摘要:

Synthesis of multi-substituted furan rings the role of silicon Brian A. Keay Department of Chemistry University of Calgary Calgary Alberta Canada T2N 1N4. E-mail keay@ucalgary.ca Received 3rd December 1998 Silanes have proven extremely useful for the synthesis of multi-substituted furan rings. Their judicious placement on a furan ring can be used to control either the placement of new groups around the furan ring or the silane can be directly replaced via an ipso-substitution. This paper will briefly review the role silicon has played in the preparation of furan rings having a variety of substituent patterns. 1 Introduction Compounds that have a multi-substituted furan ring as their main component or have a furan ring fused with other rings have exhibited a vast array of activity.1 Thus the design of synthetic routes towards furan rings having specific substitution patterns has interested chemists for decades.2 Of the many strategies developed for the preparation of a multi-substituted furan ring one of the simplest involves starting with either furan itself or a mono-substituted furan followed by the introduction of functional groups at various positions around the furan ring.This approach however has some drawbacks. First the introduction of functional groups into the 3- and 4-positions is difficult as furan rings lithiate and add electrophiles preferentially at the 2 and 5 positions. Second placement of a group at C-3 results in the 2 and 5 positions becoming regioisomeric so that either regioselective methods or blocking groups must be used to introduce new moieties.2 The use of an organosilyl group as a blocking group has been the most useful in furan chemistry.Not only are they easy to introduce and remove but they can be Brian A. Keay was born in 1955 and received his PhD degree in 1983 from the University of Waterloo (Canada) working with Professor R. Rodrigo. After an NSERC postdoctoral fellowship with Professor E. Piers (University of British Columbia Canada) he joined the faculty at the University of Windsor (Canada) as an assistant Professor. In 1989 he moved to the University of Calgary where he is now a Professor. In 1996 he was awarded the Merck Frosst Centre for Therapeutic Research Lecture Award by the Canadian Society for Chemistry.His research interests include the design and synthesis of asymmetric ligands for use with Lewis acids and transition metals palladium-catalyzed polyene cyclizations intramolecular Diels–Alder reactions of furans and the synthesis of natural products. replaced by electrophiles via an ipso-substitution. In addition their size has had a pronounced influence on the regioselectivity for introduction of other groups on the furan ring and on some rearrangements. This paper will briefly review the role organosilyl groups have played in the synthesis of multisubstituted furan rings. This will not be an exhaustive review but will concentrate on how silyl groups have been used to prepare specific substitution patterns with furan rings.Scheme 1 2 The role of silyl groups on furan rings In 1948 Benkeser and Currie reported3 not only the introduction of a trimethylsilyl group at the C-2 position of furan but that the silyl group could later be replaced by a bromine atom via an ipso-substitution (Scheme 1). Treatment of furan (1) with BuLi followed by the addition of trimethylsilyl chloride provided 2-(trimethylsilyl)furan (2) which when treated again with BuLi followed by the addition of CO2 gave 5-(trimethylsilyl)- 2-furoic acid (3). Subsequent treatment of 3 with Br2 in CCl4 gave 5-bromo-2-furoic acid (4 no yield given). This work also confirmed that furan (1) lithiated at the C-2 position when treated with BuLi. To my knowledge this was the first report of the introduction of an organosilyl group onto a furan ring and the first example of an ipso-substitution on a furan ring.In order to introduce functional groups regioselectively around the furan ring it is sometimes prudent to use a group to block a particular position. Organosilyl moieties have been used as blocking groups as they are easily introduced and removed. Carpenter and Chadwick4 required the synthesis of 2,3-disubstituted furan rings but as mentioned above the introduction of groups into the 3 position of a furan ring is difficult. They achieved the introduction of a group at the C-3 position of 2-furoic acid (5) by first blocking the C-5 position with a trimethylsilyl group which afforded acid 3 according to the protocol developed by Knight5 (Scheme 2).Treatment of 3 with BuLi (THF 278 °C 0.5 h) resulted in a regiospecific lithiation at C-3. Quenching the anion with CO2 followed by an acidic workup gave 6. The trimethylsilyl group was then removed to give furan-2,3-dicarboxylic acid (7). The yield for the overall sequence was 78%. A similar sequence (but using different reagents) was also reported by Chadwick6 using 2-(4,4-dimethyl-4,5-dihydro-1,3-oxazol-2-yl)furan 8 instead of 2-furoic Chem. Soc. Rev. 1999 28 209–215 209 Scheme 2 acid (5) and has been expanded upon by Gilchrist7 and Queguiner.8 Keay and co-workers9,10 have developed a strategy for the synthesis of a variety of polysubstituted furan rings by taking advantage of the silyl groups 1) ability to be a blocking group; 2) tendency to undergo silyl migrations; and 3) size.In this manner they have been able to develop routes towards the synthesis of 2,3- 2,3,4- 2,4- and the more difficult 3,4-substituted furan systems starting from a common 3-substituted furan ring. First it was shown that a variety of silyl protected 3-hydroxymethylfurans 9 undergo an intramolecular [1,4] O?C silyl migration to produce 2,3-disubstituted furans 10 in good to excellent yield by treating 9 with a mixture of BuLi and HMPA in THF (Scheme 3).9 Treatment of 10 with 2.2 equiv. of BuLi in Scheme 3 DME resulted in a regiospecific C-4 lithiation. Quenching the anion with a variety of electrophiles provided 2,3,4-trisubstituted furans 11 in good to excellent yields.10 To our knowledge this was the first example of a direct C-4 lithiation in the presence of an unprotected C-5 position on a furan ring.This regiospecific lithiation was a result of the presence of the silyl blocking group at C-2 and also was a result of the size of the silyl group forcing the hydroxymethyl group to reside very Chem. Soc. Rev. 1999 28 209–215 210 close to the C-4 hydrogen atom so that when BuLi was added the C-4 hydrogen atom was preferentially abstracted instead of the C-5 hydrogen atom. This chemistry also provided a route to directly access 3,4-disubstituted furan rings which of all the possible furan patterns are the most difficult to prepare.2 Treatment of 11 with TBAF in THF resulted in the direct removal of the C-2 silyl group to provide 3,4-substituted furans 12.Interestingly the silyl group could be reused to protect the hydroxy group at C-3 by treating 11 with NaH in DMF. This resulted in an intramolecular [1,4] C?O silyl migration to form 13.9 The complete sequence in Scheme 3 illustrates the importance and usefulness of a silyl group. This approach was used recently by Scott and co-workers to prepare furan 12a.11 It is interesting to note that if the hydroxymethyl group of 10 is protected as a triethylsilyl12 or tert-butyldimethylsilyl group,13 to give 14 and 14 is then treated with 1.3 equiv. of BuLi lithiation occurs exclusively at C-5 (Scheme 4). Trapping Scheme 4 the C-5 anion with electrophiles provided 2,3,5-trisubstituted furans 15. When the triethylsilyl group was used it could be selectively removed (AcOH–THF–H2O; 8 8 1) to provide alcohol 16.The hydroxy group in 16 could be re-protected by the silyl group at C-2 via a [1,4] C?O silyl migration thereby providing access to 2,4-disubstituted furan rings 17. In addition both silyl groups could be removed by treatment with TBAF in THF to provide furans like 18. Recently Katsumura used this chemistry to prepare tetrasubstituted furan rings as precursors towards butenolides.14 One of the more difficult transformations with 2- or 3-furaldehyde is their direct oxidation to 2- or 3-furoic acid since the furan ring is very sensitive to acidic reagents and normally most oxidations of aldehydes to acids involve using acidic reagents. Thus attempts to oxidize 12 to the corresponding furoic acid failed.To overcome this difficulty Keay and coworkers showed that silyl esters of 3-furoic acid 19 also undergo a [1,4] O?C silyl migration to provide 2-silylated-3-furoic acids 20 in moderate yield (Scheme 5).9 Acids 20 when treated with BuLi resulted in a regiospecific C-4 lithiation. Trapping the anion with a variety of electrophiles provided 2,4-disubstituted-3-furoic acids 21 in good to excellent yield.10 The silyl group was easily removed with TBAF and the acid converted to the methyl ester (for ease of workup) by treatment with diazomethane to provide 4-substituted-3-furoic esters 22. Quayle and co-workers have used the silyl migration of silylesters 19 to prepare 2-silylated tetrahydrofurans 23.15 Scheme 5 In order to illustrate the usefulness of our silyl strategy we investigated the introduction of other groups into the C-4 position of furan 10 (Scheme 6).16 The C-4 anion of furan 10 Scheme 6 could be trapped with tributyl- or trimethylstannyl chloride providing furans 24a and 24b.The direct Stille coupling with 24a or b provided disappointing yields of furans 24c; however when the silyl group at C-2 was migrated to the hydroxy group at C-3 (to give 25) the Stille coupling proceeded in yields ranging from 55–90% to provide 26. This gave access to a variety of 3,4-disubstituted furans in which one of the substituents was an aromatic ring. Since the Stille coupling with 24a and 24b was sluggish we investigated trying a Suzuki coupling.This required the preparation of boronic acid 27 (Scheme 7). Unfortunately the isolation of boronic acid 27 after quenching the C-4 anion of 10 with trimethylborate was extremely difficult. Thus we developed an in situ variant of the Suzuki reaction in which the boronic acid component does not have to be isolated.17,18 The reaction is easy to perform. Trimethylborate is added to the anion and the reaction is stirred for 1 h. Then instead of working up the mixture an aryl or vinyl halide or triflate is added along with Pd(PPh3)4 water and Na2CO3 and the mixture is refluxed for 1–20 h. A variety of products 28 were isolated in yields ranging from 20–93%. The in situ variant of the Suzuki reaction is a general reaction and is not limited to just furan systems; a variety of vinyl and aryl carbanions can be treated in a similar manner.18 Scheme 7 Finally a tetrasubstituted furan ring can be generated very easily by the metalation of furan 29 (Scheme 8) which was prepared using the strategy illustrated in Scheme 7.For Scheme 8 example furan 29 when treated with BuLi provided the C-5 anion 30 which when quenched with acid chloride 31 provided tetrasubstituted furan 32. Both silyl groups in 32 were easily removed to provide a 2,3,4-tribsubstituted furan 33 which is structurally different than the 2,3,4-trisubstituted furan that can be prepared via Schemes 3 6 and 7. Furan 33 was required as an intermediate towards the synthesis of (+)-xestoquinone (34 Scheme 8).19 Conversion of the hydroxy group in 33 into a triflate (quantitative) followed by an asymmetric palladium-catalyzed polyene cyclization formed rings C and D (82% yield 68% ee).Two additional steps afforded (+)-34. Tanaka and co-workers20 have been investigating how silylated furan rings behave towards conditions that favor ipsosubstitution. Treatment of 35a or 35b (prepared via Diels–Alder chemistry) with sulfuryl chloride bromine or iodine monochloride afforded 36a–e in yields ranging from 64–76% (Scheme 9). In related work they also reported an interesting [1,2] C?C silyl migration.21 When furan 37 was treated with sulfuryl chloride in acetonitrile the expected ipso-substituted product 38 was obtained.22 However when the solvent was changed to CH2Cl2 a 24:76 mixture of 38 39 was obtained in which the major compound 39 had the silyl group at C-4 and a 211 Chem.Soc. Rev. 1999 28 209–215 Scheme 9 chlorine atom at C-5. This is the first example of a [1,2] C?C silyl migration with a furan ring. Wong and co-workers have reported some very interesting furan chemistry starting from 3,4-bis(trimethylsilyl)furan (42 Scheme 10). A Diels–Alder reaction between oxazole 40 and Scheme 10 bis(trimethylsilyl)acetylene (41) followed by a retro-Diels– Alder reaction of the initial adduct liberates benzenenitrile and furan 42.23 From this relatively simple furan they have designed routes to 3,4- 2,4- and 2,3,5-substituted furan rings. For example heating 42 with acetylenic dienophiles like dimethyl acetylenedicarboxylate provided 3,4-disubstituted furans like 43 (54% yield).Wong has extended this chemistry by developing a route to the preparation of 3,4-disubstituted furans starting with 42 by performing two successive sequences.24,25 The first reaction of each sequence was an ipso-substitution with one equivalent of BCl3 to provide boroxine 44 in 98% yield (Scheme 11). The second reaction in the sequence was a Suzuki coupling between 44 and a variety of aryl and heteroaryl halides to provide 45. Repeating this sequence on 45 provided 3,4-disubstituted furans 46 in which two aryl or heteroaryl groups are present (they could be the same group or could be different). A year later Wong26 expanded upon this sequence by showing that when 44 or 47 was treated with o-bis(bromomethyl) arenes in the presence of a palladium catalyst 48 and/or 49 were formed in various ratios depending on the arene that was used.For example when the reaction between 44 and 2,3-bis(bromomethyl)quinoxaline (50) was performed only 49 Chem. Soc. Rev. 1999 28 209–215 212 Scheme 11 was obtained in 90% yield. He used this unique result with quinoxaline 50 to his advantage by homo-coupling a variety of boroxines 47 (R = CH2C6H4-p-CO2Me n-Bu C6H4-o-Me C6H4-p-Me and C6H5) in yields ranging from 52–80%. One interesting application of the chemistry shown in Scheme 12 Scheme 12 was to take compound 49 (R = SiMe3) and treat it with one equivalent of BCl3 to form the boroxine 51 (Scheme 13). When 51 was homo-coupled in the presence of 50 and a palladium catalyst quaterfuran 52 was formed.A repeat of the sequence provided octifuran 53. In addition to the above interesting chemistry Wong and coworkers27 have reported a mono-ipso-iodination of 42 and its use in the preparation of 3,4-disubsituted furan rings (Scheme 14). Treatment of 42 with iodine in the presence of silver trifluoroacetate provided furan 54 in 80% yield. With 54 in hand the authors performed Heck Stille and Suzuki reactions that provided a large number of 3-substituted-4-(trimethylsilyl) furans 55. The second trimethylsilyl group was then replaced using Wong’s boroxine protocol (see Scheme 11) to Most recently Wong and colleagues28 have reported new strategies towards the synthesis of 2,3- 2,4- 2,3,4- and 2,3,5-substituted furan rings.The approach towards the preparation of 2,3-disubstituted furan rings was developed by observing that when furan 42 is heated in a sealed tube at 160 °C in the presence of trifluoroacetic anhydride containing a catalytic amount of trifluoroacetic acid one of the silyl groups undergoes a [1,2] C?C silyl migration to provide 2,4-bis(trimethylsilyl) furan (60) in 80% yield (Scheme 15). Direct lithiation of 60 Scheme 15 with BuLi and trapping of the anion with electrophiles provided furan 61. The TMS group at C-5 could be selectively ipsoiodinated to give 62. The iodine was easily removed by treatment of 62 with LAH providing 2,3-disubstituted furan 63. Finally 2,3-disubstituted furans 64 were prepared by using Wong’s boroxine protocol followed by a Suzuki coupling.An alternative strategy was developed towards compounds like 64 as the overall yield of 64 was low starting with 42.28 Furan 6523,29 was lithiated exclusively at the less hindered C-2 position with t-BuLi (Scheme 16). Trapping of the anion with Scheme 13 Scheme 14 Scheme 16 electrophiles yielded 66 which could be proto-desilylated with trifluoroacetic acid to give 67 in high overall yield. give 56. Furan 56 could be treated with I2 in the presence of AgBF4 to provide iodide 57 which was used to prepare acetylenic furans like 58. Or 56 could be used directly in a Suzuki coupling to give 3,4-disubstituted furans 59. The monoreplacement of the trimethylsilyl groups in 42 clearly expands the scope of the use of furan 42 in synthesis.213 Chem. Soc. Rev. 1999 28 209–215 Starting from furan 68 Wong was able to prepare 2,4-disubstituted furans (Scheme 16).28 Treatment of 68 with t-BuLi in THF at 0 °C and trapping of the anion with benzyl bromide provided only furan 69 in which a regiospecific lithiation at the C-5 position of 68 had occurred. Using his boroxine chemistry followed by a Suzuki coupling resulted in the preparation of 2,4-disubstituted furans 70. As might be expected the boroxine generated and isolated by the treatment of 69 with BCl3 can either be homo-coupled to provide 71 or iodo-deboronated to provide 72 which can be used further to prepare other 2,4-disubstituted furan rings. Compound 54 has been used to prepare 2,3,4-trisubstituted furan rings (Scheme 17).28 A nickel catalyzed cross-coupling Scheme 17 reaction with butylmagnesium chloride and 54 gave 73.Regiospecific lithiation of 73 at C-2 with t-BuLi and trapping the anion with an electrophile provided 74 in which the silyl group could be replaced by an iodine atom albeit in low yield thereby providing a route to 2,3,4-trisubstituted furans 75. Finally Wong28 has developed a route to 2,3,5-trisubstituted furans 77 (Scheme 18). Furan 62 prepared according to Scheme Scheme 18 15 was treated with p-tolylmagnesium bromide in the presence of a nickel catalyst to provide 76. The trimethylsilyl group in 76 was then replaced with aromatic rings via Wong’s boroxine chemistry followed by a Suzuki reaction giving 77.The last few schemes clearly show the important role that a trimethylsilyl group plays in the preparation of a variety of furan substitution patterns. Not only can the silyl group be ipsosubstituted by boron and iodine but the steric size of the group can help direct lithiation to a site remote (i.e. less hindered) from the silane. So far the silyl groups have been used for ipso-substitution directing metalation reactions and removed completely by Chem. Soc. Rev. 1999 28 209–215 214 treatment with TBAF through a silyl migration or through a proto-desilylation reaction with acid. An additional use of the trimethylsilyl group is to treat 2-substituted silylated furans with either a peracid or singlet oxygen. This results in a regiospecific conversion of the silylated furan to a butenolide in which the carbonyl group is attached to the carbon atom that the silyl group was initially attached.For example Kuwajima30 reported that furan 78 when treated with peracetic acid formed but- 3-enolide 79 in yields ranging from 36–84% (Scheme 19). Scheme 19 Goldsmith and Liotta31 later showed that 2-silylated-4-substituted furan ring 80 also underwent an oxidation reaction with peracetic acid but provided but-2-enolide 81 in 78% yield. Tannis32 has also shown that 2-silylated-3-substituted furan rings 82 undergo a similar regiospecific oxidation to give a 1 1 mixture of but-2- and -3-enolides 83 and 84 in 78% yield. Adam and Rodriguez33 have reported that 5-substituted-2-(trimethylsilyl) furan 85 can be oxidized to 4-hydroxy-4-substitutedbut-2-enolide 86 in quantitative yield.Similarly 3- or 4-substituted-2-(trimethylsilyl)furans 87 and 88 can be oxidized regiospecifically to 2- or 3-substituted-4-hydroxy-but-2-enolides 89 and 90 respectively in yields ranging from 89–94%.34 Conclusions This brief review clearly shows the important role that silicon has played in the development of new strategies for the preparation of multi-substituted furan rings. The size migratory aptitude tendency to undergo ipso-substitutions and their ease of attachment to and removal from furan systems of silyl groups has been used to the fullest. Some future endeavors might include using fluorinated silanes so that they can be directly replaced by other functionalities using palladium-coupling reactions.35 This would eliminate the need for an ipsoreplacement of the silane with an iodine or boron atom.Very little work has been done with the more robust tert-butyldimethylsilyl group. Methods for its replacement with other functionalities rather than just removing it would offer additional alternatives to those involving the trimethylsilyl group. References 1 B. A. Keay and P. W. Dibble in Comprehensive Heterocyclic Chemistry II ed. C. W. Bird Elsevier New York 1996 vol. 2 ch. 2.08 pp. 395–436. 2 X. L. Hou H. Y. Cheung T. Y. Hon P. L. Kwan T. H. Lo S. Y. Tong and H. N. C. Wong Tetrahedron 1998 54 1955 and references therein. 3 R. A. Benkeser and R. B. Currie J. Am.Chem. Soc. 1948 70 1780. 4 A. J. Carpenter and D. J. Chadwick Tetrahedron Lett. 1985 26 1777. 5 D. W. Knight and A. P. Nott J. Chem. Soc. Perkin Trans. 1 1983 791. 6 D. J. Chadwick M. V. McKnight and R. Ngochindo J. Chem. Soc. Perkin Trans. 1 1982 1343. 7 D. S. Ennis and T. L. Gilchrist Tetrahedron Lett. 1990 46 2623. 8 J.-Y. Lenoir P. Ribereau and G. Queguiner J. Chem. Soc. Perkin Trans. 1 1994 2943. 9 E. Bures P. G. Spinazze G. Beese I. R. Hunt C. Rogers and B. A. Keay J. Org. Chem. 1997 62 8741. 10 E. Bures J. A. Nieman S. Yu P. G. Spinazze J.-L. J. Bontront I. R. Hunt A. Rauk and B. A. Keay J. Org. Chem. 1997 62 8750. 11 R. E. Danso-Danquah A. I. Scott and D. Becker Tetrahedron 1993 49 8195. 12 J. A. Nieman and B. A. Keay Tetrahedron Lett.1994 35 5335. 13 E. J. Bures and B. A. Keay Tetrahedron Lett. 1988 29 1247. 14 S. Katsumura K. Ichikawa and H. Mori Chem. Lett. 1993 1525. 15 R. L. Beddoes M. L. Lewis P. Gilbert P. Quayle S.P. Thompson S. Wang and K. Mills Tetrahedron Lett. 1996 37 9119. 16 B. A. Keay and J.-L. Bontront Can. J. Chem. 1991 69 1326. 17 W. A. Cristofoli and B. A. Keay Tetrahedron Lett. 1991 32 5881. 18 S. P. Maddaford and B. A. Keay J. Org. Chem. 1994 59 6501. 19 S. P. Maddaford N. G. Andersen W.A. Cristofoli and B.A. Keay J. Am. Chem. Soc. 1996 118 10766. 20 K. Nakayama Y. Harigaya H. Okamoto and A. Tanaka J. Heterocycl. Chem. 1991 28 853. 21 K. Nakayama and A. Tanaka Chem. Pharm. Bull. 1992 40 1966. 22 K. Nakayama and A. Tanaka Chem. Express 1991 6 699. 23 M. S. Ho and H. N. C. Wong J. Chem. Soc. Chem. Commun. 1989 1238. 24 Z. Z. Song Z. Y. Zhou T. C. W. Mak and H. N. C. Wong Angew. Chem. Int. Ed. Engl. 1993 32 432. 25 Z. Z. Song M. S. Ho and H. N. C. Wong J. Org. Chem. 1994 59 3917. 26 Z. Zhong and H. N. C. Wong J. Org. Chem. 1994 59 33. 27 Z. Z. Song and H. N. C. Wong Liebigs Ann. Chem. 1994 29. 28 M. K. Wong C. Y. Leung and H. N. C. Wong Tetrahedron 1997 53 3497. 29 Z. Z. Song and H. N. C. Wong J. Chin. Chem. Soc. 1995 42 673. 30 I. Kuwajima and H. Urabe Tetrahedron Lett. 1981 22 5191. 31 D. Goldsmith D. Liotta M. Saindane L. Waykole and P. Brown Tetrahedron Lett. 1983 24 5835. 32 S. P. Tannis and D. Head Tetrahedron Lett. 1984 25 4451. 33 W. Adam and A. Rodriguez Tetrahedron Lett. 1981 22 3505. 34 S. Katsumura K. Hori S. Fujiwara and S. Isoe Tetrahedron Lett. 1985 26 4625 and Tetrahedron Lett. 1988 29 1173. 35 Y. Hatanaka S. Fukushima and T. Hiyama Tetrahedron 1992 48 2113 and references therein. Review 8/09439J 215 Chem. Soc. Rev. 1999 28 209–215

ISSN:0306-0012    

DOI:10.1039/a809439j

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Physico-chemical treatment methods for the removal of microcystins (cyanobacterial hepatotoxins) from potable waters Linda A. Lawton and Peter K. J. Robertson School of Applied Sciences The Robert Gordon University St Andrew Street Aberdeen UK AB25 1HG Received 6th January 1999 The incidence of cyanobacterial blooms in freshwaters including drinking water reservoirs has increased over the past few decades due to rising nutrient levels. Microcystins are hepatotoxins released from cyanobacteria and have been responsible for the death of humans as well as domestic and wild animals. Microcystins are chemically very stable and many processes have only limited efficacy in removing them. In this paper we review a range of water treatment methods which have been applied to removing microcystins from potable waters.1 Introduction Cyanobacteria (blue–green algae) produce several types of toxins that can be harmful to humans. The most frequently Dr Linda Lawton is a Lecturer in Environmental Science within the School of Applied Sciences of The Robert Gordon University. She has eleven years experience in the field of toxic cyanobacteria with her key areas of expertise in the chemical analysis of cyanotoxins including extraction purification and chemical characterisation. Over a number of years she has focused much of her attention on the detection and removal of cyanotoxins from potable water including the development of non-mammalian bioassays to determine the toxicity of degradation products. Currently much of her research effort in conjunction with Dr Peter Robertson has been targeted towards the use of photocatalysis as potential treatment method for the removal of microcystins from drinking water.Dr Lawton is Secretary to the Department of the Environment’s Standing Committee of Analysts Toxins Panel (6.11) and has recently been elected to be a member of the Society for General Microbiology Environmental Microbiology Group Committee. Finally she is currently acting as an editorial advisor for a World Health Organisation Publication Toxic cyanobacteria in water a guide to public health monitoring. Linda A. Lawton occurring are the microcystins a group of at least 60 heptapeptides which share the common structure cyclo(d-Alal-X-erythro-b-d-methylaspartic acid-l-Y-ADDA-d-isoglutamic acid-N-methyldehydroalanine) where X and Y are variable l amino acids and ADDA is a unique 20-carbon amino acid (3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4,6-dienoic acid).Microcystins are named according to the variable amino acids that they contain. Microcystin-LR (Fig. 1) one of the most commonly occurring and frequently studied microcystins contains leucine (L) and arginine (R) in the variable positions. It has been established that microcystin toxicity is due to their potent and irreversible inhibition of important regulatory enzymes (protein phosphatase 1 and 2A) which can cause both acute and chronic effects. Of great concern is the tumour promoting activity of microcystins and hence there have been recent moves to minimise exposure through the publication of a guideline value (1 mg dm23 in potable water) by the World Health Organisation.A full discussion and extensive On completing his D.Phil. at the University of Ulster in October 1989 Dr Peter Robertson joined the Faraday Centre in Carlow Ireland. In this post he was involved in a wide range of electrochemical research projects including bulk electrolysis processes and environmental treatment systems. In 1991 Dr Robertson moved to the Industrial Research and Technology Unit just outside Belfast where he was a Higher Scientific Officer. This government body acted as an industrial consultancy and research unit for government and industry in Northern Ireland.In this position he managed a range of research projects on photocatalytic and electrochemical waste treatment for local industry. Dr Robertson also assisted several companies in solving contamination problems that they were experiencing and aided them with product developments. Dr Robertson joined the staff of the School of Applied Sciences at Robert Gordon University in September 1995. The main areas of his research centre around photocatalytic treatment of potable water and industrial waste effluents heavy metal recovery and homogenous photocatalysis. He is also involved in the development of heavy metal and hydrocarbon sensors for the marine estuarine and fresh water environments. Peter K. J. Robertson 217 Chem. Soc.Rev. 1999 28 217–224 Fig. 1 Microcystin-LR. referencing of many aspects of microcystins can be found in reference 1.1 Cyanobacteria typically occur in high numbers (often referred to as a bloom) in nutrient-rich water bodies. These blooms are usually found in lakes and reservoirs but more recently very slow flowing rivers have also been affected. Increasing awareness and/or occurrence has focused attention on the potential impact on human health of microcystins in drinking water. Furthermore the recent fatalities of dialysis patients2 due to contaminated water highlight the essential need for reliable treatment and monitoring methods. Microcystins are known to be relatively stable compounds possibly as a result of their cyclic structure. They are reported to withstand many hours of boiling and may persist for many years when stored dry at room temperature.It is therefore not surprising that microcystins are not readily removed from drinking water by conventional treatment methods. The investigation of suitable methods for their removal spans over twenty years beginning even before microcystins were structurally characterised. It is only in recent years that the use of a broad range of advanced techniques for toxin removal has been applied and evaluated. Microcystins are produced and retained within healthy cyanobacterial cells and are only released into the surrounding water when cells die and break open. They can therefore enter water treatment works either as dissolved compounds in the raw water or within cells.Partitioning of this nature presents a unique challenge to water treatment processes. Removal of toxin containing cells at the early stages of treatment can be a relatively straightforward way of greatly reducing the microcystin burden. In doing so however great care must be taken to prevent cell lysis or toxins will move into the water phase and the benefit is negated. In contrast it may be more desirable to release microcystins into the liquid phase where they can be removed by treatment processes. Where whole cells are permitted to pass through the treatment system the toxins retained within may be protected from adsorption or chemical destruction. These issues are primarily a system management problem and water treatment engineers must be aware of how cells behave within individual treatment plants.For the purpose of this review we will focus on the removal of microcystins dissolved in water. 2 Physical treatment methods for the removal of microcystins Once the stability and resistance to traditional treatment methods had been established the use of alternative methods for Chem. Soc. Rev. 1999 28 217–224 218 the removal of microcystins was investigated. Early treatment studies focused primarily on the physical removal of microcystins by adsorption technologies particularly the application of activated carbon with detailed studies into the use of this technique continuing to the present day. Other removal methods including reverse osmosis have also been assessed.2.1 Activated carbon Two general types of activated carbon are utilised by the water industry for the removal of trace organics. Granular activated carbon (GAC) is typically used in flow-through column reactors whereas powdered activated carbon (PAC) can be added directly to water prior to coagulation or filtration. The removal of taste and odour problems also caused by cyanobacteria were one of the initial applications of activated carbons in the water industry with their use now expanded to a wide range of contaminants. Systematic studies by Hoffman in 1976 revealed that flocculation/sedimentation sand filtration and chlorination all failed to eliminate the toxicity associated with the peptide toxins from cyanobacteria.3 The same study however repeatedly found that filtration through activated carbon was capable of removing the toxins as was the addition of PAC although this was dependent on dosing levels.The required dosing rate appeared to vary depending on the microcystin variant under investigation but no further elucidation regarding the factors influencing this mechanism can be made since the toxins were uncharacterised at the time. Removal of both microcystins under investigation was achieved at 800 g m23 which was a relatively high PAC dosing rate (typical range 1–100 g m23). Only a tenth of this dose however was necessary to eliminate one of the toxin variants which was being studied. It can only be assumed that the two uncharacterised toxins were different microcystin variants which exhibited marked differences in sorption behaviour probably due to their differing polarities.Subsequent studies confirmed that activated carbon can be effective to varying degrees and for most carbon types high doses of PAC are required. Falconer et al.4 reported that many of the PACs in their investigation only removed a small percent of the microcystins when used at a dosing rate of 1 kg m23 with at least 50 kg m23 required for many of the carbon types to be effective. A number of investigations have assessed the suitability of a range of carbons both PAC and GAC produced from differing starting materials e.g. wood coal peat and coconut. It has however been concluded that the surface characteristics and not the parent carbon source were more important in determining sorption performance.5 Surface characteristics and therefore sorption properties result from both the nature of the starting material and the mode of activation.This study reported variation in microcystin-LR adsorption levels between 20 to 280 mg mg21 depending on the physical characteristics of PAC. It is generally assumed that the greater the available surface area the greater the adsorption efficiency of an activated carbon. Surface area is typically determined using the Brunauer Emmett and Teller (BET) method but this was not found to correspond to the effectiveness of microcystin-LR removal. Previous work on humic acids had suggested that the volume of pores in the mesopore range (2–50 nm diameter) was the main factor influencing their adsorption.Likewise adsorption of microcystin-LR (estimated diameter 1.2–2.6 nm) was greatest for carbon types which were shown to have high mesopore volumes and this would appear to be the dominant factor influencing adsorption. No correlation was found between micropore volume (diameter < 2 nm) and adsorption which suggests toxin molecules are too large to enter these pores hence preventing adsorption. Competitive adsorption can also greatly influence the amount of microcystin removed by activated carbon. The removal of microcystins in high purity water (e.g. Milli-Q) was compared to that in raw water. In each case as might be expected removal efficiencies decline due to competition for binding sites on the carbon.Problems of high natural organic matter (NOM) are particularly relevant to microcystin removal since their presence in water is accompanied by the occurrence of the many components of cyanobacterial cells. Considering that microcystins only account for around at most 1% of the dry weight of cyanobacterial cells it is only to be expected that the death of a bloom and subsequent release of microcystins will be associated with high carbon loading of the water. The incorporation of GAC filtration beds is increasingly common in modern treatment works however the cost and effort of recharging is high. This reduces the willingness to recharge the beds at relatively short intervals (i.e. months compared to years).It is therefore important to determine GAC capacity and hence breakthrough of toxin into drinking water. It has been demonstrated that breakthrough can occur after filtration of a relatively small water volume.4 When a 70 g GAC column was utilised a number of the carbon types tested removed less than 50% of the toxicity after 10 dm3 had been filtered and even the most efficient material failed after only 50 dm3. This suggests that the useful life of GAC for removal of microcystins is relatively short. Furthermore tests carried out with GAC which had been used in a water treatment plant for 5 months prior to evaluating adsorption of microcystin revealed the adsorption capacity (Kf) of the used carbon was reduced to 6.2 compared with 50 for virgin GAC of the same type.6 Some products are available for the treatment of water by the consumer.These include filter jug systems which filter water through a pre-packed cartridge which contains a mixture of GAC along with ion exchange resin. It was found that with some filter types percentage removal was highly variable and the performance of the filters declined with increasing volume.7 None of the filters removed all of the microcystin present in the water. It is not clear which component of the cartridge is removing the microcystin although it could be expected that the GAC adsorbs at least some of the toxins. This is one of the only studies to compare the removal of a number of microcystins and it was consistently found that a slightly greater percentage of the more hydrophobic microcystins (e.g.microcystin-LW) was removed compared with microcystin-LR. The findings discussed demonstrate that activated carbon can be successfully employed to ensure microcystin removal however performance may be reduced with normal water treatment practices. Furthermore little is known about the fate of microcystins adsorbed onto activated carbon. It may be biodegraded and hence rendered safe or it may be desorbed over time with associated hazards to health. 2.2 Reverse osmosis Reverse osmosis (RO) has been used in water treatment for the desalination of brackish or estuarine waters to provide a fresh potable supply. RO is similar to filtration methods in that it separates a liquid from a mixture of suspended or dissolved components.The two methods do however differ in their mode of operation. Filtration excludes components from the filtrate due to their size whereas the primary function of the RO membrane is to exclude dissolved salts through the use of a semi-permeable membrane although it does exclude organic compounds with a molecular weight above 100. As RO is currently in use in a number of countries to provide drinking water the ability of this process to remove microcystins has also been assessed. It can be assumed that microcystins will be retained by RO membranes since the molecular weight of microcystins is around 1000. Studies with microcystin-LR and -RR gave retention rates greater than 95%.8 Nodularin a closely related pentapeptide produced by the brackish water cyanobacteria Nodularia spumigina was also retained during RO water treatment.9 As the salt and toxin concentration increased in the retentate traces of nodularin were however detected in the treated water.This was paralleled by an increase in salinity of the permeate which was deemed to be a good indicator of water quality. Since brackish waters where nodularin occurs may be desalinated by RO to provide drinking water it is reassuring that correctly operated systems will also remove this potential health risk. Although RO appears to be a suitable method of producing safe drinking water it must be remembered that this process is also retaining toxin enriched water which has to be disposed safely since the toxins are not destroyed by this treatment.3 Chemical treatment methods for the removal of microcystins The first report of chemical treatment of microcystins was in 1976 when Hoffmann3 studied the effect of a range of reagents on these toxins. Since then there have been several investigations on this topic using a variety of oxidising agents which are commonly employed for treating potable water. Table 1 displays the most frequently used water treatment reagents together with their oxidation potentials in decreasing order of reactivity. Table 1 Oxidation potentials of chemical oxidants commonly used by the water industry Oxidation Potential/Volts vs. NHE Reagent Hydroxyl Radical Ozone Hydrogen Peroxide Perhydroxyl Radical Permanganate Hypochlorous Acid Chlorine Chlorine Dioxide 2.80 2.07 1.78 1.70 1.68 1.49 1.36 1.28 A common feature of the toxin destruction work reported in the literature is a lack of characterisation of decomposition products.The effectiveness of a treatment method tends to be reported in terms of how quickly the parent microcystin disappears. Occasionally toxicity testing by mouse or brine 219 Chem. Soc. Rev. 1999 28 217–224 shrimp bioassay is performed but frequently the investigation goes no further. A few researchers have carried out mass spectral (MS) analysis of products but there has been little attempt to rationalise mechanisms for the treatment processes. Most of the reagents listed in Table 1 will react readily with unsaturated bonds such as those included in the ADDA moiety of microcystin.This group is commonly used as a chromophore for analysis of the toxin (UV absorption 238 nm). Therefore if it is modified the parent toxin will seem to disappear. This does not however provide information on the by-products that are generated and the potential hazards that may be presented by these materials. The main treatment methods are considered below together with mechanisms that we propose as the probable initial modes of attack on the toxin. These mechanisms are only tentative since there is little detail available in the literature on decomposition products in order to validate our proposals. Much further work therefore will have to be performed in order to determine their validity.3.1 Chlorination Chlorine has been used as a reagent for water treatment since the last century replacing sand filtration as the disinfection method for potable water in the United States at the beginning of the 20th century. Chlorine dissolves in water forming hypochlorous acid. Above pH 5 this starts to dissociate forming hypochlorite ions with 100% dissociation above pH 10. The undissociated hypochlorous acid molecule is established as the most effective disinfecting agent. Initial work on the application of chlorination for the removal of microcystins carried out by Hoffmann,3 Keijola et al.10 and Himberg et al.11 indicated that the process was ineffective. This may have been due to the work being performed at a pH where the free chlorine concentrations were relatively low.Although Keijola and Himberg did not report the pH levels Hoffman performed his investigation at pH 8.5. The effect of pH on chlorination of microcystins was subsequently investigated and a dependence on destruction efficiency was observed.12 Nicholson et al. investigated the use of chlorine and chloramine for the destruction of microcystin and the pentapeptide toxin nodularin.12 Both toxins were destroyed provided a residual level of 0.5 mg dm23 chlorine was maintained for 30 minutes. A pH dependence was observed with NaOCl and Ca(OCl)2 being less effective at higher pH values. Table 2 Table 2 Destruction efficiencies of chlorinating agents for the destruction of microcystin-LR % Destruction after 30 minutes treatment Concentration/mg dm23 Reagent 2 > 95 > 95 ~ 40 70–80 Aqueous Cl2 Ca(OCl) NaOCl NaOCl 1 1 1 5 shows the relative efficiencies of chlorinating agents for the destruction of toxin concentrations in the range of 130–300 mg dm23 which represents highly contaminated water.The rates of toxin destruction were greatly reduced at pH levels above 8 varying from 79% at pH 7 to 0.4% at pH 10. This was believed to be due to decreasing concentrations of hypochlorous acid which is a more effective oxidant than the hypochlorite ion. No by-products of destruction were reported or mechanism for the process discussed. Chloramination was found to be an ineffective treatment for both toxins with little effect on the level of either compound after a 5 day treatment with a 20 mg dm23 chloramine solution.Chem. Soc. Rev. 1999 28 217–224 220 Tsuji et al. examined the effect of sodium hypochlorite on microcystin-LR and -RR.13 They found that the toxin disappeared with little difficulty and was dependent on the free chlorine dose. Following a 30 minute treatment 99% toxin removal was observed for free chlorine concentration of 2.8 mg dm23 while only 35% destruction was achieved at a level of 0.7 mg dm23 with contact time of 60 minutes. Several reaction products were observed by HPLC one of which was identified as a dihydroxy-microcystin. They proposed that this compound was generated by the action of the free chlorine on either of the unsaturated bonds of the ADDA group to form a dichloromicrocystin followed by hydrolysis.Stereo- and regioisomers may also have been produced. The formation of the dichloromicrocystin postulated by this group is an unlikely intermediate of decomposition. It is thought that the more plausible mechanism to occur is that the halogen dissolves in water forming hypochlorous acid. Hypochlorous acid may in turn be protonated and become a powerful electrophile which subsequently reacts with unsaturated bonds to form a chlorohydrin. The halogen in the chlorohydrin may then undergo a nucleophilic substitution reaction with the solvent to form a dihydroxy-microcystin (Scheme 1). + H2O Cl + Cl OH2 + H2O OH – OH OH Cl OH Scheme 1 Decomposition of microcystin-LR via chlorination.To date there have been no reports of the use of chlorine dioxide for the treatment of microcystins. This has been applied effectively to the treatment of many organic compounds and it is believed that it may have applications in the destruction of microcystins. Chlorination does appear to provide an effective method of removing microcystins from potable water although this is dependant on dose and the maintenance of adequate residual chlorine levels. Unfortunately the mechanism and products of decomposition remain to be fully characterised and therefore the possible formation of harmful by-products is as yet unknown. 3.2 Ozonation Ozone is an unstable gas with a relatively high oxidation potential and is used for water treatment by dispersing the gas in aqueous media.It is widely utilised for treating drinking water but it is an expensive and sometimes unpredictable reagent. Ozone combined with UV light has proved to be an effective reagent and is very efficient in destroying a wide range of organic compounds. Ozone is highly reactive towards double bonds forming carbonyl compounds via ozonides. Either of the double bonds in the ADDA group in microcystin would be susceptible to such an attack so decomposing the parent toxin (Scheme 2). The toxicity of microcystins has been shown to be associated with the ADDA olefin groups therefore cleavage of this moiety would be expected to eliminate toxicity. Harada et al.14 used ozone to cleave one of the double bonds on the ADDA group which would support this hypothesis.Early studies by Keijola et al.10 reported a 100% efficiency of 1 mg dm23 ozone for the removal of up to 60 mg dm23 microcystin. O O O O + – O O O O + O O O Scheme 2 Decomposition of microcystin-LR by ozone. Rositano and co-workers have found that 99% of microcystin was removed in 15 seconds when treated with 0.05 mg dm23 ozone.15 Their investigations found that ozone was more effective than chlorine hydrogen peroxide and potassium permanganate for the destruction of microcystin-LR which corresponds to the higher oxidation potential of ozone. They found combining with hydrogen peroxide further enhanced the efficiency of ozone treatment.A pH dependence on the destruction efficiency was observed with alkaline conditions being less favourable. This was rationalised by the fact that the oxidising potential of ozone is 1.24 volts vs. NHE (normal hydrogen electrode) in basic solutions compared to 2.07 volts vs. NHE in acidic solutions. The reported studies all suggest that ozonation is an effective method for the removal of microcystins from drinking water. Although it is still important to characterise the decomposition products and their potential health implications the overall effectiveness of ozonation as a suitable water purification method appears promising. 3.3 Permanganate The large scale application of permanganate began in the 1960s in the United States and Europe.It is a strong oxidising agent capable of destroying organic compounds and micro-organisms. Permanganate generally attacks functional groups with multiple bonds and will cleave benzene rings. Permanganate is commonly used by organic chemists for the hydroxylation of alkenes with the formation of diols (Scheme 3). It should O O O O Mn Mn O – O – O O OH – OH O HO + n M OH O – HO Scheme 3 Decomposition of microcystin-LR by permanganate. therefore be effective in decomposing microcystin and removing the toxicity by attacking the unsaturated bonds in the ADDA group. A 1 mg dm23 solution of potassium permanganate removed 95% of a 200 mg dm23 microcystin-LR solution in 30 minutes.15 A comparison of the oxidation of microcystin-LR using chlorine and potassium permanganate (both at 2 mg dm23) demonstrated that the toxin followed a similar decay curve for both oxidants.Permanganate however was observed to bring about more rapid removal of the microcystin which is consistent with the known oxidation potentials (Table 1). No by-products of decomposition were characterised. This reagent appears to show much promise but there has been little detailed study of the system for microcystin destruction. 3.4 Hydrogen peroxide The use of hydrogen peroxide in water treatment is limited. Although thermodynamic data suggest that this should be an effective oxidant (Table 1) the kinetics for many water treatment applications are unfavourable. Hydrogen peroxide has however been applied to the oxidation of phenolic wastewater and for the treatment of paper mill effluent drilling muds and other types of organic wastewater.The oxidising effectiveness of hydrogen peroxide can be enhanced by irradiating it with UV light. The light dissociates the molecule generating two highly reactive hydroxyl radicals. A range of organics including aromatics alcohols aliphatics and haloaliphatics has been successfully treated using UV/H2O2. 2O2 for the treatment of microcystin-containing water. Research has shown that hydrogen peroxide is relatively ineffective in degrading microcystin-LR with only 17% removal after 60 minutes treatment with a 20 mg dm23 solution of peroxide.16 Further work investigated the effect of a 2 mg dm23 hydrogen peroxide solution on 1 mg dm23 solution of microcystin-LR.15 Virtually no toxin destruction was observed after 10 minutes.Combination of H2O2 with ozone was however extremely effective with virtually all the toxin being removed within 30 seconds. There are no reports of the use of UV/H This reagent shows little promise as an effective treatment method for microcystins although there may be scope for the use of combined treatments such as ozone/H2O2 which has already been shown to be very powerful and UV/H2O2 which requires research to determine its suitability for microcystin removal. 3.5 Photolysis Photolytic oxidation can occur by a number of methods. To bring about photodegradation it is necessary that the compound absorbs light at a wavelength which is the same as that emitted by the light source.If this is not the case the photolytic breakdown may still be achieved using a sensitiser or relay which will absorb the light and initiates the decomposition reaction. In many cases there is an electron transfer from the excited state of the molecule undergoing degradation to ground state oxygen generating the superoxide radical anion.17 The radical cation formed by this process may then be hydrolysed. In some cases for example the photolytic degradation of halogenated aliphatic and aromatic compounds radicals may form via homolysis which subsequently react with oxygen. This process has been applied to a wide range of organic substrates including chlorinated aromatics and aliphatics nitrotoluenes phenols and oil products.A number of research groups have examined the photolysis of microcystins although few have attempted to provide any mechanistic explanation for the process.18–21. The main chromophores in the molecule are in the ADDA group and are utilised in the analytical methods. These groups are the aromatic ring and the unsaturated bonds at the 4–5 and 6–7 positions. When an electron is promoted to a p* orbital in the unsaturated bond the p bond will uncouple leaving only the s bond between the carbon atoms. The electrons in the carbon 2p orbitals will repel each other and to minimise this a rotation 221 Chem. Soc. Rev. 1999 28 217–224 about the single bond may occur. This mechanism is believed to be the process involved in the cis–trans photoisomerisation in olefins.Photoisomerisations of microcystins have been reported by Tsuji et al. and Kaya and Sano.18,19 Cycloaddition reactions with the unsaturated bond are possible where the weakened p bond in the excited state forms two s bonds with a suitable substrate. An internal cycloaddition on the ADDA grouping would be possible between the unsaturated bonds and the aromatic ring. It is important to consider the geometry of the ADDA group on the microcystin molecule. It has been proposed that this group is U-shaped and curled over the microcystin structure.22 Such a geometry would bring the benzene ring and double bonds into close proximity with one another (Scheme 4). This type of internal cycloaddi- H H H H3CO Microcystin ring H3C H CH H 3 H H H3CO Microcystin ring H3C H hn H CH3 Scheme 4 Internal photosensitised cycloaddition in microcystin-LR.tion has recently been reported for microcystin-LR.19 It should be noted however that a number of other reports have suggested a solution structure where the ADDA group is not bent in such a fashion.23,24 This suggests that the geometry is not appropriate for an internal cycloaddition. However Bagu found that the ADDA side chain is sufficiently flexible23 that in principle it could momentarily position itself favourably for the cycloaddition to occur. On photoexcitation microcystin could transfer an electron to molecular oxygen generating a microcystin radical cation and a superoxide radical anion.These species may then react with one another so generating oxidation products such as carbonyl compounds. The radical cation of the toxin could also form oxidation products by reaction with molecular oxygen. It is interesting to note that superoxide reacts 30 times faster with olefin radical cations than does molecular oxygen.25 In the aqueous systems superoxide is rapidly protonated26 generating the hydroperoxide radical. This radical may attack the double bonds of microcystin generating hydroperoxide products. In the presence of suitable sensitisers singlet oxygen may also be generated. Singlet oxygen is highly reactive towards unsaturated bonds generating peroxides hydroperoxides dioxetanes and carbonyl compounds.27 Scheme 5 displays speculative mechanisms for the photochemical oxidation of microcystin in aqueous solution.Tsuji et al. investigated the effect of sunlight on microcystin-LR and -RR.18 Irradiation alone had little effect on either microcystin with 86% of the toxin remaining after 26 days photolysis. Addition of naturally photosynthetic pigments greatly enhanced the destruction process with over 95% decomposed following 29 days photolysis. The rate of destruction depended on the pigment concentration. A photo-induced geometric isomerisation of microcystin to a less toxic isomer was also reported. At higher pigment concentrations the rate of decomposition was found to be faster than the rate of isomerisation. The pigments involved in the photolytic process are phycocyanins which on photolysis in oxygenated solutions generate both singlet oxygen Chem.Soc. Rev. 1999 28 217–224 222 Sens + O 1O2 2 + hn O O + 1O2 O + O OOH + 1O2 • + + O2 + O • – 2 HO2 • O2 • – + H + OOH + HO2 • hn Scheme 5 Potential modes of photochemical decomposition of microcystin- LR. and superoxide.28 It has been proposed that these molecules catalyse the toxin destruction via the generation of both singlet oxygen and superoxide which subsequently forms peroxide and hydroxyl radicals.21 Tsuji reported from MS data that a dihydroxy-microcystin had formed resulting from hydroxyl radical attack on the diene group of the ADDA. This would be consistent with an electron transfer type mechanism with the eventual formation of peroxide which subsequently disproportionates forming hydroxyl radicals.Further work by Tsuji et al. showed that microcystin-LR could be completely destroyed within 10 minutes under a UV light intensity of 2550 W cm22.28 When a weaker light source was used isomerisation was again observed. Welker and Steinberg have studied the effect of humic substances in the indirect photolysis of microcystins-LR -YR and -RR.29 Following 8 hours photolysis the concentrations of these toxins were 5 53 and 44% of the initial concentration. The active agents for the decomposition process were again believed to be singlet oxygen hydrogen peroxide and hydroxyl radical generated by the humic substances.Kaya and Sano reported the formation of three major nontoxic products when microcystin-LR was irradiated with UV light.19 Two of these products were geometric isomers of the ADDA group of the toxin [4(E),6(Z)-ADDA] and [4(Z),6(E)- ADDA] microcystin. The third product was a previously unidentified tricyclo-ADDA-microcystin formed by a [2+2] addition between the aromatic ring and the unsaturated bond at the 6-7 position of the ADDA group of the microcystin. While the isomeric products were stable to further photolysis the cyclic product degraded although the products of this decomposition were not identified. The authors suggested that the photolytic destruction of microcystin proceeded via the tricyclo-ADDA-microcystin-LR. UV irradiation resulted in a 50% reduction in concentration of microcystin-LR in 75 minutes.Much of the work on photolysis is inapplicable for the removal of microcystin from potable supplies. The production of modified microcystins in the natural environment through the actions of photosensitising compounds like humic acids and pigments gives valuable insight into breakdown/detoxification mechanisms. It also identifies other microcystin related compounds which may be entering water treatment systems from natural raw waters. 3.7 Semiconductor photocatalysis The use of semiconductor photocatalysis for the destruction of environmental pollutants is a well established technique with the mineralisation of a wide range of materials being reported. Semiconductors have a band structure with a filled valence band and an empty conductance band separated by a band gap.When semiconductors (SC) are illuminated with light of energy greater than the band gap electrons are promoted from the occupied valence band to the unoccupied conductance band (Fig. 2). This generates oxidising sites in the valence band and Fig. 2 The process of semiconductor photocatalysis for water purification. reducing sites in the conductance band. It is believed that preadsorbed hydroxide ions are oxidised to hydroxyl radicals at the valence band. These radicals subsequently oxidise the polluting material while at the conductance band an electron is donated to oxygen thereby generating the superoxide radical anion. The superoxide is then protonated and eventually forms hydrogen peroxide which also acts as an oxidising agent.The use of a titanium dioxide photocatalyst (1% w/v) for the removal of microcystin-LR in water has been investigated.30 It was found that even at levels of 200 mg dm–3 the toxin rapidly degraded when illuminated in the presence of the catalyst. The process appeared to follow Langmuir–Hinshelwood kinetics although a discrepancy was observed between the adsorption constants determined for the photocatalytic process and those obtained from dark isotherms. No breakdown was observed in light alone. A subsequent study investigated the factors which influenced the rate of the toxin destruction at the photocatalyst surface.31 A primary kinetic isotope effect of approximately 3 was observed when the destruction was performed in a heavy water solvent.Hydroxylated compounds were observed as products of the destruction process. No destruction was observed when the process was investigated under a nitrogen atmosphere. Oxygen must be present in order to react with electrons in the conductance band and products of this reduction e.g. peroxide may be involved in the destruction of microcystin. A more detailed mechanistic study of the photocatalytic destruction of microcystin showed that the toxin disappearance was accompanied by the appearance of seven UV detectable compounds.32 Spectral analysis revealed that some of these compounds retained spectra similar to the parent compound suggesting that the ADDA moiety thought to be responsible for the characteristic spectrum remained intact whereas the spectra of some of the other products were more radically altered.Six of the seven observed reaction products did not appear to undergo further degradation during prolonged photocatalysis (100 min). The degree to which microcystin-LR was mineralised by photocatalytic oxidation was determined and it was found that less than 10% mineralisation occurred. Since more than 90% of the toxin is not mineralised it is important to fully characterise the products of decomposition and assess their health impact. Shepard et al.33 have also reported the use of TiO2 photocatalysis for the degradation of microcystin-LR and two other microcystins YR and YA. Rapid decomposition was observed with a half life of less than 5 minutes for each toxin.These findings suggest that photocatalytic destruction of microcystins may be a suitable method for the removal of these potentially hazardous compounds from drinking water. As for all the other treatment processes detailed characterisation of the decomposition by-products is necessary. 4 Conclusion The water treatment methods reviewed here could be applied to the removal of microcystins with varying degrees of ease and effectiveness. A number of oxidative methods have been discussed and although most appear to be suitable many exhibit limitations. Chlorination a well established technique may appear effective but the relatively high dosing rates and contact time required is outside commonly adopted protocols.Its use would therefore have to be carefully monitored and it is essential that further work is carried out to determine the potential hazards associated with by-products. Ozonation is very effective for the rapid removal of cyanotoxins from potable water and could provide an appropriate treatment method. Cost may prove to be prohibitive particularly since contamination with microcystins is typically seasonal and unpredictable. Of the other oxidative methods discussed photocatalytic degradation does appear to be very promising. This is a relatively new treatment method and it remains to be seen how well it will perform but initial findings are very favourable. Its great strengths are that application should be relatively simple and easy to operate it should mineralise most organics with a limited chance for the production of harmful by-products and it is potentially a sustainable and clean technology.Both established and new advanced technologies appear to render microcystin contaminated water safe to drink. It is important however that a greater effort is made to understand the mechanisms by which the disappearance of microcystins is achieved. Furthermore work must be done to structurally characterise the breakdown products generated from chemical treatment methods. Where possible the effects of acute and chronic exposure to common by-products should be carried out. By their nature methods which physically remove microcystins can be used without concern about by-product generation but their safe use must be monitored closely to ensure systems are functioning as expected.It was lack of vigilance in the use of both GAC and RO that allowed over 50 dialysis patients in Brazil to die as a result of microcystin exposure2. Most work has centred on microcystin-LR. Considering there are over 60 different variants of microcystin it will be important to investigate the suitability of each process on representative variants. The effective use of physico-chemical methods will be hampered by increased NOM associated with cyanobacterial blooms. This will greatly reduce the efficiency of the treatment processes due to competition for active species or binding sites. 5 References 1 Toxic Cyanobacteria in Water Eds.I. Chorus and J. Bartram E & FN Spon London 1999. 223 Chem. Soc. Rev. 1999 28 217–224 2 S. Pouria A. deAndrade J. Barbosa R. L. Cavalcanti V. T. S. Barreto C. J. Ward W. Presser G. K. Poon G. H. Neild and G. A. Codd The Lancet 1998 352 21. 3 J. R. H. Hoffman Water SA 1976 2 58. 4 C. Donati M. Drikas R. Hayes and G. Newcombe Water Res. 1994 28 1735. 5 I. R. Falconer M. T. C. Runnegar and V. L. Huynh Tenth Federal Convention of the Australian Water and Wastewater Association Sydney 1983 p. 26–1. 6 T. W. Lambert C. F. B. Holmes and S. E. Hrudey Water Res. 1996 30 1411. 7 L. A. Lawton B. J. P. A. Cornish and A. W. R. MacDonald Water Res. 1998 32 633. 8 U. Neumann and J. Weckesser Environ. Toxicol. Water Qual. 1998 13 143.9 E. Vuori A. Pelander K. Himberg M. Waris and K. Niinivaara Water Res. 1997 31 2922. 10 A. M. Keijola K. Himberg A. L. Esala K. Sivonen and L. Hiisvirta Toxic. Assess. 1988 3 643. 11 K. Himberg A. M. Keijola L. Hiisvirta H. Pyysalo and K. Sivonen Water Res 1989 23 979. 12 B. C. Nicholson J. Rositano and M. D. Burch Water Res 1994 28 1297. 13 K. Tsuji T. Watanuki F. Kondo M. Watatabe S. Suzuki H. Nakazawa M. Suzuki H. Uchida and K. Harada Toxicon 1997 35 1033. 14 K.-i. Harada H. Murata Z. Quiang M. Suzuki and F. Konda Toxicon 1996 34 701. 15 J. Rositano B. C. Nicholson P. Pieronne Ozone Sci. Technol. 1998 20 223. 16 M. Drikas Control and/or removal of toxins in Toxic Cyanobacteria Current Status of Research and Management Eds D.A. Steffensen and B. C. Nicholson Australian Centre for Water Quality Salisbury Australia 1994 p. 93. Chem. Soc. Rev. 1999 28 217–224 224 17 O. Legrini E. Oliveros and A. M. Braun Chem. Rev 1993 93 671. 18 K. Tsuji S. Nalto F. Kondo N. Ishikawa M. F. Watanabe M. Suzuki and K.-I. Harada Environ. Sci. Technol. 1994 28 173. 19 K. Kaya and T. Sano Chem. Res. Toxicol. 1998 11 159. 20 T. Lanaras C. M. Cook J. E. Eriksson J. A. O. Meriluoto and M. Hotokka Toxicon 1991 901 906. 21 J. R. Bagu F. D. Sönnichsen D. Williams R. J. Andersen B. D. Sykes and C. F. B. Holmes Struct. Biol. 1995 2 114. 22 J. Goldberg H.-B. Huang Y.-G. Kwon P. Greengard A. C. Nairn and J. Kuriyan Nature 1995 376 745. 23 M. Julliard and M. Chanon Chem. Rev. 1983 83 425. 24 E. R. Carraway A. J. Hoffmann and M. R. Hoffmann Environ. Sci. Technol. 1994 28 494. 25 C. S. Foote in Singlet Oxygen Reactions with Organic Compounds and Polymers B. Rånby and J. F. Rabek Eds. John Wiley and Sons New York 1978 pp. 135. 26 J.-A. He Y.-Z. Hu and L.-J. Jiang Biochim. Biophys. Acta 1997 1320 165. 27 P. K. J. Robertson L. A. Lawton and B. J. P. A. Cornish J. Porphyrins Phthalocyanines 1999 accepted for publication. 28 K. Tsuji T. Watanuki F. Kondo M. Watatabe S. Suzuki H. Nakazawa M. Suzuki H. Uchida and K. Harada Toxicon 1995 33 1619. 29 M. Welker and C. Steinberg Water Res. 1999 33 1159. 30 P. K. J. Robertson L. A. Lawton B. Münch and J. Rouzade Chem. Commun. 1997 4 393. 31 P. K. J. Robertson L. A. Lawton B. J. P. A. Cornish and M. Jaspars. J. Photochem. Photobiol. A. Chem. 1998 116 215. 32 L. A. Lawton P. K. J. Robertson B. J. P. A. Cornish and M. Jaspars Environ. Sci. Technol. 1999 33 771. 33 G. S. Shepard S. Stockenstrom D. De Villiers W. J. Engelbrecht E.W. Sydenham and G. F. S. Wessels Toxicon 1998 36 1895. Review 8/05416I

ISSN:0306-0012    

DOI:10.1039/a805416i

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Nucleophilic routes to selectively fluorinated aromatics Dave J. Adams and James H. Clark Chemistry Department University of York Heslington York UK YO10 5DD Received 6th November 1998 Selectively fluorinated aromatic compounds are of interest in many sectors. Of the methodologies used for their formation halogen exchange is the only industrial rival to the diazonium based routes. Although many compounds can be formed by halogen exchange the formation of metafluorinated species by this route is difficult. One possible method for overcoming this is fluorodenitration. Although traditionally plagued by side reactions recent reports suggest that with careful control of the reaction conditions fluorodenitration could well provide a viable industrial alternative for the formation of selectively fluorinated aromatic compounds.1 Introduction Selective fluorination and its commercial importance Selectively fluorinated compounds are used in many areas including the pharmaceutical agrochemical and dye industries. Such compounds have applications as herbicides and fungicides as well as being used in the treatment of cancer. Interest in selectively fluorinated aromatic compounds arises from the unique properties of the C–F bond and the effect of the incorporation of fluorine on the physical and chemical properties of organic molecules.1,2 The high strength of the C–F bond can result in the inhibition of metabolism when fluorine is incorporated at or near a reactive site. Fluorine is also one of the smallest available substituents which can be significant for molecules in which molecular conformation is important.Thirdly the electron-withdrawing effect of fluorine can have profound effects on the reactivity of other functional groups in the molecule. While organofluorine compounds are rare in nature the interest in such compounds is demonstrated by the vast numbers of synthetic products which are now available. Dave Adams graduated from the University of Leeds in 1995. He obtained his D.Phil. from the University of York in 1998 for work on the fluorination and trifluoromethylation of aromatic molecules via denitration. He is currently undertaking post-doctoral research at York with James Clark. James H. Clark Dave J. Adams O F Cl NO NO2 NO2 F F Selectively fluorinated compounds can be formed by a number of methods.1,2 Of the nucleophilic methodologies available Balz–Schiemann HF-diazotisation and halogen exchange are well established routes to such compounds.Both Balz–Schiemann and HF-diazotisation are based around the conversion of a diazonium salt to a fluoroaromatic with hazardous and toxic reagents being required. In halogen exchange reactions a chloro group is displaced by a fluoride ion to yield the corresponding fluoroaromatic Scheme 1. First 2 2 225 NO2 Scheme 1 The halogen exchange of 2,4-dinitrochlorobenzene. discovered in 1936 by Gottlieb3 for the conversion of 2,4-dinitrochlorobenzene to 2,4-dinitrofluorobenzene (Sanger’s reagent used for the labelling of peptides and terminal amino acid groups in proteins) it is now routinely used on an industrial scale to make a wide range of compounds.While the chloro moiety is a good leaving group it is known that aromatic nitro groups are much more labile.4 Indeed in a few cases the nitro group has been shown to be more labile than fluorine. There are many reports of the displacement of a nitro group as a means of synthesising substituted aromatics.4 This coupled with the ready availability and relatively low cost of many nitroaromatics makes these compounds attractive starting materials for the synthesis of fluoroaromatics via fluorodenitration. Finger and Kruse were the first to report fluorodenitration initially as an unwanted side reaction.5 In 1956 whilst attempting to fluorodechlorinate 2,4-dinitrochlorobenzene using KF at 150 °C brown fumes presumably due to the liberation of NO were observed and low boiling products were detected on work- James Clark is a graduate of Kings College London.He joined the academic staff at York in 1979 and now holds the position of Professor of Industrial and Applied Chemistry. His research interests include the catalysis of organic reactions using mesoporous solids new materials and fluorine chemistry. Chem. Soc. Rev. 1999 28 225–231 up. Further investigation revealed the presence of 1,2-difluoro- 4-nitrobenzene indicating that displacement of a nitro group was also possible under these conditions Scheme 2. Although F F Cl NO F 2 NO2 + NO2 NO2 NO2 Scheme 2 First observation of fluorodenitration.fluorodenitration is currently not widely used for the formation of fluoroaromatics recent reports indicate that it should be possible to exploit this methodology to complement the currently available routes. F Cl Cl Cl Cl Cl Scheme 3 The fluorodenitration of 2,3,5,6-tetrachloronitrobenzene. ever the greater inductive effect of the nitro group compared to that of a chlorine atom makes its displacement more favourable under the same activation. For example fluorination of 1-chloro-3-nitrobenzene only gives a low yield of 1-fluoro- 3-nitrobenzene but the fluorodenitration of 1,3-dinitrobenzene 2 Applications and importance Halogen exchange is an important procedure and represents the only real industrial rival to the diazonium-based methods for the synthesis of selectively fluorinated aromatic compounds.Starting materials are generally readily available and there is a major advantage in avoiding the use of hydrogen fluoride. Activation of the chloro group by other moieties on the ring is necessary for substitution to occur. This can be achieved through the presence of inductively activating 2I groups on the ring or mesomerically activating groups ortho- and para- to the group to be displaced. Generally cost restricts this to the use of nitro groups although CF3 CN CHO and COOMe have also been used. A wide range of fluorinated compounds can be formed by halogen exchange. However there are areas where halogen exchange is generally inappropriate or inefficient.One of the limitations of halogen exchange reactions is that while good yields can be obtained from compounds with chlorine ortho- or para- to an electron-withdrawing group meta-chloro compounds without any other activating groups generally give poor yields. A rare example of good yields of meta-fluoroaromatic products being achieved by halogen exchange is the formation of 3,4-difluorobenzonitrile from the corresponding dichlorobenzonitrile in 65% yield but even in this case the reaction had to be carried out in 1,3-dimethylimidazolidin-2-one at 290 °C in a pressure reactor.6 Fluorodenitration could provide a useful alternative route to meta-fluorinated aromatic compounds.Activation by other groups is still necessary for substitution to occur but in addition to inductive and mesomeric activation bulky ortho-groups are thought to assist by twisting the nitro group out of the plane of the molecule aiding fluorodenitration.7 So just as nitro groups were used to activate the chlorine towards displacement in 2,4-dinitrochlorobenzene the nitro group was found to be activated by the inductive effects of the chlorine substituents as well as by steric congestion enabling the quite facile fluorodenitration of 2,3,5,6-tetrachloronitrobenzene Scheme 3.7 How- NO2 Cl Cl 226 Cl Chem. Soc. Rev. 1999 28 225–231 to produce this product in yields above 80% is welldocumented. 8 Similarly Maggini et al.9 describe how 1,2-difluoro-4-(trifluoromethyl)benzene an industrial intermediate could not be formed via the fluorination of the 1,2-dichloro analogue under any reaction conditions.However it was found that the required product could be formed by halogen exchange and subsequent fluorodenitration of 1-chloro-2-nitro-4-(trifluoromethyl) benzene. Traditionally many commercial organofluorine compounds have been based on fluoroaromatics with the fluorine in an ortho or para position due to the availability of the precursors from halogen exchange chemistry. Increasing the availability of meta-fluoroaromatics would open the door to a wider range of product molecules for pharmaceutical agrochemical and other applications. One possible advantageous aspect of fluorodenitration is in the inherently greater leaving group ability of the nitro group as compared to the chloro group which can mean that milder reaction conditions are required.The reduction of reaction temperatures from the high ( > 100 °C) values often required with halogen exchange has clear safety advantages and the substitution of high boiling point dipolar aprotic solvents could also reduce concerns over toxicity side reactions and solvent recovery. The better leaving ability of the nitro group compared to the chloro group can also be exploited in the direct nucleophilic fluorinations of weakly activated substrates. Thus the halogen exchange route to 4,4A-difluorobenzophenone (the key intermediate in the manufacture of the speciality polymer poly(ether ether ketone) PEEK) starting from the 4,4A-dichloro analogue only occurs under very forcing conditions.However it has been shown that by starting from the dinitro analogue good yields of the difluorinated product can be obtained under more moderate conditions (see later).10 One area where fluorodenitration is of particular use is in radiolabelling. Positron emission tomography a non-invasive technique which is used in the in vivo visualisation of free radicals requires the use of radiolabelled compounds. Unlike 11C 13N and 15O 18F has a sufficiently long half-life to allow the synthesis and administration of such radiolabelled species. 11,12 Fluorodenitration is an attractive means of forming 18F-labelled fluoroaromatics due to both the one-pot and rapid nature of the synthesis.For example fluorodenitration of 2-nitrobenzonitrile with Rb18F in DMSO gave an 85% yield of the 18F-labelled benzonitrile after only 20 minutes at 150 °C.12 An added advantage of this methodology is that any other nitro group present to activate the group towards displacement can be easily converted into other functionalities via the diazonium salt. A variety of substituted nitroaromatics can be converted to the 18F-labelled compounds both with electron-donating and electron-withdrawing groups on the ring. 3 Solvents and reagents Although it is possible to carry out both halogen exchange and fluorodenitration reactions in neat substrate these reactions require high temperatures and long reaction times.Finger and Kruse were the first to recognise that the use of solvents was advantageous.5 Reactions were found to occur at lower temperatures than in neat substrate.5 Another advantage was that the fluorodenitration of 3-nitrophthalic anhydride proceeded safely without thermal runaway when carried out in a solvent.13 However the choice of solvent is important. In protic solvents strong hydrogen bonds are formed between the solvent and the fluoride anion.14 This results in a reduction in the nucleophilicity of the fluoride and also activates the solvent such that this can then act as a nucleophile leading to unwanted side-products. However in dipolar aprotic solvents such as dimethyl sulfoxide (DMSO) and N,N-dimethylformamide (DMF) hydrogen-bonding does not occur to any significant extent and the nucleophilicity of fluoride is much increased.14 It is also known that the nucleophilicity and basicity of a fluoride is solvent dependent as well as being sensitive to the level of hydration and the identity of the counter-cation.When not hydrogen-bonding to solvent molecules fluoride salts are known to be strongly basic which again can lead to undesired side reactions. The high polarity of some dipolar aprotic solvents is thought to be a major advantage in carrying out these reactions. Such solvents are thought to stabilise the organic reaction intermediates. Clark and Macquarrie15 found that for a range of organic substrates the rates of both halogen exchange and fluorodenitration with KF in aprotic solvents were subject to substrate dependent solvent effects.Under otherwise identical experimental conditions rates were found to follow the order DMSO � DMAc (N,N-dimethylacetamide) > sulfolane » acetonitrile. DMSO was found to be the solvent in which the rate of fluorodenitration is greatest despite the low solubility of KF in this solvent. Indeed there was found to be a poor correlation between the rates of reaction of aromatic compounds in different solvents and the solubility of the fluoride. However it was found that the order of solvent dependent rate enhancement did correlate with the degree of resonance stabilisation of the intermediate s-complex. Like other SNAr reactions,16 fluorodenitration and halogen exchange are thought to proceed via an intermediate in which the a-carbon is sp3 hybridised.Benzenoid resonance is thus lost in the intermediate NO2 – NO2 NO2 –NO F– 2 – NO F 2 F NO2 Scheme 4 Mechanism by which fluorodenitration occurs. Scheme 4. Electron-withdrawing groups ortho and para to the a-carbon therefore stabilise the intermediate resonance forms. The greatest solvent dependencies were found with those substrates with no effective 2M groups whereas those with two 2M groups gained very little in going from acetonitrile to the more polar DMSO. This suggests that the stabilisation of the sintermediate is of major importance in determining the overall rate of reaction. From an industrial point of view the high cost of the solvent makes both use and recovery of major importance to process design.Environmental concerns also make recovery important. As such there is a growing trend towards the development of low- or no-solvent processes. For example Chambers and Edwards have recently described the use of perfluorocarbons to “bulk-up” dipolar aprotic solvents and found that it is possible to carry out halogen exchange reactions in 75% perfluoroperhydrophenanthracene 25% sulfolane using KF and 18-crown-6.17 The perfluorocarbon could easily be recovered although the recycled solvent was of reduced efficiency. It is entirely feasible that similar solvent systems may be suitable for fluorodenitration. 4 Fluoride sources In general only monovalent fluorides have a sufficiently low lattice energy to be reactive in halogen exchange and fluorodenitration reactions.There are many examples of the use of alkali metal fluorides although while lithium and sodium fluoride would be preferred on the grounds of cost they are inactive in these systems. Alkali metal fluorides (apart from lithium) are hygroscopic and must be dried prior to use (as mentioned above the nucleophilicity of the fluoride is sensitive to the hydration level of the system). The high thermal stability of these salts allows drying by heating under vacuum at 100 °C for several hours.14 KF is the most often used alkali metal fluoride for these reactions providing a compromise between reactivity and cost. Alkali metal fluorides are only significantly soluble in some protic solvents (notably water HF and the lighter carboxylic acids and alcohols).In dipolar aprotics solubility is low (a value of 0.6 3 1023 mol dm23 for KF in DMSO has recently been reported18). This often necessitates long reaction times and high temperatures which can lead to a multiplicity of products. Many attempts have been made to overcome this low solubility. The use of spray-dried KF has been shown to be advantageous. This is thought to be due to the increase in surface area of the salt which is achieved using this drying method. Smyth et al. have also shown that increases in yield can be achieved by slowly recrystallising the fluoride from methanol.19 Another successful methodology is the use of phase transfer catalysts. Several onium salts have been successfully used such as tetraphenylphosphonium bromide8 and tetramethylammonium chloride.9 Interestingly Suzuki et al.8 report that tetramethylammonium chloride is poor as a phase transfer catalyst for the fluorodenitration of ma-substituted nitroaromatics with KF with only 12% fluorodenitration occurring in the reaction of 3-nitrobenzonitrile compared to an 81% yield of this product when tetraphenylphosphonium bromide was used instead.On the other hand Maggini et al.9 found that tetramethylammonium chloride could successfully be used to achieve high yields of the required fluoroaromatics. Supported phase transfer catalysts have also been used which have the further advantage of being easily recyclable.20 One reported method of increasing the solubility of KF is to use a crown ether such as 18-crown-6.21 It has been shown that this even solubilises KF in non-polar solvents such as benzene.Yoshida et al. have successfully used a combination of KF–tetraphenylphosphonium bromide–18-crown-6 to convert chlorobenzaldehydes to the corresponding fluorobenzaldehyde.22 This is normally very difficult to achieve due to the low level of activation induced by the CHO group. However harsh reaction conditions are often still necessary to achieve an acceptable rate of reaction. The solubility of ionic fluorides generally increases with increasing size of the countercation. Quaternary ammonium fluorides have greater solubilities than their alkali metal counterparts in dipolar aprotic solvents but generally exist in a stable form as hydrated species.Such fluorides have been used for both halogen exchange and fluorodenitration reactions. Tetra-n-butylammonium fluoride (TBAF) was found to be capable of fluorodenitrating many substrates under relatively mild conditions. However hydrolysis by-products were observed in these systems which were inherently wetter than those based on KF. Similar by-products were observed in the reaction of 3,4-dichloronitrobenzene with TBAF. Unfortunately complete drying of most quaternary ammonium fluorides has been found to be impossible with the corresponding hydrogen difluoride (bifluoride) and amine being formed from the base-induced decomposition of the salt Scheme 5. How- ( n-C 2 ( n-C4H9)4N+F– 4H9)4N+HF2 – + ( n-C4H9)3N + CH3CH2CH=CH2 Scheme 5 Decomposition of TBAF on drying.ever it is interesting to note that similar yields of 2-chloro- 4-fluorotrifluoromethylsulfone were formed from the corresponding nitroaromatic when either KF with a phase transfer catalyst at 230 °C or TBAF in THF at 278 °C were used.23 The elimination reactions leading to decomposition of the onium salt require the presence of b-hydrogens in the quaternary ammonium cation which the simplest tetraalkylammonium fluoride tetramethylammonium fluoride (TMAF) does not have. TMAF has been found to be thermally stable to 227 Chem. Soc. Rev. 1999 28 225–231 high temperatures. While being commercially available as the tetrahydrate it is the only tetraalkylammonium fluoride which can be dried to give a completely anhydrous salt.24 It has been reported that it can be dried successfully by the azeotropic removal of water using cyclohexane or toluene,25 although recent work has highlighted difficulties with this method.26 TMAF has been used to fluorodenitrate many nitroaromatics with the reported yields often being very high.25 Reaction times are much shorter than those reported with KF even at the lower temperatures used in these studies.Perhaps most importantly it has been reported that fluorodenitration with TMAF results in the absence of the hydrolysis by-products in these systems. Other fluoride sources have been used for nucleophilic fluorination reactions. Tetraphenylphosphonium hydrogendifluoride was found to be capable of successfully fluorodenitrating simple substituted nitroaromatics despite the expected lack of reactivity of the hydrogendifluoride anion with its very strong hydrogen bond.27 For example 2-fluoronitrobenzene was formed in 70% yield from 1,2-dinitrobenzene in sulfolane at 100 °C.However quantitative fluorination with this fluoride source required two equivalents of the hydrogendifluoride. Similar results were obtained when using a tetraalkylammonium hydrogendifluoride for halogen exchange reactions.28 Polyfluorides were formed resulting from the scavenging of a molecule of HF from one hydrogendifluoride by another. Tetrabutylammonium hydrogendifluoride and dihydrogentrifluoride have also been used for the fluorination of a range of chloro and nitroaromatics.29 Interestingly the reactions were carried out in toluene as solvent with DMSO a more traditional solvent giving both a lower yield of the required product and unspecified by-products.R Cl NO2 Cl 2 II 5 Side reactions Of course despite both halogen exchange and fluorodenitration being used specifically to generate fluoroaromatics each can be considered as a side reaction when the other methodology is desired. Hence while halogen exchange may be desired substitution of a nitro group (present in order to activate the chlorine towards displacement) may also occur. There is very little information available as to the relative leaving group abilities of the two groups within the same molecule. Clark et al.have examined the reaction of some chloronitroaromatics of the form I and II.23 In all cases the nitro group was found to be R NO I R = CF3 CN SO2CF3 COOMe preferentially displaced by both TBAF and KF. A semiempirical study of the halogen exchange and fluorodenitration of 2,4-dichloronitrobenzene was carried out by Smyth.30 It was found that despite experimental evidence that the orthochlorine is preferentially displaced compared to the parachlorine the semi-empirical methods indicate that there is no clear preference for displacement by free fluoride ion. It was thus concluded that the favouring of ortho-substitution must be due to the nature of the nucleophile. A semi-empirical examination of the reaction with TMAF indicated that orthohalogen exchange is preferred agreeing with experimental findings.Similarly for this molecule halogen exchange is favoured over fluorodenitration. Further information about the Chem. Soc. Rev. 1999 28 225–231 228 relative leaving group ability of chlorine and nitro groups within the same molecule can be gleaned from the reaction of a fluoride with 2-chloro-6-nitrobenzonitrile. The fluorination of this substrate has been examined by many groups almost all of which have found that denitration is preferred. For example with both TMAF26 and tetrabutylphosphonium hydrogendifluoride, 29 fluorodenitration was achieved in high yield with subsequent rather than competitive dehalogenation occurring. However with Rb18F a mixture of products was observed with 55% denitration occurring and 17% dehalogenation,12 which demonstrates that the outcome of these reactions may be difficult to predict varying with the system used.Perhaps the major drawback of both halogen exchange and fluorodenitration is the prevalence of by-products in the reaction mixture. For example Maggini et al.9 describe the fluorination reactions of 1-chloro-2-nitro-4-(trifluoromethyl)- benzene. The halogen exchange of the chlorine for fluorine is facile and can be achieved in high yield. However fluorodenitration leads to a complex reaction mixture with other products including 1-hydroxy-2-nitro-4-(trifluoromethyl)benzene 1-methoxy-2-nitro-4-(trifluoromethyl)benzene and bis[2-nitro- 4-(trifluoromethyl)phenyl] ether.Such side-reactions are thought to occur because of the good leaving group ability of fluorine. Fluorine is one of the best leaving groups available for SNAr reactions.16 Its high electronegativity can stabilise transient carbanions allowing relatively facile displacement by nucleophilic attack. Phenols and ethers are the by-products most often reported and are thought to arise from two routes. Firstly simple hydrolysis of the product fluoroaromatic leads to by-products Scheme 6. All the fluoride sources used are known to be ArOH + HF ArF + H2O ArOAr + HF Scheme 6 Formation of phenolic and ether by-products via hydrolysis. hygroscopic and hence complete drying of the reaction systems is extremely difficult. The ability of fluoride to strongly hydrogen-bond to any remaining water results in the formation of phenolic products in the system.The reactivity of the phenols produced is also likely to be greatly enhanced by hydrogen bonding to the fluoride in these systems leading to further side products. The methoxy substituted products observed by Maggini et al.9 in the fluorodenitration of 1-chloro-2-nitro(trifluoromethyl) benzene were attributed to attack by such an activated phenol on the tetramethylammonium chloride phase transfer catalyst used in their systems. They also postulated that the nitroaniline formed in the reaction arose from nucleophilic attack by trimethylamine formed by phenol attack on the ammonium salt on the nitrofluorobenzene. Secondly for fluorodenitration reactions it has also been suggested that the displaced nitrite ion can become involved in the organic chemistry acting as a nucleophile Scheme 7.8 The ArF + NO ArF + ArOH 2 – ArONO + F– ArONO + NO ArO– + N2O3 2 – ArO– + ArF ArOAr + F– Scheme 7 Formation of phenolic and ether by-products via nitrite back attack.nitrite ion is known to be an ambident nucleophile i.e. it has two potential reaction sites. The relative rates of O- and N-attack on aryl halides have been examined and it has been shown that O-attack occurs preferentially on aryl fluorides in dipolar aprotic solvents resulting in the formation of nitrite esters. These are thought to have a very short lifetime decomposing to form phenoxides which can go on to form ethers.31 As mentioned above the use of TMAF for fluorodenitration reactions results in the absence of such by-products.This is thought to be due to strong ion-pairing in tetramethylammonium nitrite stabilising the nitrite anion and so reducing its nucleophilicity.25 Suzuki et al.8 and Maggini et al.9 both concluded that this re-attack of the displaced nitrite anion was responsible for ether formation in fluorodenitration reactions and independently added phthaloyl dichloride (PDC) to their reaction systems. This led to greatly improved yields of the desired fluoroaromatic. For example reaction of KF in sulfolane with 3-nitrobenzonitrile in the presence of tetraphenylphosphonium bromide gave 10% of the 3-fluorobenzonitrile with 21% of the bis(3-cyanophenyl) ether.The inclusion of PDC into the reaction mixture increased the yield of the fluoroaromatic to 86%. It was assumed that PDC was acting as an in situ trap for the displaced nitrite ion Scheme 8. Initially the PDC is converted to phthaloyl difluoride which O O F Cl + 2KCl F Cl O O O O ONO F + 2KF 2 – + F– F F O O O O ONO + N + NO + NO 2 – 2O3 + F– O F O O Scheme 8 PDC as a nitrite trap. then reacts with nitrite to form the nitrite ester. This is thought to react with a further equivalent of nitrite to generate higher nitrogen oxides phthalic anhydride (PA) and to regenerate fluoride. There is relatively little evidence for the mechanism of reactions involving PDC described above. Passudetti et al.32 describe how the reaction of KF with 3-nitrophthaloyl dichloride gives an 82% isolated yield of 3-fluorophthalic anhydride a useful intermediate for the preparation of many compounds.However as above the mechanism described for the conversion of the phthaloyl dihalide to the anhydride requires two equivalents of nitrite. Since the only source of nitrite in this system is the 3-nitrophthaloyl dihalide itself the implication is that the maximum yield of the anhydride obtainable should be 50%. Hence PDC must be reacting with another source of oxygen in these systems to affect the conversion to the anhydride. Suzuki et al.8 recognised this lack of stoichiometry and suggested a slightly different mechanism whereby FNO was evolved rather than N2O3.However they gave no evidence for the existence of this product in their system. PDC will react with water to give PA. Dipolar aprotic solvents are notoriously difficult to dry and it is not inconceivable that such a reaction may be occurring in these systems. This would have two effects. Firstly the number of hydrolysis byproducts from the aromatic fluorination would be expected to be decreased as observed. Secondly the reaction would form HCl which might be expected to react with KF to form HF which would then react with a further KF to form the corresponding bifluoride KHF2. This would necessitate the use of excess fluoride as observed by Suzuki et al.8 3 The high basicity of the fluoride in fluorodenitration and halogen exchange reactions can also result in the formation of by-products arising from reaction with the solvents.–SCH incorporated products (derived from DMSO) were observed by Finger and Kruse.5 Recently it has been reported that solventincorporated products occur in the fluorodenitration of 1,3-dinitrobenzene in DMAc.26 Such addition products are known to occur by attack of a nucleophile on the nitroaromatic generating an anionic s-complex which is then oxidised by further parent nitroaromatic Scheme 9. However such a NO2 – NO2 H + (CH3)2NCOCH3 + F– NO2 NO2 NO –2e –H+ 2 CH2CON(CH3)2 NO2 Scheme 9 Attack of a carbon nucleophile on a nitroaromatic. product requires deprotonation of DMAc demonstrating the high basicity of the fluoride in these systems.A recent report details the formation of such compounds using acetone and acetonitrile and other carbon-nucleophiles with 1,3-dinitrobenzene. These reactions necessitate the presence of fluoride; remarkably the conventional strong base potassium tertbutoxide was found to be ineffective.33 CH2CON(CH3)2 + HF 229 6 Recent developments Recent work has concentrated on developing a greater understanding of the chemistry occurring in these systems. The role of water in these systems is now better understood. Sasson et al. have shown that there is a critical amount of water necessary for successful halogen exchange.18 In excess water no reaction proceeded and ethers and phenols were detected when a hydrated tetraalkylammonium fluoride was used for the attempted fluorination of 3,4-dichloronitrobenzene.However with KF the use of 0.2% w/w water in the system was found to be necessary for effective transport of the fluoride from the surface of the solid salt to the organic phase. In this case no hydrolysis products were observed. Interestingly although the nature of the fluoride and the inorganic chemistry is often considered there is little discussion in the literature on the chemistry of nitroaromatics in such reaction mixtures. Nitroaromatics are known to be capable of interacting with basic species in a variety of ways.34 As well as forming anionic s-complexes as precursors to nitro displacement s-complexes can be formed which do not lead to denitration. In some cases these are known to be stable and can be isolated from the reaction mixture.34 It is also known that nitrocompounds can accept electrons thus forming radical anions.These can go on to form products arising from reduction of the nitro groups. Deprotonation of both the aromatic ring and benzylic hydrogens can also occur. In the light of these possibilities it is perhaps surprising that the majority of reported side products are attributed to nucleophilic displacement of fluorine. Chem. Soc. Rev. 1999 28 225–231 While the quantity of side products found to occur in the fluorodenitration of 1,3-dinitrobenzene was reported to be unaffected by the addition of water the use of tetramethylammonium hydrogendifluoride (TMAHF2) instead of TMAF·4/3H2O (formed by drying the tetrahydrate under vacuum at 60 oC for 48 hours) resulted in improved yields.26 As mentioned earlier both the nucleophilicity and the basicity of fluoride are strongly affected by the level of hydration the counterion and the reaction medium.The nucleophilicity of TMAHF2 is lower than that of TMAF. Correspondingly the reaction rates for fluorodenitration using TMAHF2 are lower than those using TMAF as the fluoride source. The basicity of the two salts is also different with TMAF leading to a more basic reaction system and for those nitroaromatics containing relatively acidic hydrogens yields were found to be greatly increased where the less basic TMAHF2 was used. In line with this hypothesis anhydrous TMAF (expected to be highly basic) was found to give a somewhat lower product yield than TMAF·4/3H2O Scheme 10.26 Interestingly the inclusion of NO2 NO2 NO2 + F NO2 NO2 CONMe2 26 % 10 % TMAF (anhydrous) 27 % 12 % TMAF•4/3H2O 66 % TMAHF2 Scheme 10 The fluorodenitration of 1,3-dinitrobenzene using different fluoride sources in N,N-dimethylacetamide.PDC was reported to be most beneficial when added to reaction systems involving the fluorodenitration of meta-nitroaromatics. Suzuki et al.8 specifically targeted such molecules and the majority of the substrates fluorinated by Maggini et al.9 had a nitro group meta to an electron-withdrawing group. Such compounds are expected to have relatively acidic hydrogens due to the strong electron-withdrawing nature of the nitro group and other such substituents.As postulated earlier the addition of PDC may simply be acting as a method of removing excess water from the system but with the added effect of forming hydrogendifluorides in situ. Hydrogendifluorides are much less basic than the corresponding fluoride which may be the reason behind the increased fluoroaromatic product yields observed. It is also interesting to note that 4-nitrophthalic anhydride is known to react with the displaced nitrite from a fluorodenitration reaction35 but in the presence of TMAF PA was found to react to form the tetramethylammonium salt of the acid and tetramethylammonium bifluoride Scheme 11.26 The use of this O O O–TMA+ + TMAHF2 + H2O O OH O 2 TMAF•H2O + 230 O Scheme 11 Reaction of TMAF with phthalic anhydride to give the corresponding bifluoride.bifluoride formed in situ to fluorodenitrate 1,3-dinitrobenzene led to an increased yield compared to that obtained when TMAF was used. The fluorodenitration of nitroaromatics with TMAF and TMAHF2 has been examined in more detail in an attempt to identify the side products occurring. With substrates giving high yields when fluorodenitrated ethers were detected. For the more demanding substrate 1,3-dinitrobenzene solvent addition products were observed.26 Chem. Soc. Rev. 1999 28 225–231 The high basicity of typical fluorodenitration reaction systems has also been found to have other effects. Azeotropic drying of the TMAF–DMSO system was shown to lead to basecatalysed solvent decomposition giving the methylsulfinyl anion.36 It was found that the attempted halogen exchange of 4,4A-dichlorobenzophenone with TMAF in DMSO leads to the formation of 1,1-di(4-chlorophenyl)ethene.Although no other products directly attributable to the methylsulfinyl anion were detected in fluorodenitration reactions it is expected that such reactions occur given the known high reactivity of this species. Azeotropic drying of the TMAF–DMSO system was also found to lead to unexpected hydrolysis by-products when used in the attempted fluorodenitration of 4-chloro-3-nitrobenzonitrile. These included the benzamide the acid and a second amide Scheme 12 all presumed to arise from base-catalysed hydroly- CN CN CN CONH2 + + 2 NO2 NO2 NO2 Cl OH F Cl CN O COOH NH + + NO2 Cl NO2 NO NO 2 Cl Scheme 12 The reaction of 4-chloro-3-nitrobenzonitrile with azeotropically-dried TMAF in DMSO.sis reactions. Detection of these products indicates that this method of drying the system is perhaps not as effective as suggested in the earlier reports. This work also thus shows another possible previously unreported route to side-products again indicating the complexity of the systems is greater than has been generally assumed. While the high basicity of the fluoride is thought to lead to by-products in fluorodenitration reactions it is possible to exploit this in a number of ways. Fluoride catalysed H–D exchange was also found to be possible between d6-DMSO with 4,4A-dinitrodiphenylmethane.10 In an oxygenated system the base-catalysed oxidation of this substrate and other nitroaromatics was found to be possible.Interestingly the carbonyl group thus formed produced sufficient activation for fluorodenitration to subsequently occur providing a new one-pot synthesis of the industrially important monomer 4,4A-difluorobenzophenone Scheme 13.10 However it was found that O NO2 O2N NO2 O2N O F F Scheme 13 Formation of 4,4A-difluorobenzophenone by oxidation followed by fluorodenitration. a similar attempted base-catalysed oxidation followed by fluorodenitration of 4-nitroethylbenzene led to low yields of the corresponding fluoroacetophenone with indications that aldol condensations were occurring.7 Conclusions and future trends Halogen exchange is well developed and widely used as an industrial method for the formation of fluoroaromatics. Recent improvements have centred around the development of new reagents to overcome the poor solubility of the commonly used alkali metal fluorides in the solvents used. Economic and cost concerns have driven work investigating the use of alternatives for the dipolar aprotic solvents which can degrade in the presence of the strongly basic fluoride and are difficult to recover. However in recent years fluorodenitration has been shown to be a valid preparative route to a wide range of fluorinated aromatic compounds. It is possible to fluorinate nitroaromatics carrying a variety of functional groups including cyano nitro trifluoromethyl chloro and trifluoromethylsulfone.Unlike with halogen exchange reactions it is possible to fluorodenitrate meta-substituted nitroaromatics successfully even those containing only a single activating group. Recent results have highlighted that the chemistry occurring in fluorodenitration systems involving fluorides activated nitroaromatics and dipolar aprotic solvents is not as straightforward as has often been assumed. Side reactions do occur but are not simply as a result of nucleophilic displacement of the fluorine or the nitro groups. The recognition that the success of reactions can be compromised if the fluorinating reagent is too basic is unsurprising when one considers the known interactions of nitroaromatics with bases.However the results do show that the reaction conditions can be tuned to suit the particular substrate with improved results being obtained for the more demanding meta-substituted substrates such as 1,3-dinitrobenzene by using TMAHF2 rather than TMAF. Clearly it is essential to be aware of all of the side reactions that can occur with any particular nitroaromatic substrate. While the use of onium fluorides such as TMAF has enabled the methodology to be extended to a potentially wide range of substrates it must be appreciated that its greater nucleophilicity will also generally be accompanied by a higher system basicity. In many cases it will be sensible to tolerate a reduction in the overall activity through the avoidance of over drying or the use of the hydrogen difluoride salt so as to reduce the formation of unwanted byproducts.Thus by tuning the activity of the system through mild KF-based reactions to those employing TMAF to suit the particular substrate and desired product combination it should prove possible to usefully exploit fluorodenitration in an increasing range of applications. 8 References 1 M. R. S. Gerstenberger and A. Haas Angew. Chem. Int. Ed. Engl. 1981 20 647. 2 J. H. Clark D. Wails and T. W. Bastock Aromatic Fluorine Chemistry CRC New York 1997. 3 H. B. Gottlieb J. Am. Chem. Soc. 1936 58 532. 4 J. R. Beck Tetrahedron 1978 34 2057. 5 G. C. Finger and C. W. Kruse J. Am. Chem. Soc. 1956 78 6034. 6 H. Suzuki and Y. Kimura J. Fluorine Chem.1991 52 341. 7 J. H. Clark and D. K. Smith J. Fluorine Chem. 1985 26 2233. 8 H. Suzuki N. Yazawa O. Furusawa and Y. Kimura Bull. Chem. Soc. 9 M. Maggini M. Passudetti G. Gonzales-Trueba M. Prato U. Quintily Jpn. 1990 63 2010. and G. Scorrano J. Org. Chem. 1991 56 6406. 92 127. 1995 36 103. 1983 108. 10 D. J. Adams J. H. Clark and H. McFarland J. Fluorine Chem. 1998 11 G. Bormans and M. R. Kilbourn J. Labelled Compd. Radiopharm. 12 M. Attina F. Cacace and A. P. Wolf J. Chem. Soc. Chem. Commun. 13 D. J. Milner Synth. Commun. 1985 15 485. 14 J. H. Clark Chem. Rev. 1980 80 429. 15 J. H. Clark and D. J. Macquarrie J. Fluorine Chem. 1987 591. 16 J. Miller Aromatic Nucleophilic Substitution Elsevier Amsterdam 1968. 17 R. D. Chambers and A. R. Edwards J. Chem. Soc. Perkin Trans. 1 1997 3623. 18 Y. Sasson S. Neguissie M. Royz and N. Mushkin Chem. Commun. 1996 297. 19 T. P. Smyth A. Carey and B. K. Hodnett Tetrahedron 1995 51 6363. 20 Y. Yoshida Y. Kimura and M. Tomoi Chem. Lett. 1990 769. 21 C. L. Liotta and H. P. Harris J. Am. Chem. Soc. 1974 96 2250. 22 Y. Yoshida and Y. Kimura Chem. Lett. 1988 1355. 23 A. J. Beaumont J. H. Clark and N. A. Boechat J. Fluorine Chem. 1993 63 25. 24 K. O. Christe W. W. Wilson R. O. Wilson R. Bau and J. Feng J. Am. Chem. Soc. 1990 112 7619. 25 N. Boechat and J. H. Clark J. Chem. Soc. Chem. Commun. 1993 921. 26 D. J. Adams J. H. Clark and D. J. Nightingale Tetrahedron in the press. 27 S. J. Brown and J. H. Clark J. Fluorine Chem. 1985 30 251. 28 C. Rieux B. Langlois and R. Gallo C. R. Seances Acad. Sci. 1990 310 25. 29 Y. Uchibori M. Umeno H. Seto Z. Qian and H. Yoshioka Synlett 1992 345. 30 T. Smyth and A. Carey Tetrahedron 1995 51 8901. 31 D. H. Rosenblatt W. H. Dennis Jnr and R. D. Goodwin J. Am. Chem. Soc. 1973 95 2133. 32 M. Passudetti M. Prato U. Quintily and G. Scorrano J. Fluorine Chem. 1990 50 251. 33 M. Cervera and J. Marquet Tetrahedron Lett. 1996 37 7591. 34 E. Buncel in The Chemistry of the amino nitroso and nitro compounds and their derivatives S. Patai Ed. J. Wiley and Sons New York 35 R. L. Markezich O. S. Zamek P. E. Donahue and F. J. Williams J. Org. 1982. Chem. 1977 42 3435. 36 D. J. Adams J. H. Clark H. McFarland and D. J. Nightingale J. Fluorine Chem. 1999 94 51. Review 8/08707E 231 Chem. Soc. Rev. 1999 28 225–231

ISSN:0306-0012    

DOI:10.1039/a808707e

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Core level photoelectron spectroscopy for polymer and catalyst characterisation A. Paul Pijpers and Robert J. Meier DSM Research PO Box 18 6160 MD Geleen The Netherlands Received 7th October 1998 Electron spectroscopy for chemical analysis or ESCA is a powerful tool for the characterisation of chemical species at surfaces. ESCA can therefore play a major role in the study of heterogeneous catalysts and catalytic processes. In the polymer field more recently an increase in sensitivity combined with high spectral resolution has revealed details on the chemical environment around the probed atom. Information on nearest and next-nearest neighbours sometimes even including conformational information has become accessible. In catalysis the valence state of the metal centre can be estimated and charge–binding energy relations can be used to establish charge–(catalytic) performance relations.1 Introduction Electron spectroscopy for chemical analysis (ESCA) has been an important analytical tool from the days it was introduced by Siegbahn and coworkers.1,2 The name is equivalent to what is known among physicists as X-ray photoelectron spectroscopy or XPS. The technique is about the determination of the binding energy (BE) of an electron in an atom or molecule. A key feature of the technique is its high surface sensitivity it probes the top few nanometres of a sample from which qualitative and quantitative chemical information can be obtained. Whereas ESCA is formally core level spectroscopy only XPS addresses both core level and valence band electron spectroscopy.In this paper we primarily limit discussion to core level spectroscopy. The reason is the straightforwardness of the interpretation of the basic features in core level spectra and the relatively high Paul Pijpers is a senior researcher at the DSM Research laboratories. In the period 1962–1978 he worked as an X-ray diffraction specialist. In 1978 he introduced XPS at DSM Research. In this field he is truly an autodidact. His research interests focus on the characterisation of polymers and issues related to surface science in general with particular attention to high resolution spectroscopy on real-life polymeric and catalytic samples in order to reveal detailed chemical information.A. Paul Pijpers intensities compared to valence band spectra. Moreover for solid state samples once the vibrational structure in the valence band spectrum is lost the latter loses most of its significance as an analytical tool. Siegbahn introducing ESCA first pointed out that chemical shifts resulting from different chemical environments were a general phenomenon and the shifts can be related to the valence state of an atom. For core level spectroscopy relative intensities of lines arising from different atoms provide direct information on their relative abundance. More recent instrumental developments in conjunction with theoretical computation of shifts in binding energy have formed the basis for the retrieval of more detailed information on the chemical environment.Examples include next-nearest neighbour induced shifts effects due to conformational changes and the recovery of coexisting valencies for metal ion centres. Angle resolved XPS enables depth profiling in the top 1–100 Å. These developments have further increased the significance of photoelectron spectroscopy in chemistry mostly based on its specific surface sensitivity. (1) Robert J. Meier 2 Theoretical background 2.1 Basic principles The basic equation describing the photoelectron experiment is given in eqn. (1). The X-ray photon energy hn is known by the KE = hn 2 BE source employed the kinetic energy (KE) of the photoelectron is measured in the XPS experiment yielding the electron binding energy (BE).However there are complicating factors. Rob Meier is a Research Fellow in the DSM Research laboratories and an Honorary Visiting Professor in the Chemistry Department of the University of York. He studied Theoretical Chemistry at the University of Amsterdam and graduated from the same university on a subject in condensed matter physics. Early in 1985 he joined DSM Research where he was a group leader in Theoretical Chemistry Raman Spectroscopy and Electron Spin Resonance. Since 1993 he has been a Research Fellow for Modelling and Spectroscopy within DSM Research and in 1997 was given an Honorary Professorship from the University of York UK. His main interests are computational chemistry and experimental spectroscopy.Chem. Soc. Rev. 1999 28 233–238 233 In an experiment on a solid non-conducting sample eqn. (1) should be recast as eqn. (2) with f the work function and S the (2) KE = hn 2 BE 2 f 2 S sample charging effect. The work function is the energy required for the electron when already removed from the atom on which it resided to be taken out of the material. This involves overcoming a barrier resulting from interaction of that electron with the rest of the material. For many samples the work function lies in the range around 5 eV. It is not ultimately necessary to know the work function very accurately for the XPS spectra can be analyzed in terms of shifts with respect to a reference signal rather than by absolute binding energies. The precise binding energy of an electron in a molecule is determined by the net interaction of the electron with all nuclei and all other electrons.Siegbahn introduced the following description. Upon bonding of an atom and the formation of a molecule the core level energies of the atom are influenced by the valence density on the atom as described by eqn. (3) in (3) BE = E0 + kq which E0 is the electron binding energy in the isolated atom q the valence charge and k a proportionality constant to be determined. When a strong potential field is induced by the environment it is necessary to add a Madelung term Si­ jqj/rij to the right-hand-side of eqn. (3). The effect of neighbouring chemically bonded atoms comes from their effect on the valence electron density of the atom monitored in the XPS experiment.Different bond types lead to different effects on the valence band electron density and thereby induce a somewhat different shift in the core level binding energy. For many cases in particular organic systems shifts due to various types of chemical bonds take typical values. A detailed interpretation of core level shifts can be obtained by comparing the experimental data to appropriate quantum mechanically calculated binding energies. In addition from such calculations a relation of the form given by eqn. (3) can be established. Sets of charge–BE relations based on quantum mechanically evaluated charge values have been collected by Sleigh et al.,3 and may be applied to yield shifts in electron density upon variation of chemical (sub)structure.2.2 Spectral assignment 1s The identification of the atom type corresponding to an observed core level binding energy is accomplished using standard tables comprising atomic core level binding energies. 2,4,5 More detailed information can be retrieved when the precise peak position and when present the composition of a line profile are analyzed. As mentioned above the chemical environment influences the BE’s in a distinct pattern. Here it is useful to introduce the distinction between primary and secondary substituent effects also referred to as nearest neighbour and next-nearest neighbour effects. Nearest neighbours often have a sizeable effect on the core level spectrum whereas secondary shifts are usually small.A fine example of primary and secondary shifts is given by the various fluorinated polyethylenes including polytetrafluoroethylene (“Teflon”) illustrated in Fig. 1. Fluorine is probably the element exhibiting the largest secondary shift on a next-nearest neighbouring C level 0.9 eV while it induces a primary substituent effect of about 3 eV. The smaller secondary shifts induced by oxygen require high resolution charge compensated (see Section 3) XPS spectra. Fig. 2 shows the C1s XPS spectrum of poly(methyl methacrylate) (PMMA). The attribution of a secondary shift ( Å 1.0 eV) to the next-nearest neighbour effect induced by the Chem. Soc. Rev. 1999 28 233–238 234 Fig. 1 Shifts in the C1s binding energy due to nearest neighbour (primary shift) and next-nearest neighbour (secondary shift) effects of fluorine atoms.An example of a primary shift is indicated by A a secondary shift by B. Experimental data from Pireaux et al.6 (+) are shown along with theoretical data (x) taken from ref. 7. Fig. 2 C1s XPS spectrum for PMMA revealing primary (C1s–O and C1s(NO)O) and secondary (C1s–C(NO)O) shifts due to oxygen. The secondary shift is only apparent after curve fitting but the physical significance including the magnitude of this small shift has been confirmed by theoretical calculation. two oxygen atoms bonded to the Cb atom has been confirmed by quantum mechanical calculations.8 Photoexcitation of a core level may be accompanied by a valence electron excitation (shake-up) or a valence electron ionisation (shake-off).These phenomena are in fact rather common in XPS spectra but because of their relatively low intensity are not always apparent. When present in the experimental spectrum shake-up satellites can be used to characterise specific chemical groups. In particular for aromatic ring systems the p?p* shake-up satellite is used for identification (e.g. see the weak broad line in the range 290–292 eV in the C1s spectrum of poly(ethylene terephthalate) (PET) shown in Fig. 4). 2.3 More on charge–BE energy relations Because chemical reactivity is so much related to charge and distribution of charge it is important to realise that charge– binding energy relations viz. eqn. (3) may be applied to characterize species subjected to chemical reaction.This may find particular use in both homogeneous and heterogeneous catalysis. Analysis of the XPS spectrum of a metal based catalyst leading to the charge on the metal centre may reveal a relevant charge–catalytic activity relation. It is however not a priori trivial to obtain the necessary charges for there is not such a thing as absolute atomic charge except for ionised atomic species in the gas-phase. The atomic charges in a molecule may be calculated using quantum mechanical methods but each method or level of calculation will yield a different magnitude of the charges. Apart from the level of calculation there is an ambiguity in the way to calculate atomic charge. The charge assigned to an atom depends on how the total charge distribution in the molecule is partitioned over the atoms.The popular Mulliken analysis definitely leads to large errors for organometallics due to the complex bonding involving diffuse d- and f-orbitals. Moreover referring to the possible application to catalytically active species quantum chemical calculations can only be carried out on well-defined species. For this reason Folkesson and Larsson have proposed an elegant alternative.9,10 First charge–BE relations are established for the elements e.g. C N and O that comprise the ligands. Then the full XPS spectra of a series of known organometallic compounds are recorded. For each of the neutral organometallic species the charge on the metal is then assumed to have the same magnitude but opposite sign as the sum of charges residing on the ligands.From an analysis of the XPS spectra of the light elements and the already established charge–BE relations for the light elements the charge–BE relations for metal elements have been established. 3 3 The XPS experiment The basic design of a photoelectron spectrometer involves four components illustrated in Fig. 3. First one needs an X-ray source. Common choices are based on an Al anode with a photon energy around 1487 eV or a Mg anode providing 1254 eV photon energy. For heavy elements the BE’s for the 1s 2s and 2p levels are much higher in energy than for light elements and also higher than the named X-ray photon energies. Therefore one selects an appropriate set of energy levels for the various elements of the periodic table which lie within a certain range of BE’s.When using either the Al-Ka or the Mg-Ka excitation radiation a suitable choice is given by C1s F1s Ti2p Cr3d Zr3d etcetera. When required for high resolution work the line width of the X-ray source can be further reduced using a monochromator. The next component is the high vacuum chamber. The high vacuum is required in order to have a sufficiently long mean free path to allow the emitted photoelectrons to reach the detector but also to keep the surface of the sample clean of any adsorbents that may otherwise contaminate the sample during handling and measurement. In XPS being a highly surface sensitive technique the smallest disruption of the original surface will affect the recorded spectrum.For example species that are sensitive to oxidation like many catalysts require a glove box with a wellconditioned atmosphere before being transferred into the highvacuum section of the XPS-apparatus. Before the detector the fourth component is reached which is usually an electron multiplier electrons have to be selected according to their kinetic energy. This is accomplished by an analyzer which is usually a combination of electrostatic lenses and a hemispherical analyzer. By modifying the field exerted by this lens system different electron kinetic energies are probed consecutively providing the total photoelectron spectrum. Some of the modern instruments allow for XPS imaging experiments.11,12 Imaging is a highly popular technique in optical spectroscopies.The possibility to view a spectroscopic image of a spatial domain of a material is of particular interest Fig. 3 Basic scheme showing the major components of an XPS instrument. The X-ray source (1) the high vacuum chamber (2) the analyzer (3) and detector (4). to industrial applications as is well-known from micro-infrared and micro-Raman spectroscopy. We are likely to see an increase in the number of applications involving XPS imaging in forthcoming years. The X-rays used to excite the electrons penetrate deeply into most materials. The kinetic electrons generated have however a limited mean free path depending on the material and the electron kinetic energy. Because the electron kinetic energy depends on the binding energy of the probed energy level in the atom cf.eqn. (1) with hn fixed the profiling depth may also be different for different element types in the same sample. Furthermore the X-rays are high-energy electromagnetic radiation and may cause damage to the sample.4,13,14 The energy of the X-rays is much higher than the bond energy between atoms and can thus lead to bond scission loss of components crosslinking and reduction. Moreover the secondary electrons generated in the sample have a broad band energy spectrum and are therefore capable of breaking most types of chemical bonds present in the sample. In order to avoid this a possibly low dose of X-ray photons should be applied in combination with a sensitive detector system.When in doubt radiation damage can be monitored by following the evolution of the XPS spectrum during X-ray exposure in real time.13 The X-ray beam causes electrons to be removed from the atoms in the sample leading to (positive) charging of the surface of non-conducting samples. This surface charging leads to an additional shift in the XPS spectrum and often further broadening of the lines in the spectrum. The shift may be corrected for by considering a reference signal in the sample (see below) but the broadening may obscure valuable chemical information contained in a small shift now hidden under a broad line profile. One way to compensate surface charging is to apply a flood gun which spreads out low energy electrons homogeneously over the sample surface.A fine and more extensive discussion of the problems involved has been presented by Briggs and Seah.4 235 Chem. Soc. Rev. 1999 28 233–238 Effects such as surface charging and the act of a workfunction introduce an apparent shift in the BE’s which is often in the order of 5 eV. In addition there may be instrumental factors that add to this problem. The way this problem is usually tackled is to use a reference peak in the recorded spectrum. There are several possibilities in choosing a proper reference signal e.g. the C1s line from a hydrocarbon chain which has an accepted value of 284.8 eV.4 Fig. 4 Spectra from Beamson et al. on PET conformational dependence [Reproduced from Polymer 37 G. Beamson D. T.Clark N. W. Hayes D. S.-L. Law V. Siracusa and A. Recca p. 379 Copyright 1997 with permission from Elsevier Science]. 4 XPS applied to polymers Much information on polymer XPS spectra is available through sources such as Briggs and Seah4 and Beamson and Briggs14 and Briggs.15 Over the last couple of years the higher resolution attainable with modern instruments and adequate charge compensation has allowed for the retrieval of more detail from polymer XPS spectra. Examples of polymer spectra showing next-nearest neighbour effects were shown in Figs. 1 and 2. High resolution monochromatized XPS may reveal asymmetry in line-shape due to vibrational excitation.16 XPS has been found to be extremely useful for the study of biocompatibility of polymers;17 the characterisation of apolar polymer surfaces after flame treatment (partial oxidation in order to improve adhesion properties) surface composition and surface segregation of block copolymers have been studied,18,19 as well as acid–base interactions in relation to adhesion.20 Whereas many polymers can be discriminated using core level XPS the important class of polyolefins poses serious problems with this technique.For this particular case valence band XPS can be used in order to distinguish between ethylene and propylene and characterize e.g. ethylene–propylene copolymers.21 Apart from the study of the polymer XPS can be employed to detect additives like release agents lubricants and anti-statics. Beamson et al.22 have reported high-resolution core level spectra of crystalline and amorphous poly(ethylene terephthalate).Spectra are shown in Fig. 4. The assignment of a small shift of about 0.1 eV to the trans–gauche difference of the glycol fragment was based on a similar shift in poly(acrylic acid) which was predicted several years ago8 on the basis of theoretical calculation. More recent calculations on PET have revealed23 a theoretically predicted shift of 0.11 eV in very good agreement with the experimental shifts reported in the range 0.1–0.14 eV. Recognising the limited escape depth of the electrons angleresolved XPS spectroscopy can be used to probe either the top layer of the surface or more of the bulk of the sample. When the electron take-off angle takes a small value ( < 10°) i.e.the Chem. Soc. Rev. 1999 28 233–238 236 sensitivity. Fig. 5 shows C electrons leave almost parallel to the surface the limited escape depth of the photoelectrons results in an extreme surface 1s spectra of a polymer covered with Fig. 5 Angle-resolved XPS spectra in the C1s range originating from a polymer sample coated with a very thin layer of silicon oil. The take-off values are indicated. For the lowest take-off value surface sensitivity is highest and only the aliphatic carbons from the silicon oil are recovered. a thin layer of a silicon compound measured at electron take-off angles ranging between 90° and 5°. Using Mg-Ka radiation with l = 3.2 nm the analysis depth obtained from d = 3lsinq ranges from 10 nm down to less than 1 nm.The spectra clearly show the decrease of the polymer contribution at low take-off values. Analysis of these spectra allows differentiation between the signal arising from the very thin surface layer from the 5° spectrum and the signals arising from the bulk polymer at 90° take-off angle. The “bulk” spectrum can be attributed to carbonyl and ether or alcohol functionalities whereas the outermost layer (1 nm) is characteristic for an aliphatic group in the present case arising from the silicon oil. (4) > COH + F When different surface species cannot be discriminated by their straightforwardly recorded XPS spectra e.g. because of insufficient resolution within the envelope of the peak derivatisation of specific groups may be pursued in order to enable differentiation between the various surface groups.After treatment of the surface of apolar polymers such as polyethylene and polypropylene by flame or plasma with the objective to incorporate oxygen at the surface in order to improve adhesive properties it is not possible to discriminate an ether from an epoxide or an alcohol since all carbons and oxygens present in C–O bonds exhibit mutually minor differences in core level BE’s. Derivatisation with a reactant which is selective with respect to one of the functionalities allows for discrimination in the XPS spectrum. An example is illustrated in Fig. 6 showing the result of treating the alcohol functionality in polyvinyl alcohol by trifluoroacetic acid (eqn. (4)). The overview XPS 3CCOOH ? > COC(NO)CF3 + H2O spectra shown in Fig.6 show that the alcohol function is very well accounted for by the strong fluorine related signal in the derivatised sample. Several studies have been presented though the method is not without pitfalls,16,24 in particular with respect to the chemistry involved in derivatisation. In particular the efficiency of the reactions and the occurrence of side reactions require attention. 5 XPS applied to catalysis The potential application of ESCA to catalysis involving metal containing compounds has been known for quite some time.25,26 Application was demonstrated to both homogeneous and heterogeneous catalysis. Correlations between acidity/basicity and XPS shifts have been reported.27,28 Like for polymers more recently higher spectral resolution has enabled more detailed Fig.6 XPS overview spectra showing the change when an alcohol functionality is derivatised with trifluoroacetic acid. The upper spectrum originates from polyvinyl alcohol the lower spectrum from the same sample after treatment with trifluoroacetic acid. The specificity of the derivatisation which is accompanied by the introduction of a clearly visible fluorine related line allows for the specific detection of the alcohol functionality. assignments. Angle resolved studies on dispersed catalysts on support e.g. silica support showed that the dispersed catalyst centres are really located at the surface.29 Olefin polymerisation catalysts of the metallocene type are of huge contemporary interest both academically as well as industrially.A simple model system is the biscyclopentadienyl zirconium species Zr(C5Me5)2Me2 which requires activation in order to yield the active catalyst Zr(C5Me5)2Me+. The XPS spectrum of the neutral starting complex shows the typical Zr 3d3/2–Zr 3d5/2 doublet with these components separated by 2.2 eV. After activation a shift of the Zr3d BE of about 1 eV was observed which is of the correct magnitude expected for a +1 change in formal charge. The well-known Phillips catalyst for olefin polymerisation involves chromium dispersed on silica. From the chemistry it seems unlikely that Cr(ii) and bulk Cr(vi) are stable under calcination conditions whereas signals due to isolated Cr(vi) isolated Cr(iii) and bulk Cr(iii)2O3 may be expected.Isolated here means that the Cr is linked via an oxygen bridge to a silicon atom rather than to another Cr as in bulk chromate. The Cr2p region in the XPS of some calcined samples are shown in Fig. 7a. The lower spectrum shows the characteristic doublet (2p1/2 2p3/2) of isolated Cr(vi) on silica. In the upper spectrum another doublet is visible with its 2p3/2 component at 575.8 eV which can be attributed to bulk Cr(iii). Curve fitting however shows at least one more component is present. This component might be either bulk Cr(vi) or isolated Cr(iii) on silica a hypothesis which can be tested assuming an analogy with secondary substituent effects in polymers. For this particular case we have verified by quantum mechanical calculations applied to the models depicted in Fig.7b that indeed there is a next-nearest neighbour effect on the Cr2p BE of 2.0 eV which compares very well to the experimental difference of 2.1 eV (Fig. 7a) between the (fitted) peaks at 581.6 ± 0.1 eV (isolated Fig. 7 (a) Cr2p region of the XPS spectrum of a calcined Phillips catalyst sample. (b) Models for Cr(vi) bulk and Cr(vi) on silica studied by quantum mechanical calculation to reveal the difference in 2p BE’s. Cr(vi)) and 579.5 ± 0.5 eV (bulk Cr(vi)) in the experimental spectra of the shown and other Cr-on-silica samples. XPS spectra of pure metal surfaces recorded at sufficiently Fig. 8 showing decomposition of the Pd 3d high resolution may show differences between the surface core level shifts and the bulk atom shift.30 An example is shown in 5/2 core level into Fig.8 Decomposition of the Pd 3d5/2 core level into bulk and surface components [Reproduced from J. Phys. Condens. Matter 4 R. Nyholm M. Quarford J. N. Andersen S. L. Sorensen and C. Wigren p. 277 Copyright 1992 with permission from IOP Publishing Limited]. bulk and surface atom components. This decomposition could be physically justified by varying the excitation energy (390 and 450 eV respectively) which causes a variation in the ratio between contributions from surface and bulk Pd atoms as a result of the change in escape depth as a function of electron kinetic energy (see the upper two spectra in Fig. 8). The surface character of one of the peaks in the decomposition was further substantiated by showing that the presence of adsorbates on the 237 Chem.Soc. Rev. 1999 28 233–238 Pd surface made this peak disappear and by theoretical support. The vibrational fine structure and orientation of ethylene and ethylidyne on Rh(111) have been studied32 by monitoring the C1s spectra of C2H3 and C2D3 overlayers on the Rh substrate see Fig. 9. In order to reduce broadening due to low energy Fig. 9 C1s XPS spectra from Rh(111)-C2H3 and C2D3 overlayers. Dots represent experimental data points the thinner (broken) lines are results from curve fitting [Reproduced from Chem. Phys. Lett. 269 J. N. Andersen A. Beautler S. L. Sorensen R. Nyholm B. Setlik and D. Heskett p. 371 Copyright 1997 with permission from Elsevier Science].vibrational excitations all measurements were carried out at 100 K. The existence of two main peaks is consistent with the upright geometry of the adsorbate molecule a feature which was independently found from a quantitative LEED (low energy electron diffraction) study. In addition the C2H3 spectrum shows at least two shoulders on the high binding energy side whereas in the C2D3 spectrum the first shoulder has moved considerably closer to the main peak and the second peak is hardly discernable. The combination of these observations leads to the interpretation of C–H vibrational fine structure in these C1s spectra. It should be emphasized that the vibrational energy involved is that corresponding to the core ionised molecule and thus cannot be interpreted on the basis of frequencies known from traditional infrared or Raman vibrational spectroscopy.6 References 1 K. Siegbahn C. Nordling G. Johansson J. Hedman P. F. Heden K. Hamrin U. Gelius T. Bergmark L. O. Werne R. Manne and Y. Baer ESCA Applied to Free Molecules 1969 North-Holland Publishing Company Amsterdam London. 2 K. Siegbahn C. Nordling A. Fahlman R. Nordberg K. Hamrin J. Hedman G. Johansson T. Bergmark S. E. Karlsson I. Lindgren and B. Chem. Soc. Rev. 1999 28 233–238 238 Lindberg ESCA - Atomic Molecular and Solid State Structure Studied by Means of Electron Spectroscopy Uppsala 1967. Nova Acta Regiae Soc. Sci. Ups. Ser. IV 20 1967. 3 C. Sleigh A. P. Pijpers A. Jaspers B. Coussens and R. J.Meier J. Electron Spectrosc. Relat. Phenom. 1996 77 41. 4 D. Briggs and M. P. Seah Practical Surface Analysis vol. 1 2nd edn. Wiley Chichester 1990. 5 W. L. Jolly K. D. Bomben and C. J. Eyermann At. Nucl. Data Tables 1984 31 433. 6 J. J. Pireaux J. Riga R. Caudano J. J. Verbist J. M. Andre J. Delhalle and S. Delhalle J. Electron Spectrosc. Relat. Phenom. 1974 5 531. 7 R. J. Meier J. Mol. Struct. (THEOCHEM) 1988 181 81. 8 R. J. Meier and A. P. Pijpers Theor. Chim. Acta 1989 75 261. 9 P. Sundberg C. Andersson B. Folkesson and R. Larsson J. Electron Spectrosc. Relat. Phenom. 1988 46 85. 10 B. Folkesson and R. Larsson J. Electron Spectrosc. Relat. Phenom. 1990 50 267. 11 M. Keenlyside and P. Pianetta J. Electron Spectrosc. Relat. Phenom. 1993 66 189.12 C. Coluzza and R. Moburg J. Electron Spectrosc. Relat. Phenom. 1997 84 109. 13 A. P. Pijpers in Scientific Methods for the Study of Polymer Colloids and their Applications eds. F. Candau and R. H. Ottewill Kluwer Academic Publishers 1990 p. 291. 14 G. Beamson and D. Briggs High Resolution XPS of Organic Polymers The Scienta ESCA300 Database 1992 Wiley Chichester. 15 D. Briggs Surface Analysis of Polymers by XPS and Static SIMS Cambridge University Press Cambridge 1998. 16 G. Beamson D. T. Clark J. Kendrick and D. Briggs J. Electron Spectrosc. Relat. Phenom. 1991 57 79. 17 L. Sabbatini and P. G. Zambonin J. Electron Spectrosc. Relat. Phenom. 1996 81 285. 18 C. M. Kassis J. K. Steehler D. E. Betts Z. Guan T. J. Romack J. M. DeSimone and R.W. Linton Macromolecules 1996 29 3247. 19 L. Li C.-M. Chan and L. T. Weng Macromolecules 1997 30 3698. 20 S. R. Leadly and J. F. Watts Polymer-Solid Interfaces from Model to Real Systems ICPSI-2 Proceedings of the Second International Conference eds. J.-J. Pireaux J. Delhalle and P. Rudolf Presses Universitaires de Namur 1998. 21 A. A. Galuska and D. E. Halverson Surf. Interface Anal. 1998 26 425. 22 G. Beamson D. T. Clark N. W. Hayes D. S.-L. Law V. Siracusa and A. Recca Polymer 1996 37 379. 23 A. Tarazona E. Koglin A. P. Pijpers and R. J. Meier Polymer 1997 38 2615. 24 C. D. Batich Appl. Surf. Sci. 1988 32 57. 25 New Catalytic Materials Volume XI. State-of-the-art techniques for catalyst characterisation 1984 Catalytica Associates Inc. California 26 S. C. Avanzino H.-W. Chen C. J. Donahue and W. L. Jolly Inorg. 27 M. Casamassima E. Darque-Ceretti A. Etcheberry and M. Aucouturier 28 H. J. M. Bosman A. P. Pijpers and A. W. M. A. Jaspers J. Catal. 1996 29 F. Verpoort A. R. Bossuyt and L. Verdonck J. Electron Spectrosc. 30 J. N. Anderson D. Hennig E. Lundgren M. Methfessel R. Nyholm and 31 R. Nyholm M. Qvarford J. N. Andersen S. L. Sorensen and C. Wigren 32 J. N. Andersen A. Beutler S. L. Sorensen R. Nyholm B. Setlik and D. USA. Chem. 1980 19 2201. J. Mater. Sci. 1993 28 3997. 161 551. Relat. Phenom. 1996 82 151. M. Scheffler Phys. Rev. B 1994 50 17525. J. Phys. Condens. Matter 1992 4 277. Heskett Chem. Phys. Lett. 1997 269 371. Review 8/07826B

ISSN:0306-0012    

DOI:10.1039/a807826b

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

A rigid cyclohexane-based polyamino-polyalcohol as a versatile building block for tailored chelating agents Kaspar Hegetschweiler Universität des Saarlandes Anorganische Chemie Postfach 15 11 50 D-66041 Saarbrücken Germany Received 11th March 1999 Selective tailored chelators are of importance in medicine for the treatment of metal intoxication or to stabilise metal cations in diagnostic radiopharmaceuticals and paramagnetic contrast agents. In this report the co-ordination chemistry of polyamines polyalcohols and polyaminopolyalcohols is examined and some general prerequisites for a successful design of tailored chelators are summarised. The metal binding properties of 1,3,5-triamino-1,3,5-trideoxy-cis-inositol (taci) a rigid cyclohexane-based polyamino-polyalcohol are reviewed.Concepts for the design of selective chelators which are based on the taci structure are presented and the potential of such ligands for medical applications is briefly discussed. 1 Design and application of selective chelating agents The design of tailored chelating agents represents one of the basic challenges in the field of synthetic co-ordination chemistry. Tailored chelators can be used to govern the stability of a complex or to modulate the electronic properties of the metal cation (redox potential spin state).1 Specific control of such properties is of particular importance in medicine where selective ligands are used to treat metal intoxication (chelation Kaspar Hegetschweiler received his Diploma (1979) and doctoral degree (1984) from the Federal Institute of Technology (ETH) in Zürich (Switzerland).He then joined Paul Saltmans’ group at the University of California San Diego for a postdoctoral appointment (1984–1985). After some teaching activities (chemistry and physics at the Gymnasium Interlaken) he worked as a scientific researcher at the ETH in Zürich where he received his Habilitation (1995). In 1995 he spent six months as a visiting fellow at the Research School of Chemistry Australian National University Canberra (Prof. A. M. Sargeson) and was then appointed as a Professor (C3) for Inorganic Chemistry at the University of Saarbrücken (1996). His main interest is directed to co-ordination chemistry in aqueous solution with a particular focus on the design and development of novel tailored ligands and co-ordination polymers.therapy) or to stabilise metal cations in diagnostic radiopharmaceuticals and paramagnetic contrast agents.1 A number of prerequisites must be fulfilled for the successful use of a chelating agent in medicine. For diagnostic applications where the entire complex is introduced into a patient sufficiently high kinetic and thermodynamic stability is required to prevent decay of the complex in the body. This is of particular importance for diagnostic agents with toxic metal cations such as GdIII or radionuclides. In addition rapid clearance and suitable biodistribution of the complex are of importance. In the therapy of metal intoxication a high selectivity of the ligand towards the target metal is one of the key properties.Again a suitable biodistribution is necessary to reach the deposits of the toxic metal in the body. The free ligand itself must be non-toxic and should not be metabolised or excreted too rapidly. On the other hand the complex itself should be excretable and should be cleared promptly. A variety of chelators have been developed for the treatment of metal intoxication.1 2,3-Dimercaptopropan-1-ol (BAL British Anti-Lewisite) is one of the first examples which has been developed specifically with respect to the treatment of heavy metal poisoning. d-Penicillamine is used to cure copper overload and [Ca(EDTA)]22 is an antidote for lead intoxication. The well known microbial siderophore desferriferrioxamine B is currently used for the treatment of iron overload.Due to the increasing contamination of the environment by other toxic heavy metals such as Ni Cd or Pu the general interest in tailored chelating agents has further grown within the last few years. The metal binding properties of a multidentate ligand can be influenced by the choice of the donor atoms and by the steric demands of the ligand backbone and a systematic optimisation of these two properties must be considered to be the major tool for the design of selective chelators. The affinity of a specific metal cation for a donor atom may be described in terms of empirically derived correlations such as the HSAB principle.2 For a given donor the efficacy in metal binding in a defined aqueous medium can be tuned by optimising its basicity.Most common donors in a chelating agent are Brønsted bases and competition with the proton must be taken into account. As long as the same type of donor atom is involved this competition can be described by simple linear free energy correlations.1 In terms of absolute stability more basic ligands form more stable ligand–metal bonds. However the apparent (or ‘conditional’) stability in a given aqueous medium may be different.3 Under acidic conditions the more basic ligand may in fact be less effective due to the strong competition with the proton (Fig. 1). For example the strongly basic catecholate forms a tris complex with FeIII having an exceptionally high stability whereas the corresponding stability of the less basic oxalate is much lower.Consequently in an alkaline solution of about pH 10 the triscatecholate complex is readily formed and is stable while in the 239 Chem. Soc. Rev. 1999 28 239–249 3)3 Fig. 1 Species distribution for an equilibrated solution of Fe(NO catechol (H2cat) and oxalic acid (H2ox) with total concentrations of 1 mmol dm23 (FeIII) and 3 mmol dm23 (cat and ox). pKa values used 1.1 and 3.6 for H2ox 9.3 and 13.1 for H2cat. Overall formation constants bi (1 � i � 3) used 7.6 13.8 18.6 for [Fe(ox)i]3 2 2i and 21.0 35.9 45.1 for [Fe- (cat)i]3 2 2i (data from NIST Standard Reference Database 46 Critically Selected Stability Constants of Metal Complexes Gaithersburg MD 20899 USA).Reduction of FeIII by the organic ligands and formation of mixed oxalato–catecholato complexes are not considered. sole presence of oxalate solid FeOOH precipitates. In an acidic solution of about pH 3 the tris-catecholate complex decomposes completely due to protonation of the ligand. The trisaverage pK oxalato complex is however stable. One can show that for a given ligand optimal conditions are reached if pH equals the a values of the donor atoms (Fig. 2). Fig. 2 Sequestering ability of a hypothetical hexadentate ligand H6L (pK1 = 5 pK2 = 6 pK3 = 7 pK4 = 7 pK5 = 8 pK6 = 9). The ratio [FeL] total [L] is shown assuming excess of solid FeOOH and variable stability for FeL a) log KFeL = 23 b) log KFeL = 25 c) log KFeL = 30; Kw = 10214 Kso (FeOOH) = 10242.Steric strain within the complex can be used to stabilise or destabilise specific co-ordination geometries.4 It is for example well known that due to Jahn–Teller distortion CuII preferentially forms complexes with a tetragonal 4 + 2 coordination environment and consequently tridentate ligands with a tripodal arrangement of the donor atoms (which is restricted to a facial co-ordination) show a comparably low affinity for this cation. Molecular mechanics methods have been established as a useful and efficient tool to analyse this type of strain in terms of simple force field calculations. Other effects Chem. Soc. Rev. 1999 28 239–249 240 such as the influence of solvation are much more difficult to predict and are seldom considered in these discussions.Various synthetic routes to different types of multidentate ligands have been established in the literature. The use of suitable molecular building blocks of lower denticity which can be coupled to the target ligand is one straightforward strategy. It is particularly successful if a template method is used where the donor set is already preorganised for the coupling reaction by co-ordination to a suitable metal catioThe topic of this review is the fascinating co-ordination chemistry of polyaminopolyalcohols (sections 2 and 3) and the possibility of using these compounds as building blocks for the design of selective tailored ligands (section 4). This is illustrated by a brief overview of some selected biomedical studies (section 5) which underline the potential of these compounds for such applications.A comprehensive discussion of these applications will however be given elsewhere and is not a subject of the present report. 2 Metal binding of polyamines polyalcohols and polyamino-polyalcohols 2.1 Polyamines Organic saturated polyamines are a well known group of complexing agents for transition metal cations and innumerable reports of their co-ordinating properties have appeared in the literature.6 These ligands are of interest since the donor atom can either be present as a primary secondary or tertiary amine. An N-donor can thus act as a ramification point and can be used to build up branched structures. The systematic investigation of metal complex formation with aliphatic polyamines in aqueous solution started with the pioneering work of Schwarzenbach and co-workers.7 Macrocyclic polyamines have been of particular interest in the last two decades owing to their ability to form metal complexes of exceptionally high thermodynamic and kinetic stability.8 Saturated polyamines are fairly strong bases in water and competition with H+ must generally be considered in acidic aqueous solutions [eqn.(1)]. (1) [L–M]z+ + n H+ " HnLn+ + Maq z+ For a strong oxophilic Lewis acid the protonation of the amine can be coupled with the formation of hydroxo complexes (or of the solid metal hydroxide) and eqn. (1) changes to eqn. (2). (2) [L–M]z+ + m H2O " HmLm+ + [M(OH)m](z 2 m)+ This reaction does not depend on pH.2.2 Polyalcohols Interactions of aliphatic or alicyclic saturated polyalcohols such as the sugar alcohols or the cyclitols with metal ions are generally weak and complex formation of a neutral polyalcohol is usually not significant in aqueous solution. However deprotonated polyalcohols represent rather strong and efficient metal binding agents.9–11 Polyols are only weak acids (pKa > 12) i.e. a deprotonated polyolato ligand represents an even stronger base than a polyamine and by analogy with eqn. (1) polyolato metal complexes decompose readily in the presence of acid. It is however possible to reduce the basicity of the alkoxo group by the introduction of electron withdrawing substituents. As has been reported perfluoropinacol has a pKa of about 6 and forms a stable complex with FeIII in neutral aqueous solution whereas pinacol itself does not bind FeIII in water.12 The ability of a variety of naturally occurring polyalcohol ligands such as glycerol or sorbitol to act as sequestering agents, preventing formation of solid metal oxides or hydroxides in alkaline aqueous solution has been well known for many years.13 However the correct structures of the complexes that were formed could not be elucidated until very recently.In the last few years a number of crystal structures analyses have been reported,14,15 allowing a correct assignment of the structure of such polyolato complexes. These studies showed that the coordinated alkoxo group has a pronounced tendency to bind an additional metal cation and to form polynuclear alkoxo-bridged species.With regard to the design of tailored ligands a significant difference between the polyalcohols and the polyamines should be considered. A negative oxygen donor can only bind one Catom of the ligand backbone and can thus not be used as an interconnecting structural motif for the construction of extended ligand architectures. 2.3 Polyamino-polyalcohols Although the co-ordinating properties of polyamino-polyalcohols (papas) have only scarcely been investigated,11,16,17 these compounds exhibit a rich and interesting co-ordination chemistry. Both the oxygen and the nitrogen donors of a papaligand are regarded as hard Lewis bases in terms of the HSAB concept. The hydroxy group is however apparently harder than the amino group and soft metal cations are preferentially bound to the nitrogen donors.(3) The co-ordination chemistry of the papa-ligands is an illustrative example that a complex system cannot be discussed simply in terms of the combination of the single components. Papas can of course act as simple polyamines or polyols by binding a metal cation using the oxygen donors the nitrogen donors or a combination of both (mixed N–O-co-ordination). However the interplay of amino groups and hydroxy groups opens the possibility of further interactions which are not possible for the two individual components. As shown in eqn. (3) the amino groups can act as internal bases facilitating the deprotonation of the hydroxy groups.R2N–RA–OH " R2N(H)+–RA–O2 This reaction represents an intramolecular proton transfer and results in the formation of a zwitterionic form of the ligand. By analogy to eqn. (2) the equilibrium constant of this reaction does not depend on H+ concentration and (in contrast to a pure polyol-ligand) the generation of the deprotonated oxygen donor required for metal binding does not depend on the addition of an external base. Although the negative charge on the oxygen donor is stabilised by the positive charge of the ammonium group the intrinsic acidity of the aliphatic hydroxy group is usually not sufficient to generate the zwitterionic form of the free ligand to a substantial extent. An estimate of this acidity for a fully protonated papa-ligand can be obtained by using a corresponding model compound with permanent positive charges (Scheme 1).Comparison of the 1,3,5-trideoxy- 1,3,5-tris(trimethylammonio)-cis-inositol (H3ttci3+) cation with the triply protonated 1,3,5-trideoxy-1,3,5-tris(dimethylamino)- cis-inositol (H3tdci3+) revealed a pK1 of 8.1 and 6.5 respectively (25 °C 1 M KCl).18 If we accept these two values as the intrinsic (microscopic) acidity of the hydroxy group and the dimethylammonio group of the two trications the zwitterionic form of H2tdci2+ does not form in a greater amount. Coordination of a highly charged metal cation to the oxygen donors increases however the acidity of the hydroxy groups and the generation of the zwitterion will then predominate. This implies a dramatic stabilisation of the complex in terms of an increased conditional stability at low pH (see section 1 and Scheme 2 competitive protonation of such a papa ligand is significantly reduced since it occurs at the amino groups having a comparably weak basicity whereas metal binding occurs at Scheme 1 OH HO NR2 R2N NR2 L + M z + + 3 H+ M OH O O HO HO O NHR2 HR2N NHR HO 2 2 NHR2 3L3+ HR2N Scheme 2 NHR 241 MLz + Chem.Soc. Rev. 1999 28 239–249 H the highly nucleophilic alkoxo groups). The dramatic increase in conditional stability can be exemplified by comparing complex formation of cis-inositol ( = ino) and tdci with FeIII.15,19 The former forms the hexanuclear complex [OFe6{(ino)6 2 21H}]52 in alkaline aqueous solution.In acidic media this complex is not stable and hydrolyses completely. For tdci however stable mononuclear species are observed over the entire pH range (Fig. 3)! Fig. 3 Molecular structure of [Fe(tdci)2]3+ and species distribution of the Fe3+–tdci system (charges omitted) in equilibrated aqueous solutions. Total Fe = 1023 and total tdci ( = L) = 2 3 1023 mol dm23 (reproduced with permission from Chem. Eur. J. 1995 1 74). The peripheral (non co-ordinated) ammonium groups in such a papa-complex can of course also be deprotonated and these groups represent a reservoir of weakly acidic protons. The degree of protonation can be altered simply by varying the pH allowing the overall charge of the complex to cover a broad range.For example tdci forms a bis complex with TaV with the metal centre bound to the six alkoxo groups of the two ligand molecules (Fig. 4).20 In acidic solution the cation [Ta(tdci)2]5+ Fig. 4 Molecular structure of [H22Ta(tdci)2]Cl3 showing a trigonal prismatic rather than octahedral structure (reproduced with permission from Angew. Chem. 1997 109 2052; Angew. Chem. Int. Ed. Engl. 1997 36 1964) and pH-dependent species distribution (25 °C 0.1 mol dm23 KCl) for [H2xTa(tdci)2]5 2 x (pKa-values 4.5 5.7 7.2 9.3 10.5 11.2). is observed and the two tdci-ligands are present as neutral zwitterions. Increase of the pH results in successive deprotonation and in sufficiently alkaline medium the anion [Ta(H23tdci)2]2 is formed. It is thus possible to alter the overall charge of this complex from +5 to 21 without changing the basic structure of the Ta(tdci)2-unit.Such behaviour could be of importance with regard to medical applications where the generation of uncharged species is often required for the transfer across biomembranes. OH NH2 3 The co-ordination chemistry of 1,3,5-triamino-1,3,5-trideoxy-cis-inositol The rigid 1,3,5-triamino-1,3,5-trideoxy-cis-inositol (taci) is an illustrative representative for a papa ligand.21 As the two chair conformations are of similar energy this ligand offers a total of four different modes for metal binding (Scheme 3). Mode (i) (i) (iv) (iii) (ii) H2N HO H2N taci OH NH2 HO H2N OH HO H2N Scheme 3 and mode (iv) represent complex formation as a pure polyamine or a pure polyalcohol while modes (ii) and (iii) represent mixed N–O co-ordination.All four binding modes are restricted to a Chem. Soc. Rev. 1999 28 239–249 242 facial co-ordination of the ligand. It is however interesting that the steric requirements for a symmetric adamantane-like type (i) and type (iv) structure are distinctly different from the asymmetric type (ii) and type (iii) structures. It has been shown by a series of molecular mechanics calculations that the adamantane-like structures are particularly favourable for small metal cations whereas large cations are bound preferentially to the asymmetric type (ii) or type (iii) sites (Fig. 5).22 This ability Fig. 5 Molecular mechanics calculation for the M(taci)-fragment where M denotes a generalised metal ion with a constant force field having a type (i) type (ii) or type (iii) structure.The strain energy SU is shown as a function of the metal ionic radius (reprinted from K. Hegetschweiler M. W�orle M. D. Meienberger R. Nesper H. W. Schmalk and R. D. Hancock Inorg. Chim. Acta 250 35 Copyright 1996 with permission from Elsevier Science). to bind a metal centre either by oxygen or nitrogen donors together with the different steric requirements of the four coordination modes and the possibility of co-ordination in either a zwitterionic or a non-zwitterionic form make taci a very versatile ligand. The co-ordinating properties of this ligand have been investigated thoroughly.23 Complex formation with more than 30 different elements has been verified by crystal structure analysis (Fig.6) and the stability constants of most of the Fig. 6 Section of the periodic table. The bold faced symbols represent elements where complex formation with taci has been verified by X-ray structure analysis. complexes have been determined. For a systematic discussion of these results the different affinity of a metal cation for oxygen or nitrogen donors the different steric requirements of the symmetric and asymmetric binding sites and the charge of the metal cation must all be taken into account. A classification of metal cations Mz+ into five categories proved helpful in reviewing the co-ordination properties of taci:24 (1) With the sole exception of Be2+ metal cations of Group 1 and Group 2 elements having a d0 configuration with z � 2 form mononuclear type (iv) bis complexes.These cations are weak acids and consequently a transfer of the protons from O to N (generation of the zwitterionic form) is not observed i.e. taci acts in these complexes as a simple polyalcohol. Li+ Na+ and Mg2+ show octahedral co-ordination. For the heavier elements such as K+ Ca2+ Sr2+ and Ba2+ higher co-ordination numbers (up to 9) are realised by co-ordination of additional ligands such as H2O or a counter ion. As mentioned in section 2.2 neutral polyalcohols are only poor ligands and consequently these complexes are of rather low stability and they dissociate readily in aqueous solution (Mg2+ b mol21 dm3). 1 Å 1 (2) Small d0 and d10 metal cations with z ! 3 also form mononuclear bis complexes [M(taci)2]z+ with exclusive coordination via the oxygen donors in an octahedral fashion.However due to the higher charge the ligands co-ordinate in the zwitterionic form. This mode has been established for [Al(taci)2]3+ [Ti(taci)2]4+ [Ge(taci)2]4+ and [Sn(taci)2]4+. As discussed in section 2.3 the adoption of the zwitterionic form results in a significant stabilisation and consequently these complexes exhibit high stability in aqueous solution. (3) Divalent transition- and d10-metal cations generally tend to adopt a bis-type (i) structure. This has been found for Mn2+ Fe2+ Co2+ Ni2+ Cu2+ Zn2+ Cd2+ and also for the trivalent Tl3+. The ligand behaves as a simple tripodal amine and the stabilities of these complexes in aqueous solution fall in the range observed for a triamine.Maximum stability is observed for Ni2+ (log b2 = 20.94 Cu2+ log b2 = 18.79). As mentioned in section 1 this deviation from the Irving–Williams behaviour is a consequence of the Jahn–Teller distortion of CuII. (4) Some of the trivalent transition metal ions and small trivalent d10 metal ions form mononuclear bis complexes with a mixed type (i)–type (iv) co-ordination. The oxygen-bound ligand co-ordinates as a zwitterion. This mixed MN3O3 coordination mode represents an intermediate case between the two categories (2) and (3). It has been established for FeIII and GaIII in the solid state by crystal structure analysis of the isomorphous [M(taci)2](NO3)3 ·3H2O salts and also in aqueous solution by NMR and Mössbauer spectroscopy.(5) Large sufficiently hard trivalent metal cations such as Bi3+ and the lanthanides form trinuclear species of the composition [M3(H 2 3taci)2]3+ by co-ordinating simultaneously the three type (iii) sites of taci.25,26 Each of the three metal cations binds to an amino group and to two deprotonated alkoxo groups (Fig. 7). The two ligands which are present as trianions Fig. 7 Molecular structure of [Gd3(H23taci)2(OH2)6]3+ and species distribution of the Gd3+–taci system in equilibrated complex solutions with total Gd = 3 3 1023 and total taci ( = L) = 2 3 1023 mol dm23 (reprinted with permission from R. Hedinger M. Glisletta K. Hegetschweiler � E. T�oth A.E. Merbach R. Sessoli D. Gatteschi and V. Gramlich Inorg. Chem. 1998 37 6698. Copyright 1998 American Chemical Society). encapsulate the equilateral triangle formed by the metal centres and give rise to a cage structure with six m2-alkoxo bridges. In the lanthanide complexes the co-ordination sphere is completed by peripheral ligands such as H2O or counter ions. In the Bi complex only very weak interactions with Cl2 counter ions are observed probably due to the presence of a stereochemically active lone pair at the BiIII centres. Although this classification is useful for an overview of the rich co-ordination chemistry of taci some metal centres do not fit into this scheme. Be2+ for example is a very small and strong Lewis acid having a strictly tetrahedral co-ordination geometry.With taci a trinuclear complex of the composition [Be3(taci)3]6+ is formed in solution and in the solid state (Fig. 8).21 In this complex the three ligands co-ordinate the three Fig. 8 Molecular structure of [Be3(taci)3]6+. Be2+ centres exclusively via the oxygen donors and by analogy with category (2) the ligands adopt a zwitterionic structure. However to realise a tetrahedral environment around the Be2+ centre one of the oxygen donors undergoes bridging to a neighbouring Be2+ centre giving rise to a six-membered Be–O– Be–O-Be–O–ring. CoIII is known to be rather inert and it is thus possible to isolerent isomers which may not correspond to the most stable form of this complex. A variety of CoIII–taci complexes have been characterised in which type (i) type (ii) and type (iii) co-ordination modes have been established.23 Type (ii) co-ordination is actually very rare and besides CoIII it has only been observed in the ReI complex [(CO)3Re- (H21taci)],27 and in a trinuclear cluster of the Mo3S4 4+ unit.22 For PbII a trinuclear species of the composition [Pb3(H23- taci)(m3-OH)]2+ was identified in solution,28 and this trinuclear structure was also observed in the solid state where OH2 was replaced by a bridging nitrate.25 The structure of the [Pb3(H23taci)]3+ moiety corresponds to the M3(H23taci)2-cage described in category (5) with one of the ligands dissociated.4 Derivatisation of taci the design of tailored ligands 4.1 Synthetic strategies Because of the high number of functional groups present in the taci ligand a variety of synthetic modifications can be performed.These modifications are of particular interest with regard to a systematic modulation of the co-ordinating properties. However the multifunctionality may render a specific synthetic modification difficult to control and several strategies have been developed for a selective substitution either at the Chem. Soc. Rev. 1999 28 239–249 243 oxygen or nitrogen atoms.18,29 The synthesis of O-alkylated derivatives for example proved particularly troublesome since direct treatment of taci with alkylating agents resulted in Nrather than O-alkylation. A variety of classical protecting groups for the primary amine functions were investigated.Although such groups could readily be attached to the nitrogen atoms the resulting compounds showed no significant reactivity towards the alkylating agent. Obviously the large steric demands of the protecting groups enforced equatorial orientation of the nitrogen donors and consequently the hydroxy groups were in the less reactive axial positions. The combination of an equatorial position for the hydroxy groups and effective protection of the amino groups could be achieved by using [Ni(taci)2]2+ as starting material.29 This procedure allowed an almost quantitative and selective alkylation of the oxygen atoms. A systematic investigation of such substituted taci derivatives has been undertaken and these compounds were shown to have a number of interesting properties.19,29–32 Some examples will be discussed in the following sections (4.2–4.6).4.2 Methylated derivatives 2]3+ FeL2 The introduction of methyl groups on either the oxygen or nitrogen donors generates a set of rather selective ligands.19,29 In 1,3,5-trideoxy-1,3,5-tris(dimethylamino)-cis-inositol (tdci Scheme 1) the conformer with three axial dimethylamino groups is destabilised by non-bonding (intra-ligand) repulsion and only the conformation with three axial hydroxy groups is available for metal binding. Also additional inter-ligand repulsion between co-ordinated dimethylamino groups substantially increases steric strain in a bis-complex of tdci with MN6 co-ordination. Similar considerations are valid for all-cis- 2,4,6-trimethoxycyclohexane-1,3,5-triamine (tmca) where preferential co-ordination by three axial nitrogen atoms is observed.The FeIII complexes of taci tdci and tmca illustrate this behaviour (Scheme 4) [Fe(tdci)2]3+ has an FeO6 structure whereas [Fe(tmca)2]3+ is one of the rare examples of a ferric hexaamine complex.30 As mentioned in section 3 [Fe(taci) exhibited a mixed FeN3O3 co-ordination which demonstrates that for high spin FeIII the mixed O–N environment is of lowest energy. This was further confirmed by the facile formation of [Fe(tmca)(tdci)]3+ which was obtained almost quantitatively by a simple metathesis reaction of [Fe(tdci)2]3+ and [Fe(tmci)2]3+. The two types of methylated ligands can therefore be used to direct co-ordination modes which would not be preferred with taci.Since the steric properties of the syn-triaxial binding sites are similar for the three ligands it is possible to elucidate the intrinsic electronic influence of oxygen and nitrogen donors on the properties of the central metal cation. [Fe(taci)2]3+ and [Fe(tdci)2]3+ as well as the mixed [Fe(tdci)(tmca)]3+ are highspin and labile whereas [Fe(tmca)2]3+ is low-spin and inert. Clearly it is the N-ligand which generates the low spin coordination and more than three N-donors are required. With respect to selectivity tdci is an efficient ligand for hard and highly charged cations such as Fe3+ Al3+ or Ti4+ (log b = 32.6);19 tmca on the other hand represents an effective ligand for late divalent transition metal cations such as Ni2+.29 It is of particular interest that compared to taci (which carries both coordination sites combined in one molecule) the stabilities of the bis complexes of tmca with late divalent transition metal cations and of tdci with the highly charged oxophilic centres exceed those of taci by up to seven orders of magnitude even though no significant structural differences were observed within the coordination spheres or the ligand backbones of corresponding structures.It is not obvious how this increase of stability can be explained. At least for tdci the donor groups have a higher degree of preorientation than in taci since the possibility for chair conversion is inhibited. However an improved preorienta- Chem. Soc.Rev. 1999 28 239–249 244 3+ OH HO NH2 H2N NH2 Fe O O O NH3 NH3 H3N NH OH 2 HO H2N H2N HO OH NH2 NH HO 2 OH HO H2N taci NH2 OH H2N HO H2N OH N CH3O N N OCH3 CH3O tdci tmca 3+ 3+ OCH3 CH3O NH NH CH3O NH O NH2 H2N O O Fe Fe O O H2N NH NH2 2 O NH H2N OCH3 NH HN CH3O OCH3 3+ OCH3 CH3O CH3O NH2 H2N NH2 Fe O O O NH HN NH Scheme 4 Adapted with permission from K. Hegetschweiler M. Weber V. Huch M. Veith H. W. Schmalle A. Linden R. J. Geue P. Oswath A. M. Sorgeson A. C. Willis and W. Angst Inorg. Chem. 1997 36 4121. Copyright 1997 American Chemical Society. tion of the donor groups alone cannot explain such a large increase in stability for both cases (i.e. tdci and tmca) and other reasons e.g.a more favourable solvation of the complex (in terms of a more favourable entropy of solvation) must probably account for this observation. 4.3 Substituents with additional donor groups If substituents are used which carry additional donor groups the taci ligand can be extended to a potentially penta- or hexadentate chelator. As an example the condensation of taci with three equivalents of salicylic aldehyde followed by hydrogenation of the CNN double bonds generates the tris-Nalkylated 1,3,5-trideoxy-1,3,5-tris[(2-hydroxybenzyl)amino]- 3thci) providing three aliphatic amino groups cis-inositol (H three aliphatic hydroxy groups and three phenolic hydroxy groups for the co-ordination of a cation.31 The ReV complex [ReO(thci)] is an example where all three types of donor atoms are involved in metal binding (Fig.9). In this complex the Fig. 9 Formation and molecular structure of [ReO(thci)] (Reprinted with permission from K. Hegetschweiler A. Egli R. Alberto and H. W. Schmalle Inorg. Chem. 1992 31 4027. Copyright 1992 American Chemical Society). ReNO unit binds to the ‘asymmetric’ type (ii) site. The coordinated aliphatic oxygen is deprotonated. In addition two amino groups together with their pendant o-phenoxy substituents are co-ordinated to the cation. It is of particular interest to note that the third amino group together with its pendant aromatic substituent is not involved in metal binding and the ligand thci32 exhibits pentadentate co-ordination. This clearly illustrates the ability of this ligand to adapt its co-ordination behaviour to a specific metal cation.4.4 Template methods 2]3+ Coupling of two taci molecules by linking via nitrogen donors represents an additional promising route to novel hexadentate chelators. The synthesis of such bis-taci ligands is straightforward if a template method is used as mentioned in section 1. For this purpose the two hexaamine complexes [Co(taci) and its O-methylated derivative [Co(tmca)2]3+ have been prepared and they were subsequently converted to the corresponding hexa-imino derivatives.32 This method was originally developed by Sargeson and co-workers (as reviewed in ref. 5). The two hexa-imines proved to be stable in acidic aqueous solution and could be isolated as solids.With nucleophiles such as a hydride or a carbanion a rapid addition reaction was observed and in the presence of acetaldehyde or nitromethane and base the two tridentate ligands could readily be coupled yielding corresponding CoIII complexes of new macrocyclic ligands (Scheme 5). Demetalation of these complexes and isolation of the free ligands have however not been described as yet. 3+ CH OCH3 3O CH3O NH2 NH NH2 Co NH NH2 NH OCH3 OCH3 CH3O R = CH3 CH3CHO CH2O NEt3 3+ 3+ OR RO RO OR RO N RO H2N CH2 N N NH2 H2N H CH2 2C Co Co NEt3 CH2 H2C N N NH2 H2N CH2O H2C N NH2OR OR OR RO RO R = H CH3NO2 NEt3 + HO OH HO HN N NO2 – OR – NH CH3 Co O2N NH N CH3 NH OH OH HO Scheme 5 Adapted with permission from K.Hegetschweiler M. Weber V. Huch R. J. Geue A. D. Rae A. C. Willis and A. M. Sorgeson Inorg. Chem. 1998 37 6136. Copyright 1998 American Chemical Society. 4.5 Variation of lipophilicity Replacement of the H–O or H–N hydrogen atoms by organic substituents inherently increases the lipophilicity of the ligand. Already tdci the tris-N,N-dimethylated derivative of taci is clearly a much more lipophilic molecule than its parent. The unsubstituted taci behaves similarly to a sugar and is only soluble in water and to a very limited extent in MeOH while tdci is readily soluble in chloroform or even in boiling hexane.18 The difference in lipophilicity is also of relevance to the metal binding properties as exemplified by the significantly different behaviour of the PbII complexes of taci and tdci.The solution chemistry of the PbII–taci system is restricted to water. In aqueous media the trinuclear [Pb3(H23taci)(OH)]2+ (see section 3) is formed.28 A related species was observed for tdci in MeOH having the composition [Pb3(H23tdci)(m3-OCH3)]2+. Here hydroxide is replaced by a m3-methoxo ligand. However tdci also forms the neutral species [Pb3(H23tdci)2] which has a rather unpolar surface.33 Due to the high lipophilicity and the lack of charge this complex is insoluble in water but soluble in organic solvents and despite the rather high molecular weight it is volatile and sublimes at 250 °C (14 mbar N2). Volatile metal complexes are of interest for metal organic chemical Chem.Soc. Rev. 1999 28 239–249 245 vapour deposition (MOCVD) and in a recent study it has been shown that this compound can be used as a precursor for the deposition of thin films of lead and lead(ii) oxide on various substrates by the MOCVD technique.33 4.6 Bifunctionalisation Finally the introduction of suitable substituents could be of use to generate bifunctionalised derivatives where one of the substituents serves as a linker to attach the complex covalently to a protein or to an ion exchange resin. Such modifications are of particular interest with regard to medical applications (see section 5.3). The potentially pentadentate H3bhci–glu–H provides an example of bifunctionalisation where one type of substituent is used to extend the ligand system (the two ohydroxybenzyl substituents) and another is used for the introduction of a linking unit (the glutaric acid moiety).34 The different synthetic routes shown in Scheme 6 once again demonstrate the difficulties encountered with the high multifunctionality of this system.Although the ligand H3bhci could be synthesised by condensation of two equivalents of salicylic aldehyde with taci and subsequent reduction of the imines to amines this process gave rise to a mixture of 1 1 1 2 and 1 3 products. Subsequent derivatisation of H3bhci with glutaric anhydride resulted in the formation of a crude mixture of various compounds and the desired product was only obtained in poor yield. The reverse procedure in which the monoamide was formed first and was further derivatised with salicylic aldehyde was more successful since the intermediate amide is of low solubility and precipitates out of solution.The subsequent reduction of the imines was easily performed and good yields of the bifunctional ligand H3bhci–glu–H were achieved. The two ReV complexes [ReO(bhci)] and [ReO(bhci–glu–H)] have been prepared for potential radiopharmaceutical applications (Scheme 7).34 Inspection of the structural features of the two complexes revealed a remarkable difference in the conformation of the cyclohexane rings. In [ReO(bhci)] an H2N NH2 NH2 OH OH HO taci HO NH2 OH NH HN OH OH H3bhci O NH HO O 186 ReV NH2 NH O OH OH O Scheme 7 Adapted with permission from A.Kramer R. Alberto A. Egli I. Novak-Hofer K. Hegetschweiler U. Abram P. V. Bernhard and P. A. Schubiger Bioconjugate Chem. 1998 9 691. Copyright 1998 American Chemical Society. [186ReO(bhci-glu)]-mAB [186ReO(bhci)] Chem. Soc. Rev. 1999 28 239–249 246 H2N NH2 NH2 OH OH HO O H taci OH O O O 1 eq. 2 eq. H2 PtO2 O O HO OH 2 H HN 2N H2N OH OH NH NH HN OH OH HO OH HO not isolated H3bhci O H OH O O 2 eq. O 1 eq. overall yield 2% 61% H2 PtO2 O O HO OH OH NH HN HN OH HO OH H3bhci-glu-H Scheme 6 Reprinted with permission from A. Kramer R. Alberto A. Egli I. Novak-Hofer K. Hegetschweiler U. Abram P. V. Bernhard and P. A. Schubiger Bioconjugate Chem.1998 9 691. Copyright 1998 American Chemical Society. almost ideal chair is observed while [ReO(bhci–glu–H)] exhibited a twisted boat conformation. This difference can be O O HO OH NH OH HN HN OH OH HO O NH H3bhci-glu-H O 186 ReV NH O OH OH O O O NH OH [186ReO(bhci-glu-H)] O NH O 186 ReV NH O OH OH O O Lys O NH NH explained on the basis of the different types of intramolecular hydrogen bonds which are formed in the two complexes underlining though the importance of hydrogen bonding between different substituents for the structural properties of complexes with such cyclohexane-based polyamino-polyalcohol ligands. 50 > 5 Applications 5.1 Iron overload Iron overload is mainly caused by regular blood transfusions which are necessary to treat genetically imposed disorders of haemoglobin production such as Thalassaemia.35 The ability of the body to excrete excess iron is very limited and repeated blood transfusions inevitably lead to an accumulation of iron in the body.In a first step excess iron is accumulated in specific storage proteins such as ferritin. Further uptake of iron results in the deposition of solid FeOOH in tissues and organs and the patients will finally die owing to severe damage to these organs. Administration of a specific iron binding chelator is the method of choice to solve this problem. The use of desferrioxamine B for the removal of excess iron has already been mentioned in the introduction (section 1).However this hydroxamate-based ligand has some severe disadvantages. It is inactive when administered orally and has a very short biological half life. Thalassaemia is responsible for some 100 000 child deaths per year and consequently the search for selective iron chelating agents is of particular importance. Although considerable efforts have been made to develop more suitable iron chelators in the last two decades,35 a completely satisfactory compound has not yet been found. The high affinity of tdci for hard and highly charged Lewis acids and its low tendency to form complexes with divalent transition metal cations make it an ideal candidate for the treatment of iron overload. The excellent selectivity of tdci can be attributed to (a) the restriction to only one chair conformation which disfavours the competitive binding of Cu2+ and Zn2+ and (b) the poor donor capacity of the (neutral) hydroxy groups which renders the interactions with the ubiquitous Mg2+ and Ca2+ very weak.In fact the oxygen donors of tdci are only effective in metal binding if they are deprotonated and the divalent Mg2+ and Ca2+ are not acidic enough to cause the required proton transfer from the hydroxy groups to the amino groups (see section 3).19 In the complex with the trivalent FeIII however co-ordination to alkoxo groups is possible even in acidic media (the generation of the tautomeric form is not dependent on pH see Scheme 1 Fig. 3 and also section 2.3). Animal studies revealed a rather low toxicity of tdci (LD 1 g kg21) and a preliminary in vitro screening exhibited very favourable kinetics for the dissolution of solid FeOOH.36 Moreover a recent animal study with the 59Fe labelled bis-tdci complex showed excellent clearance when the complex was administered subcutaneously.Over 80% of the injected [59Fe(tdci)2]3+ was excreted through the kidneys within 24 h.37 Other ligands such as the currently used desferrioxamine B have considerably longer retention times. However studies with iron overloaded rats revealed an almost negligible removal of iron when the free ligand was administered subcutaneously. This disappointing result clearly indicates that tdci itself cannot be used as a therapeutic agent to treat iron overload.The reason for the failure of tdci to sequester substantial amounts of Fe is not clear. The ligand could be metabolised too rapidly or an improper biodistribution might impede rapid loading of the ligand with Fe. These problems may be circumvented by adding different substituents to the nitrogen donors however to date such efforts have not brought success. 5.2 Contrast agents for magnetic resonance imaging (MRI) In the last two decades nuclear magnetic resonance (NMR) imaging has been developed as a powerful non-invasive diagnostic tool for acquiring images of tissues as topological representations of NMR parameters.38 The lack of ionising radiation is a particularly attractive advantage of this method. Although the use of a contrast enhancing agent is not an essential prerequisite for MRI it increases lesion detection and diagnostic accuracy.Since nearly all NMR clinical imaging involves proton magnetic resonance and the principal proton species that generates the NMR signal is the water molecule an effective contrast enhancer basically operates by affecting the NMR properties of body water (faster proton relaxation). For this purpose a variety of paramagnetic metal complexes have been studied. Owing to the high number of unpaired electrons Gd3+ complexes are of particular interest. The free Gd3+ aqua ion is however toxic and it must therefore be administered in the form of a complex of sufficient stability. The ability of such a complex to decrease the relaxation time of the surrounding water protons is governed by the number of inner sphere water molecules the rate of water exchange rotation and electronic relaxation.Currently GdIII complexes with suitable polyamino polycarboxylates are utilised as contrast agents for MRI.38 As has been shown in section 3 the trivalent lanthanide cations form trinuclear complexes of the composition [M3(H23taci)2(H2O)6]3+. The corresponding GdIII complex could be of particular interest in the context of MRI owing to the compact structure the high amount of paramagnetism induced by three GdIII centres and the total of six water ligands which are attached to the metal atoms. A comprehensive study has been performed to elucidate the structure as well as the electronic and solution properties of this complex.26,39 Magnetic susceptibility measurements showed that only weak antiferromagnetic coupling is present within the trinuclear core and at room temperature the full magnetic moment corresponding to 21 unpaired electrons was observed.The water exchange has been investigated by 17O-NMR measurement. It is considerably slower than in the Gd3+ aqua ion and has much more associative character. The difference in the exchange rate can be explained by the rigidity of [Gd3(H23taci)2(H2O)6]3+ which slows down the transition from the eight co-ordinate ground state to the nine co-ordinate transition state. The rotational correlation time of the complex is unexpectedly low and this has been interpreted in terms of the spherical structure with a large hydrophobic surface which prevents the formation of a substantial hydration sphere around the trinuclear complex molecule.The stability of this complex in water was investigated by an NMR study and pH-metric titrations. These measurements showed that [Gd3(H23taci)2(H2O)6]3+ readily forms in alkaline media but decomposes immediately if the pH falls below 6 (Fig. 7). The inability of taci to bind lanthanide cations in acidic aqueous media is based on the very unfavourable proton balance of the formation reaction (4). (4) 2 H3taci3+ + 3 Eu3+ + 12 OH2 ? [Eu3(H23taci)2]3+ + 12 H2O In the context of MRI the potential use of [Gd3(H23- taci)2(H2O)6]3+ as a contrast agent is thus limited by its insufficient stability under physiological conditions.The trinuclear complex is however an interesting model for studying the effects of intramolecular Gd–Gd interaction on electronic relaxation and consequently on proton relaxivity. Increased stability could be achieved by connecting the two taci frameworks by an N–(CH2)2–NH–(CH2)2–N or N–(CH2)2–O– (CH2)2–N bridge (section 4.4) or by introducing substituents onto the nitrogen atoms which carry additional donor groups 247 Chem. Soc. Rev. 1999 28 239–249 (section 4.3). The investigation of such ligands is currently being carried out in our laboratory. 5.3 Radiopharmaceutical applications Labelling of tumour seeking monoclonal antibodies (MAB) with radioactive nuclides is another well established diagnostic tool and is of particular use for the location of cancer metastases.In addition this method is relevant to therapeutic applications.40 Due to favourable g-decay characteristics and ready availability radioimmunoconjugates labelled with 99mTc are used for diagnostic imaging of specific organs. The bemitting rhenium isotopes 186Re and 188Re are suited for therapeutic purposes (radioimmunotherapy RIT).40 Based on the chemical relationship between rhenium and technetium the same ligands that are used for 99mTc labelling may also be applicable for 186/188Re. Currently MAG3 a tetradentate N3S ligand originally developed for renal function measurement with 99mTc is used as a ligand in tumour therapy with 186/188Re. The Re–MAG3 system showed some promising results although the labelling protocols have not proved to be convenient for routine application.There is thus considerable demand for new suitable chelators which form stable complexes with Re and which could be attached to a protein. A broad range of possible oxidation states is known for Re. ReV has proved to be suitable as it forms stable chelate complexes which can be prepared readily from ReO42.1 The design of a specific pentadentate bifunctionalised chelator for the ReVNO moiety has been described in section 4.6 (Schemes 6 and 7).31,34 The derivatisation of H3bhci with glutaric acid not only offers the possibility of linking the ligand to a protein but also prevents coordination of a cation to the three nitrogen atoms. In this way a competitive binding of biologically relevant divalent transition metal cations such as Cu2+ or Zn2+ can be avoided.The ligand bhci32 and its bifunctionalised analogue bhci–glu42 both encapsulate the ReV centre efficiently yielding a distorted octahedral co-ordination geometry and the rigid cyclohexane backbone together with the preorientation of the N2O3 donor set lead to the high stability of the complex. Neither [ReO(bhci)] nor [ReO(bhci–glu)]2 showed any detectable decomposition in human serum or under in vivo conditions while ligands with an Fig. 10 Tumour to organ ratios of [186ReO(bhci–glu)]-labelled mAb-35 in tumour bearing nude mice. The ratios were determined at 4 8 12 24 and 48 h post injection (a) for the labelled F(abA)2 fragment and at 24 96 and 144 h post injection (b) for the labelled intact antibody (Reprinted with permission from A.Kramer R. Alberto A. Egli I. Navak-Hofer K. Hegetschweiler U. Abram P. V. Bernhard and P. A. Schubiger Bioconjugate Chem. 1998 9 691. Copyright 1998 American Chemical Society). Chem. Soc. Rev. 1999 28 239–249 248 analogous N3O3 donor set but lacking the cyclohexane core decomposed in blood serum over a period of three days. The bifunctionalised H3bhci–glu–H was applied in a prelabelling protocol to 186/188Re (Scheme 7) and subsequently the anti colon cancer antibody mAb-35 was labelled with [186/188ReO(bhci–glu–H)] with full retention of immunoreactivity. Biodistribution of 186Re labelled mAb-35 in tumour bearing mice revealed good tumour uptake with no significant accumulation of radioactivity in normal tissue.34 Since it is known that F(abA)2 fragments may present advantages for RIT compared to intact antibodies in terms of better tumour penetration and faster clearance from the body both intact mAb-35 and its F(abA)2 fragment were labelled and tested as radio immuno conjugates in a comparative study.Animal experiments showed promising results in terms of biodistribution and clearance from the body especially of the labelled mAb-35 F(abA)2 fragments (Fig. 10). The results with this novel [ReO(bhci–glu)] label clearly indicate its potential for 186/188Re labelling of antitumour antibodies and for their use in RIT. 6 Acknowledgements The author thanks Dr Peter Osvath (Melbourne) for his valuable advice and support and his co-workers whose names appear in the references for intense and fruitful collaboration.Parts of this work were financially supported by Vifor (international) St. Gallen the ETH-Zürich the Paul Scherrer Institute (PSI Villigen) the Studienstiftung des deutschen Volkes and the Deutsche Forschungsgemeinschaft. 7 References 1 A. E. Martell and R. D. Hancock Metal Complexes in Aqueous Solutions Plenum Press New York 1996. 2 R. G. Pearson Chemical Hardness Wiley-VCH Weinheim Germany 1997. 3 J. N. Butler and D. R. Cogley Ionic Equilibrium—Solubility and pH Calculations Wiley New York NY 1998. 4 P. Comba and T. W. Hambley Molecular Modeling of Inorganic Compounds VCH Weinheim Germany 1995. 5 G. A. Lawrance M. Maeder and E.N. Wilkes Rev. Inorg. Chem. 1993 13 199 and references therein. 6 D. A. House Comprehensive Coordination Chemistry G. Wilkinson ed. Vol. 2 Pergamon Press Oxford UK 1987 pp. 23–72. 7 P. Paoletti R. Walser A. Vacca and G. Schwarzenbach Helv. Chim. Acta 1971 54 243 and references therein. 8 A. Bianchi M. Micheloni and P. Paoletti Coord. Chem. Rev. 1991 110 17. 9 K. Burger and L. Nagy in Biocoordination Chemistry Coordination Equilibria in Biologically Active Systems K. Burger ed. Ellis Horwood Chichester 1990 pp. 236–283. 10 J.-F. Verchère S. Chapelle F. Xin and D. C. Crans Prog. Inorg. Chem. 1998 47 837. 11 S. Yano Coord. Chem. Rev. 1988 92 113. 12 M. Allan and C. J. Willis J. Am. Chem. Soc. 1968 90 5343. 13 P. Goldschmidt E. Weingardt and W.Bachmann Kolloid-Z. 1929 47 14 J. Burger C. Gack and P. Klüfers Angew. Chem. 1995 107 2950; 15 K. Hegetschweiler L. Hausherr-Primo W. H. Koppenol V. Gramlich 49. Angew. Chem. Int. Ed. Engl. 1995 34 2647. L. Odier W. Meyer H. Winkler and A. X. Trautwein Angew. Chem. 1995 107 2421; Angew. Chem. Int. Ed. Engl. 1995 34 2242. 16 D. A. House V. McKee and P. J. Steel Inorg. Chem. 1986 25 4884. 17 W. Plass Eur. J. Inorg. Chem. 1998 799 and references therein. 18 K. Hegetschweiler I. Erni W. Schneider and H. Schmalle Helv. Chim. Acta 1990 73 97. 19 K. Hegetschweiler T. Kradolfer V. Gramlich and R. D. Hancock Chem. Eur. J. 1995 1 74. 20 K. Hegetschweiler T. Raber G. J. Reiß W. Frank M. Wörle A. Currao R. Nesper and T. Kradolfer Angew. Chem.1997 109 2052; Angew. Chem. Int. Ed. Engl. 1997 36 1964. 21 M. C. Ghisletta Thesis Diss. No. 10891 ETH-Zürich Switzerland 1994. 22 K. Hegetschweiler M. Wörle M. D. Meienberger R. Nesper H. W. Schmalle and R. D. Hancock Inorg. Chim. Acta 1996 250 35. 23 M. Ghisletta L. Hausherr-Primo K. Gajda-Schrantz G. Machula L. Nagy H. W. Schmalle G. Rihs F. Endres and K. Hegetschweiler Inorg. Chem. 1998 37 997 and references therein. 24 K. Hegetschweiler R. D. Hancock M. Ghisletta T. Kradolfer V. Gramlich and H. W. Schmalle Inorg. Chem. 1993 32 5273. 25 K. Hegetschweiler M. Ghisletta and V. Gramlich Inorg. Chem. 1993 32 2699. 26 R. Hedinger M. Ghisletta K. Hegetschweiler � E. T�oth A. E. Merbach R. Sessoli D. Gatteschi and V. Gramlich Inorg. Chem. 1998 37 6698. 27 A. Egli Thesis Diss. No. 10936 ETH-Zürich Switzerland 1994. 28 R. Hedinger Thesis Diss. No. 12611 ETH-Zürich Switzerland 1998. 29 M. Weber D. Kuppert K. Hegetschweiler and V. Gramlich Inorg. Chem. 1999 38 859. 30 K. Hegetschweiler M. Weber V. Huch M. Veith H. W. Schmalle A. Linden R. J. Geue P. Osvath A. M. Sargeson A. C. Willis and W. Angst Inorg. Chem. 1997 36 4121. 31 K. Hegetschweiler A. Egli R. Alberto and H. W. Schmalle Inorg. Chem. 1992 31 4027. 32 K. Hegetschweiler M. Weber V. Huch R. J. Geue A. D. Rae A. C. Willis and A. M. Sargeson Inorg. Chem. 1998 37 6136. 33 R. Hedinger T. Kradolfer K. Hegetschweiler M. Wörle and K.-H. Dahmen Chem. Vap. Deposition 1999 5 29. 34 A. Kramer R. Alberto A. Egli I. Novak-Hofer K. Hegetschweiler U. Abram P. V. Bernhard and P. A. Schubiger Bioconjugate Chem. 1998 9 691. 35 S. Singh Chem. Ind. 1994 452. 36 W. Schneider I. Erni and K. Hegetschweiler all-cis-1,3,5-Triamino- 2,4,6-cyclohexanetriol derivatives their use processes for their preparation and pharmaceutical applications containing them European Patent EP 0 190 676 B1 1988. 37 A. Wong M. J. Pippard K. Pattanapanyasat B. Faller and H. P. Schnebli In Vivo Distribution of Iron Chelator Complexes and Routes 38 S. Aime M. Botta M. Fasano and E. Terreno Chem. Soc. Rev. 1998 39 É. Tóth L. Helm A. E. Merbach R. Hedinger K. Hegetschweiler and 40 J. R. Dilworth and S. J. Parrott Chem. Soc. Rev. 1998 27 43 and of Iron Excretion Novartis Pharma AG Internal Report 1995. 27 19 and references therein. A. Jánossy Inorg. Chem. 1998 37 4104. references therein. Review 8/02638F 249 Chem. Soc. Rev. 1999 28 239&ndash

ISSN:0306-0012    

DOI:10.1039/a802638f

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

3-Ylidenepiperazine-2,5-diones as versatile organic substrates Jürgen Liebscher and Shangde Jin Institut für Chemie Humboldt-Universität Berlin Hessische Str. 1-2 D-10115 Berlin Germany Received 22nd June 1998 3-Ylidenepiperazine-2,5-diones and 3,6-diylidenepiperazine-2,5-diones are cyclic dipeptides consisting of one or two didehydroamino acid moieties respectively. Some compounds of this series occur in nature. They can easily be synthesised by several methods also in optically active form and are prone to addition reactions to the C–C double bond by electrophiles (enamine reactivity) nucleophiles (Michael reactivity) radicals oxidising reagents or 1,3-dipoles usually in a stereoselective manner. The resulting adducts can further be transformed to natural products and analogues or serve as precursors for interesting a-amino or a-keto acid derivatives by cleavage of the diketopiperazine ring.1 Introduction 3-Ylidenepiperazine-2,5-diones 1 (hereafter piperazine- 2,5-dione is abbreviated as PDO some authors prefer the term 2,5-diketopiperazine abbreviated as DKP) and 3,6-diylidene- PDOs 5 have important functions in biochemistry and synthetic chemistry. Such compounds are cyclic dipeptides composed of a didehydroamino acid and an a-amino acid or of two didehydroamino acids respectively. 3-Ylidene-PDOs and 3,6-diylidene-PDOs were found as natural products1 often produced by fungi such as Actinomyces strains and Penicillium species. Some of these compounds exhibit antibacterial activity Shangde Jin was born in Jilin province P.R. China in 1964. He studied chemistry at Lanzhou University and obtained his BSc in 1986. Afterwards he worked at Changchun Institute of Applied Chemistry Chinese Academy of Sciences where he became assistant researcher in 1992 and finished his MSc in 1996. Since the end of 1996 he has been working in the group of Professor J. Liebscher at Institute of Organic and Bioorganic Chemistry Humboldt- University Berlin. His current research is in asymmetric synthesis of natural products and their analogues. Shangde Jin J�urgen Liebscher and inhibit tumour growth in mice. Their biochemical origin is obviously from peptides and proteins. From the synthetic point of view 3-ylidene-PDOs 1 and 3,6-diylidene-PDOs 5 have gained wide interest because of their various reactive functional groups allowing reactions at different positions of the ring and at the exocyclic carbon atom of the C–C double bond (Scheme 1).Nucleophilic attack at the carbonyl carbon atoms could be used for hydrolytic cleavage of the piperazine ring. If the ring nitrogen atoms are unsubstituted or are substituted by alkyl groups they can be attacked by electrophiles allowing the introduction of substituents at these positions. The carbon atom a to the carbonyl group at position 6 can be acidic. The C–C double bonds have been found to be particularly synthetically useful because they allow addition reactions affording saturated products 2 or 7 or partially unsaturated products 6.Since chirality can be created at position 6 of 3-ylidene-PDOs 1 such additions can also be carried out in a stereoselective manner. After further hydrolytic cleavage of the PDO-ring of 2 interesting a-amino acids 3 could be synthesised. This hydrolytic cleavage is possible under acidic conditions without racemisation2,3 and the resulting amino acids can be esterified for better separation.3 Since the C–C double bonds of 3-ylidene-PDOs 1 or 5 are part of a didehydroamino acid addition reactions can follow a different mechanism. Thus the exocyclic position of the C–C double bond can be attacked by nucleophiles (Michael acceptor reactivity formation of enolates 8 and products 9) or by electrophiles (enamine reactivity formation of N-acyliminium Jürgen Liebscher is Professor of Organic and Bioorganic Chemistry at Humboldt-University Berlin.He was born in Freital Germany in 1945. He received his MS (1969) PhD (1973) and DSc (1977) degrees from the Technical University Dresden Germany. He became senior assistant and moved to the Department of Chemistry Addis Ababa University Ethiopia as Associate Professor (1979 to 1982). He joined Humboldt- University Berlin Department of Organic Chemistry in 1982 as Dozent and was appointed Professor in 1992. He stayed at the University of Würzburg and at University of Texas as Visiting Professor. His research interests comprise organic synthesis heterocyclic chemistry natural products and analogues and asymmetric synthesis. He has published more than 160 publications.251 Chem. Soc. Rev. 1999 28 251–259 O on several factors such as reaction conditions substituents at the piperazine ring or the presence of a second ylidene group N attached to the PDO ring such as in compounds 5. N O 1 O O NH X N + HO X OH Y Y N NH O O 3 4 2 O O X N N Y N N O O 5 6 O X N Y N Y O X 7 Scheme 1 structures 10 and products 11 Scheme 2). Furthermore radical attack is possible at this carbon since a radical intermediate 12 is formed which is stabilised by the capto-dative effect and cycloaddition reactions are also feasible. Semi-empirical calculations as well as the comparison of 13C NMR chemical shifts revealed a good deal of enamine character for 3-ylidene-PDO 1.Practical investigations however showed that the reaction behaviour i.e. Michael system versus enamine system depends O O – Nu N N Nu– N N O O 8 X • E + O O Nu N X E N 1 C • N O O 2 Synthesis of 3-ylidenepiperazine-2,5-diones 3-Ylidene-PDOs have been synthesised with various substituent patterns in the non-chiral racemic and optically active series by different synthetic pathways. In the following just the basic principles will be addressed briefly. Syntheses of 3-ylidene-PDOs either established the piperazinedione ring from an open-chained didehydroamino acid precursor or started with a preformed PDO and introduced the ylidene group. Optically active 4-acyl-3-ylidene-PDOs 16 are conveniently available by an Erlenmeyer-type route by ring opening of 4-ylideneoxazolones 13 (Scheme 3) with a-amino carboxylates 14 and subsequent cyclisation of the resulting dipeptide 15.2 The N-acyl group of the 3-ylidene-PDO 16 could be removed by reaction with amino compounds (formation of 17) and alkyl groups were introduced by further N-alkylation.In an alternative route racemic or achiral 3-alkylidene-PDOs 21 could be synthesised by cyclisation of N-(a-aminoacyl)dehydroamino acid esters 20 obtained either from didehydroamino acid ester 18 and a-phthalimidoacyl chlorides 19 or by reaction of a-haloacetamides 23 with a-ketoesters 224 (Scheme 4). For establishing the ylidene group of PDOs 255 and 276 a Wittig or a Wittig–Horner reaction is suitable if either 2,3,5-triketopiperazines 24 (derived from an a-amino acid and oxalate) or PDO-3-phosphonate 26 are used as starting materials (Scheme 5).The latter method allowed optically active 3-ylidene-PDOs 27 to be obtained. As an alternative to the Wittig reaction 3-ylidene groups could be introduced into PDOs by aldehydes using aldol reactions. Thus a one- or two-fold aldol condensation of 28 gave access to 3-ylidene-PDOs 29 and 3,6-diylidene-PDOs 30 (Scheme 6).7 Often N,NA-diacetylated 28 (R3 = R4 = Ac) were used losing the acetyl group attached to the N-atom at position 4 by an intramolecular acyl transfer thus affording 3-alkylidene- PDOs 31.7–10 While aldol condensations of PDOs were usually implemented as one-pot procedures in a basic medium7–11 eventually under PTC conditions or accelerated by ultrasound or in the presence of AcONa–Ac N 2O sometimes intermediate O E N E + N+ O 10 Nu– O E N Nu N O 11 9 12 " Y • " 2 Scheme 2 Chem.Soc. Rev. 1999 28 251–259 252 R5 R3NH O R1 O R1 R3 COONa N R2 O R2 N R5 R4 R4CONH NaO2C 15 13 R1 R1 O O 2 R3 R3 N N R2 R2 N N R5 H R5 R4CO O O 17 16 Scheme 6 SR) could alternatively be obtained by cyclisation of dipeptides derived from sere or cysteine. b-Elimination of 32 gave access to optically active 3-ylidene-PDOs 33 in particular to 3-methylidene derivatives (R2 = H) and to 3,6-diylidene-PDOs 33 (R1 represents NCHR) (Scheme 7).12–15 Leaving groups for elimination to 3-ylidene-PDOs 33 and to corresponding 3,6-diylidene-PDOs can also be situated at the ring position (e.g.reactants 34).15 Such 3-alkyl-3-hydroxy-PDOs 34 (X = OH) acted as intermediates in the transformation of a-(aAketoacylamino) amides 35 into 3-methylidene-PDOs 33 (R2 = H).16–18 Finally the debromination and dehydrogenation of 3-alkyl-PDOs 36 (X = Br or H respectively) are to be mentioned as entries to 3-alkylidene-PDOs 33 (see also Scheme 16). 19 Scheme 4 Scheme 5 14 + RNH – R4CONHR Scheme 3 3-(a-hydroxyalkyl)-PDOs 32 (X = OH) were isolated and dehydrated in a separate step. Similar 3-(a-hydroxyalkyl)-PDOs or derivatives 32 (X = OH OSO2Me OAc) or corresponding sulfur analogues 32 (X = 3 Isomerisation of 3-ylidenepiperazine-2,5-diones For steric reasons 3-ylidene-PDOs are usually more stable in the Z-configuration 37 (Scheme 8).Thus isomerisation of the Eisomers 38 to 37 was reported by treatment with acids.20 Such a transformation is also possible in the opposite direction i.e. from 37 to 38 by irradiation with light (R2 = aryl)4,8,10,21 or if R2 and R4 are large enough (e.g. R2 = i-Pr R4 = Me) by treatment with acids. Under acidic or basic conditions migration of the C–C double bond of 3-ylidene-PDOs 37 can occur affording racemic isomers 39.13,17,22 This rearrangement is likely to occur either via 2,5-dihydroxypyrazines 41 (if R3 = R4 = H) or via N-acyliminium structures when acids are present. The former intermediates could be synthesised by treating N,NAunsubstituted 3-ylidene-PDOs 40 with sodium hydroxide.21 Corresponding 2,5-dialkoxypyrazines 41 were obtained with trialkyloxonium salts.253 4 Hydrogenation of CNC and reduction of CNO Catalytic hydrogenation of optically active 3-ylidene-PDOs 42 (Scheme 9) in the presence of Pd or Pt catalysts and polar solvents occurs in a highly stereoselective syn-fashion from the opposite side with respect to the substituent R2 attached to the chiral ring position 6.23 cis-PDOs 43 were obtained usually in excellent yields (but for R1 = aryl) and with a wide scope of substituents. They could be cleaved by acid hydrolysis thus Chem. Soc. Rev. 1999 28 251–259 Scheme 7 Scheme 8 giving access to enantiomerically pure amino acids.Mechanistic investigations of the catalytic deuteration of 3-ylidene- PDOs 45 gave similar results (formation of 46).23 But remarkably in some cases of N,NA-diBoc-protected 45 (R2 = Boc) with R1 = Ph no syn-addition was observed but instead anti-addition was seen i.e. only a D-atom was added from the opposite side of the i-Pr group found at the ring position while the exocyclic C-atom of the C–C double bond was attacked from the same side. Racemic trans-3,6-dialkyl-PDOs 44 were obtained as major products if optically active 3-ylidene-PDOs 42 were heated with t-BuI. Probably HI was formed under the thermal conditions acting as hydrogenating reagent.24 Hydrogenation of one of the two C–C double bonds of 3,6-diylidene-PDOs 47 was achieved by catalytic hydrogenation with zinc under acidic conditions or using HI.The hydrogenation could be stopped at the stage of 3-ylidene-PDOs 48 somehow demonstrating the higher reactivity of diylidene- PDOs 47 as compared with the resulting monoylidene-PDOs 48. Naturally the latter could further be hydrogenated by the same reagents under more forcing conditions to racemic cis- 3,6-dialkyl-PDOs 49. LiAl(Ot-Bu)3H left the C–C double bond of the 3-benzylidene-PDOs 50 unchanged (Scheme 10) but reduced the conjugated carbonyl group affording hemiaminal structures 51 which were employed in synthetic approaches to ecteinascidines via benzazocines 52.9 5 Radical addition to the C–C double bond Although 3-ylidene-PDOs are composed of a didehydroamino acid moiety and thus should be prone to formation of radicals at the 3-position according to the concept of the capto-dative effect radical reactions seem not to be favoured by such systems.Addition of organomercurates to the 3-methylidene- Chem. Soc. Rev. 1999 28 251–259 254 Scheme 9 PDOs 53 gave low yields of 3-alkyl-PDOs 54 in the presence of sodium borohydride (Scheme 11).12 The approach of the hydrogen to the 3-position occurred stereoselectively from the phase opposite to the alkyl substituent R at position 6. 6 Nucleophilic addition to the exocyclic position of the C–C double bond In general Michael acceptor properties of 3-ylidene-PDOs seem to be weak i.e. addition of nucleophiles to the exocyclic Scheme 10 Scheme 11 position of the C–C double bond is not preferred.Thus proline derived 74 turned out to be reluctant to react with various nucleophiles such as alkoxides mercaptanes amines or CHacidic compounds. But reactivity of 3-ylidene-PDOs with nucleophiles is increased in methylidene-PDOs such as 55 or in 3,6-diylidene-PDOs (e.g. 57 59 or 64). Thus thioacetic acid could be added to the methylidene-PDOs 55 and sodium methoxide to the dibenzylidene-PDO 57 affording monoadducts 5617 20 and 58 (Scheme 12).15 Unlike the latter product the analogous monoadducts 60 and 62 derived from the 3,6-dimethylidene-PDO 59 and thiols or aziridine still possess a reactive 3-methylidene-PDO moiety. Thus further nucleophilic addition was possible under more forcing conditions affording functionalised 3,6-dialkyl-PDOs 61 and 63 respectively.14 Grignard addition to 3,6-diarylidene-PDOs 64 afforded monoadducts 65 even with an excess of Grignard reagent or corresponding O-methylated products 66 after quenching with dimethyl sulfate (Scheme 13).25 7 Acid catalysed addition of nucleophiles to the ring position of the C–C double bond If thioacetic acid or hydrosulfide were reacted with 3-methylidene-PDOs in the presence of an acid or Lewis acid the nucleophilic attack did not occur at the exocyclic position of the Scheme 12 Scheme 13 C–C double bond (see previous chapter) but at the ring carbon due to formation of intermediate N-acyliminium salts 10 (E = H).16,17 20 3-Mercapto-PDOs 68 were formed (Scheme 14) which are particularly interesting if a second mercapto function is found at position 6 (R2 = SH) i.e.if either R2 = SH in the starting material 67 or a suitable leaving group (R2 = Cl OH OMe SAc) is substituted by SH. Such 3,6-dimercapto-PDOs 68 (R2 = SH) allowed an oxidative ring closure (I2–pyridine) to the epidithiapiperazinedione moiety (69 70) a common class of fungal metabolites with interesting pharmacological properties. 16 An alternative route to 70 is possible from 3,6-dimethylidene-PDOs 72 by two-fold addition of hydrosulfide in the presence of ZnCl2 via dithiols 73. Similarly ethanethiol was added twice to 3,6-diylidene-PDOs.20 Bridged dithioacetals 71 255 Chem. Soc. Rev. 1999 28 251–259 Scheme 14 useful for the total synthesis of Sporidesmin A a compound causing facial eczema a serious disease of sheep could be obtained by BF3-catalysed addition of hydrosulfide in the presence of anisaldehyde or with a 5-anisyl-1,2,3,4-oxatrithiane to dimethylidene-PDO 72 or 5-acetylthio-3-methylidene-PDO 67 (R2 = SAc) respectively.26 Some natural products such as gliotoxin comprising the disulfide moiety 70 exhibit very high antifungal activity.In the presence of catalytic amounts of acids it is also possible to add C-nucleophiles to the endocyclic position of the C–C double bond of 3-ylidene-PDOs. Thus treatment of 74 with 1,1-diphenylethene gave optically active 3-alkyl-3-diphenylvinyl-PDOs 75 mostly accompanied by racemic cycloadducts 76 and 77 (Scheme 15).27 While 75 are derived from intermediate N-acyliminium salts 10 (E = H) again the formation of cycloadducts 76 and 77 requires alternative enolised Nacyliminium intermediates 78 which give a Diels–Alder Scheme 15 Chem.Soc. Rev. 1999 28 251–259 256 reaction with diphenylethene. p-Excess N-heterocycles such as pyrrole or indole can also act as C-nucleophiles in acid catalysed addition to 3-ylidene-PDOs 74. Enantiopure PDOs 79 were obtained consisting of proline and an a-quaternary amino acid.22 Both PDOs 75 and 79 represent promising candidates for hydrolytic cleavage forming novel optically active a-amino acids. Introduction of indole to the 3-position of 3-ylidene- PDOs could also be achieved in an intramolecular fashion giving rise to interesting bridged PDOs 81 which were obtained in synthetic approaches towards naturally occurring phytotoxines.28 8 Addition of electrophiles to the C–C double bond and oxidation The enamine character of 3-ylidene-PDOs enables the addition of electrophilic halogen atoms or phenylselanyl to the exocyclic position (formation of 10 with E = halogen or PhSe) followed by addition of a nucleophile e.g. halide hydroxide or alkoxide to the ring position of the starting C–C double bond. Thus bromine or chlorine gave two-fold addition to both C–C double bonds of 3,6-diylidene-PDOs 82 affording tetrahalo-PDOs 83 (Scheme 16).15 29 The ring halogen atoms of 83 could be further substituted by alcohols or water to corresponding 3,6-dialkoxy or 3,6-dihydroxy-PDOs such as 86 (Scheme 17). Interaction of sodium iodide or thiols with 83 did not cause substitution but elimination to the starting 3,6-diylidene-PDOs 82.29 PDOs with alkoxyhalo or hydroxyhalo structures 86 were also directly obtained by reaction of 3,6-diylidene-PDOs 84 with Nhalosuccinimides and alcohols or water respectively.15 Just one C–C double bond was affected (formation of monoaddition products 85) if equimolar quantities of the reagents were used.15 The halogen atoms of haloalkylalkoxy-PDOs 8515 8615 and others28 could reductively be removed with H2/Pd affording alkylalkoxy-PDOs such as 88 and 89.This reaction was also used in the synthesis of thaxtomines which are herbicide toxins produced by Streptomyces and Actinomyces species.28 A haloalkylhydroxy-PDO 85 (R1 = H) served as precursor for the Scheme 16 oxirane 87 by intramolecular nucleophilic substitution.15 The bromine–alkoxide addition also happened with racemic monoarylidene-PDOs28 and the chiral 3-benzylidene-PDO 9027 allowing the synthesis of optically active addition products 91 in the latter case.The NBS–methoxide addition was further possible to the wfunctionalised 3-propylidene-PDOs 92 if the terminal OH group is protected (R = Ac tetrahydropyranyl; formation of 93). In the presence of a free hydroxy group intramolecular halohydroxyalkylation occurred in the presence of NBS or t-BuOCl (Scheme 18).30 The resulting spiro-PDOs 94 possess the main skeletons of the natural products bicyclomycin and aspirochlorin. In a similar manner the phenol and aniline derivative 95 gave the spirobenzofuran 96 (X = O) or spiroindole 96 (X = NAc) upon treatment with NBS.10 As has been shown in the transformation of the 6-hydroxybutenyl-3-methylidene-PDO 97 into the bicyclic product 98 an intramolecular attack of a hydroxy group after prior attack of bromine or phenylselenyl chloride at the exocyclic position of the double bond can also occur if the hydroxyalkyl group is found at position 6 of a 3-ylidene-PDO.The product 98 represents a substructure of Bicyclomycin.31 Reaction of the 3-ethylidene-PDO 99 with benzeneselenenyl chloride in the presence of AcOK–AcOH gave 3-hydroxy- 3-(1-phenylselanylethyl)-PDO 100 (Scheme 19).32 Similar additions of benzeneselenenyl choride and alcohols in pyridine were achieved in a stereoselective manner if chiral 3-alkylidene-PDOs 74 were used.27 Remarkably treatment of condensed 3-ylidene-PDOs 101 with NBS or Br2 in the presence of THF–water gave diols 102 together with corresponding monohydroxy products obviously because of hydrolysis of the expected bromohydrin.19 33 34 Diols 102 are found in naturally occurring Fumitramorgins.Scheme 17 cis-Hydroxylation with OsO4 was applied to 3-ylidene-PDOs affording diols 102-A34 or 103.11 This reaction was stereoselective and could be applied to the synthesis of fungal metabolites. Bisacetoxylation of the non-chiral 3-ylidene-PDO 104 with lead tetraacetate afforded the diacetoxyproduct 105 (X = OAc) (Scheme 20) which allowed further substitution of the endocyclic acetoxy group or of both acetoxy groups by nucleophiles such as H2O H2O2 or H2S.35 Epoxidation of 3-ylidene-PDOs 106 was achieved with mchloroperbenzoic acid or with dimethyldioxirane and could also Scheme 18 257 Chem.Soc. Rev. 1999 28 251–259 Scheme 19 Scheme 20 Scheme 21 be used to synthesise optically active oxiranes 107 (Scheme 21).27 36 In some cases diols 108 (R3 = H) were obtained instead due to easy hydrolysis.27 Corresponding alkoxysubstituted ring opening products 108 (R3 = alkyl) were available by treatment of oxiranes 107 with alcohols.27 Epoxidation of a 3-(o-hydroxybenzylidene)-PDO led to the spiro-PDO 109 via an intermediate oxirane.8 Reaction of 3-benzylidene-PDOs 110 with singlet oxygen disrupts the benzylidene group affording piperazinetriones 111 (Scheme 22).35 Whether this reaction runs via intermediate peroxides or iminium salts is not clear.Similar C–C bond cleavage to piperazinetriones was achieved by ozonolysis.13 9 Dipolar cycloadditions 3-Ylidene-PDOs do not seem to be favoured for 1,3-dipolar cycloadditions. The highly reactive diazomethane however Chem. Soc. Rev. 1999 28 251–259 258 Scheme 22 Scheme 23 could be reacted with the proline derived PDOs 112 affording pyrazolines 113 (Scheme 23). These products were further changed to optically active allo-coronamic acids 114.37 The cycloaddition was highly stereoselective but remarkably occurred with opposite phase selectivity as compared with the epoxidation (see Scheme 21). 10 References 1 P.G. Sammes Prog. Chem. Org. Nat. Prod. 1975 32 51. 2 H. Poisel and U. Schmidt Chem. Ber. 1973 106 3408. 3 S. D. Bull S. G. Davies and M. D. O’Shea J. Chem. Soc. Perkin Trans. 1 1998 3657. 4 C. Shin Heterocycles 1983 20 1407. 5 D. Person and M. le Corre Bull. Soc. Chim. Fr. 1989 673. 6 A. Lieberknecht and H. Griesser Tetrahedron Lett. 1987 28 4275. 7 C. Gallina and A. Liberatori Tetrahedron 1974 30 667. 8 K. Itoh M. Kasami R. Yamada T. Kubo M. Honda and A. Sera Heterocycles 1997 45 1345. 9 N. Saito K. Tashiro Y. Maru K. Yamaguchi and A. Kubo J. Chem. Soc. Perkin Trans. 1 1997 53. 10 Y. Sato Y. Nakajima and C. Shin Heterocycles 1992 33 589. 11 M. Yamaura T. Suzuki H. Hashimoto J. Yoshimura and C. Shin Bull. Chem. Soc. Jpn. 1985 58 2812.12 C. L. L. Chai and A. R. King Tetrahedron Lett. 1995 36 4295. 13 M. Bergmann and A. Miekeley Liebigs Ann. Chem. 1927 458 40. 14 K. H. Ongania Arch. Pharm. (Weinheim Ger.) 1979 312 963. 15 S. M. Marcuccio and J. A. Elix Aust. J. Chem. 1985 38 1785. 16 J. D. M. Herscheid H. P. H. Scholten M. W. Tijhuis and H. C. J. Ottenheijm Recl. Trav. Chim. Pays-Bas 1981 100 73. 17 J. A. Marshall T. F. Schlaf and J. G. Csernansky Synth. Commun. 1975 5 237. 18 W.-R. Li and S.-Z. Peng Tetrahedron Lett. 1998 39 7373. 19 M. Nakagawa H. Fukushima T. Kawate M. Hongu T. Une S. I. Kodato M. Taniguchi and T. Hino Chem. Pharm. Bull. 1989 37 23. 20 P. J. Machin and P. G. Sammes J. Chem. Soc. Perkin Trans. 1 1974 698. 21 K. W. Blake and P. G. Sammes J. Chem. Soc.(C) 1970 980. 22 S. Jin and J. Liebscher Synlett 1999 459. 23 M. Oba S. Nakajima and K. Nishiyama Chem. Commun. 1996 16 1875. 24 S. Jin and J. Liebscher J. Prakt. Chem. 1998 340 390. 25 M. A. F. Elkaschef K. E. Mokhtar F. M. E. Abdel-Megeid and S. A. A. Khallaf J. Chem. Soc. (C) 1969 622. 26 Y. Kishi S. Nakatsuka T. Fukuyama and M. Havel J. Am. Chem. Soc. 1973 95 6493. 27 A. Bartels S. Jin and J. Liebscher unpublished results see also A. Bartels Dissertation Humboldt-University Berlin 1997 and S. Jin Dissertation Humboldt-University Berlin 1999 under preparation. 28 J. Moyroud J. Gelin A. Chene and J. Mortier Tetrahedron 1996 52 8525. 29 J. Yoshimura Y. Sugiyama and H. Nakamura Bull. Chem. Soc. Jpn. 1973 46 2850. 30 C. Shin Y. Sato S. Honda and J. Yoshimura Bull. Chem. Soc. Jpn. 1983 56 2652. 31 T. Fukuyama B. D. Robins and R. A. Sachleben Tetrahedron Lett. 1981 22 4155. 32 Y. S. Oh and H. Kohn J. Org. Chem. 1992 57 3662. 33 S. I. Nakatsuka K. Teranishi and T. Goto Tetrahedron Lett. 1986 27 6361. 34 S. I. Kodato M. Nakagawa M. Hongu T. Kawate and T. Hino Tetrahedron 1988 44 359. 35 P. J. Machin and P.G. Sammes J. Chem. Soc. Perkin Trans. 1 1976 628. 36 A. Bartels P. G. Jones and J. Liebscher Tetrahedron Lett. 1995 36 3673. 37 C. Alcaraz M. D. Fernandez M. P. deFrutos J. L. Marco and M. Bernabe Tetrahedron 1994 50 12443. Review 8/04743J 259 Chem. Soc. Rev. 1999 28 251–259

ISSN:0306-0012    

DOI:10.1039/a804743j

出版商:RSC    

年代:1999

数据来源: RSC