Practical synthesis of Schollkopf's bis-lactim ether chiral auxiliary: (3S)-3,6-dihydro-2,5-dimethoxy-3-isopropyl-pyrazine

Article abstract of DOI:10.1016/S0957-4166(97)00623-X

Practical methodology for the bis-O-methylation of (3S)-isopropyl-piperazine-2,5-dione 4 on a 45 g scale to generate Schollkopf's bis-lactim ether chiral auxiliary (3S)-3,6-dihydro-2,5-dimethoxy-3-isopropyl-pyrazine 1 has been developed. Monomethylated intermediates 5 and 6 are reported for the first time. The gelling effects of 4 in a range of common solvents are also described.

Full text of DOI:10.1016/S0957-4166(97)00623-X

Pergamon
Tetrahedron: Asymmetry 9 (1998) 321–327
Practical synthesis of Schöllkopf’s bis-lactim ether chiral
auxiliary: (3S)-3,6-dihydro-2,5-dimethoxy-3-isopropyl-pyrazine
Steven D. Bull,a Stephen G. Davies a,∗ and William O. Mossb
aDyson Perrins Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QY, UK
bZeneca Pharmaceuticals, Hurdsfield Industrial Estate, Macclesfield, Cheshire SK10 2NA, UK
Received 18 November 1997; accepted 5 December 1997
Abstract
Practical methodology for the bis-O-methylation of (3S)-isopropyl-piperazine-2,5-dione 4 on a 45 g scale to
generate Schöllkopf’s bis-lactim ether chiral auxiliary (3S)-3,6-dihydro-2,5-dimethoxy-3-isopropyl-pyrazine 1 has
been developed. Monomethylated intermediates 5 and 6 are reported for the first time. The gelling effects of 4 in a
range of common solvents are also described. © 1998 Elsevier Science Ltd. All rights reserved.
1. Introduction
A large number of methods have been developed for the asymmetric synthesis of non-proteinogenic
α-amino acids.1 Many of these approaches are based on stoicheiometric chiral auxiliaries where dia-
stereoselective alkylation of a masked glycine enolate is controlled by an attached homochiral fragment.
Schöllkopf’s bis-lactim ether methodology was first reported in 19812 and this class of auxiliary 1 has
proven to be particularly successful for the synthesis of small quantities of novel homochiral α-amino
acids.3 We have found that the greatest problem in using this methodology for synthesis is the difficulty
in preparing sufficient quantities of the parent auxiliary 1.4 We now report on a thorough investigation
that enables 1 to be reliably prepared on a 45 g scale.
The original report of the preparation of 1 lacks experimental detail.2 Although the synthetic protocol
appears straightforward, it is often problematical and we have encountered difficulty in obtaining
reproducible yields whilst following this procedure.5 The synthesis of 1 is conveniently divided into
the preparation of parent piperazine-2,5-dione 4 (diketopiperazine, DKP) followed by bis-O-methylation
to afford the bis-lactim ether 1 (Scheme 1).
∗
Corresponding author.
0957-4166/98/$19.00 © 1998 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(97)00623-X

322
S. D. Bull et al. / Tetrahedron: Asymmetry 9 (1998) 321–327
Scheme 1. (i) COCl2; (ii) glycine methyl ester·HCl, 2.1 eq. triethylamine, THF/CHCl3, −78°C; (iii) toluene, ∆; (iv) Me3O·BF4,
CH2Cl2
2. Results and discussion
A suspension of homochiral L-valine was stirred with a 1.97 M solution of phosgene in toluene and
THF at 50°C to give the desired N-carboxy-Leuch’s anhydride 2, as a white crystalline solid in 90–95%
yield. Careful control of dilution, reaction temperature and time were necessary for good yields of 2 to
be obtained otherwise significant quantities of polymeric material began to accumulate.
Formation of dipeptide H2N-L-Val-Gly-OMe 3 was best carried out at relatively high concentration
and at low temperature in order to minimise competing polymerisation. Two equivalents of triethylamine
were added to a solution of glycine methyl ester hydrochloride in chloroform at −78°C followed by the
addition of a solution of 2 in THF. It was essential that the reaction was stirred efficiently during this
addition since triethylamine hydrochloride forms a gelatinous precipitate which can result in inefficient
mixing of reagents and decreased yields of 3. After stirring for 4 hours at −78°C the reaction mixture was
stored overnight at −20°C before removal of the triethylamine hydrochloride by filtration and evaporation
of the solvent to afford the thermally unstable dipeptide 3 as a crude oil. Subsequent cyclisation to DKP
4 was achieved by heating crude 3 in toluene at reflux for 24 hours. DKP 4 was purified by refluxing with
charcoal in boiling water and after evaporation of solvent the resulting powder was dried for 24 hours at
90°C under vacuum to give an overall yield of 70–80% from L-valine.
Our initial attempts to methylate DKP 4 with Me3O·BF4 were unsuccessful affording 1 in poor, non-
reproducible yields (<20%). A detailed investigation into this reaction revealed a number of experimental
conditions which must be followed if good yields of 1 are to be obtained:
(1) DKP 4 must be totally free of solvent. Any adventitious water results in the premature cleavage
of the bis-lactim ether bonds of 1 to afford the methyl esters of valine and glycine. Severe conditions
are required to obtain solvent-free DKP 4 as a result of both its hygroscopic nature and its remarkable
solvent gelling properties. Our preliminary studies show that as little as 1 mg of DKP 4 can completely
gel 1 ml of a wide range of commonly available solvents (acetone, acetonitrile, benzene, chloroform,
dichloromethane, dioxane, ethanol, ethyl acetate, nitromethane, tetrahydrofuran and toluene). Many of
these gels are stable up to the boiling points of the pure solvents and simple calculations reveal that
this gelation process requires a single molecule of DKP 4 to be associated with at least 3000 molecules
of solvent.6 Solvents in which DKP 4 does not gel include butan-2-one, chlorobenzene, cyclohexane,
dimethoxyethane, dimethylformamide, di-n-butyl ether, diethyl ether, pentane and water.
(2) Good yields of 1 were only obtained when freshly prepared Me3O·BF4 was used. Commercial
samples were of inferior quality and gave poor yields.7 Fresh Me3O·BF4 was prepared by modification
of the Organic Synthesis procedure which involved the dropwise addition of epichlorohydrin to a solution
of Me2O and BF3·Et2O in dichloromethane at low temperature.8 This original procedure was highly
wasteful of the gaseous component Me2O and a protocol was therefore developed where sequential,
portionwise additions of epichlorohydrin and BF3·Et2O ensure that all of the costly Me2O is consumed.
This modification enables kilogrammes of this valuable oxophilic alkylating agent to be prepared
economically in excellent yield.9
(3) The sampling of aliquots from the reaction mixture revealed that initial methylation of DKP 4

S. D. Bull et al. / Tetrahedron: Asymmetry 9 (1998) 321–327
323
resulted in the formation of two monomethylated intermediates 5 and 6 in a ratio of 2:1 respectively.
These structural isomers could be prepared in good yield by treatment of DKP 4 with one equivalent
of Me3O·BF4 followed by chromatographic separation. Structural assignment of these isomers proved
difficult due to the similarity and simplicity of their 1H NMR spectra. Steric arguments suggested that the
C2 carbonyl of DKP 4 was hindered by the C3 isopropyl group and therefore less available for methylation
than the C5 carbonyl. This reasoning was confirmed by selective hydrolysis of the lactim bond of major
isomer 5 into the known dipeptide NH2-L-Val-Gly-OH 7.10 A 1H NMR NOE experiment was carried out
on the more polar minor isomer which revealed a 0.4% enhancement between the OMe resonance at δ
3.74 and the Pri methyl substituents at δ 0.82 and δ 1.08 thus confirming the structure of this compound
as 6.
It is clear that the second methylation of the diketopiperazine ring is relatively slow since treatment of
DKP 4 with one equivalent of Me3O·BF4 afforded 5 and 6 with only small quantities of 1 being isolated
(<2%). This observation reflects both the heterogeneous nature of the reaction and the inherent difficulty
in methylating 8 and 9. This methylation is slow because it requires the introduction of a second positive
charge into the ring system to give the highly reactive species 10. Since this second methylation step is
inefficient it is important that a small portion of the crude reaction mixture is always analysed for the
presence of monomethylated isomers 5 and 6 before workup. Additional Me3O·BF4 may then be added
as required.
(4) The reaction mixture is highly acidic and it is therefore essential that the biphasic reaction mixture
is added, as a slurry, to an aqueous bicarbonate solution (pH >7.5) during workup in order to minimise
acid catalysed hydrolysis of bis-iminium cation 10.
This information enabled us to devise the following experimental protocol which enabled the reliable
preparation of 1 on a 45 g scale. Anhydrous dichloromethane was added to a mixture of DKP 4 and
Me3O·BF4 under an atmosphere of nitrogen and the heterogeneous reaction mixture was stirred for 48
hours. The reaction was quenched by slowly pouring the mixture into a rapidly stirred saturated solution
of sodium bicarbonate which was maintained at pH >7.5 by portionwise additions of solid sodium
bicarbonate. Extraction of the aqueous layer with CH2Cl2, and concentration under vacuum, afforded
a crude oil which was essentially pure 1 in 70–80% yield.
In conclusion, we have reported on a series of modifications which enable Schöllkopf’s auxiliary 1 to

324
S. D. Bull et al. / Tetrahedron: Asymmetry 9 (1998) 321–327
be reliably prepared from valine on a 45 g scale in 60% yield. Bis-lactim auxiliaries which are based on
α-amino acids other than valine may also be prepared using this general procedure.11
3. Experimental
3.1. General methods
Melting points were measured on a Gallenkamp hot stage apparatus and are uncorrected. Optical
rotations were measured on a Perkin–Elmer 241 polarimeter. 1H NMR spectra were recorded on a Bruker
WH 300 and AM 500 and the chemical shifts referenced to CHCl3 (δ 7.27) in CDCl3. IR spectra were
recorded on a Perkin–Elmer 781 spectrophotometer. Mass spectra were recorded using a VG MASSLAB
VG 20-250 instrument while HRMS were determined on a VG Autospec. All solvents were purified and
dried according to the procedures described in Purification of Laboratory Chemicals, D. D. Perrin and
W. L. F. Armarego, Pergamon Press, Oxford, 1988. All chemicals used for synthetic procedures were of
reagent grade or better. Merck 70–320 mesh silica gel was used for chromatography.
3.2. L-Valine-N-carboxyanhydride [(4S)-4-isopropyl-oxazolidin-2,5-dione] 2
Phosgene (243 ml, 20% solution by weight in toluene, 0.47 mol) was added to a suspension of L-
valine (50.0 g, 0.43 mol) in THF (500 ml) and the mixture stirred at 50–60°C for 4 hours to afford a
homogeneous solution. The solution was cooled and purged of excess phosgene by bubbling N2 through
the reaction mixture and passing the exhaust gases through aqueous sodium hydroxide (2 M). The solvent
was removed in vacuum to afford a crude solid which was recrystallised from petrol:ether (2:1, 500 ml)
to afford 2 as a white crystalline solid (58.62 g, 96%). M.p. 70°C. [α]2D3 −44.2, (c 1.0 in THF) [lit.12
m.p. 62°C, [α]2D3 −42.9 (c 1.0 in THF)]. 1H NMR (CDCl3) δ: 1.03 (3H, d, J 6.8, CH(CH3)2), 1.09 (3H,
d, J 6.8, CH(CH3)2), 2.26 (1H, m, J 6.8 and 4.0, CH(CH3)2), 4.23 (1H, d, J 4.0, H4), 6.99 (1H, s br,
NH).
3.3. cyclo-L-Val-Gly [(3S)-isopropyl-piperazine-2,5-dione] 4
Triethylamine (87.55 g, 0.87 mol) was added to a mechanically stirred suspension of glycine methyl
ester hydrochloride (54.44 g, 0.43 mol) in chloroform (500 ml) under nitrogen and the reaction mixture
cooled to −78°C. A solution of 2 (58.62 g, 0.41 mol) in THF (300 ml) was added dropwise over a
period of 4 hours resulting in the precipitation of triethylamine hydrochloride as a gelatinous white
solid. After stirring for a further hour at −78°C, the reaction mixture was stored at −20°C for 12 hours.
The triethylamine hydrochloride precipitate was removed by filtration through Celite® and the filtrate
concentrated in vacuum at room temperature to afford a crude oil which was redissolved in THF (300
ml). Filtration through Celite® and evaporation of the solvent at room temperature afforded a colourless
unstable oil 3 which was dissolved in toluene (300 ml) and refluxed for 24 hours. Residual solvent was
removed by decantation and the remaining toluene removed under vacuum to afford a crude pink solid
which was dissolved in water (300 ml) and refluxed with decolourising charcoal for 1 hour. Charcoal was
removed by filtration through Celite®, and ethanol (300 ml) was added to the filtrate in order to facilitate
removal of the water in vacuum. The resulting crude solid was powdered and dried for 24 hours at 90°C
under high vacuum, to afford DKP 4 as a white powder (53.1 g, 79%). An analytically pure sample was
obtained by recrystallisation from ethanol:chloroform (3:1). M.p. 256°C. [α]2D3 +23.7 (c 1.0 in H2O),

S. D. Bull et al. / Tetrahedron: Asymmetry 9 (1998) 321–327
325
[lit.2 m.p. 254°C, [α]2D3 +20.2 (c 0.9 in H2O)]. 1H NMR (D2O) δ: 0.90 (3H, d, J 7.0, CH(CH3)2), 0.99
(3H, d, J 7.0, CH(CH3)2), 2.24 (1H, m, J 3.8 and 7.0, CH(CH3)2), 3.89 (1H, d, J 3.8, H3), 3.94 (1H, d,
J 19.2, H6), 4.12 (1H, dd, J 19.2 and 1.2, H6). 13C NMR (D2O) δ: 16.6 (CH(CH3)2), 18.8 (CH(CH3)2),
33.7 (CH(CH3)2), 44.6 (C6), 60.7 (C3), 169.7 (C_O), 171.1 (C_O).
3.4. Solvent gelling reactions
DKP 4 (20 mg) was dissolved in refluxing acetone:water (20:1, 2 ml) and cooled to room temperature.
A portion of this solution (0.1 ml) was then added to the appropriate solvent (1 ml) at 0°C and the resulting
solution allowed to stand for 5 minutes. Those solvents whose gels were stable at room temperature are
detailed above.
3.5. Trimethyloxonium tetrafluoroborate
Trimethyloxonium tetrafluoroborate was prepared by modification of the method of Curphey.8
Dimethyl ether (400 g) was condensed into a three neck flask containing dichloromethane (500 ml)
and epichlorohydrin (50 g) at −78°C under an atmosphere of nitrogen. The reaction vessel was fitted
with a dry-ice condenser and the mixture allowed to warm to −20°C. BF3·Et2O (50 g) was then added
dropwise with cooling and stirring. After addition was complete, more epichlorohydrin (50 g) was added
to the reaction vessel and the dropwise addition of another portion of BF3·Et2O (50 g) was repeated. This
alternating stepwise addition of epichlorohydrin and BF3·Et2O was repeated until no more dimethyl ether
was seen to be refluxing from the cold finger (6–7 additions). The reaction mixture was allowed to warm
to room temperature and the white solid filtered off under a stream of nitrogen, washed with chloroform
(500 ml), and dried under vacuum to afford Me3O·BF4 (590 g, 92% yield).
3.6. Formation of monomethylated lactim ether intermediates 5 and 6
DKP 4 (1 g, 6.10 mmol) and Me3O·BF4 (0.95 g, 6.44 mmol) were stirred together in dichloromethane
(20 ml) under an atmosphere of nitrogen for six hours. The reaction mixture was quenched into phosphate
buffer (50 ml) and the aqueous layer extracted with ethyl acetate (2×50 ml). The combined organic
1
layers were dried (MgSO4) and the solvent removed under vacuum to give a crude oil (627 mg). H
NMR analysis revealed a 2:1 mixture of the two monomethylated isomers 5 and 6 and the mixture was
separated by flash column chromatography (silica, ethyl acetate).
3.7. (6S)-6-Isopropyl-2-methoxy-5-oxo-3,4,5,6-tetrahydropyrazine 5
Rf 0.35 (EtOAc). M.p. 134°C (Dec). [α]2D3 +9.1 (c 1.0 in H2O). IR (KBr) 3246 (NH), 2969 (CH),
1
1705 (C_N), 1673 (C_O), 1633 (NH). H NMR (CD3OD) δ: 0.88 (3H, d, J 6.8, CH(CH3)2), 0.98
(3H, d, J 6.8, CH(CH3)2), 2.16 (1H, m, CH(CH3)2), 3.71 (3H, s, OMe), 3.90 (1H, m, H6), 4.02 (2H,
JAB 21, JAX 2.3, δAB 0.06, 2×H3). 13C NMR (CD3OD) δ: 16.9 (CH(CH3)2), 18.6 (CH(CH3)2), 34.1
(CH(CH3)2), 50.5 (C3), 53.7 (OMe), 59.4 (C6), 162.7 (C2), 171.8 (C5). HRMS for C8H15N2O2 (M+1)+,
calcd 171.1139, found 171.1134.

326
S. D. Bull et al. / Tetrahedron: Asymmetry 9 (1998) 321–327
3.8. (3S)-3-Isopropyl-2-methoxy-5-oxo-3,4,5,6-tetrahydropyrazine 6
Rf 0.25 (EtOAc). M.p. 138°C (Dec). [α]2D3 +8.5 (c 1.0 in H2O). IR (KBr) 3264 (NH), 2948 (CH),
1699 (C_N), 1680 (C_O), 1649 (NH). 1H NMR (CD3OD) δ: 0.82 (3H, d, J 6.9, CH(CH3)2), 1.08 (3H,
d, J 6.9, CH(CH3)2), 2.33 (1H, m, CH(CH3)2), 3.74 (3H, s, OMe), 3.88 (1H, m, H3), 3.92 (2H, JAB
1
17.4, JAX 2.5, δAB 0.03, 2×H6). H NMR NOE, 2×0.4% enhancement between δ 3.74 to δ 0.82 and
δ 3.74 to δ 1.08. 13C NMR (CD3OD) δ: 15.6 (CH(CH3)2), 17.4 (CH(CH3)2), 32.4 (CH(CH3)2), 48.8
(OMe), 52.1 (C6), 57.9 (C3), 161.8 (C2), 172.4 (C5). HRMS for C8H15N2O2 (M+1)+, calcd 171.1139,
found 171.1133.
3.9. H2N-L-Val-Gly-OH 7
Compound 5 was dissolved in 2 M HCl for 12 hours and the solvent removed under vacuum to afford,
after treatment with epoxypropane, 7 in quantitative yield. This compound was identical to material
obtained from treating dipeptide 3 with 2 M HCl. [α]2D3 +94.8 (c 2.0 in H2O), m.p 253°C [lit.10 m.p
258°C, [α]2D3 +101.5 (c 2.0 in H2O)]. This compares with literature data for NH2-Gly-L-Val-OH [lit.10
m.p. 248°C, [α]2D3 −20.2 (c 2.0 in H2O)].
3.10. (3S)-3,6-Dihydro-2,5-dimethoxy-3-isopropyl-pyrazine 1
Me3O·BF4 (135 g, 915 mmol) and DKP 4 (50 g, 320 mmol) were efficiently mixed together, as
solids, then covered with dichloromethane (500 ml) under an atmosphere of nitrogen. After 24 hours
the heterogeneous reaction mixture was stirred vigorously using a mechanical stirrer. After a further 48
1
hours a small portion of the lower phase was removed and H NMR spectral analysis carried out to
determine whether any 5 or 6 was remaining. An extra equivalent of Me3O·BF4 (35 g) was then added if
required and the reaction mixture stirred for a further 24 hours. The biphasic reaction mixture was then
poured slowly, with rapid stirring and cooling, into a saturated solution of sodium bicarbonate (750 ml)
[caution — effervescence], ensuring that the solution remained above pH 7.5 at all times by the addition
of solid bicarbonate. The biphasic solution was extracted with dichloromethane (2×200 ml), filtered
through Celite®, the organic layers dried (MgSO4), and concentrated in vacuum to afford 1 (45 g, 245
mmol) as a pale yellow oil. [α]2D3 +108.9 (c 1.0 in EtOH) [lit.2 [α]2D3 +106.3 (c 1.0 in EtOH)]; 1H NMR
(CDCl3) δ: 0.77 (3H, d, J 6.9, CH(CH3)2), 1.04 (3H, d, J 6.9, CH(CH3)2), 2.23 (1H, m, CH(CH3)2),
3.69 (3H, s, OMe), 3.73 (3H, s, OMe), 3.90 (3H, m, H3 and 2×H6). 13C NMR (CDCl3) 16.8, 18.9
(2×CH(CH3)2), 32.3 (CH(CH3)2), 46.4 (C6), 52.3 (OCH3), 52.34 (OCH3), 60.8 (C3), 162.2, 164.7 (C2
and C5). MS: m/z 185 (M+1)+.
Acknowledgements
Under the LINK Asymmetric Synthesis Programme, financial support for this work is gratefully
acknowledged from the EPSRC, DTI and Zeneca Pharmaceuticals.
References
1. R. M. Williams, Synthesis of Optically Active α-Amino Acids, Pergamon Press, 1989; H. Heimgartner, Angew. Chem. Int.
Ed. Engl., 1991, 30, 238; R. O. Duthaler, Tetrahedron, 1994, 50, 1539; D. Seebach, A. R. Sting, M. Hoffmann, Angew.

S. D. Bull et al. / Tetrahedron: Asymmetry 9 (1998) 321–327
327
Chem. Int. Ed. Engl., 1996, 35, 2708; T. Wirth, Angew. Chem. Int. Ed. Engl., 1997, 36, 225; R. Chinchilla, L. R. Falvello,
N. Galindo, C. Nájera, Angew. Chem. Int. Ed. Engl., 1997, 36, 995.
2. C. Deng, U. Groth, U. Schöllkopf, Angew. Chem. Int. Ed. Engl., 1981, 21, 798; U. Schöllkopf, Pure Appl. Chem., 1983, 55,
1799; U. Schöllkopf, U. Busse, R. Lonsky, R. Hinrichs, Liebigs Ann. Chem., 1986, 2150.
3. Recent examples where 1 has been successfully employed for the asymmetric synthesis of complex non proteinogenic α-
amino acids include V. Ojea, C. Fernández, M. Ruiz, J. M. Quintela, Tetrahedron Lett., 1996, 32, 5801; A. J. Pearson, H.
Shin, J. Org. Chem., 1994, 59, 2314; R. Mueller, L. Revesz, Tetrahedron Lett., 1994, 35, 4091; M. Ohba, T. Mukaihira, T.
Fujii, Chem. Pharm. Bull., 1994, 42, 1784; J. Paladino, C. Guyard, C. Thurieau, J. Fauchère, Helv. Chim. Acta., 1993, 76,
2465; J. E. Baldwin, R. M. Adlington, M. B. Mitchell, J. Chem. Soc., Chem. Commun., 1993, 1332.
4. 1 is commercially available in limited quantities at high cost (£219 for 5 ml) from Fluka/Merck–Schuchardt.
5. J. E. Rose, P. D. Leeson, D. Gani, J. Chem. Soc., Perkin Trans. 1, 1995, 157.
6. These figures compare favourably with other gelling agents of low molecular weight; K. Hanabusa, Y. Matsumoto, T. Miki,
T. Koyama, H. Shirai, J. Chem. Soc., Chem. Commun., 1994, 1401; H. T. Stock, N. J. Turner, R. McCague, J. Chem. Soc.,
Chem. Commun., 1995, 2063.
7. Methylation of both amide functionalities on oxygen is highly desirable because bis-methylated auxiliaries exhibit simple
1H NMR spectra which enable subsequent alkylation diastereoselectivities to be easily determined.
8. T. J. Curphey, Organic Synthesis. Coll. Vol. VI, 1988, John Wiley, New York.
9. In situ methylation of DKP 4 using methyl trichloroacetimidate and BF3·Et2O afforded inferior yields of 1 although this
method is an excellent way of preparing bis-benzylated auxiliary 11; U. Groth, C. Schmeck, U. Schöllkopf, Liebigs. Ann.
Chem., 1993, 321.
10. R. Appel, G. Bäumer, W. Strüver, Chem. Ber., 1975, 108, 2680.
11. Penicillamine-based 12 is particularly noteworthy because of the ease of separation of the parent α-amino acids methyl
esters by extraction of the penicillamine ester trifluoroacetate salt, into dichloromethane enabling the desired α-amino acid
to be recovered from the aqueous layer; L. S. Richter, T. R. Gadek, Tetrahedron: Asymmetry, 1996, 6, 427.
12. T. Weihrauch, D. Leibfritz, Liebigs. Ann. Chem., 1985, 1917.