcome by performing an intramolecular Mannich reaction using an a,bunsaturated ketone. Thus after alkylation of the anion of 40 with bromoacetaldehyde diethyl ketal and decyanation using Li in liquid ammonia a Wittig reaction permitted the preparation of a,b-unsaturated aldehyde 70 (Scheme 22).This latter product was easily cyclized with H2SO4 in methanol to give methoxylated bicyclic derivative 71. This compound was debenzylated and methylated in a one-pot reaction to 72 using H2 with Pd/C in the presence of an aqueous formaldehyde solution. Acidic treatment of 72 in benzene achieved this short (6 steps) and efficient (20% overall yield) synthesis of (+)-ferruginine 73. This strategy starting from commercially available materials compares favorably to previously described synthetic routes.20 3 Enamine reactivity Simple a,b-unsaturated piperidines are notoriously unstable and the presence of electron-withdrawing groups at the C-3 position is often a requirement for their synthetic exploitation.It will be shown that building blocks 1 and 40 have proved to be stable and reactive forms of the parent enamine. 3.1 Electrophilic substitution The potential enamine reactivity of 1 was used in an intramolecular cyclization to form the perhydroquinoline skeleton of pumiliotoxin-C,21 an alkaloid present in the skin secretion of the neotropical frogs the Dendrobatidae. It would appear that the synthesis of the natural enantiomer needs to start with enantiomeric (+)-1 obtained from (S)-(+)-phenylglycinol. The anion of (+)-1 was easily alkylated to give compound 74 bearing a pentanone side chain (Scheme 23). Ketone 74 gave the cyclized compound 75 upon simple treatment with alumina. In this transformation the departure of CN2 assisted on alumina induced the formation of the iminium salt and then the corresponding enamine which reacted with the ketone to give an a,b-unsaturated iminium salt on which 1,4-addition of cyanide afforded 75.The oxazolidine function of 75 allowed the addition of the propyl side chain. The last two steps consisted of reductive removal of the cyano group (Na/NH3) and hydrogenolysis (H2 Pd/C). A 7:3 mixture of (+)-2R,5R-transdecahydroquinoline 78a and (2)-pumiliotoxine-C 78b was then obtained in a 95% combined yield. Among the numerous syntheses of this alkaloid this scheme represents the second successful synthesis of the natural enantiomer. 3.2 Electrochemical halogenation22 In order to further develop the enamine reactivity we planned to oxidize one of the potential enamine functions of 1 and 40 by electrochemical means.It was found that anodic oxidation of 1 and 40 in the presence of Cl2 and Br2 as an oxidative mediator gave respectively the chlorinated derivatives 79 and 80 in high yields (Scheme 24). It is noteworthy that oxidation exclusively occurred at the C-5 position of piperidine and C-4 of pyrrolidine. Moreover in both series the configuration of the Ncyanomethyloxazolidine system remained unchanged. Nevertheless it has also been shown that it was a specific reaction of the latter system since under these conditions oxazolidine 82 did not afford the corresponding di-halo substitution. If the Cl2 source was omitted the reaction occurred with 3CN or of lactam formation of dibromo compound 81 in dry CH 83 in the presence of water.These two latter compounds are useful starting materials for the synthesis of polyfunctionalized piperidines and are currently under investigation.23 4 Cyano group reactivity As mentioned above the a-aminonitrile function of 5-cyano- 3-phenylhexahydro-5H[1,3]oxazolo[3,2-a]pyridine 1 and its derivatives can be used as an iminium equivalent by addition of a nucleophilic reagent a Grignard or a hydride. In specific cases additions can occur onto the CN group depending upon the experimental conditions. Chem. Soc. Rev. 1999 28 383–394 391 Scheme 23 4.1 Nucleophilic addition24 When 1 was reacted with BuLi in ether bicyclic imine 85 was obtained as a rearranged product from imine 84 (Scheme 25). The NaBH4 reduction of 85 followed by hydrogenolysis gave the diamine 86 as a single diastereomer.4.2 Reduction24 The LiAlH4 treatment of 1 in ether led to the reduction of the cyano group as well as the reductive opening of the oxazolidine to furnish the amino alcohol 87a in high yield (Scheme 25). 1,2-Diamine 88a was easily obtained after hydrogenolysis. The same strategy applied to the C-2 alkylated compound 25 afforded ((2S)-2-methylpiperidin-2-yl)methanamine 88b difficult to obtain otherwise. 4.3 Solvolysis Scheme 24 Compound 1 appeared as a good precursor of (2)-pipecolic acid 92 an important natural but non-proteinogenic amino acid for the preparation of which only few asymmetric methods are known. However the solvolysis of 1 was not straightforward; indeed adsorption of a toluene solution of 1 on silica gel was necessary Scheme 25 Chem.Soc. Rev. 1999 28 383–394 392 Scheme 26 prior to a dry HCl treatment in ethanol. These conditions allowed ester 89 to be isolated in excellent yield (95%) but with some epimerization at C-2 (Scheme 26). Bicyclic lactone 90 obtained upon reduction of the oxazolidine ring of ester 89 was expected to allow a controlled epimerization at C-2 through a deprotonation–reprotonation sequence. Treatment of 90 (de 50–60%) by LDA in THF at 278 °C followed by quenching the enolate with AcOH gave 91 in 96% de raised to 100% after a single recrystallisation. Pure (S)-(2)-pipecolic acid (92)25 was obtained by hydrogenolysis of 91 in the presence of AcOH.The phosphonic analogue 93 of pipecolic acid was also obtained starting from 1 by using an Arbusov reaction as the key step.26 The abovementioned intermediates in the pipecolic acid synthesis also enabled access to 2- and 6-substituted pipecolic acids according to previously established chemistry (Scheme 27). Recently 3-substituted derivatives were also obtained from Scheme 27 a,b-unsaturated pipecolic intermediates.27 The anion of lactone 94 quenched with an alkyl halide gave rise to 2-alkylated amino acids 95 in both high yield and diastereomeric excess. On the other hand the oxazolidine function of ester 96 after activation with BF3·O(C2H5)2 could be alkylated with a Grignard reagent. The 6-alkylated lactones 97 were then obtained in excellent yield.The diastereoselectivity was not as good as for the 2-alkylated products since the starting material consisted of epimeric mixtures. However if purified (2S)-ester 96 was used cis 6-alkylated compounds 97 were formed with de ranging from 70 to 98%.25 This synthetic strategy was recently exploited by Zabriskie et al.28 for the diastereoselective preparation of pipecolic acid derivatives deuterated at C-6 necessary for the study of the stereochemical course of their enzymatic oxidation in mammalians. For instance 96 was sequentially treated with BF3·O(C2H5)2 and NaBD4 to furnish the deuterated lactone precursor of stereoselectively deuterated pipecolic acid 98 (Scheme 27). 4.4 Formation of a spiro center29 Natural histrionicotoxin extracted from skins of the Columbian poison frog Dendrobates histrionicus and its analogues are active in neuromuscular transmission.The synthesis of a depentyl analogue of this alkaloid afforded an example for the formation of a spiro carbon center a to the nitrogen of the piperidine ring. The novelty of this synthesis is the use of the cyano group as an integral feature of the target molecule. Introduction of the propanal chain protected as a ketal was conducted as usual and followed by addition of a methyl Grignard in ether to furnish imine 100 in high yield (Scheme 28). Several chemical transformations led to 101 precursor of Scheme 28 spiro derivative 102 through an aldol reaction. The a,bunsaturated system permitted the introduction of the butyl chain and the hydroxy group in the required configuration.5 Conclusion Among the great number of synthetic routes for pyrrolidines and piperidines only two strategies allow a large variety of substitution patterns and enantioselectivity the substitution of chiral bicyclic lactams developed by Meyers;30 and the CN(R,S) method based on the reactivity of the N-cyanomethyloxazolidine system. It should be mentionned that Comins31 has extended his method of functionalization of 1-acylpyridinium salts to the corresponding chiral series allowing the efficient synthesis of a large variety of piperidine alkaloids. This method is based on 393 Chem. Soc. Rev. 1999 28 383–394 the use of a series of chiral auxiliaries whereas a b-amino alcohol derived from the chiral pool is the source of both nitrogen and chirality in the two others.A good way to compare the respective advantages of each strategy in the piperidine series is to examine their potential as synthetic equivalent i.e. cationic or anionic synthons (Fig. 3). Fig. 3 An interesting characteristic of the pyridinium salt is a facile functionalization at C-4 after the first nucleophilic attack at C-2 generating a ketone. The chiral bicyclic piperidine and pyrrolidine lactams have been successfully employed for the asymmetric construction of quaternary centers in carbocyclic series. They have more recently received applications for the synthesis of alkaloids although their potential has not been extensively exploited.It should be noted that the C-4 position of the piperidine ring can also be substituted by alkylation of a-cyano enamines derived from bicyclic lactams.32 As far as the CN(R,S) method is concerned the presence of a cyano group offers additional synthetic possibilities. Indeed the a-aminonitrile function includes both potential and masked iminium reactivities. Furthermore the cyano group itself is a precursor for carbonyl and methylamino functions. The challenge of functionalization of the C-4 position of piperidines could be possible by bromination at C-5 followed by elimination providing a potential conjugated iminium system. Thus substitution of any or all of the ring atoms becomes possible. Interestingly one might consider that most of the syntheses of alkaloids achieved according to the CN(R,S) method probably follow part of a biomimetic route that is to say reproduce a key step of what is occurring in nature.The efficiency and the shortness of some syntheses carried out might confirm the likelihood of this hypothesis. In many examples it has been shown that complicated things can be simple according to this strategy. Chem. Soc. Rev. 1999 28 383–394 394 6 Acknowledgements The authors are pleased to thank all their colleagues collaborators and students whose names are listed along the references. Their intellectual and practical participation was outstanding for the development of the CN(R,S) method. 7 References 1 L. Guerrier J. Royer D. S. Grierson and H.-P. Husson J.Am. Chem. 2 M. Bonin D. S. Grierson J. Royer and H.-P. Husson Org. Synth. 1992 3 V. Ratovelomanana L. Vidal J. Royer and H.-P. Husson Heterocycles Soc. 1983 105 7754. 70 54. 1991 32 879. 4 J. M. McIntosh and L. C. Matassa J. Org. Chem. 1988 53 4452. 5 V. Ratovelomanana J. Royer and H.-P. Husson Tetrahedron Lett. 6 P. Lienard J. Royer J.-C. Quirion and H.-P. Husson Tetrahedron Lett. 7 C. Yue I. Gauthier J. Royer and H.-P. Husson J. Org. Chem. 1996 61 1985 26 3803. 1991 32 2489. 4949. 8 C. Yue J. Royer and H.-P. Husson unpublished results. 9 D. S. Grierson J. Royer L. Guerrier and H.-P. Husson J. Org. Chem. 1986 51 4475. 10 H. Poerwono K. Higashiyama T. Yamauchi H. Kubo S. Ohmiya and H. Takahashi Tetrahedron 1998 54 13955. 11 C.Agami F. Couty and H. Mathieu Tetrahedron Lett. 1998 39 3505. 12 J. Royer and H.-P. Husson Tetrahedron Lett. 1987 28 6175. 13 S. Arseniyadis P. Q. Huang and H.-P. Husson Tetrahedron 1988 44 2457. 14 A. R. Katrizky G. Qiu B. Yang and P. J. Steel J. Org. Chem. 1998 63 6699; A. R. Katrizky X.-L. Cui B. Yang and P. J. Steel J. Org. Chem. 1999 64 1979. 15 J. Royer and H.-P. Husson J. Org.Chem. 1985 50 670. 16 J. Royer and H.-P. Husson Tetrahedron Lett. 1985 26 1515. 17 E. Zeller and D. S. Grierson Heterocycles 1988 27 1575. 18 S. Arseniyadis P. Q. Huang and H.-P. Husson Tetrahedron Lett. 1988 29 1391. 19 C. Yue J. Royer and H.-P. Husson J. Org. Chem. 1992 57 4211. 20 I. Gauthier J. Royer and H.-P. Husson J. Org. Chem. 1997 62 6704. 21 M. Bonin J. Royer D. S. Grierson and H.-P. Husson Tetrahedron Lett. 1986 27 1569. 22 F. Billon-Souquet T. Martens and J. Royer Tetrahedron 1996 52 15127. 23 F. Billon-Souquet T. Martens and J. Royer Tetrahedron Lett. 1999 40 3731. 24 O. Froelich P. Desos M. Bonin J.-C. Quirion and H.-P. Husson J. Org. Chem. 1996 61 6700. 25 J.-F. Berrien J. Royer and H.-P. Husson J. Org. Chem. 1994 59 3769. 26 C. Maury Q. Wang T. Gharbaoui M. Chiadmi A. Tomas J. Royer and H.-P. Husson Tetrahedron 1997 53 3627. 27 A. Zaparucha M. Danjoux A. Chiaroni J. Royer and H.-P. Husson Tetrahedron Lett. 1999 40 3699. 28 T. M. Zabriskie W. L. Kelly and Xi Liang J. Am. Chem. Soc. 1997 119 6446. 29 J. Zhu J. Royer J.-C. Quirion and H.-P. Husson Tetrahedron Lett. 1991 32 2485. 30 A. I. Meyers and G. P. Brengel Chem. Commun. 1997 1. 31 D. L. Comins and S. P. Joseph in Advances in Nitrogen Heterocycles ed. C. J. Moody vol. 2 pp. 251–294 JAI Press Inc. 1996. 32 J. B. Schwarz and A. I. Meyers J. Org. Chem. 1998 63 1619. Review 9/00153K

ISSN:0306-0012    

DOI:10.1039/a900153k

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Biotechnology for the production of commodity chemicals from biomass Herbert Danner and Rudolf Braun Department for Environmental Biotechnology IFA-Tulln Konrad Lorenz Str. 20 A-3430 Tulln Austria. E-mail danner@ifa-tulln.ac.at; Fax (43-2272)66280-503 Received 11th March 1999 Current developments especially in fermentation technologies membrane technologies and genetic manipulation open new possibilities for the biotechnological production of market relevant chemicals from renewable resources. This article reviews possible fermentation strategies actual research activities and future research demands to enhance the use of renewable resources as a source of chemicals. Special focus is put on commodity chemicals which have a considerable market volume or which are believed to be key chemicals of the future.1 The starting point Up to the 19th century mankind used renewable resources not only for food but also for functional applications. At the turn of the century the situation changed completely as coal became the key raw material for the production of coke ammonia tar and gas. Finally the developments and improvements in the area of fossil oil processing resulted in today’s chemical industry mainly based on oil and natural gas. Nowadays more than 2500 different oil-based products are on the market. Crude oil represents the major source for plastics fibers and colours. At present 10% of world-wide natural gas consumption 21% of the natural gas liquids and 4% of crude oil are used for chemicals.On a combined basis the chemical industry requires about 7–8% of the total consumed liquid and gaseous hydrocarbons. Although this value seems to be quite low it still represents a total market volume of more than 215 billion US dollars.1 Fig. 1 shows the most important low carbon organic chemicals by volume manufactured worldwide. Fossil resources mainly oil and natural gas are converted into key chemicals like ethylene propylene or the C4-fraction which are Herbert Danner born 1966 graduated from the University of Agricultural Sciences Vienna Austria. He is now working at IFA-Tulln which is part of the University of Agricultural Sciences and is operating a fermentation pilot plant. His research interest is the application of biotechnology for the production of chemicals and food additives.Herbert Danner This journal is © The Royal Society of Chemistry 1999 Fig. 1 Sources and production routes of the most important organic C1–C4- chemicals (Mio = 106). intermediates for the synthesis of the respective chemicals. Ethylene represents the organic chemical consumed in largest quantities worldwide. The total volume of ethylene based chemicals and ethylene-based polymers amounted to about 79 Mio tons in 1995 and this is expected to increase by 5.4% annually. Propylene serves as a raw material for the production of C3-chemicals with considerable market volumes as indicated Rudolf Braun born 1947 graduated from the University of Agricultural Sciences Vienna Austria.He is now head of the Environmental Biotechnology Department at the IFA-Tulln in Tulln. Since 1997 he has been an Associate Professor. His research interest is the degradation of organic waste materials and methane fermentation. Rudolf Braun 395 Chem. Soc. Rev. 1999 28 395–405 Fig. 2 Biorefinery industry based on renewable resources versus petroleum based industry. in Fig. 1. The third major raw material methanol is derived from natural gas or from synthetic gas. Methanol is used as fuel blends (as methyl tert-butyl ether MTBE or in various mixing ratios with conventional petroleum products) or for the chemical synthesis of various C1 C2 and C3 compounds like formaldeyhde acetaldehyde acetic acid or glycol. Some of these C1–C4-chemicals can be synthesized from renewable resources like starch cellulose or other carbohydrates such as sugar from sugar cane or sugar beet.The petroleum crisis of the seventies resulted in a shift from total reliance on fossil resources and simultaneously triggered research into biomass based technologies. Renewable resources became a popular phrase and as a result of the oil crisis countries like Brasil and the United States initiated national ethanol programmes to partly substitute fossil fuels (Proalcool and Gasohol programmes respectively). Due to the decrease in oil prices in the eighties most of the efforts for the production of substitutes for petrochemicals were given up. Nowadays agricultural surpluses in Europe and worldwide efforts to reduce atmospheric CO2 emmissions are the major driving forces for the implementation of renewable resource based technologies.The use of biomass as a source of energy and chemicals enables closed cycle material fluxes as indicated in Fig. 2 The objective of this article is to summarize the actual ongoing activities in the field of biotechnology for the production of bulk chemicals. Special focus is put on commercial processes and on genetical strain improvement to broaden substrate and product ranges of microorganisms. 2 State of the art Today only a small number of chemicals are produced from renewable resources via fermentation. In Europe the biotechnological production of lactic acid acetic acid and ethanol are the only processes which are currently applied in technical scales and which can compete with petrochemical routes.While the fermentation routes are limited by pathways of the microorganisms together with chemical synthesis a wide range of chemicals can be produced as indicated in Fig. 3. 2.1 Present industrial fermentations 2.1.1 Ethanol from carbohydrates. Ethanol represents the highest volume organic chemical produced predominantly via Chem. Soc. Rev. 1999 28 395–405 396 Fig. 3 Chemicals currently produced by fermentation from carbohydrates. fermentation. While Brasil is producing all its ethanol (15.4 billion liters per annum) from sugar cane the extent of fermentation ethanol in the United States amounts to 94% (5.3 billion liters per annum) based on starch hydrolysates and sucrose from sugar beet.In Europe synthetic ethanol still has a considerable market and fermentation ethanol accounts for 60% (1.5 billion liters per annum) of total ethanol produced. In Brasil approximately 40% of the automobiles run on pure ethanol fuel. The remainder utilize a blend of 22% ethanol/78% gasoline. In the United States most of the ethanol is applied as fuel blend. However the use of ethanol in high percentages such as E85 (85% ethanol 15% gasoline) remains limited. In european countries like France the production of ETBE (ethyl tert-butyl ether) as a substitute of MTBE as a fuel additive is more favourable than direct blending with gasoline. Spain Sweden and the Netherlands envisage national programmes for the production of bio-ethanol-ETBE.Traditionally ethanol fermentation is based on the conversion of sucrose from sugar cane and sugar beet or of glucose from corn or cereal starch hydrolysates by yeasts mainly Saccharomyces cerevisiae. One glucose molecule is degraded to two molecules of carbon dioxide and two molecules of ethanol. This results in a weight loss of approximately 50% as carbon dioxide which represents less than 5% of the energy content. Product recovery is done by distillation. Steam requirement has been reduced significantly during the past years resulting in steam consumption of less than 0.1 kg per kg of ethanol. 2.1.2 Acetic acid from carbohydrates. At present most of the demand for technical acetic acid is met synthetically which seemed to be the most economic way in the last 100 years.The fermentation by a species of Acetobacter which converts ethanol to acetic acid with final concentrations of a small percentage (4–6%) has been used almost exclusively for vinegar production. Because of the loss of one carbon in the form of CO2 in the glucose–ethanol fermentation the theoretical maximum yield of the whole reaction is 2 moles of acetic acid from one mole of glucose or 0.67 g acetic acid per gram of glucose. In commercial practice the actual yield is 0.50–0.55 g acetic acid per gram of glucose or roughly 75–80% of the theoretical yield. 2.1.3 Lactic acid from carbohydrates. Lactic acid represents a chemical with a small world market volume of 54500–59000 tonnes per annum.While the market for traditional applications of lactic acid is estimated to be growing at about 3–5% annually new products based on lactic acid may increase the world market share significantly. New applications for lactic acid include the use of derivatives such as ethyl esters to replace hazardous solvents like chlorinated hydrocarbon solvents in certain industrial applications. Furthermore lactic acid may be polymerized to biodegradable plastics as demonstrated by Danone Inc. in the form of yoghurt cups. Industrial production of lactic acid is based on the conversion of pure sugar like glucose or sucrose with bacteria of the genus Lactobacillus at temperatures around 40–45 °C under anoxic conditions. One mole of glucose results in almost two moles of lactic acid.Unlike in ethanol fermentations the recovery process for lactic acid is much more sophisticated involving various precipitation chromatographic and/or distillation steps. 2.1.4 Acetone–butanol–ethanol fermentation. An acetone– butanol–ethanol (ABE) blend (ratio 3+6+1) may serve as an excellent car fuel which can be easily mixed not only with Fig. 4 Production of chemicals from fermentation ethanol in India. ‘super’ fuels but also with diesel. ABE as a fuel additive has the advantage of a similar heat of combustion to hydrocarbons and perfect miscibility with hydrocarbons even when water is present. The fermentative production of ABE used to be the second largest industrial fermentation after ethanol production.Especially during the World Wars acetone production increased while butanol was considered to be an undesirable byproduct. In 1945 60% of butanol demand of the United States was produced via fermentation. Due to the shortage of raw materials namely corn and molasses and to decreasing prices of oil all plants were closed in the years that followed. In 1981 the last plant in the world closed down in South Africa. The traditional fermentative production of acetone–butanol– ethanol is a batch fermentation with Clostridia a strictly anaerobic bacteria. The substrate consists of molasses and phosphate and nitrogen sources. The substrate fomulation is sterilized prior to fermentation. The fermentation itself takes place under strictly anaerobic conditions and lasts 40–45 h followed by product separation by distillation.Solvent yields based on the fermentable sugars were usually around 29 to 33%. Instead of molasses other sugar sources like maize mash or sugar from polysaccharide type plant material may serve as a raw material for fermentation.2 2.2 Products derived from fermentation ethanol In the seventies ethanol was used to act as a basic building block in the organic chemical industry when chemicals like ethylene and acetaldehyde were synthesized from fermentation ethanol. Currently no ethylene plant based on this technology is in operation but in a few developing countries some other ethanol based technologies are still applied mainly because of relatively cheap ethanol and high-cost petroleum.In India for instance more than 20 companies are producing not only ethanol from sugar cane or molasses but also chemicals like acetic acid acetic anhydride or ethyl acetate from fermentation ethanol. Fig. 4 gives an example of the two most versatile organic companies Somaiya Organo Chemicals Ltd and VAM Organic Chemicals Ltd. Other enterprises are mainly focusing on the production of acetic acid and ethyl acetate. These 397 Chem. Soc. Rev. 1999 28 395–405 activities meet 100% of the Indian demand for acetic anhydride and ethyl acetate and almost 80% of the acetic acid demand.3 Production in Brasil has been decreasing during the last years leaving Cloroetil Solventes Acéticos S.A. as the only bioethanol based chemical company focusing on acetaldehyde acetic acid and ethyl acetate.2.2.1 Acetaldehyde and derivatives. Acetaldehyde is obtained from ethanol by catalytic dehydration or catalytic oxidation. The dehydration process involves Cr/Cu/Si-catalysts at temperatures of 350 °C resulting in acetaldehyde and hydrogen which may be used as hydrating agents.4 The oxidative process which is still applied in India involves silver catalysts at 550 °C. Yields of acetaldehyde from ethanol for both processes vary from 87 to 92% (w/w). Aldol condensation can lead to butanediol which is an important intermediate for polyesters. However no plant could be found applying this process on a large scale today with the exception of crotonaldehyde (but-2-enal) production by aldol condensation.Somaiya Organo Chemicals Ltd. India is converting acetaldehyde to aldol which is then distilled with acetic acid as catalyst to produce crotonaldehyde. Further conversion to n-butanol with hydrogen from ethanol dehydration was performed until a few months ago when rising substrate costs (molasses) altered the economic feasibility of the process and resulted in a shut down of the production. To produce 100 kg of butanol 145 kg acetaldehyde were required (79.8% of the theoretical yield).5 The cyclic trimer of acetaldehyde paraldehyde is still produced with mineral acids as catalysts at temperatures of 5–6 °C.5 Butadiene which is almost entirely used in rubbers like styrene–butadiene rubber for the auto tyre industry may be manufactured from ethanol and acetaldehyde at 310 °C with tantalum oxide on silica gel.Union Carbide Corporation was operating plants using this process during World War II. Until the late seventies India Brasil and Russia were using this process but today no plant could be found which is still in operation. 2.2.2 Acetic acid and derivatives. Acetic acid is obtained by oxidation of acetaldehyde. This oxidation is performed at moderate temperatures (50–60 °C) in the presence of catalysts like Mn/Co or mixtures of KMnO4 and Mn(CH3COO)2. Yields are around 97% of the theoretical yield. The overall yield of acetic acid based on consumed sugar is 49–52 kg of acetic acid per 100 kg of glucose which represents 74–78% of the theoretical yield (2 moles acetic acid per mole of glucose).Acetic anhydride can be obtained by the reaction of acetic acid at 700 °C under vacuum or by synthesis of acetic acid and acetic anhydride from acetaldehyde at 5 bar with cobalt(ii) acetate at moderate temperature (55 °C). Today both processes are still applied in India by VAM Organic Chemicals Ltd. and The Dhampur Sugar Mills Ltd respectively (Fig. 5). One major application of acetic anhydride is the production of cellulose acetate for textile yarns or cigarette filters. Fig. 5 Synthesis of acetic anhydride. To produce 1000 kg of acetic anhydride 1260 kg of acetic acid are required. The direct synthesis from 100 kg acetaldehyde gives practical yields of 50 kg acetic acid and 50 kg acetic anhydride at theoretical yields of 58 and 68.2 kg.Vinyl acetate monomer (VAM) is an intermediate in poly(vinyl acetate) production and therefore has a considerable Chem. Soc. Rev. 1999 28 395–405 398 market in the polymer industry (resins and lattices). In India 10000 tonnes per annum are produced from fermentation ethanol by VAM Organic Chemicals Ltd. Acetaldehyde and acetic anhydride are converted at 90 °C with toluene-p-sulfonic acid as catalyst to vinyl acetate monomer and acetic acid as byproduct. Separation of products is achieved by distillation. 600 kg acetaldehyde and 1400 kg acetic anhydride are required to give 1000 kg vinyl acetate and 800 kg of acetic acid.6 Acetic acid esters have a considerable market as solvents. In Brasil the market is increasing because ethyl esters may replace toxic cyclic solvents as toluol or benzol.Therefore companies often synthesize esters from their products namely ethanol butanol and acetic acid giving ethyl acetate and butyl acetate as products. The technology for the production of ethyl acetate is still applied in Brasil (Cloroetil Solvents S.A.) and India (VAM Organic Chemicals Ltd.). The well known Tischtschenko process is currently not used on a technical scale although an ongoing research project in Brasil is addressing this topic.4 Somaiya Organo Chemicals Ltd produce butyraldehyde. The process utilises 600 kg of acetic acid and 600 kg of butanol to produce 1000 kg of butyraldehyde. 2.2.3 Other chemicals derived from fermentation ethanol. In India considerable amounts of pyridine and b-picoline are produced from acetaldehyde formaldehyde and ammonia.The production of b-picoline is especially attractive because further conversion to nicotinamide and nicotinic acid is possible.6 Fig. 6 Production of novel chemicals from lignocellulosic raw materials. 3 Research activities There are a few major hurdles in the use of renewable resources like the availability of biomass at a constant quality the whole year over the fractionation technology of lignocellulosic materials limitations due to the metabolism of microorganisms and the lack of integrated biomass based technologies. Recent research has tried to address these hurdles using different approaches. Of interest is the genetic engineering of microbial metabolism which yields various opportunities to produce new chemicals out of novel sugar resources (Fig.6). 3.1 Raw material Traditionally hexose sugars have been the major fermentation feedstocks. In Europe the availability of these substrates is limited to sugar beet or sweet sorghum and starch hydrolysates from corn cereals or potatoes. In order to reduce the cost of fermented products it is essential to expand the range and form of raw materials to produce fermentation feedstocks. In addition reductions in the cost of feedstocks could be achieved through the use of all the available sugars in the raw material sources. Therefore the development of innovative techniques to hydrolyse hemicellulose and cellulose creates opportunities to use a range of non-traditional biomass resources as fermentation raw materials.Fractionation of lignocellulosic materials may be achieved by various physical chemical and biological methods like milling extrusion steam explosion or enzymatic hydrolysis. Combination of different methods may lead to sugar solutions with both pentose and hexose sugars from the hemicellulose and cellulose fraction of lignocellulosic materials.7 Developments in the field of plant breeding will not only increase the content of fermentable sugar but also will influence the processing of lignocellulosic materials. Research results on viable mutant plants with altered lignin synthesis capability are quite promising. This might allow a more extensive exploitation of plants.8 3.2 Microbiological approaches Traditional fermentation processes have been based on hexose sugars.However a large number of organisms have been screened and shown to be able to also use pentoses like xylose or arabinose. In general these microorganisms have the disadvantage of low product tolerance.9 Considerable work especially in the United States UK and Sweden has been carried out to improve pentose fermentation. On the other hand genetic engineering has enabled scientists to broaden the spectrum of fermentatively produced chemicals. Due to the metabolic restrictions in microorganisms only a few bulk products like ethanol lactic acid or acetone–butanol can be produced via fementation. Changing fermentation technologies together with genetic engineering can broaden the product spectrum of microorganisms.Recombinant microorganisms with altered sugar metabolism are able to ferment sugar to chemicals which the corresponding wild type strain does not produce. The following chapter will summarize research in these areas. 3.2.1 Ethylene from sugar based resources. A wide variety of fungi and bacteria have been found in soil and on the surface of fruits that directly produce ethylene. These organisms are able to form ethylene from renewable resources like hydrolysates of organic biomass. In the last 20 years quite a lot of Fig. 7 Microbial synthesis of ethylene from renewable resources. research has been done on different biosynthetic pathways of ethylene forming microorganisms. However biotechnological production of ethylene in large industrial scale remains negligible.In principle two biosynthetic pathways for the production of ethylene in microorganisms have been described (Fig. 7). In one pathway ethylene is produced via 2-oxoglutarate by an ethylene forming enzyme as in Penicillium digitatum and Pseudomonas syringae. This ethylene forming enzyme has been sequenced and was cloned and expressed in Escherichia coli and Pseudomonas syringae and Pseudomonas putida.10 The ability of recombinant cells to synthesize the ethylene forming enzyme increased with specific activities 41 times higher than that of the ethylene forming enzyme in the parental P. syringae. In the second pathway ethylene is produced via S-methyl 2-keto-4-thiobutyric acid a deaminated derivative of l-methionine by an NADH-Fe(iii)EDTA oxidoreductase as in Escherichia coli strain B SPAO.11 Substrates for the fermentation are glucose glycerol or lmethionine.Sakai et al.11 succeeded in transferring the gene for the ethylene forming enzyme into the cyanobacterium Synechococcus sp. PCC 7942. The enzyme catalyses the conversion of atmospheric CO2 to ethylene. Sakai et al.11 reported a maximum specific ethylene forming activity of 323 nl ml21 OD73021 h21 a maximum ethylene formation of 16.4 ml ml21 and a carbon recovery of 5.84% for ethylene (percentage of total carbon fixed incorporated into ethylene). 3.2.2 Ethanol from lignocellulosic raw materials. Genetic engineering could reduce the production costs for ethanol dramatically.Research is mainly focused on the fermentation of pentoses and other unusual sugars. The pentose sugars of importance are d-xylose and l-arabinose which comprise up to 30% of the neutral carbohydrates derived from agricultural crop residues wood and other plant materials. However the conversion of pentose sugars is difficult to achieve because of the lack of suitable biocatalysts. Research activities in the field of simultaneous bioconversion of cellulose and hemicellulose to ethanol have been summarized recently.12 Activities in the United States focus on recombinant strains of Zymomonas mobilis E. coli and on recombinant Saccharomyces cerevisiae. Zymomonas mobilis a bacterium has a high ethanol yield and product tolerance.Furthermore it has considerable tolerance to inhibitors formed during hydrolysis of lignocellulosic materials. Zhang et al.13 introduced four genes encoding xylose assimilation and pentose phosphate pathway enzymes into Zymomonas mobilis to enable it to grow on xylose as the sole carbon source with efficient ethanol production (xylose isomerase xylulokinase transketolase and transaldo- 399 Chem. Soc. Rev. 1999 28 395–405 lase see Fig. 8). Yields of ethanol produced were 0.44 g per gram of xylose consumed which corresponds to 86% of the theoretical yield (5 moles of ethanol from 3 moles of xylose). Ingram et al.14 introduced the genes encoding for alchoholdehydrogenase and pyruvate-decarboxylase from Zymomonas mobilis into Escherichia coli KO11 which enables fermentation of hemicellulose hydrolysates of agricultural wastes like bagasse corn stover and corn hulls.Ethanol concentrations of over 40 g l21 within 48 h could be attained with yields ranging from 86% to over 100% of the maximum theoretical yield of 0.51 g ethanol per gram of sugar. BC International Corporation together with the US Department of Energy are currently building the first large scale biomass-to-ethanol plant in Jennings La. based on this strain.15 Genetic modifications of Saccharomyces strains to enhance its sugar utilization range are reported by the insertion of genes encoding xylose reductase xylitol dehydrogenase and xylulokinase. 16 Strain LNH-ST with genes integrated into the chromosome can co-ferment glucose and xylose with yields of 63.5% of theoretical yield on pretreated corn biomass and 70.4% on synthetic glucose–xylose mixtures.In Europe the company Agrol Ltd. (UK) focuses on the screening of thermophilic strains like Bacillus stearothermophilus, 17 which are naturally capable of utilizing pentoses. Genetic strain improvement is applied to lower byproduct formation by decreasing the activities of lactate dehydrogenase and pyruvateformate lyase. 3.2.3 Acetaldehyde. The production of acetaldehyde has been investigated by applying different approaches. In the first ethanol can be oxidized to acetaldehyde using Candida utilis or Pichia pastoris. Production of acetaldehyde by this method must be carefully regulated to limit the conversion of acetaldehyde to acetic acid.However there is no significant advantage compared to the chemical oxidation described earlier. Acetaldehyde can also be fermented directly from sugar. Changing the fermentation strategy from ethanol to acetaldehyde has some major advantages. First of all acetaldehyde has a boiling point of 20.8 °C which enables easy separation from the fermentation broth by stripping. Ethanol boils at 78.5 °C and separation makes up two-thirds of total production Fig. 8 Enzymes of interest for increasing biotechnological ethanol production. Chem. Soc. Rev. 1999 28 395–405 400 costs. On the other hand acetaldehyde forms no azeotropic mixture with water which again reduces purification costs. Direct fermentation of acetaldehyde with Zymomonas mobilis is described with two different strategies.Wecker and Zall18 were mutating and screening for strains with decreased alcohol dehydrogenase activities (Fig. 9). Strain selection was done by adding allyl alcohol which is normally converted by alcohol dehydrogenase to acrylaldehyde a substance highly toxic to cells. Therefore only strains without alcohol dehydrogenase activities are able to grow in the presence of allyl alcohol making it easier to screen for mutants lacking alcohol dehydrogenase activity. On the basis of the amount of glucose utilized the level of acetaldehyde production represents nearly 40% of the maximum theoretical yield. Tanaka et al.19 were investigating the performance of Zymomonas mobilis under various oxygen supply conditions.Aeration increases the activity of the NADH-oxidase and consequently the availability of NADH for the alcohol dehydrogenase is decreased. With optimised aeration rates yields around 55% of the theoretical yield based on utilised glucose could be observed. Acetaldehyde may also be produced with Saccharomyces cerevisiae using a well known process employed by the Germans during World War I. In this case acetaldehyde is fixed by sodium sulfite and therefore can not act as hydrogen acceptor like in conventional ethanol fermentation. Instead of acetaldehyde dihydroxyacetone functions as the main hydrogen acceptor and becomes reduced to glycerol. Alcohol and acetaldehyde are separated by distillation glycerol may be recovered after precipitation of sulfite by the addition of calcium oxide or calcium hydroxide followed by filtration.Technical glycerol is then obtained by distilling the supernatant liquor. 100 g of hexose theoretically yields 51 g of glycerol and 24.4 g of acetaldehyde practical yields were about 20–25% glycerol 30% alcohol and 5% acetaldehyde. It required almost 12 kg of refined sugar to produce 1 kg of dynamite glycerol on industrial basis. Losses of glycerol are mainly due to inefficient recovery processes.20 3.2.4 Acetic acid. The traditional biological production of acetic acid by a species of Acetobacter has been used almost exclusively for making vinegar. Because of the loss of one Fig. 9 Pathways for the production of acetaldehyde by Zymomonas mobilis and Saccharomyces cerevisiae.Fig. 10 Production of acetic acid by Clostridium thermoaceticum. 1 Mole glucose yields 3 moles of acetic acid 2 mole of pentoses yield 5 moles of acetic acid. carbon through CO2 in the glucose to ethanol fermentation the theoretical maximum yield of the whole reaction is 2 moles of acetic acid from one mole of glucose or only 0.67 g acetic acid per gram of glucose. In commercial practice the actual yield is 0.50–0.55 g acetic acid per gram of glucose or roughly 75–80% of the theoretical yield. Fermentation with Clostridium thermoaceticum a spore forming thermophilic bacterium (optimum 55–60 °C) offers a significant advantage in terms of acetate yield compared to the conventional vinegar fermentations because this strain can theoretically produce 3 moles of acetic acid from 1 mole glucose (Fig.10). In practice 85% of the sugar may be converted to acetic acid. Furthermore Clostridium thermoaceticum is strictly anaerobic and therefore no costly aeration is required. Additionally Clostridium thermoaceticum is able to ferment various sugars like fructose glucose and xylose. Unfortunately the toxicity of acetate pH optima and production rates do not suggest successful applications of this strain for industrial production of acetic acid. Clearly new mutant strains must be obtained with improved properties for these applications as demonstrated by Parekh and Cheryan.21 Another possibility would be the isolation from nature of new strains or species of acetogens with properties suitable for industrial use.Leigh et al.22 isolated a new chemolithoautotroph homoacetogenic thermophilic anaerobic microorganism Acetogenium kivui that oxidizes hydrogen and reduces carbon dioxide to acetic acid. The temperature optimum for growth is 66 °C and the optimum pH is 6.4. Suitable growth substrates include glucose mannose fructose pyruvate and formate. Acetogenium kivui theoretically produces 3 moles acetic acid from 1 mole of glucose. Because of strong product inhibition maximum acetate concentration is 30–40 g l21. In batch fermentation 280 mM glucose are converted to 625 mM acetic acid in 50–60 h which is a yield of 2.55 mol per mole of glucose (85% of theoretical). 3.2.5 Propane and propylene.Although propylene has to be considered as a byproduct of ethylene production research on the biotechnological production of propylene has been carried out. According to Fukuda23 C3-hydrocarbon-producing strains are widely distributed. Among the 178 strains tested 87 (49%) produced propane and 37 propylene from glucose media. The C3-hydrocarbons are usually produced together with other hydrocarbons such that a mixture of propane propylene butane butene pentane etc. is obtained. Unless the hydrocarbons are used for fuel purposes some purification (usually distillation) will be required. Compared to C2-hydrocarbons the production rates are rather low. Highest productivity for propane is reported for Cryptococcus albidus (6.5 nl ml21 h21) and Brevibacterium ammoniagenes (8.6 nl ml21 h21) while the most efficient propylene producing strains are Gliocladium roseum (3.0 nl ml21 h21) and Schizosaccharomyces octosporus (1.2 nl ml21 h21) respectively.Levy et al.24 describe a process for the production of propylene from industrial waste streams. Propylene is produced from the electrolytic oxidation of butyric acid which is formed by the suppressed-methane anaerobic fermentation of carbohydrates in waste water and recovered by liquid–liquid extraction. Besides propylene methane and hydrogen may be recovered as byproducts. Yields are 15.1% propylene 8.9% methane 0.6% hydrogen and more than 50% CO2 depending on waste sugar. Although yields seem to be quite low very optimistic figures on return on investments are presented.3.2.6 Acrylic acid. Although acrylic acid and chemicals associated with acrylics have a world market volume of almost 3 3 106 tonnes per annum relatively few attempts have been made to produce them with microorganisms. Currently 100% of acrylic acid is produced from fossil fuels. The production from renewable resources is propagated via lactic acid fermentation and subsequent chemical conversion to acrylic acid. Un- 401 Chem. Soc. Rev. 1999 28 395–405 fortunately the chemical conversion gives rather low yields due to decarbonylation decarboxylation and condensation reactions which mainly lead to acetaldehyde or penta-2,3-dione.25 Only a few microorganisms have been described that produce acrylic acid as a biochemical intermediate substance but observations of free acrylic acid in biological systems are rare.Anaerobic formation of acrylic acid is found in the direct reduction pathway of lactic acid of microorganisms like Clostridium propionicum. This conversion is a dehydration reaction. When this microorganism uses lactic acid as the energy source the main metabolic products are propionic acid (2/3) and acetic acid (1/3). The propionic pathway may be blocked with 3-butynoic acid. Nevertheless acrylate concentrations never exceeded 1% of the initial substrate concentration. These low yields are due to the enrichment of reducing equivalents like ferredoxin rubredoxin and flavodoxin which inhibit further growth of cells. These reducing equivalents can be regenerated by providing the cells with an external electron acceptor.Research activities in this area are limited and were summarized recently.26 3.2.7 Propanediol. The only industrial process for manufacturing propane-1,2-diol (propylene glycol) is direct hydrolysis of propylene oxide with water. Dipropylene glycol and tripropylene glycol are obtained as byproducts. Biotechnological research concerning propane-1,2-diol concentrates upon the formation of enantiomers especially (R)-propane-1,2-diol. This compound can be used in organic synthesis e.g. for the preparation of (R)-propylene oxide optically active polymers and chiral crown ethers. Different strategies are followed to obtain (R)-propane- 1,2-diol from renewable resources via fermentation involving Clostridium sphenoides and several strains of Thermoanaerobacterium thermosaccharolyticum.27 The formation of propane-1,2-diol starts from dihydroxyacetone a common intermediate in sugar metabolism.Conversion to methylglyoxalate and subsequent reduction to lactaldehyde or hydroxyacetone and further reduction gives propane-1,2-diol as the end product (Fig. 11). As substrate various sugars like glucose mannose xylose and cellobiose have been used. Fig. 11 Metabolic pathways to propane-1,2-diol and propane-1,3-diol (Cameron et al.28) Unlike propane-1,2-diol propane-1,3-diol (1,3-PD) is produced on a much smaller scale. Due to the difficulties in manufacturing the product the price is too high compared to other diols. As a result the use of 1,3-PD is limited to applications requiring very specific performance characteristics.A potential large scale application represents the Chem. Soc. Rev. 1999 28 395–405 402 manufacturing of polyesters like poly(propylene terephthalate) PPT used in carpet fibers. Several microorganisms are known to ferment glycerol to 1,3-PD. The conversion of glycerol consists of three steps involving 3-hydroxypropionaldeyde (3-HPA hydracrylaldehyde) as an intermediate. Due to the high price of glycerol this fermentation is not economically attractive however it provides the basic knowledge for constructing recombinant microorganisms. Cameron et al.27 cloned and expressed the genes for the conversion of glycerol to 1,3-PD from Klebsiella pneumoniae in Escherichia coli and Saccharomyces cerevisiae.As S. cerevisiae produces glycerol from glucose as a byproduct these gene transfers enable the direct fermentation of 1,3-PD from glucose. Theoretical product yields are 2 moles of propanediol per mole of glucose (0.84 g g21). As propanediol is more reduced than glucose another compound has to be oxidised to maintain the overall electron balance. Assuming that the oxidized byproduct is CO2 from glucose the maximum theoretical yield is 1.5 mol mol21 (0.63 g g21). Practical yields are lower but nevertheless DuPont and Genencor are working on the production of propane-1,3-diol from sugar.28 3.2.8 Glycerol. Glycerol is a component of all plants and animal fats and oils. However it is not found in its free form but as a component of fatty acid esters.The glycerol content of fats and oils varies between 8 and 14%. In the USA 74% of the glycerol capacity is made up by production from fats and oils. Fermentation of sugar (from molasses) by Saccharomyces cerevisiae in presence of sulfite (Neuberg’s second type of fermentation) has been described earlier in this article (Section 3.2.3 and Fig. 9). Another way to produce glycerol with Saccharomyces cerevisiae but without sulfite has been described by Compagno et al.29 The triosephosphate isomerase gene was inactivated by genetical engineering. Due to this inactivation the accumulated dihydroxyacetone phosphate was reduced to glycerol. Glycerol was obtained in high molar yields (90%) as the major fermentation product.As the process is carried out in a medium consisting mainly of glucose and phosphate problems with the recovery of glycerol have been reduced significantly. Rehm30 refers to investigations on the green algae species Dunaliella tertiolecta and Dunaliella bardawil which produce glycerol in a medium with high concentrations of NaCl. The possible advantages of these algae are that CO2 as a cheap and renewable resource is used as carbon source solar energy is the main energy source and protein as well as b-carotene may be obtained as valuable byproducts. 3.2.9 Lactic acid. Improvements on the economics of the process are mainly focused on broadening the substrate range increasing the lactic acid tolerance reducing the requirements for complex and cost intensive growth supplements and on product recovery.New processes include simultaneous saccharification and fermentation thermophilic fermentation with Bacillus stearothermophilus continuous fermentation and product recovery with membrane bioreactor systems and electrodialysis and novel recovery technologies with ion exchange chromatography.31 3.2.10 Butanediol. Butane-2,3-diol has been known as a bacterial fermentation product since early this century. During World War II fermentative butane-2,3-diol production was of great interest mainly because it may serve as a precursor for the manufacture of buta-1,3-diene. However none of the developed processes have been applied on a commercial scale. The interest in butane-2,3-diol production has been renewed because various sugar sources including both hexoses and pentoses from lignocellulosic hydrolysates can be fermented.Furthermore apart from butane-1,3-diol a broad range of chemicals and products may be synthesized out of butane-2,3-diol namely polyurethane g-butyrolactone or octane booster for gasoline. Maximum theoretical product yields from 1 mol glucose are 0.67 mol butanediol and 0.33 mol CO2 while maximum observed yields of both glucose and xylose range between 60 and 70% of the theoretical value. On a mass basis this is equivalent to 0.30 to 0.35 g butanediol per gram of glucose. Besides CO2 acetate formate ethanol and lactic acid are the major byproducts. Cao et al.32 have reported yields of 0.31 g butanediol and 0.088 g ethanol per gram of corn cob cellulose which was hydrolysed by fungal cellulases.3.2.11 Butadiene. Unfortunately there is no method for the direct production of butadiene from renewable resources involving microorganisms as catalysts. Anyhow there are some syntheses described for the chemical production of butadiene out of acetaldehyde and ethanol which may be derived from fermentation processes as highlighted earlier. Butadiene from fermentation of butanediol has been described especially in very old literature.20 It is rather difficult to convert butane-2,3-diol to compounds with ethylenic double bonds by catalytic dehydration due to the strong tendency towards formation of the methyl ethyl ketone instead. The most successful process for the manufacture of butadiene from butane-2,3-diol is pyrolysis of the diacetate.The process is known to give butadiene of over 99% purity. The reaction is usually accomplished by passing the vapor of the diacetate through an unpacked tube at 585 °C to 595 °C. Yields of 82% are obtained in a single pass with a cumulative yield of about 87% of the theoretical yield. According to McCutchan and Hickey20 butadiene could also be produced from n-butanol which was one of the driving forces for the industrial development of the acetone–butanol fermentation process at the beginning of the century. Unfortunately no yields for the conversion of n-butanol to butadiene are available and data on the technical process are scarce. 3.2.12 Succinic acid.Succinic acid is of special interest because it may serve as an intermediate in the production of chemicals like butane-1,4-diol tetrahydrofuran g-butyrolactone or adipic acid (precursor to nylon). Therefore succinic acid has numerous potential uses in textile plastics and resins detergents and the food industry. It is currently produced by hydrogenation of maleic anhydride derived from petrochemical feedstocks. Nghiem et al.33 present a novel fermentation process for the production of succinic acid from renewable resources. The process is based on a recombinant Escherichia coli strain ATCC 202021 which can convert corn sugar into succinic acid. The process consists of two stages the first is the growth phase under aerobic conditions while the second is the main production phase under anaerobic conditions during which glucose is converted mainly into succinic acid with acetic acid and ethanol as byproducts.The developed process seems to be Table 1 Maximum allowable sugar prices for selected commodity chemicals. Theoretical yield/kg kg21 Fermentation product Process 0.32 0.51 0.70 1.00 1.00 via ethanol-fermentation direct fermentation via ethanol fermentation direct fermentation direct fermentation Ethylene Ethanol Acetic acid Acetic acid Lactic acid Observed yield/kg kg21 0.27–0.28 0.43–0.46 0.54–0.58 0.85 0.95 4 Economical aspects It has been demonstrated that various biotechnological processes for the production of chemicals from renewable resources are technically feasible however they remain uneconomical under current conditions.As a result only a limited number of processes have made their way from laboratory scale to commercial industrial scale until now. At the same time processes based on biomass lost economic competiveness as a direct result of cheaper petrochemical resources. Different factors are responsible for this development and these are discussed below. 4.1 Substrate costs The cost of raw materials for fermentations can represent up to 70% of the total value of the fermentation product particularly for commodity chemicals. Hence substrate costs are the most critical costs in the production of bulk products from renewable resources.As stated earlier in this article the availability of cheap substrates from wood forest wastes paper mill wastes crop plants agricultural residues municipal wastes or algae is imperative for economically viable processes. Changes in the price of raw materials can be achieved through the following strategies • Increasing the yield per hectare and/or producing crops with higher content of desirable ingredients • Increasing the value of byproducts • Fractionation of whole plant biomass by hydrolysis and natural lignin recovery • Modifying crops or cultivation of new crops grown especially for industrial purposes • Using wastes as raw material Conventional fermentation processes often consider just one single product from the whole crop resulting in a scenario whereby the final product is burdened by the whole raw material cost.Biomass based plants have to convert all components of the starting material into useful products just as a petroleum refinery utilises all of the raw material to achieve low production costs. Hence not only sugar has to be considered as raw material for fermentation but also lignin or fibers as raw material for other industrial applications.34 The maximum allowable raw material costs for various chemicals can be calculated from the final product price and the corresponding fermentation yields as shown in Table 1. Product prices are taken from the Chemical Marketing Reporter.35 The maximum sugar price (theoretical) refers to the maximum allowable price for substrate under theoretical yields while the maxium sugar price (practical) equals the maximum allowable sugar price under observed product yields.At a given sugar price of approximately 0.4 Euro kg21 only the direct fermentation of acetic acid and lactic acid can be economically feasible. Economically viable processes can only be achieved if substrate Max. sugar price (theory)/ Euro kg21 Product price/ Euro kg21 0.18–0.19 0.35 0.54–0.62 0.77–0.88 1.48 0.55–0.60 0.68 0.77–0.88 0.77–0.88 1.48 Chem. Soc. Rev. 1999 28 395–405 economically feasible and will be commercialised by Applied CarboChemicals Inc. (USA). Max. sugar price (pract.)/ Euro kg21 0.15–0.17 0.30–0.31 0.42–0.51 0.65–0.75 1.41 403 Table 2 Major bioconversion parameters determining production costs (rec = recombinant efficiency = observed yield/theoretical yield nk = not known) Autoselective Process Product Ethanol 2 + 22 +/2 Acetaldehyde 22 Acetic acid + + 22+ 2 Acrylic acid Propanediol Lactic acid Butanediol Succinic acid Saccharomyces Zymomonas rec.13 E.coli KO1114 Saccharomyces rec.16 Bac. stearothermophilus17 Zymomonas mobilis18 Zymomonas mobilis19 Acetobacter Cl. thermoaceticum21 Cl. propionicum E. coli rec.27 Lactobacilli Klebsiella oxytoca ATCC 872432 E. coli ATCC 20202133 2 costs are lower than the given maximum practical sugar price or if the yields are increased towards the theoretical yield.4.2 Cost effective bioconversion parameters Besides the corresponding product yields product titers and volumetric productivities other parameters have to be considered to end up with economically feasible bioconversion processes. First of all all byproducts have to be considered as valuable products. This includes not only the fermentation end products but also the applied microorganism itself. In most cases microorganisms represent a valuable source of proteins for animal feed. Therefore generally recognized as safe (GRAS) microorganisms should be promoted for industrial applications. Another critical cost parameter is the requirement of sterility of the substrate. While fermentations for lactic acid ethanol or acetic acid can be performed without sterilization of the substrate others like acetone–butanol–ethanol or ethanol fermentation with recombinant microorganisms require aseptic conditions which catapults the cost of equipment and substrate pretreatment significantly.Furthermore it makes a great difference whether the desired process requires aerobic or anaerobic conditions. If the process is aerobic one is faced with additional costs for aeration and cooling equipment thus increasing the overall production costs. Thermophilic strains growing at temperatures around 60 °C might help to overcome costs for cooling and sterilization. The major parameters determining the costs for bioconversion processes are summarized in Table 2 for a few selected processes. 4.3 Product recovery Most biotechnological processes have the major disadvantage of low product concentrations.Therefore separation of the product from the water stream contributes significantly to the overall production costs. Many products can be recovered by simple distillation a method which has been optimized during the past years. Novel methods like pervaporation or electrodialysis enable continuous product separation simultaneously eliminating product inhibition. 5 Concluding remarks Even today the potential of microorganisms for the production of bulk chemicals is far from being fully exploited. Microorgansims may be applied for the production of typical metabolic end products like ethanol lactic acid acetone butanol or butanediol. Some of these processes have been used during periods when fossil resources were in short supply and as Chem.Soc. Rev. 1999 28 395–405 404 Efficiency (%) Commercial scale Pentose utilization Anaerobic metabolism 2 2 + +/2 + + + 2 + + + + + +/2 84–90 86 86–100 70.4 80 39 55 22+ + 222 222 + 2 + + +/2 +/2 +/2 + 2 22 2 + 2 75–80 85 1 12–16 95 62.5 nk + + 2 +/2 such can still be resuscitated under changing circumstances namely environmental legislative instruments. On the other hand novel genetic engineering methods may increase the importance of microorganisms and their biosynthetic capabilities. New metabolic endproducts like acrylic acid acetaldehyde or propanediol may be obtained by simple methods like mutation and advanced techniques of recombinant DNA technology.However the cost of feedstocks still remains one of the crucial points if biotechnological processes are to succeed. The fermentation of pentoses may lower the product costs significantly and research in this area is therefore justified. Current developments in ethanol production from pentoses demonstrate that metabolic engineering is worth further investigation and may even play an important role in the development of economic processes. On the other hand developments with thermophilic bacteria enable cheaper fermentations due to lower risks of contaminations and higher productivities and conversion rates. Some of these organisms have the capability of using pentoses and therefore genetic engineering may be avoided.Unfortunately the potential of thermophilic microorganisms is still not yet fully recognized. Low product tolerances high byproduct formation and high growth supplement requirements could be overcome by screening and mutation strategies. Finally process engineering has to be considered. New fermentation strategies may lead to new chemicals as demonstrated with acetaldehyde. The combination of product formation and product separation enables increased and sustained synthesis of toxic or inhibitory fermentation products. Parameters like process stability and process reliability have to be improved through long term experiments and up-scaling initiatives.Only a demonstration of process viability may convince potential industrial partners. 6 Acknowledgements Substantial financing of the work was obtained by the Austrian Ministry of Economics. Special thanks to Cecilia LaLuce from UNESP in Araraquara Brasil and to L.N. Batra from the All India Sugar Distilleries New Dehli India for opening doors to production facilities in their countries. 7 References 1 Proceedings of the conference ‘Chemical Feedstock Alternatives’ ed. F. J. Van Antwerpen American Institute of Chemical Engineers New York 1977. 2 D. T. Jones and D. R. Woods Microbiol. Rev. 1986 50 484. 3 L. N. Batra All India Distiller’s Association New Dehli 1998 Personal Communication. 4 V. Zanardi Cloroetil Solventes Aceticos S.A. Mogi Mirim SP Brasil 1998 Personal Communication.5 S. Sing Somaiya Organics (India) Ltd. Barabanki India 1998 Personal Communication. 6 A. Kumar VAM Organic Chemicals Ltd. Uttar Pradesh India 1998 Personal Communication. 7 M. Galbe G. Zacchi and A. S. Schmidt Novel Fermentation Feedstocks. The Monograph EC DGXII SSMA Concerted Action NFF 1998 p. 1. 8 A. M. Boudet Biofutur 1996 158 27. 9 P. Mishra and A. Singh Adv. Appl. Microbiol. 1993 39 91. 10 J. E. Ince and C. J. Knowles Arch. Microbiol. 1986 146 151. 11 M. Sakai T. Ogawa M. Matsuaoka and H. Fukuda J. Ferment. Bioeng. 1997 84 434 and references therein. 12 P. Chandrakant and V. S. Bisaria Crit. Rev. Biotechnol. 1998 18 295. 13 M. Zhang C. Eddy K. Deanda M. Finkelstein and S. K. Picataggio Science 1995 267 240.14 L. O. Ingram X. Lai M. Moniruzzaman B. E. Wood and S. W. York ACS Symp. Ser. 1997 666 57. 15 M. L. Wald The New York Times 1998 October 25. 16 S. T. Toon G. P. Philippidis N. W. Y. Ho Z. D. Chen A. Brainard R. E. Lumpkin and C. J. Riley 1997 Appl. Biochem. Biotechnol. 1997 63 243. 17 B. S. Hartley and N. Baghaei-Yazdi Proceedings of the ‘Non-wood fibres for Industry’ conference 1994 Silsoe Research Institute Silsoe Bedfordshire UK. 18 M. A. Wecker and R. R. Zall Appl. Environ. Microbiol. 1987 53 2815. 19 H. Tanaka H. Ishikawa K. Osuga and Y. Takagi J. Ferment. Bioeng. 1990 69 234. 20 W. N. McCutchan and R. J. Hickey in Industrial Fermentations eds. L. A. Underkofler and R. J. Hickey Chemical Publishing Co. Inc. New York 1954. 21 S. R. Parekh and M. Cheryan Appl. Microbiol. Biotechnol. 1991 36 384. 22 J. Leigh F. Mayer and R. Wolfe Arch. Microbiol. 1981 129 275. 23 H. Fukuda Y. Kawaoka T. Fujii and T. Ogawa Agric. Biol. Chem. 1987 51 1529. 24 P. F. Levy A. P. Leuschner J. H. Stoddart and D. L. Wise International Biosystems ed. D. L. Wise CRC Press Florida vol. 1 p. 237 1989. 25 G. C. Gunter R. H. Langford J. E. Jackson and D. J. Miller Ind. Eng. Chem. Res. 1995 34 974. 26 H. Danner M. Ürmös M. Gartner and R. Braun Appl. Biochem. Biotechnol. 1998 70 887. 27 D. C. Cameron N. E. Altaras M. L. Hoffman and A. J. Shaw Biotechnol. Prog. 1998 14 116. 28 C. Potera Genet. Eng. News 1997 17 17. 29 C. Compagno F. Boschi and B. M. Ranzi Biotechnol. Prog. 1996 12 591. 30 Biotechnology ed. H.-J. Rehm and G. Reed 2nd edition Vol. 6 VCH Weinheim New York 1996. 31 H. Danner M. Neureiter L. Madzingaidzo M. Gartner and R. Braun Appl. Biochem. Biotechnol. 1998 70 895. 32 N. Cao Y. Xia C. S. Gong and G. T. Tsao Appl. Biochem. Biotechnol. 1997 63 129. 33 N. P. Nghiem B. H. Davison B. E. Suttle and G. R. Richardson Paper 12 presented at the 20th Symposium on Biotechnology for Fuels and Chemicals Gatlingburg Tennessee May 3–7 1998. 34 Organic chemicals from biomass ed. I. S. Goldstein CRC Press Inc. Boca Raton Florida 1981. 35 M. McCoy Chem. Mark. Rep. 1997 251 26. Review 8/06968I 405 Chem. Soc. Rev. 1999 28 395–405

ISSN:0306-0012    

DOI:10.1039/a806968i

出版商:RSC    

年代:1999

数据来源: RSC