“. \ w 1 1W NIH l “WW 1 \ fig tWIHMHUUUIHIIH‘JWl If? 1.183.937 . mchlgansmto [1:332:31 V' —r This is to certify that the thesis entitled INVESTIGATION INTO THE SYNTHESIS AND REACTIVI TY OF 2 , 2-DIORGANO-2-SILA- 1 , 3-DIOXANE-4 , 6-DIONES presented by James Alan Rabe has been accepted towards fulfillment of the requirements for Mas te rs degree in Chemi 3 try , Major professor Date July 20, 1982 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution MSU LIBRARIES m V RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. INVESTIGATION INTO THE SYNTHESIS AND REACTIVITY OF 2,2-DIORGANO-2-SILA-l,3-DIOXANE-4,6-DIONES BY James Alan Rabe A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemistry 1982 ABSTRACT INVESTIGATION INTO THE SYNTHESIS AND REACTIVITY OF 2,2-DIORGANO-2-SILA-l,3-DIOXANE-4,6-DIONES by James Alan Rabe This investigation sought to demonstrate the synthesis of 2,2-diorgano-2-sila-l,3-dioxane-4,6-diones, compounds expected to improve the yield of a recently described mild procedure for converting acyl chlorides to methyl ketones. The silylation of malonic acid with either dimethylbis(iso- propenyloxy)silane or diphenylbis(isopropenyloxy)silane gave solutions whose proton and 13C NMR spectra were con- sistent with the desired compounds. No pure 2,2-dimethyl- 2-sila-l,3-dioxane-4,5-dione or derivative of it could be isolated from reactions employing the former silane, and mass spectra of all solids (isolated by simple stripping, sublimation, recrystallization, or preparative GLC) showed high molecular weight species. Therefore, the possibility that that product was actually a cyclic oligomer could not be discounted. However, the mass spectrum of a solid pre- cipitate from a reaction using the latter silane was con- sistent with 2,2-diphenyl-2-sila-l,3-dioxane-4,6-dione. Crude products from both reactions were tried in the acyl chloride conversion procedure under various conditions without success. ACKNOWLEDGEMENTS The author expresses his appreciation to Dr. Michael Rathke for suggesting the topic of this study and for his guidance during its investigation. He also thanks the members of Dr. Rathke's research group for their helpful discussions and encouragement, and Dow Corning Corp. for providing financial assistance. ii TABLE OF CONTENTS List of Tables List of Figures List of Symbols I. II. III. IV. Introduction Background Results A. Reactions of Malonic Acid with Dimethyl- bis(iSOpropenyloxy)silane B. Reaction of Malonic Acid with Diphenyl- bis(isopropenyloxy)silane C. Acylations of Diorganosilylated Malonic Acid Discussion A. Reactions of Malonic Acid with Dimethyl— bis(isopropenyloxy)silane B. Reaction of Malonic Acid with Diphenyl- bis(isopropenyloxy)silane C. Acylation Reactions Experimental A. Materials B. Preparation of Dimethylbis(isopr0penyl- oxy)silane C. Preparation of Diphenylbis(isopropenyl— oxy)silane iii vi vii viii 14 17 20 20 28 30 37 37 38 38 Reactions of Malonic Acid with Di- methylbis(isopropenyloxy)silane 1. General Procedure 2. Concentration Variation 3. Cold Reaction Malonic Acid - Dimethylbis(isopropenyl- oxy)silane Reaction Product Purification Attempts and Reactions 1. Sublimation 2. Recrystallization 3. Preparative GLC 4. Precipitation from Carbon Tetra- chloride 5. Reaction with Bis(trimethylsilyl)- acetamide 6. Reaction with Triethylamine a. Derivative Formation b. Gas Generation Measurement 7. Hydrolysis Reactions of Malonic Acid With Di- phenylbis(isopropenyloxy)silane 1. Room Temperature Reactions 2. Reactions at Reflux Reactions with Malonic Acid - Diphenyl- bis(isopropenyloxy)silane Reaction Product 1. Reaction with Triethylamine 2. Hydrolysis Acylations 1. Of the Malonic Acid - Dimethylbis- (iSOpropenyloxy)silane Reaction Product iv 39 39 39 4D 40 40 41 41 42 42 42 42 43 43 44 44 44 45 45 45 46 46 2. Of the Malonic Acid - Diphenylbis- (isoprOpenyloxy)silane Reaction Product 3. Acylation in Tetrahydrofuran I. Reaction of Benzaldehyde with the Malonic Acid - Diphenylbis(isopropenyl- oxy)silane Reaction Product J. Analyses Bibliography 46 47 48 48 49 1. LIST OF TABLES Summary of Acylation Experiments vi 18 LIST OF FIGURES Effect of Triethylamine Addition on the Malonic Acid - Dimethylbis(isopr0penyl- oxy)silane Reaction Product Effect of Triethylamine Addition on the Malonic Acid - Diphenylbis(isopropenyl- oxy)silane Reaction Product vii 12 16 LIST OF SYMBOLS I. - 2,2-Dimethy1-l,3-dioxane—4,6-dione (Meldrum's Acid) 9 0 II - X P3739 R O L 0 O HO 0 III - )< R o O Q\\ l/gi VI - ‘/§i O \\R 0 Me \ ,/ V _ SI \\Me 0 O o O \ ./ VI - SI 0/ \ viii VII - R'C Sl PyH 0/ \R L 0 _l O 0 VIII - R'CCHZCOH CH . // 2 IX " M3251 OC\ 2 CH3 \ [CH2 X - /Si OC\ 2 o XI - Stripped Reaction Product of Malonic Acid and Dimethylbis(iSOpropenyloxy)silane XII - Precipitated Reaction Product of Malonic Acid and Diphenylbis(isoprOpenyloxy)silane O 0 II it XIII - Me SiO CCH CO 0 Me \\~.’/’ XIV - //§1\\ 0 Me O H C O R 3 \ / XV _ /SJ. 0 \R H3C O o o o 0 II II N I XVI - RCOCCHZCO-Si-Cl U .. \; EX 0 I . INTRODUCTION A recently described technique for converting acyl chlorides to methyl ketones1 via reaction with Meldrum's acid (2,2-dimethy1-l,3—dioxane-4,6-dione) followed by hy- drolysis and decarboxylation, avoids using organometallic reagents which are necessary in most other such methods, but sometimes gives poor yields, apparently because of hy- drolysis at the acyl carbon. Replacing Meldrum's acid with a 2,2-diorgano-2-sila-l,3-dioxane-4,6-dione, in which a silicon atom is substituted for the ketal carbon of the former compound, should permit a more selective hydrolysis and improve the yield of this method. However, no such silicon-substituted Meldrum's acid analogues have been re- ported. Therefore, this effort was undertaken to demon- strate a method for their preparation. When indications of a successful synthesis were obtained but difficulties were encountered in isolating pure samples, some experi- ments were run to determine whether the crude products could be used in the acyl chloride conversion reaction. I I . BACKGROUND Many methods have been discovered for synthesizing ketones in which an alkyl group replaces the halogen of an acyl halide. In most of these techniques, the alkyl group is introduced by an organometallic reagent. The classical reagents have contained cadmium, zinc, lithium, or mag- nesium. Because these compounds, particularly the latter two classes, can react with the product ketone to form tertiary alcohol, they often give low yields.2 In the last few years, a variety of other organometallic compounds have been shown to transform acyl halides into ketones without attacking the product.1 Through careful choice of organo- metallic reagent and experimental conditions, a variety of acyl chloride substrates may be converted successfully to ketones. However, few single reagents are mild and speci- fic enough for general use. Most are also expensive and they often require careful handling because of their re- activity. Recently, an exceptionally mild, general reaction for preparing B-keto esters from inexpensive reagents has been disclosed.3 This method takes advantage of the high acidity (pKa = 4.97) of the methylene protons in Meldrum's acid, 3 2,2-dimethyl-l,3-dioxane-4,6-dione (I), to effect acylation even in the absence of a strong base: 0 o " p ri din 0 X o RCCl + —X—————§ PyH o o ‘T““" R 2 eq) L. O n (I) (II) 0 (II) HCl HO 0 __._.___L X R'OH ‘(aq) §==:::é R 0 A 0 (III) 0 o o RCCHZCOR' + CH3CCH3 + coz Only a weak base, for instance pyridine, is required to serve as an acid acceptor and to drive the initial reaction to completion by forming an enolate anion salt. The method provides good yields (70-80%) of B-keto ester with a variety of acyl chloride substrates. A modification of this procedure by Base and Salonen1 has extended its use to the preparation of methyl ketones. In place of the final alcoholysis step, they use acid hydrolysis to pro- duce a B-keto acid which readily decarboxylates to the ketone: 0 O O H ZO/HOAC\ n n RCCHzCOH TT’ RCCH3 (III) ‘reflux Yields are only moderate, however. Especially low yields are obtained when the acyl chloride contains unsaturation conjugated with its carbon-oxygen double bond, e.g. cin- namoyl or benzoyl chlorides. In these cases, an acylated Meldrum's acid derivative forms, but apparently hydro- lysis conditions stringent enough to destroy the ring also cleave the bond between the acyl group and ring to give back the acid analogue of the starting acyl chloride. Improved ketone yields should result if the Meldrum's acid ring could be hydrolyzed under less vigorous con- ditions. One way to make it more susceptible to hydro- lysis would be to replace the ketal carbon with a silicon atom. This would create silyl ester linkages, which hydrolyze swiftly under mildly acidic or even neutral conditions without heating.4 Thus the reaction of an acyl chloride with a 2,2-diorgano-2-sila—l,3-dioxane-4,6-dione (IV) followed by gentle hydrolysis, then warming: F W 9 O O R R " \ ./ . . O o\ ./ o R'CCl + SI Pyridine R'C SI PyH o/ \R " 0/ \R O 0 IV VII V R=Methyl VI R=Phenyl o 0 H20 ,. ,. . VII :—_:_: R'ccnzcon +co2 +R281(OH)2 H+ VIII A ° 0! VIII ———\ R'CCH3 +c02 ‘___ should provide an exceptionally mild, non-hazardous and relatively inexpensive method for converting the acyl chloride to a methyl ketone. In addition, keeping the mixture cold might step the reaction after the first de- carboxylation. This would give a new path to B-keto acids, another class of compounds for which most existing syn- thesis require organometallic reagents. The procedure requires the ready availability of IV, but since no such compounds have previously been reported, the first step in demonstrating the method's feasibility must be the development of a practical synthesis for them. The standard way to prepare silyl esters is to react a carboxylic acid or its salt with a silylating agent - a silane with a hydrolyzable substituent: o 0 II I! noon + R' 351x ——> RCOSiR' 3 + HX v“— Although the silylating agents employed have almost always been monofunctional, this reaction might be adapted to the desired synthesis by condensing malonic acid with a di- functional silylating agent. 6 No such reaction with malonic acid has been reported. A few similar reactions involving difunctional silylating agents and dibasic acids have been performed, but the goal has usually been to produce a linear polymer. For example, Henglein, et. al.5 reacted dimethyldichlorosilane and diphenyldichlorosilane with terephthalic, fumaric, and sebacic acids in the presence of pyridine to produce ma- terials from which fibers could be drawn, but little structural information was obtained on the products. Radosavljevic, et. a1.6 condensed adipic.acid with various amounts of dimethyldichlorosilane and determined by com- positional and endgroup analysis that the products were linear polymers with degrees of polymerization between about four and ten. Carraher7 has reacted the sodium salts of adipic and succinic acids with diphenyldichlorosilane through an interfacial technique to give linear polymers with an average degree of polymerization of about five. However, others have shown that compounds capable of form- ing five- or six-membered rings upon reaction with a di- functional silylating reagent will in fact do so under appropriate conditions. Schott and Henneberg8 have shown that oxalic acid condenses with diethyldichlorosilane to give a product which is a cyclic monomer in camphor. Also, Baburina and Lebedev9 obtained a cyclic compound by reacting o-hydroxybenzoic acid with hexamethyltrisilthiane: ‘\ {’S\../' 9 SI Si COH O /l |\+ ) i- s S OH \ \Sf/ \ / although in this case only one of the functional groups with which the silylating agents reacts is carboxyl. Reaction of malonic acid with a difunctional silylating agent might also yield a cyclic product. The silylating agent should react readily with car- boxylic acids, be relatively inexpensive and commercially available or easily prepared, and release innocuous by- products. In particular, the byproducts should not be basic, or so acidic that basic acid acceptors are necessary during the synthesis. The reason for this is that Mamer and Tjoa,lo while studying the reaction between malonic acid and a monofunctional silylating agent, bis(trimethylsilyl)- acetamide (BSA), found that bases catalyze the formation of tri-substituted product: 0 0 O OSiMe3 II II II I HOCCHZCOH + BSA -€>’ Me3SiOCC=COSiMe3 H Formation of the tris(trimethylsi1y1) product could be minimized by excluding bases, but even the acetamide by- product from BSA was basic enough to generate a significant amount of it. Thus, the leaving group should be as neutral _as possible. 8 A variety of monofunctional reagents have been em- ployed for silylating carboxylic acids. While amine com- pounds like hexamethyldisilazane and trimethylsilyldi- ethylamine, trimethylchlorosilane, and BSA have been most widely used,11 others such as hexamethylcyclotrisilthiane,9 ketene methyl trialkylsilylacetals,12 N,O-bis(trimethyl- 14 and silyl)sulfamate,l3 trimethyl(isopropenyloxy)si1ane, N,O-bis(trimethylsilyl)carbamate,15 have also shown utility in this application. Most of the leaving groups on these reagents are acidic or basic, and some would be difficult to use in difunctional silylating agents because they are difunctional themselves. However, the byproduct of the isopropenyloxy group, acetone, poses neither of these dif- ficulties, and in addition is volatile enough to be readily stripped from the product. A silylating agent containing two groups can be prepared easily and inexpensively from acetone and diorganodichlorosilane:16 O I’CHZ RZSiClz + 2 CH3CCH3 ________.&. R251 C\ 2 CH3 IX R=Methyl X R=Pheny1 Because dimethyldichlorosilane is the cheapest dichloro- silane, IX was chosen as the first silylating agent to try in the preparation of V. Later, the preparation of VI was attempted using X. I I I . RESULTS A. Reactions of Malonic Acid with Dimethylbis(isopro- penyloxy)silane When equimolar amounts of IX and malonic acid were combined at room temperature in methylene chloride, proton NMR spectrum peaks from IX diminished by 90% within thirty minutes, were nearly absent after one hour, and showed no change thereafter. Simultaneously, a singlet integrating to twelve protons at 62.1 (acetone), a six proton singlet at 60.55 (methyl-on-silicon), and a complex, approximately two proton peak at 63.4 (methylene) appeared. Also present were a small, broad acid proton absorbtion (about 0.1 pro- ton) near 611.5 and a tiny singlet at 60.3. The complex methylene peak was a singlet at 63.5, often with an incom- pletely resolved shoulder, and some tiny separate peaks just upfield. Stripping the solution one hour at 0° pro- duced a waxy white solid, XI, weighing about 105% of the amount expected for V. The solid was completely resoluble in methylene chloride. Except for almost complete absence of the acetone absorbtion at 62.1, the proton NMR spectrum was similar to that of the unstripped product. 10 The 13C NMR spectrum of a sample prepared in methylene chloride, stripped, then redissolved in deuterochloroform showed three major peaks in addition to the solvent triplet - a sharp singlet at 6167 (carbonyl), a slightly broadened, but single, peak at 645 (methylene), and another sharp singlet near 61.7 (methyl-on-silicon). Small peaks barely above noise level also appeared near 648, 632, 610, and 61. Reactions were also run in other solvents, and at various concentrations and low temperature in methylene chloride. Results in deuterochloroform and acetonitrile were similar to those obtained in methylene chloride. The reaction was slower in benzene (required four to five hours to reach equilibrium), diethyl ether (several hours to equilibrium), and tetrahydrofuran (at least four days to equilibrium), but gave a similar ultimate product as judged by 1H NMR comparison. In carbon tetrachloride, a slurry of solid remained in the flask throughout the reaction. Pro- ton NMR showed a slow disappearance of IX, but only the acetone peak at 62.1 replaced it since the product pre- cipitated as it formed. The choice of solvent had little effect on the distribution of methylene peaks or the size of the peak at 60.3 or the acid proton peak near 611.5, although the chemical shift of the latter varied somewhat. Concentration also failed to change the product signifi- cantly — experiments conducted at 3, 10, 27, and 50% in methylene chloride gave nearly identical Spectra. When a 'mixture in methylene chloride was held at -78°, no reaction 11 occurred even after one day, but as the mixture warmed in an NMR tube, a spectrum like those described above appeared. When an unstripped reaction mixture in methylene chloride was hydrolyzed with a large excess of water, the product decomposed into malonic acid and siloxanes. When an equimolar amount of triethylamine was added to XI redissolved in methylene chloride, the methylene proton peak disappeared completely from the NMR spectrum, and the methyl on silicon peak near 60.55 shifted upfield by about 0.05ppm. The peak at 60.3 became larger, and a broad low rise appeared near 66. Over a period of hours, this rise drifted downfield and broadened further until it became invisible. The peak at 60.3 grew at the eXpense of the original methyl on silicon peak until an equilibrium was reached with only about one-third of the methyl-on- silicon remaining in the original peak. After one day, a group of small unresolved methylene peaks formed between 63.0 and 63.4, while tiny single peaks appeared at 65.0 and 61.9. The combined area of these peaks was much smaller than the original methylene peak. The solution also evolved about one mole of gas for every three to four moles of reacted malonic acid originally present. Gas generation and dimethylsilyl conversion data are plotted in Figure 1. No such changes occurred when pyridine was added instead of triethylamine. Several procedures were used in an attempt to isolate 12 uosooum :oflpomom ocmawmwaoahcmm IOHQOmHVmwnahnuoEfio I baud UHCOHMS one no cowuflood mcflanhnumfluB mo pommmm .H ousmfim Amuse mzHe om oa . — d 4 q m seamnuflaH noommooonmmza mmaoa oaxo.k A oaxv : = M I 1 m O 0 ”RACE ooumuocmw wow I I ucoEcoufl>cm 3oz ou omuum>coo mzflm I 4 I. v ‘I 13 pure V from the malonic acid — IX reaction product mix- ture. Simple stripping gave a waxy white solid, XI, whose mass spectrum contained high m/e species, including peaks at 379, 367, 293, 219, and 207, among others, but only a small peak appeared at 145, and no signal at all showed at 160. Sublimation of XI gave a product whose mass Spectrum exhibited these peaks and others at m/e 441, 453, 515, 527, and 589. A preparation in carbon tetrachloride instead of methylene chloride on the assumption that V was insoluble in that solvent and crystallized on for- mation also gave a product with high m/e peaks in its mass spectrum. Attempted recrystallization from methylene chloride produced an insoluble white solid when the solu- tion was warmed slightly at high concentration. The major non-solvent peak in the GLC trace of a reaction mixture was isolated by preparative GLC, but the substance was not a single species when re-injected, and also exhibited the high m/e mass spectrum peaks seen in other samples. A GC-Mass. Spec. analysis of the reaction mixture separated species whose spectra matched those of dimethylcyclo- siloxanes and two other species whose highest m/e peaks, 293 and 367, matched peaks which appeared together in most previous mass spectra. Attempts to isolate derivatives of V from the reaction of XI with triethylamine, n-butyl lithium, or BSA all failed to produce a material which ex- hibited the expected NMR or mass spectrum. 14 B. Reaction of Malonic Acid with Diphenylbis(isopro- penyloxy)silane When 1 mmole malonic acid and l mmole diphenylbis- (isopropenyloxy)silane (X) were reacted for ten hours at room temperature in acetonitrile, proton NMR showed a single but broad methylene peak near 63.6, a phenyl ab- sorbtion between 67.1 and 67.7, and an acidic proton peak at 69.8 in a ratio of 2/10/0.4. When the reaction was conducted at reflux, a sharp methylene singlet at 63.7 with a few tiny adjacent peaks formed within five minutes, and the spectrum remained unchanged thereafter. The ratio of total phenyl protons to total methylene protons was again about 10/2, with the sharp methylene peak constituting 80-90% of the total methylene area. The acid proton peak was only 25% as large as the one produced after ten hours at room temperature. Reactions at room temperature initially produced a clear solution, but upon standing, a white solid, XII, fell out. The mass spectrum of this solid (m.p. l43-l45°) showed a molecular ion peak at 284 along with M+l and M+2 peaks as eXpected for VI, and no higher m/e peaks. The yield was only 29% after six days at room temperature, even if the solid was pure VI. However, chilling one of the refluxed solutions produced a solid whose mass spectrum matched that of the solid obtained at room temperature in 89% yield. Attempts to purify the product further by 15 sublimation formed a material whose mass spectrum exhibited high m/e peaks, and only a small amount of the original 284 molecular weight species remained. Purification by prep. GLC was impossible because only solvent peaks eluted from the column. XII was insoluble in chloroform, benzene, hexane, and diethyl ether. Traces dissolved in acetone, but the meth- ylene protons in the solution behaved like those of malonic acid when examined by proton NMR, i.e. they had the same chemical shift and moved upfield but did not disappear when triethylamine was added. The mass spectra of both the acetone-soluble portion and the material which did not dissolve matched that of a sample exposed to the laboratory atmosphere for a few minutes. All contained high molecular weight peaks and no residual peak at m/e 284. When tri— ethylamine was added to a slurry of XII in deuterochloro- form, rapid dissolution occurred. In this case, triethyl— amine completely removed the methylene proton peak from the NMR spectrum, leaving only the phenyl absorbtion at 67.1-7.7. As with XI, triethylamine caused gas evolution from this solution as shown in Figure 2. If the solid XII was essentially all VI, within one day 0.75 moles of gas formed for each mole of that species present. 16 uosooum cofiuomom ocmawmamxoamcom Ionmowwvmfloamconmwa I owed owconz may no coauwood ocflEmawnuoflua mo pommmm .m ousmflm Amumv mzHe om ca —l J _ A ll @ seamfloHcH Icemmuooufimu mmaos muoaxem.~ mm@ (EOIX) oarvaauas 3V9 saqow 17 C. Acylations of Diorganosilylated Malonic Acid Several acyl chlorides were reacted with X1 or XII at a variety of temperatures and reaction times. Most reactions were conducted in methylene chloride, but tetra- hydrofuran and acetonitrile were each used in one run. Both pyridine and triethylamine were tried as catalysts, and in several cases they were added after the acyl chlorides to minimize loss of reactant via the gas-forming reaction. In two cases, XII was used without prior stripping. In one experiment, benzaldehyde was used in- stead of an acyl chloride, as per Corey.17 The mixtures were quenched at 0° or -78°, usually with one or two equiv- alents of aqueous HCl, but in two runs water alone was used. After quenching, the methylene chloride layer was stripped one hour at 0°. The conditions for each trial are summarized in Table l. The mixtures using pyridine as catalyst quickly turned so dark and muddy that the interface between the methylene chloride and water layers could not be detected unless the reaction was run at low temperature and for only a short time. Mixtures containing triethylamine catalyst turned deep Opaque red, but lightened to clear yellow-orange after quenching. In proton NMR spectra of the residues left after stripping, the only recognizable products were silo- xanes and the acid analogue of the starting acyl chloride. All spectra except those for experiments 1, 2, and 7 in 18 .Uouoa mm unooxo Avov Hum 20 mucoaw>asuo w suaa oo um wonoaoau can Naommu ca yuan: H us can ncoauuwom .haao Honda sues conceded + .mwozuo cw ovfinoanu ahum anommn .OIN .moz cw moquoano ahom Houmo vouv< .muamam>wsvm one e +uouaHsua ma oasaoaHa I I oGG\ok+ znmno Homonmoov \; mH NHUNmo aH HHx euaaHuumaa I nH\o OH\m~I \/ «H zommu aH HHx auaaHuumap I OH\o oNH\mhI MH +Hma aH I I oo\o Hoooo~xmmuvm NH I OGHo OGHSHI HUCUNHHmonmu HH I oaHo oo\m~I _ 9, OH .mHI Ha eoaoamso I I m-\mnI zmum a I .23 3;? e w I oo\o~ oo\o znmmo HHx H .257 um 3:288 I I SHE? 9 > G Hum .am .am H nuHa emanamac I oo\o om\m~I zmum m .mHI um coauaaao I I nH\mhI a Hum .aa .am H nuHa euauamno oe\o~ oo\o m\mhI Hoououxmmuvm m Hum .aa .uu H auHa eunuaoso I nnH\o n\mkI Hoooomxmmuv H maHauauau uaoauHa emaaHuum I oa\o~ oo\o znmmo Hooommu Hx H muaoaaoo m none N can“ H Iowan meHHOHao Heme .mmmmmmmmw .qmm AuaHa\o.v maOHuHeaoo mucuafiuonxm coauuamo< mo humaasm .H manna 19 Table 1 contained a peak at 63.4-3.5 which might have been caused by methylene protons in a B-keto acid product. However, their areas indicated yields of less than 10% if they did represent product. No spectrum showed the acetyl proton peak of the methyl ketone product expected from B-keto acid decarboxylation. IV. DISCUSSION A. Reactions of Malonic Acid with Dimethylbis(isopro- penyloxysilane Malonic acid and dimethylbis(isopropenyloxy)silane (IX) react smoothly and nearly to completion as indicated by the appearance of two equivalents of acetone while the enol ether peaks of IX disappeared almost totally from the proton NMR spectrum. The product is primarily a species containing one kind of methyl-on-silicon proton, and one kind of methylene proton, which is responsible for the singlet at 63.5, in a ratio of approximately 6/2. The shoulder and other small peaks just upfield of the latter singlet, and the small peak at 60.3 show that minor amounts of other Species are formed as well. The size of the minor peaks varied, but in some cases they contributed 10% or less of the total area in the methylene region. The rate of reaction between malonic acid and IX generally varied inversely with the degree of solvation of the reactants. In diethyl ether or tetrahydrofuran, malonic acid dissolved fairly well but was highly solvated. The activation energy for nucleophilic attack on the silyl- ating agent was relatively high and the reaction proceeded 20 21 slowly. Rapid rates were achieved in solvents like methylene chloride or deuterochloroform, where malonic acid was solvated poorly but did dissolve to some extent. The behavior in carbon tetrachloride was an exception to the above pattern. In that case, even though there was a low degree of solvation the reaction proceeded slowly because malonic acid was nearly insoluble in carbon tetra- chloride and not readily available for reaction. Despite the variation in solute-solvent interactions and observed reaction rates, reaction conditions had little effect on the nature of the product. The products which are most likely to form are of four types - the desired produCt V, a cyclic polymer of V, a linear polymer, or a product in which the malonic acid portion of one of these products has been silylated three times as observed by Mamer and Tjoa,lo e.g.: o M Me\ /0 O\Si/e /Si\o 0/ \Me Me Me // o—Si—o Me Very little of the latter type of product could have formed because no vinyl proton peak appeared for it in the proton NMR spectrum. A small amount of linear polymer was pro- bably generated and may have contributed to the minor peaks. However, it cannot be the major product because the number average chain would have to be over one hundred atoms long 22 to leave so small a residue of acid and enol protons. The formation of molecules this large should cause a sig- nificant increase in the viscosity of the solution, but that does not occur. The evidence thus favors a cyclic species as the major component in XI. In order to produce the single carboxyl, methylene, and methyl-on-silicon peaks observed in 13C NMR, as well as the single principal methylene and methyl-on-silicon peaks observed in proton NMR, the major product could be a Single species, probably either V or a dimer of it, or it could be a mixture of various sized rings all so large that cor- responding features are indistinguishable by NMR. Of these possibilities, the chemical shift of the methylene protons and their response to base addition correspond best to the behavior expected for V. The methylene peak appears 0.2- 0.3ppm downfield of the ones for linear esters like bis- (trimethylsilyl)malonates (63.2) or diethylmalonate (63.25), and in nearly the same position as the one for Meldrum's acid (63.55). Also, the initial NMR spectrum changes upon addition of triethylamine are nearly identical to those which occur with Meldrum's acid. The methylene peak dis- appears as one proton is removed by the base, then ex- changes rapidly with the remaining vinylic proton to pro- duce a broad rise centered near 66-7. The methyl peak shifts upfield slightly in response to additional shielding from release of electrons by the methylene proton removal. In contrast, the methylene peaks of bis(trimethylsilyl)- 23 malonate or malonic acid itself do not disappear in the presence of triethylamine. Thus, NMR data suggest that as in Meldrum's acid, the methylene protons of the major component in XI are deshielded and exceptionally acidic because the rest of the molecule is pulled back in a small ring. However, the mass spectrometry data conflict with this conclusion. In mass spectrometry, molecules con- taining methyl groups on silicon usually show a peak at M-lS+ instead of a peak for the molecular ion itself be- cause of a facile methyl cleavage.18 The Spectra of XI or XI purified by sublimation or preparative GLC contained only a trace of the 145 peak expected for V. Of the heavier species which appeared, none matched the 160x-15 values expected for cylcic oligomers of V. GLC-MS showed some to be dimethylcyclosiloxanes, and showed that the peaks at 293 and 367 were the highest m/e peaks in two different components of the product. These values are fifteen less than the molecular weights of the species: [:(Me28i0)‘3_] '— (MeZSiO) 4 occnzc and '—- occnzc _, O O respectively. Many of the other high m/e peaks corres- pond in the same fashion to another member of the series 24 O O Y where x is from three to seven and y is from zero to two. These Species could have formed by a thermal decom- position analogous to that observed for other acyloxy- silanes, e.g. tetrapropionyloxysi1ane19 and methylphenyl- bis(acetoxy)silane,20 in which siloxane bonds form as an- hydrides are split out. Some of these compounds probably formed during purification processes, but substantial amounts of the lower molecular weight members also ap- peared in the Spectrum of XI itself. If XI was actually composed of these species, its proton NMR spectrum*would have shown multiple methylene and methyl-on-silicon peaks, and most of the methyl-on-silicon would have appeared at sou-0.2.21 Since redissolved x1 still showed essentially a single methyl on silicon peak near 60.55, the mass spec- trum analysis itself must have caused extensive decom- position. Because of this decomposition during analysis, the lack of peaks at m/e 145 or 160 in the mass spectrum of XI does not itself rule out the presence of V. However, only higher molecular weight oligomers could have formed the species observed. Originally the possibility was considered that the major product in dilute solution was V, but that it was in equilibrium with larger cyclic species, and upon concentration during stripping the 25 equilibrium shifted towards the larger rings. This would explain why high m/e peaks appeared in the mass spectrum of the stripped solid even though NMR data on solutions indicated the product to be V. However, such a shift should cause the appearance of the methylene region in the proton NMR spectrum to vary with concen- tration. Because nearly identical spectra were obtained over a concentration range of 3% to 50%, this type of shift is unlikely. Another process which would resolve the discrepancy between NMR and mass spectrum results is a rapid exchange of the acid groups on silicon to create an equilibrium between various ring sizes including a small amount of V. Addition of triethylamine could then shift the equilibrium to the monomer by selectively trapping it as its amine salt because of its acidic methylene protons. The other possibility is that the methylene protons of oligomers of V are themselves un- usually acidic. Because all attempts to isolate pure V caused de- composition and the various analyses did not clearly define the nature of the stripped solid XI attempts were made to prepare a derivative which could be isolated and char- acterized. However, none yielded the desired result. Reaction of XI with butyllithium gave a virtually intrac- table white powder. Proton NMR analysis of the small amount which did dissolve in dimethylsulfoxide showed only siloxanes. BSA was added to XI in methylene chloride 26 to produce a tri-silylated derivative: \\ :/,Me O\\Sf/,Me Me3Si XIV Although a similar reaction to give tris(trimethylsilyl)- malonate proceeds to a significant extent,10 this mixture failed to give any new GLC peak which could be identi- fied as XIV. To avoid decomposition of V, the reaction was run at room temperature (8 hours) without addition of base catalyst. These conditions might have been too mild, since the reported reaction with malonic acid was con- ducted at 60° and was accelerated by base addition. Al- ternatively, XIV may have decomposed when injected into the GLC. The attempt to isolate a triethylamine salt of V yielded a viscous yellow liquid whose NMR spectrum showed that all of the triethylamine had been stripped away. The residue was a complex mixture giving several peaks between 62.8 and 63.4, a single peak near 61.9, and at least four methyl-on-silicon peaks between 60.3 and 60.6. As mentioned, the initial behavior of XI upon tri- ethylamine addition appeared similar to that of Meldrum's acid. However, a slow subsequent reaction occurred in which much of the methyl-on-silicon shifted to a new en- vironment, gas was evolved, and several peaks gradually appeared in the methylene region of the proton NMR spectrum. 27 The methyl—on-silicon shift is not directly related to the gas generation, since as shown in Figure 1, less than one mole of gas is generated for each mole of methyl-on- Silicon which changes environment. This behavior might be explained by an opening of some V to form linear species with carboxyl groups on the end which could gen- erate gas through decarboxylation: Me 9 9 -CO Me 9 «a SiOCCHZCOH ——Z-) 'V Sioccn3 Me Me Or alternatively, if linear species containing several units formed: 0 Me OSi~ \CII q E/ Me Me Me -co ASio OSi~ / 01 f0 5 Me \c Me Si II c I Me2 H2 However, although the new peaks which appeared at 61.9 and 65.0 in the proton NMR spectrum could be due to the acyl and vinyl groups predicted to form during these rearrangements, their areas are insufficient to account for the volume of gas generated. The situation was not clarified by 13C NMR; it showed strong triethylamine signals and several types of methyl-on-silicon, but a complete absence of carboxyl, methylene, or other carbon species. Thus, the nature of this reaction is unclear. It may be responsible for the inability to isolate the triethylamine salt of V, but does not cause the initial 28 loss of the proton NMR methylene signal, since that occurs immediately after triethylamine addition, before any Significant change in the methyl-on-silicon peak occurs. B. Reaction of Malonic Acid with Diphenylbis(i30pro- penyloxy)silane Because of the difficulties encountered in isolating V, preparation of the diphenyl-substituted version (VI) was also studied. While this intermediate would be more expensive, its formation and characterization was ex- pected to be more straightforward. The steric effects of the phenyl groups should be more effective than methyls in favoring formation of VI in preference to larger rings or linear polymers. Also, the susceptibility of silicon to nucleophilic attack should be reduced by the ability of phenyl groups to donate electrons through (p+d)n interactions:7 <—> and by their steric hindrance. Thus, VI should form more '_P slowly than V, but might be more resistant to rearrange- ment during isolation and analysis. In addition, because phenyl groups cleave from silicon much less readily than methyls during mass spectrum analysis,22 it is more likely that a molecular ion could be detected for VI. 29 As expected, room temperature reaction between malonic acid and X was slower than the reaction involving IX. The diffuse nature of the methylene peak in the proton NMR spectrum of this reaction mixture indicates that a variety of species, probably mixed cyclics and short linear poly- mers, formed. However, the faster reactions which occurred at reflux generated a high proportion of a single species. Both the downfield shift of the methylene peak for this product (63.7 vs. 63.3 for malonic acid in acetonitrile) and the fact that the peak disappeared when triethylamine was added indicate that those protons are unusually acidic as expected for VI. Mass spectrometry confirmed that the solid isolated from this reaction had the correct mole- cular weight for VI (284). The insolubility of this solid product prevented an estimate of purity by NMR analysis. The solid dissolved only in the presence of triethylamine, which eliminated the methylene proton peak. .However, the solid obtained from the room temperature reaction mixture also had a mass spectrum whose highest m/e peak was at 284, even though the NMR spectrum of the reaction mixture indicated the presence of several other species which probably had higher molecular weights. Apparently, VI is the least soluble species, and precipitates selectively. The small portion of the isolated solid which dis- solved in polar solvents like acetone was not VI. Instead, two observations identified it as malonic acid and di- phenyl- containing material present as impurities in the solid and/or generated through hydrolysis of VI by traces 30 of water in the solvent. First, the proton NMR methylene peak appeared at the same position as malonic acid methyl- ene protons, did not disappear upon triethylamine addition, and was too small in comparison to the phenyl absorbtion (1:7 vs. 1:5 expected). Second, the mass spectrum was unlike that of the original solid, but matched the spectrum of solid which had been exposed to the laboratory atmo- sphere for a few minutes. It had no peak at 284, but many higher molecular weight peaks, including one at 594 which fits hexaphenylcyclotrisiloxane, a possible hydro- lysis product of VI. Contact with the solvent also hydrolyzed the undissolved solid, since its mass spectrum matched that of the material which did dissolve. Thus, VI is quite unstable to hydrolysis. However, a negligible amount of decomposition occurred during a two week period when contact with moisture was carefully avoided. C. Acylation Reactions Despite good evidence for the existance of VI and a possibility that V had been prepared, thermal and hydro- lytic instability as well as solubility problems pre- vented obtaining a demonstrably pure sample of either compound. However, the crude products should perform just as well in the acylation procedure described in the introduction as long as they contain no interfering species. The acetone byproduct of the original reaction 31 mixture might interfere by undergoing an aldol conden- sation:23 O O O O R H C O R u \ / 3 \ ./ CH3CCH3 + l/Sl Bas Sl