Document gbExL8nLwVq3LYG5E5xe4mz9e

BFG TECHNICAL DOCUMENT 6jjjhJ Chemical Division RESEARCH AND DEVELOPMENT REPORT THE COMPOSITION OF CODE IOC AND CODE 10F by J. L. Dorsch D. A. Ernes J. A. Nikora Project No. 3508/7721 .Report Type PLASTIC MATERIALS, Date CLOSURE -- STS #458 JANUARY 10, 1980 Copy Approval Authority R F KOEBEL DISTRIBUTION Cleveland Akron J. Hughes Powell, Jr. R. Wymbs *R. A. Krueger \ F. E.rRsause X.C.HblbrbuDk - ALTC CTF - (6) R. R. Bloor B. K. Mikofalvy L. B. Crider *C. E. Fleming L. Cohen D. E. Witenhafer R. L. Bowles C. A. Daniels G>>H/ Schaaf M.W.Roha - G.A .indsay *A. W. Clements *H. 0. Jacobsen *B . A. DiLiddo *J.C.Healy - E..J. Sehm F. J. Donat *J. F. Malone M. M. O'Mara *R. M. Kreager R . F . Ko eb e 1 D. J. Smith D. Hanshumaker R. J. Meyer/Int'l. - (2) J.A.TePas - R.D.Hardesty J.W.Ryan - R.C.Williams - Brecksville C.H.Lufter - J . Pausch R.Komoroski - . .Westfahl *R. J. Fawcett E.D.Truscott R r P .Lattimer E.R.Hooser * Summary Copy only. RD File - (2) BFG*tee05-A 8/78 PRINT ED IN U S A. BFG06342 20768001 39 APPENDIX I Procedure for Preparation of Code IOC--TFAA Derivatives The two derivatizing reagents used were: 1.) Hexamethyldilazane (HMDS), PCR Research Chemicals, Inc., Gainesville, Florida (Cat. 34020-8) 2.) Trifluoroacetic Anhydride (TFAA); Aldrich Chemical Co., Milwaukee, Wis. and Supelco, Bellefonte, Pa. The derivatization reactions were typically carried out in-. Supelco Micro Reaction Vessels (1,2, or 5 mL) capped with Teflon coated septa. Generally, a small quantity of material was dissolved in acetonitrile in the micro vessel and an excess of the derivatizing agent was added. The HMDS reactions were carried out at room temper ature and the ammonia gas by-products were either blown off or removed by vacuum. The TFAA derivatization reactions were carried out at 60-80C and the excess TFAA was removed by vacuum distilla tion through a syringe needle. Caution must be exercised when using the TFAA derivatization technique! TFAA is very volatile and combines with any water available to form the very strong trifluoro acetic acid. Pressure build-up in the heated micro-reaction vessels can (and did) result in rupture of the septa. 9M )89402! -i- EXECUTIVE SUMMARY Objectives 1. ) To identify the structures of the components in Code IOC and Code 10F in support of licensing activity and continued development of reactor coating materials. 2. ) 3. ) To study the relationship of structure to the required surface activity in PVC reactors. To study the chemical reactions occurring in the production of Code IOC and 10F. Significant Results 1. ) Code IOC is a dark brown-colored solid, soluble in aqueous base and dimethyl sulfoxide. It is a mixture of compounds including residual starting material, resorcinol; alkylated resorcinols; alkyl hydroxy chromanes and chromenes; alkylated hydroxy xanthenes; poly(hydroxyphenyl) oligomers; and other compounds of increasing molecular weight and complexity. 2. ) A typical sample of Brecksville pilot plant production Code IOC contains 12.4 weight percent resorcinol and 4.9 percent 9,9-dimethyl xanthene-3,6-diol as determined by liquid chromatography. The third largest component is the resor cinol dimer, 2,3',4-trihydroxybiphenyl, which is present at 5-10%. The amounts of this and remaining components can only be estimated because authentic, pure samples are not available for calibration. Approximately 30% of Code 10C is volatile enough for mass spectrometric analysis. The molecular weights and structures for most of the components in this group have been identified. The structures in the volatile and non-volatile fractions have been compared by analysis of the NMR data and mechanistic descriptions of the reactions. 3. ) The Code 10C mixture is produced by a set of competing reac tions which include the intended self-condensation of resor cinol and the unintended decomposition of resorcinol into carbonyl compounds which further react to form the alkylated phenols, xanthenes, and higher molecular weight products observed. 4. ) It has not been determined which particular components in Code 10C are the desired surface-active species. Several low molecular weight components have been shown to be not surface-active. It is not known whether the alkylated com pounds are more or less surface-active than the poly(hydroxy phenyl) compounds. BFG06343 20768002 Discussion - (continued) Louis Cohen does not feel that oxidation of the initial com ponents, by itself, is an important factor in increasing the surface acitivity of the mixtures based on numerous experiments with Code 10 materials and antioxidants1. He feels that high molecular weight is the most desirable characteristic in any Code 10 component and that oxidative coupling, which leads to an increase in molecular weight, is useful. Oxidative coupling is his rationale for the improved activity of Code 10P over Code 10C and Code 10G over 4,4*-- thiobis(phenol). Oxidative coupling has not yet been confirmed analytically in Code 10F, but has been detected in Code 10G in R. P. Lattimer's recent FD/MS spectra. Whatever the relative effects of structure, oxidation, and molecular weight on the activity of resorcinol condensation products, their vulnerability to oxidation guarantees that they will all be highly colored. JLD-DAE-JAN/dmw John L. Dorsch David A. Ernes John A. Nikora - ii - Significant Results - (continued) 5. ) 6. ) A large fraction by weight of the Code IOC mixture is com posed of components whose structures are not disclosed in the Code IOC patent. We have found an alternate route to a surface-active IOC-like mixture of oligomers starting from resorcinol and acetaldehyde using very mild conditions An invention record has been written for this preparation. Many of the alkylated compounds in Code IOC contain the xanthene structure and so are highly fluorescent. From ultraviolet excitation, they fluoresce an intense green or blue-green in dilute agueous or methanol solutions. This phenomenon may be of some use in analytical determinations. 7. ) The Code 10F mixture was found to be even less volatile and higher in molecular weight than IOC. Those components observed by mass spectrometry were found to be chlorinated versions of IOC components. Conclusions 1.) The formation of oligomers in Code IOC production proceeds to a large extent because of the thermal decomposition (at 280C) of resorcinol to the more reactive carbonyl contain ing compounds which condense rapidly with resorcinol and other poly(hydroxyphenyl) compounds. -2.) An alternate route to IOC-like oligomers exists through the condensations of resorcinol with aldehydes. These conden sations require relatively mild conditions ( 120C, atmo spheric pressure) and so have an advantage over current IOC technology. However, these products are highly colored, just like IOC; the condensation of resorcinol with an alde hyde is not a route to a "white Code 10". Recommendations 1. ) We should be prepared to answer detailed questions concern ing Code 10C composition from our licensees who may be doing their own analytical work on 10C and who may be surprised at the results. 2. ) We should pursue a patent on the use of condensation products of resorcinol with aldehydes as reactor coating materials. 20768003 BFG06344 35 Figure xiv; PROPOSED REACTION PATHWAY "ACETALDEHYDE SERIES" 0 8 hcch3 + C02 OH" * yH 'y0 w j/""'0 0 HCCH2C-0H + (CH3C-CH3) OH . H0v^\^ 0\^0 OH OH (H) 162 iff* 20768042 OH 138 H3C H 228 9 - iii - Future Work 1. ) A detailed report on the syntheses of Code IOC-like oligomers from resorcinol and acetone, acetaldehyde, benzaldehyde, 4- methoxybenzaldehyde, and 4-hydroxybenzaldehyde is in prepar ation by J. L. Dorsch and D. J. Smith. 2. ) Analytical characterization of Code 10G is proceeding with the help of R. P. Lattimer of BRDC. Acknowledgements A large amount of mass spectrometric data required for this analysis was contributed by R. P. Lattimer of BRDC. We appreciate his continuing support. D. J. Smith and D. Hanshumaker did the laboratory separations and syntheses required in this work. All of the mechanistic descriptions of the IOC reactions were developed in conversations with R. F. Koebel. We thank Louis Cohen for his many helpful comments on this manuscript. k)O 0 9 A O 2 BFG06345 33 Figure xii: PROPOSED REACTION PATHWAY "RESORCINOL SERIES" 202 CH>99 0Z TABLE OF CONTENTS Introduction.............................................................................................................................. Experimental Results........................................................................................................ A.) and NMR Analysis -- J. L. Dorsch........................ B.) Gas Chromatographic/Mass Spectrometric Results. . C.) Code 10F Results -- Ji A. Nikora............................................... D.) Liquid ChromatographicResults-- D. A. Ernes . . Discussion................................................................................................................................... References.................................................................................................................................... Appendix 1................................................................................................................................... PAGE 1 3 3 13 22 25 29 38 39 20768005 BFG06346 31 Discussion - (continued) The same argument was used to assign members and structures in the acetaldehyde group. A supporting set of evidence for this method was obtained from mass analysis of the crude products formed in attempts to synthesize compounds 4 and 9. From acetone and resorcinol, products of molecular weight lF2, 242, and 374 were obtained. From acetaldehyde and resorcinol, products of mass 138, 228, 256, 346, 360, and many others were obtained. The assignment of molecular weights to structures is further demonstrated in Figure XI. The structures of the intended resorcinol oligomers and the unintended by-products produced tell much about the chemistry and conditions inside the Code 10C reactor. The structure of the resorcinol dimer, !5, identifies the resorcinol "self-condensation" as first an aldol condensation between two moles of resorcinol in their keto-form, then a dehydration, and finally a proton rearrange ment to form the final stable tautomer, 5. This process is drawn in Figure XII. The presence of the by-products in 10C indicates a basecatalyzed decomposition reaction of resorcinol which is competitive with the condensation reactions at 280C. In Figure XIII, this reaction is shown as a double rearrangement reaction of resorcinol's diketo-form to produce acetone and 1-oxopropanoic acid. The acetone and resorcinol then produce isopropenyl resorcinol following another aldol condensation and dehydration. This intermediate then either reacts with resorcinol to form the 9,9-dimethylxanthene-3,6-diol (M.W. 242) or is reduced in an unknown process to form isopropylresorcinol . Isopropenyl resorcinol can react with any member of the resorcinol oligomer series (110 + 92n) to form any particular member of the "acetone series" 110 + 92n + 40. The isopropyl resorcinol is probably much less reactive and may be thought of as a dead end in this reaction scheme. In Figure XIV, the rest of the possibilities are indicated through reactions of the other by-product, 1-oxopropanoic acid. Condensation with resorcinol leads to 7-hydroxycoumarin, 6_, while the apparently favored further decomposition to carbon dioxide and acetaldehyde leads to a myriad of acetaldehyde/resorcinol products. Acetaldehyde appears to be the most reactive component in the system. 20768038 LIST OF TABLES Table I - Predicted Versus Found NMR Chemical Shifts for Aromatics in Resorcinol Oligomers...................... Table II - Relative Amounts of TFAA Derivatives of Code IOC Components Detected by GC/MS (represents the most volatile 30%)............................................................ Table III - Elemental Compositions of Code 10C TFAA Deriv atives Determined from Medium Resolution GC/MS Table IV - Molecular Weights of Code 10C Components Iden tified Through GC/MS and/or FD/MS Data .... Table V - Classification of Components Based on Reac tants............................................................................................................. PAGE 8 17 20 30 30 20768006 BFG06347 29 Discussion Our analysis of Code IOC was based primarily on the exact identification of a number of the lower molecular weight components with extrapolation to the structures of the higher molecular weight components through the use of the high mass MS-molecular ions, the condensed phase NMR data, and a mechanistic description of the reactions which must form all the species present. In Table IV, a list is provided of the twenty-seven molecular weights identified in the Code IOC mixture through mass spectrometry. For some of the weights, complete structures are known; for others, elemental compositions and some structural details are known. For the rest, only the nominal masses are known. However, all the com ponents can be linked directly through the building blocks of resor cinol and its decomposition products; acetone, acetaldehyde, and 1-oxopropanoic acid. This classification is shown in Table V. In the previous section on MS results, the known compounds were grouped into structural classes, such as alkylated resorcinols, xanthenes, etc., based on the available analytical data. In Table V, the com ponents are classified in a more vertical fashion as series of com pounds formed from common reactants. For example, both isopropyl resorcinol (M.W. 152) and 9,9-dimethyl-xanthene-3,6-diol (M.W. 242) belong to the group of resorcinol/acetone reaction products even though they belong to different structural classes. This rationalization can be extended to the remaining unknown components by using the numerical relationships apparent in the list of molecular weights. The molecular weights of the resorcinol oligomers with dimethyl xanthene end groups should form the series llO + 92n + 40. Just such a series is present; and the first member of the series, , M.W. 242, was completely identified. Therefore, it seems reasonable to assign unknowns 334 and 426 to this structural class. Thus, structure 8 below is proposed for unknown 334, though the exact mass is the only bit of hard data available for this component. Other molecular weights also appear to be acetone/resorcinol products, but not members of the 110 + 92n + 40 series. Besides isopropyl resorcinol (M.W. 152), unknown 374 is an example; it is assigned to structure LI, based on the likelyhood that can further condense with acetone and resorcinol to form 11. ?0 O -4 6) CO o BFG06377 LIST OP FIGURES Figure I - 60 MHz 1h NMR Spectrum of Code 10C...................... PAGE 4 Figure II - 13c NMR Spectrum of Code 10C in DMSO-dg Solution............................................................................................. 5 Figure III - Aromatic Portion of 13c NMR Spectrum of Code 10C Distillate (B.P. 220C/20ym Hg). . 6 Figure IV - Comparison of i3c NMR Spectra of and Code 10C...................................... 10 Figure V - Comparison of l3c NMR Spectra of Acetaldehyde/ Resorcinol Condensation Product and Code 10C........................................................................................................ 11 Figure VI - GC/MS Analysis of TFAA Derivatives of Code 10C............................................................................................. 18 Figure VII - GC/MS Analysis of TFAA Derivatives of Code 10C............................................................................................. 19 Figure VIII - Total Ion Current Profile of TC 9B05: Overhead from 10C Production at Brecksville (resorcinol does not elute from the GC column under these conditions)................................. 23 Figure IX - Code 10C: LC Profile....................................................... 26 Figure X - Fraction Collection: Code 10C Sublimate. . 27 Figure XI - Assignment of Molecular Weights to Series of Related Structures ....................................................... 32 Figure XII - Proposed Reaction Pathway (Resorcinol)... 33 Figure XIII - Proposed Reaction Pathway (Acetone) .... 34 Figure XIV - Proposed Reaction Pathway (Acetaldehyde). . 35 20768007 27 Figure X: Fraction Collection: Code 10c Sublimate A fr0Q90Z Introduction Code IOC (3000 x 32) is a mixture of aromatic oligomers pro duced by heating resorcinol at 280C for four hours with a catalytic amount of sodium hydroxide present. The process is called a "selfcondensation" because resorcinol is the only starting material and because water liberated by condensation is removed overhead. The solid reddish-brown product of this reaction is then dissolved in aqueous sodium hydroxide solution and used finally to treat the walls of PVC reactors to prevent build-up. This technology has been very successfully used in our own PVC production and has also generated substantial revenue from licensees. The preparation and use of Code 10C, discovered by Louis Cohen, is disclosed in U. S. Patent 4,080,173 (1978). Due to the immediate successful use of this important inven tion in PVC production and to the lack of problems associated with the synthesis of 10C, very little analytical data was obtained orT the product. The only current quality control specification for the 10C solid is a softening point of 120C or higher. The 10C composition is described in the patent as a mixture of poly(hydroxyphenyl) compounds , as shown below; however, this statement is based on literature reports of similar preparations and previous BFG analytical work on the composition of Code 10A-10B. Cohen states in the patent that the poly(hydroxyphenylenes) predominate in the mixture. This is based on the weight percent hydroxyl value reported10 for an oligomer prepared by condensing resorcinol with zinc chloride catalyst and on the observation of the higher softening point of Code 10C solid than that of Code 10A. The purpose of this report is to detail our analysis of Code 10C. In December, 1978*, Cohen requested assistance from the ALTC Analytical Group in determining the weight percent hydroxyl content in Code 10C in order to estimate the relative amounts of the two classes of oligomers. This information and other analytical data were needed to answer questions from our licensees who were being encouraged to produce their own 10C product on site. At that time, * Initial mass spectrometer data was obtained at ALTC in 1977. Because not all compounds identified by molecular weight fit the above reaction, a recommendation was made to analyze the materials on the new MAT 311A at Brecksville in early 1978. BFG06349 20766008 25 D.) Liquid Chromatographic Results -- D. A. Ernes Liquid chromatography was employed to further characterize Code IOC. Suitable resolution was achieved using a gradient elution system without derivatization. Figure IX illustrates the profile obtained for Code IOC. The sloping baseline indicated by the dashed line results from the changing solvent UV absorption during the gradient run. The early eluting components were well resolved, while the remainder were not well resolved apparently due to increasing molecular weight and similarity of structure. Due to anticipated interferences from the high-molecular weight compounds, a simpler mixture was sought for further charac terization. A fraction of IOC which sublimed in vacuum at 230C was found to be rich in the early eluting (LC) components and so was used in peak trapping experiments. Several fractions were collected during the separation of this sublimate which were then analyzed by infrared and mass spec trometry. The samples were collected at the outlet of the fluores cence detector. These samples are indicated on the chromatogram presented in Figure X. Fraction A contained primarily two components, resorcinol, which was confirmed by both IR and mass spectrometry, and an unknown of molecular weight 162. The infrared spectrum revealed that unknown 162 contained a carbonyl group and the mass spectrum revealed a fragmentation pattern very similar to that published for coumarin7. Based on this data, umbelliferone or 7-hydroxy- .coumarin, 6, was proposed as the identity of unknown 162. Later, this assignment was confirmed through an LC peak enrichment experi ment with the authentic compound and through NMR data obtained on a IOC distillate. Fraction B contained essentially one component of molecular weight 202. The infrared spectrum of this fraction compared well with the Koppers Pencolite Resin 441 distillate which was rich in the resorcinol dimer, 5. Fraction C contained several components. Mass spectral data for this fraction showed large ions at masses 152, 213, and 227. The m/e 213 and 227 ions are now known to be M-15 fragments from the M.W. 228, 9_, and M.W. 242, , components. The 152 (and 137) ions have been assigned to isopropyl resorcinol. The infra red spectrum of Fraction C and authentic 9,9-dimethyl-xanthene- 3,6-diol compared very well. Additional fractions collected did not contain a recogniz able major component and the materials had very low volatility. BFG06373 20768032 3 Experimental Results A.) 13C and -^-H NMR Analysis -- J. L. Dorsch The 1H-60 MHz NMR spectrum of Code IOC in dimethyl sulfoxide- dg solution is shown in Figure I. In addition to the expected com plex aromatic pattern, there is also a large amount of alkyl peak area indicated. The chemical shifts are consistent with the presence of both isopropylidene and ethylidene groups between rings, as well as external isopropyl groups. The 13C-20.1 MHz NMR spectrum of Code 10C,again in DMSO-dg solution,is shown in Figure II. Alkyl carbons are indicated by the peaks in the 20 to 33 ppm region, while aromatic carbons are indicated in the 100 to 160 ppm region. The four largest peaks in the spectrum are due to the residual starting material, resorcinol. The substantial amount of alkyl carbon present and the complexity of the aromatic pattern suggest a composition for the products of the 10C reaction, quite different than that described in the patent. In overall appearance, both the 13C and NMR spectra of Code 10C are reminiscent of spectra of certain commercial antioxi dants produced by condensing phenols with acetone and acetaldehyde. This observation is consistent with the weights of molecular ions detected in the FD/MS spectra3 which form such series as 110 + 92n, 110 + 92n + 26, and 110 + 92n + 40. The 110 + 92n series is the resorcinol self-condensation series , while the additions of 26 and 40 are the net results of condensing acetaldehyde and acetone, respectively, to form ethylidene (CH3-CIO and isopropylidene / (CH3) 2CC7" bridges between rings. We reported earlier3 that the Koppers resorcinol raw material is very pure and contains no impurities which can account for the alkylated structures observed in the NMR spectra. Likewise, there are no ketones, aldehydes, or hydrocarbons available in the Brecksville pilot plant to contaminate the product. Previous analyses^ of a series of Code 10C products by 3H NMR showed that the relative amount of alkyl hydrogen increased with increasing reaction temper atures and increasing amounts of sodium hydroxide catalyst used. The volatile fractions distilled from Code 10C were of great value to the interpretation of the NMR data because these fractions contained fewer components, simpler structures, and sharper NMR lines, permitting more positive relationships to be made between the NMR and mass spectrometric data. The lowest boiling fraction (120C, 20ym Hg) was found to be rich in resorcinol and isopropyl resorcinol, 2, M.W. 152, in agreement with the MS data. A higher boiling fraction (220C at 20um Hg) gave a mass spectrum indicating large amounts of molecular weight 202 and 242 components present with smaller amounts of M.W. 162 and others. The aromatic portion of the i3C NMR spectrum of this fraction is shown in Figure III. The structures of three components identified are shown along with some characteristic aromatic carbon lines of each. BFG06351 2 0 7 6 S 0 I0 23 Figure VIII: Total Ion Current Profile of TC 9B05: Overhead from IOC Production at Brecksville (resorcinol does not elute from the GC column under these conditions} Normalized Total Ion O CH3CCH3 20768030 Mass Scan Number (a retention time) SG0 3 <#> O H<*> <\ < <D id ; <D <M C 0 iH cuja eo O flj -H 1u3 U uU mu U -H o10 +j a) H m Vi MI-H E OM II << M V* * t*o- t4)3 to T3 jj <u W* a) rt-oi c0U 0uffimt<HD ,a_ij o to fo * 0aa -roH to 0 au 3 a 9 <o c u m a 0 r-t Vo-I +> >aa> ccr> -H uH -U in c cm to a) n Oh C e O O voto to 13 TJ O _ to(O0 oI Ol'O OI Uo wX ~ H O 13 Q M-t W S. Q>-H 20768012 __________ >. o i ] Ot -L-5C NMR S p e c tru m o f C ode IO C i n DMS0 - d 5 S o l u t i o n -X F ig u re I I : 21 Gas Chromatographic/!4ass Spectrometric Results - (continued) Based on these elemental compositions, five classes of com pounds are indicated as volatile components of Code IOC: Class 1: Chromenes and Chromanes HO Class 2: Resorcinol and Alkylated Resorcinols OH R = H, C2H5,' and iso-C3H7 Class 3: Xanthenes RR Class 4: Poly(hydroxyphenyl) Oligomers HO 8 2 0 8 9 /.0 2 7 13C and J-H NMR Analysis -- J. L. Dorsch - (continued) The three components identified are the expected two ring resorcinol dimer, 5_, molecular weight 202; an acetone/resorcinol reaction product -- 9,9-dimethyl-3,6-xanthene diol, , molecular weight 242; and an even stranger resorcinol by-product -- 7-hydroxycoumarin, , molecular weight 162. HO TOOa ch3n fch3 HO w 4 (242) 6 (162) 2,3/4-trihydroxybiphenyl 9,9-dimethyl-xanthene3,6-diol 7-hydroxycoumarin The 13C NMR peak assignments were made by comparison with the spectra of authentic samples of the three compounds. The 7hydroxycoumarin was obtained from Aldrich Chemical; the 9,9-dimethylxanthene-3,6-diol was prepared in the laboratory from acetone, resorcinol, and zinc chloride catalyst (m.p. 263-266; literature m.p. 266C)^. The resorcinol dimer was never obtained in pure form; however, a commercial, impure sample of the dimer was obtained as Koppers Penacolite Resin 441. As part of the analytical development work on this problem, the NMR spectra of a large number of phenols and hydroxy biphenyl compounds were obtained. From this data, a series of parameters were developed with which one can calculate, empirically, the 13C NMR aromatic chemical shifts of the poly(hydroxyphenyl) oligomers expected to be in Code 10C. This method was then used to predict the spectra of the dimer and trimer. In Table I, the predicted chemical shifts for the dimer, 5, are compared with the experi mentally determined shifts for the dimer observed as the largest component in 441 resin. The agreement is satisfactory. In the same table, the predicted chemical shifts for the trimer 1_ are listed. Notice that both compounds are predicted to have a peak or peaks near 140 ppm due to the number 3 and 15 carbons. In the general case of the poly(hydroxyphenyl) oligomer, the carbons indi cated in the drawing below are expected to be observed near 140 ppm. BFG06355 {'T 0E9 Z.0 Z 19 Figure VII: GC/MS Analysis of TFAA Derivatives of Code IOC BRDC Varian MAT 311-A Mass Spectrometer Dexsil 300 GC Column (Silicone peaks from GC column bleed erased for cosmetic purposes) BIC 02/23/79 14:36:00 SAUPLE: TFA DERIVATIZED CODE 10C DATA: MS567S1S 81 ENHANCED (S 20B 2N 0T) 20768026 9 and NMR Analysis -- J. L. Dorsch - (continued) OH * Carbons with predicted chemical shifts of 139-140 ppm. In the spectrum of the volatile fraction of 10C shown in Figure III, a large peak at 140.3 ppm is indeed visible due to the dimer . However, in the spectrum of the original Code 10C mixture, carbon peaks near 140 ppm are small. Certainly the dimer 5, trimer and several other poly (hydroxyphenyl) oligomers detected by FD/MS are present in Code 10C. However, the total weight percent of this group can not be large. The dimer, , molecular weight 20 2, is by far the largest in amount of this group as observed by mass spectrometry. The liquid chromatographic peak area for the dimer indicates that it is present in 10C at no more than ten percent. Considering the relative peak heights of the 110 + 92n molecular ions detected in the FD/MS analysis^ and the small amount of absorp tion near 140 ppm in the nmr spectrum, one can conservatively estimate that the total amount of simple poly(hydroxyphenyl) com pounds is no more than twenty percent in 10C. The remaining eighty percent consists of resorcinol and alkylated structures of one form or another. In Figure IV, the ^C NMR spectrum of authentic 9,9-dimethyl xanthene-3,6-diol, , is compared with that of Code 10C. All of the peaks of are recognizable features of the 10C pattern. It is not possible to estimate the weight percent of simply from this data, however, because other related compounds containing the di methyl xanthene structure as an end group may be contributing to these peaks. An example of a related compound which would provide carbon peaks in many of the same areas is structure below, which is our proposal for the unknown of mass 334.1183 (C21H184' <0 'O O O 0Q 20768016 2- (3 'hydroxyphenyl)-9,9-dimethyl-xanthene3,6-diol 9-methyl-xanthene3,6-diol 17 Table II Relative Amounts of TFAA Derivatives of Code IOC Components Detected by GC/MS (Represents the Most Volatile 30%? Component 0 il T = -C-CF3 Total Ion Area Percent & OT OT d; OT OT 21.0% <1 <1 20768024 TO nr to; TO OT 7.0 37.Q F ig u r e V : C o m p a ris o n o f i 3 C NMR S p e c tra o f A c e t a ld e h y d e / R e s o r c in o l C o n d e n s a tio n P r o d u c t a n d Code IOC o -b *5 15 Gas Chromatographic/Mass Spectrometric Results - (continued) OH OH OH 10 c12h103: M*W* 202 C12H103: M*w* 202 Si (CH3)3 si(CH3) 6J> ^18^263^2 M.W. 346 Si (CH3) 3 0 The mass spectral data obtained from the trimethylsilyl ethers were quite simple. The spectra were characterized by a pre dominant molecular ion peak (Mt) and a peak at (M-15)+ due to a loss of a methyl group. Very little other information was available. The gas chromatographic separations obtained were only marginally better than those obtained on the original mixture. Derivatization of the 10C components with trifluoroacetic acid anhydride (TFAA) resulted in much improved gas chromatographic separations and more complex mass spectra for each component due to extensive fragmentation of the trifluoroacyl group. The fragmenta tion patterns observed for the 10C derivatives closely paralleled those previously reported in the literature for trifluoroacyl esters of phenols and alkylphenols. The derivatized components were represented by prominent molecular ion peaks (Mt) and fragment ion peaks resulting from the following losses: CF3(m-69)+, *CF3CO(m-97) F(m-19)+, CO(m-28)+, and CF2O(m-66)t. The trifluoroacyl derivatives of Code 10C were examined by both low and medium resolution GC/MS techniques. Figures VI and VII show the total ion current profiles (TICP's) for 10C obtained on the two different mass spectrometers with prior separation on a 4.9m Dexsil-300 GC column. The individual mass spectra, not shown, were all obtained under low resolution conditions in these experiments. However, the resolution of components in the chromato- 990 D d q 20768022 13 B.) Gas Chromatographic/Mass Spectrometric Results The lower molecular weight (<800 amu's) volatile components of the Code 10C product mixture were characterized through several mass spectrometric (MS) and gas chromatographic/mass spectrometric (GC/MS) techniques. The outline below describes the approach in brief. Brecksville R&D Center -- Varian MAT 311A Spectrometer -- R. P. Lattimer Field Desorption-MS Provided nominal mass molecular ions through about mass 800. Electron Impact-MS With medium resolution (ca. 10,000), accurate elemental formulas were determined for the more volatile components from solids probe intro duction. GC/MS-Electron Impact With lower resolution (ca. 3,000), GC separation and MS elemental mapping of TFAA derivatized fractions. Avon Lake Technical Center -- duPont 21-490 Spectrometer -- J. A. Nikora Electron Impact-MS Low resolution (A*500 amu), solids probe introduction, electron impact ionization. GC/MS-Electron Impact Low resolution analysis of GC resolved components and derivatives. Mass spectral data were obtained for all of the Code 10C components which were volatile under typical MS and GC/MS conditions. To increase the volatility of the components and to facilitate their separation through gas chromatography (GC), trimethyl silyl ether and trifluoroacyl ester derivatives were prepared. In addition to these 10C fractions and derivatives, laboratory synthesized compounds and sodium hypochlorite reaction products were also examined. The Code 10C solid product represents an extremely complex mixture of products ranging from unreacted resorcinol to high molecular weight components not observable by mass spectrometry. The interpretation of the MS data was greatly simplified through prior gas chromatographic separation of the components. Several derivatization techniques were used to improve the GC resolution. BFG06361 20768020 13 B.) Gas Chromatographic/Mass Spectrometric Results The lower molecular weight (<800 amu's) volatile components of the Code 10C product mixture were characterized through several mass spectrometric (MS) and gas chromatographic/mass spectrometric (GC/MS) techniques. The outline below describes the approach in brief. Brecksville R&D Center -- Varian MAT 311A Spectrometer -- R. P. Lattimer Field Desorption-MS Provided nominal mass molecular ions through about mass 800. Electron Impact-MS With medium resolution (ca. 10,000), accurate elemental formulas were determined for the more volatile components from solids probe intro duction. GC/MS-Electron Impact With lower resolution (ca. 3,000), GC separation and MS elemental mapping of TFAA derivatized fractions. Avon Lake Technical Center -- duPont 21-490 Spectrometer -- J. A. Nikora Electron Impact-MS Low resolution (A/500 amu), solids probe introduction, electron impact ionization. GC/MS-Electron Impact Low resolution analysis of GC resolved components and derivatives. Mass spectral data were obtained for all of the Code 10C components which were volatile under typical MS and GC/MS conditions. To increase the volatility of the components and to facilitate their separation through gas chromatography (GC), trimethyl silyl ether and trifluoroacyl ester derivatives were prepared. In addition to these 10C fractions and derivatives, laboratory synthesized compounds and sodium hypochlorite reaction products were also examined. The Code 10C solid product represents an extremely complex mixture of products ranging from unreacted resorcinol to high molecular weight components not observable by mass spectrometry. The interpretation of the MS data was greatly simplified through prior gas chromatographic separation of the components. Several derivatization techniques were used to improve the GC resolution. BFG06361 20768020 15 Gas Chromatographic/Mass Spectrometric Results - (continued) OH OH 6.0j6 10 c12h103: M-W* 202 OH OH c12hio3; M.W. 202 Si. (CH3>3 si(CH3) Si (CH3) 3 990 O jq c18H263Si2: M-w- 346 The mass spectral data obtained from the trimethylsilyl ethers were quite simple. The spectra were characterized by a pre dominant molecular ion peak (Mt) and a peak at (M-15)+ due to a loss of a methyl group. Very little other information was available. The gas chromatographic separations obtained were only marginally better than those obtained on the original mixture. Derivatization of the 10C components with trifluoroacetic acid anhydride (TFAA) resulted in much improved gas chromatographic separations and more complex mass spectra for each component due to extensive fragmentation of the trifluoroacyl group. The fragmenta tion patterns observed for the 10C derivatives closely paralleled those previously reported in the literature for trifluoroacyl esters of phenols and alkylphenols. The derivatized components were represented by prominent molecular ion peaks (Mf) and fragment ion peaks resulting from the following losses: -CF3(m-69)+, CF3CO(m-97) F(m-19)+ , CO(m-28)+, and CF20(m-66)t. The trifluoroacyl derivatives of Code 10C were examined by both low and medium resolution GC/MS techniques. Figures VI and VII show the total ion current profiles (TICP's) for 10C obtained on the two different mass spectrometers with prior separation on a 4.9m Dexsil-300 GC column. The individual mass spectra, not shown, were all obtained under low resolution conditions in these experiments. However, the resolution of components in the chromato- 20768022 F ig u r e V : C o m p a ris o n o f ^-*C NMR S p e c tr a o f A c e t a ld e h y d e / R e s o r c in o l C o n d e n s a tio n P r o d u c t a n d Code IOC VMcU) O 2076801s i* U o BFG06359 17 Table II Relative Amounts of TFAA Derivatives of Code IOC Components Detected by GC/MS (Represents the Most Volatile 30%) Component______________________ OT 0 ii T = -C-CF3 Total Ion Area Percent TO TO OT <1 7.0 37.Q 20766024 9 13C and NMR Analysis -- J. L. Dorsch - (continued) OH BFG06357 * Carbons with predicted chemical shifts of 139-140 ppm. In the spectrum of the volatile fraction of 10C shown in Figure III, a large peak at 140.3 ppm is indeed visible due to the dimer 5^ However, in the spectrum of the original Code 10C mixture, carbon peaks near 140 ppm are small. Certainly the dimer trimer 1_, and several other poly (hydroxyphenyl) oligomers detected by FD/MS are present in Code 10C. However, the total weight percent of this group can not be large. The dimer, 5_, molecular weight 20 2, is by far the largest in amount of this group as observed by mass spectrometry. The liquid chromatographic peak area for the dimer indicates that it is present in 10C at no more than ten percent. Considering the relative peak heights of the 110 + 92n molecular ions detected in the FD/MS analysis2 and the small amount of absorp tion near 140 ppm in the NMR spectrum, one can conservatively estimate that the total amount of simple poly(hydroxyphenyl) com pounds is no more than twenty percent in 10C. The remaining eighty percent consists of resorcinol and alkylated structures of one form or another. In Figure IV, the ^3C NMR spectrum of authentic 9,9-dimethyl xanthene-3,6-diol, , is compared with that of Code 10C. All of the peaks of are recognizable features of the 10C pattern. It is not possible to estimate the weight percent of simply from this data, however, because other related compounds containing the di methyl xanthene structure as an end group may be contributing to these peaks. An example of a related compound which would provide carbon peaks in many of the same areas is structure 8^ below, which is our proposal for the unknown of mass 334.1183 (C21H184)* o 00 G) 2-(3'hydroxyphenyl)-9,9-dimethyl-xanthene3,6-diol 9-methy1-xanthene3,6-diol 19 Figure VII: GC/MS Analysis of TFAA Derivatives of Code IOC BRDC Varian MAT 311-A Mass Spectrometer Dexsil 300 GC Column (Silicone peaks from GC column bleed erased for cosmetic purposes) RIC 02/23/79 14:36:60 SAUPLE: TFA DERIVATIZED CODE 10C DATA: KS567S1S SI ENHANCED (S 20B 2N 0T) 20768026 7 and NMR Analysis -- J. L. Dorsch - (continued) The three components identified are the expected two ring resorcinol dimer, 5^ molecular weight 202; an acetone/resorcinol reaction product -- 9,9-dimethyl-3,6-xanthene diol, 4, molecular weight 242; and an even stranger resorcinol by-product -- 7-hydroxycoumarin, 6, molecular weight 162. (resorcinol dimer) 2,3/4-trihydroxybiphenyl 9,9-dimethyl-xanthene3,6-diol 7-hydroxycoumarin The nmr peak assignments were made by comparison with the spectra of authentic samples of the three compounds. The 7- hydroxycoumarin was obtained from Aldrich Chemical; the 9,9-dimethyl- xanthene-3,6-diol was prepared in the laboratory from acetone, resorcinol, and zinc chloride catalyst (m.p. 263-266; literature m.p. 266C)'. The resorcinol dimer was never obtained in pure form; however, a commercial, impure sample of the dimer was obtained as Koppers Penacolite Resin 441. As part of the analytical development work on this problem, the 13c NMR spectra of a large number of phenols and hydroxy biphenyl compounds were obtained. From this data, a series of parameters were developed with which one can calculate, empirically, the "c NMR aromatic chemical shifts of the poly(hydroxyphenyl) oligomers expected to be in Code 10C. This method was then used to predict the spectra of the dimer and trimer. In Table I, the predicted chemical shifts for the dimer, 5^, are compared with the experi mentally determined shifts for the dimer observed as the largest component in 441 resin. The agreement is satisfactory. In the same table, the predicted chemical shifts for the trimer 1_ are listed. Notice that both compounds are predicted to have a peak or peaks near 140 ppm due to the number 3 and 15 carbons. In the general case of the poly(hydroxyphenyl) oligomer, the carbons indi cated in the drawing below are expected to be observed near 140 ppm. BFG06355 frT 0 9 Z 0 2 21 Gas Chromatographic/!4ass Spectrometric Results - (continued) Based on these elemental compositions, five classes of com pounds are indicated as volatile components of Code IOC: Class 1: Chromenes and Chromanes HO Class 2: Resorcinol and Alkylated Resorcinols OH R = H, C2H5,' and iso-C3H7 Class 3: Xanthenes RR HO ^ 0 ^ OH R H or CH3 Class 4: Poly(hydroxyphenyl) Oligomers HO 20768028 F ig u re I I : 1 J C NMR S p e c tru m o f C ode IO C i n DMSO-dg S o lu t io n c#> O (Ti <#> o <0 <D l-l < C o c D l-l o c o m H xi u u l-l (0 u u -H p iH (0 >1 e xo rH M << t * r? ~ <rv o i 23 Figure VIII: Total Ion Current Profile of TC 9B05: Overhead from IOC Production at Brecksville (resorcinol does not elute from the GC column under these conditions) Normalized Total Ion O ii CH3CCH3 Mass Scan Number (a retention time) 20768030 3 Experimental Results A.) 13C and ^-H NMR Analysis -- J. L. Dorsch The ^H-60 MHz NMR spectrum of Code IOC in dimethyl sulfoxidedg solution is shown in Figure I. In addition to the expected com plex aromatic pattern, there is also a large amount of alkyl peak area indicated. The chemical shifts are consistent with the presence of both isopropylidene and ethylidene groups between rings, as well as external isopropyl groups. The 13C-20.1 MHz NMR spectrum of Code IOC,again in DMSO-dg solution,is shown in Figure II. Alkyl carbons are indicated by the peaks in the 20 to 33 ppm region, while aromatic carbons are indicated in the 100 to 160 ppm region. The four largest peaks in the spectrum are due to the residual starting material, resorcinol. The substantial amount of alkyl carbon present and the complexity of the aromatic pattern suggest a composition for the products of the 10C reaction, quite different than that described in the patent. In overall appearance, both the ^-3C and ^H NMR spectra of Code 10C are reminiscent of spectra of certain commercial antioxi dants produced by condensing phenols with acetone and acetaldehyde. This observation is consistent with the weights of molecular ions detected in the FD/MS spectra2 which form such series as 110 + 92n, 110 + 92n + 26, and 110 + 92n + 40. The 110 + 92n series is the resorcinol self-condensation series , while the additions of 26 and 40 are the net results of condensing acetaldehyde and acetone, respectively, to form ethylidene (CH3~CH^) and isopropylidene / (CH3)2CKJ bridges between rings. We reported earlier2 that the Kopoers resorcinol raw material is very pure and contains no impurities which can account for the alkylated structures observed in the NMR spectra. Likewise, there are no ketones, aldehydes, or hydrocarbons available in the Brecksville pilot plant to contaminate the product. Previous analyses^ of a series of Code 10C products by ^H NMR showed that the relative amount of alkyl hydrogen increased with increasing reaction temper atures and increasing amounts of sodium hydroxide catalyst used. The volatile fractions distilled from Code 10C were of great value to the interpretation of the NMR data because these fractions contained fewer components, simpler structures, and sharper NMR lines, permitting more positive relationships to be made between the NMR and mass spectrometric data. The lowest boiling fraction (120C, 20ym Hg) was found to be rich in resorcinol and isopropyl resorcinol, 2_, M.W. 152, in agreement with the MS data. A higher boiling fraction (220C at 20ym Hg) gave a mass spectrum indicating large amounts of molecular weight 202 and 242 components present with smaller amounts of M.W. 162 and others. The aromatic portion of the 13c NMR spectrum of this fraction is shown in Figure III. The structures of three components identified are shown along with some characteristic aromatic carbon lines of each. BFG06351 20763010 25 D.) Liquid Chromatographic Results -- D. A. Ernes Liquid chromatography was employed to further characterize Code IOC. Suitable resolution was achieved using a gradient elution system without derivatization. Figure IX illustrates the profile obtained for Code IOC. The sloping baseline indicated by the dashed line results from the changing solvent UV absorption during the gradient run. The early eluting components were well resolved, while the remainder were not well resolved apparently due to increasing molecular weight and similarity of structure. Due to anticipated interferences from the high-molecular weight compounds, a simpler mixture was sought for further charac terization. A fraction of IOC which sublimed in vacuum at 230C was found to be rich in the early eluting (LC) components and so was used in peak trapping experiments. Several fractions were collected during the separation of this sublimate which were then analyzed by infrared and mass spec trometry. The samples were collected at the outlet of the fluores cence detector. These samples are indicated on the chromatogram presented in Figure X. Fraction A contained primarily two components, resorcinol, which was confirmed by both IR and mass spectrometry, and an unknown of molecular weight 162. The infrared spectrum revealed that unknown 162 contained a carbonyl group and the mass spectrum revealed a fragmentation pattern very similar to that published for coumarin7. Based on this data, umbelliferone or 7-hydroxy.coumarin, 6, was proposed as the identity of unknown 162. Later, this assignment was confirmed through an LC peak enrichment experi ment with the authentic compound and through 13c NMR data obtained on a IOC distillate. Fraction B contained essentially one component of molecular weight 202. The infrared spectrum of this fraction compared well with the Hoppers Pencolite Resin 441 distillate which was rich in the resorcinol dimer, 5.. Fraction C contained several components. Mass spectral data for this fraction showed large ions at masses 152, 213, and 227. The m/e 213 and 227 ions are now known to be M-15 fragments from the M.W. 228, 9^, and M.W. 242, 4, components. The 152 (and 137) ions have been assigned to isopropyl resorcinol. The infra red spectrum of Fraction C and authentic 9,9-dimethy1-xanthene3,6-diol compared very well. Additional fractions collected did not contain a recogniz able major component and the materials had very low volatility. BFG06373 20768032 Introduction Code IOC (3000 x 32) is a mixture of aromatic oligomers pro duced by heating resorcinol at 280C for four hours with a catalytic amount of sodium hydroxide present. The process is called a "self condensation" because resorcinol is the only starting material and because water liberated by condensation is removed overhead. The solid reddish-brown product of this reaction is then dissolved in aqueous sodium hydroxide solution and used finally to treat the walls of PVC reactors to prevent build-up. This technology has been very successfully used in our own PVC production and has also generated substantial revenue from licensees. The preparation and use of Code 10C, discovered by Louis Cohen, is disclosed in U. S. Patent 4,080,173 (1978). Due to the immediate successful use of this important inven tion in PVC production and to the lack of problems associated with the synthesis of 10C, very little analytical data was obtained orf the product. The only current quality control specification for the 10C solid is a softening point of 120C or higher. The IOC composition is described in the patent as a mixture of poly(hydroxyphenyl) compounds , as shown below; however, this statement is based on literature reports of similar preparations and previous BFG analytical work on the composition of Code 10A-10B. Cohen states in the patent that the poly(hydroxyphenylenes) predominate in the mixture. This is based on the weight percent hydroxyl value reported^ for an oligomer prepared by condensing resorcinol with zinc chloride catalyst and on the observation of the higher softening point of Code 10C solid than that of Code 10A. The purpose of this report is to detail our analysis of Code 10C. In December, 1978*, Cohen requested assistance from the ALTC Analytical Group in determining the weight percent hydroxyl content in Code 10C in order to estimate the relative amounts of the two classes of oligomers. This information and other analytical data were needed to answer questions from our licensees who were being encouraged to produce their own 10C product on site. At that time, * Initial mass spectrometer data was obtained at ALTC in 1977. Because not all compounds identified by molecular weight fit the above reaction, a recommendation was made to analyze the materials on the new MAT 311A at Brecksville in early 1978. BFG06349 20768008 27 Figure X; Fraction Collection: Code IOC Sublimate A Sample: Code IOC Sublimate (210-230C) Mobile Phase: A-25/75 B-95/5 THF/water 30-100% B in 20 minutes Linear Flow Rate: 1.0 mL/minute Column: y-Bondapak C-18 (WATERS ASSOC.) (3.9 nm x 30 cm) 3 I I Detection: Fluorescence 280/370 (Xex/Aem) Sensitivity 420 TC - 1.0 sec Schoeffel 970 Fluorescence Detector C 2076803^ LIST OF FIGURES Figure I - 60 MHz 1h NMR Spectrum of Code 10C...................... PAGE 4 Figure II - 13c NMR Spectrum of Code 10C in DMSO-dg Solution............................................................................................ 5 Figure III Figure IV - Aromatic Portion of 13c NMR Spectrum of Code 10C Distillate (B.P. 220C/20ym Hg). . - Comparison of i3c NMR Spectra of and Code 10C............................................................................................. 6 10 Figure V - Comparison of 13c NMR Spectra of Acetaldehyde/ Resorcinol Condensation Product and Code 10C....................................................................................................... 11 Figure VI - GC/MS Analysis of TFAA Derivatives of Code 10C............................................................................................. 18 Figure VII - GC/MS Analysis of TFAA Derivatives of Code 10C............................................................................................ 19 Figure VIII - Total Ion Current Profile of TC 9B05: Overhead from 10C Production at Brecksville (resorcinol does not elute from the GC column under these conditions)................................. 23 Figure IX - Code 10C: LC Profile....................................................... 26 Figure X - Fraction Collection: Code 10C Sublimate. . 27 Figure XI - Assignment of Molecular Weights toSeries of Related Structures ....................................................... 32 Figure XII - Proposed Reaction Pathway (Resorcinol)... 33 Figure XIII - Proposed Reaction Pathway (Acetone) .... 34 Figure XIV - Proposed Reaction Pathway (Acetaldehyde). . 35 20768007 BFG06348 29 Discussion Our analysis of Code IOC was based primarily on the exact identification of a number of the lower molecular weight components with extrapolation to the structures of the higher molecular weight components through the use of the high mass MS-molecular ions, the condensed phase NMR data, and a mechanistic description of the reactions which must form all the species present. In Table IV, a list is provided of the twenty-seven molecular weights identified in the Code IOC mixture through mass spectrometry. For some of the weights, complete structures are known; for others, elemental compositions and some structural details are known. For the rest, only the nominal masses are known. However, all the com ponents can be linked directly through the building blocks of resor cinol and its decomposition products: acetone, acetaldehyde, and 1-oxopropanoic acid. This classification is shown in Table V. In the previous section on MS results, the known compounds were grouped into structural classes, such as alkylated resorcinols, xanthenes, etc., based on the available analytical data. In Table V, the com ponents are classified in a more vertical fashion as series of com pounds formed from common reactants. For example, both isopropyl resorcinol (M.W. 152) and 9,9-dimethyl-xanthene-3,6-diol (M.W. 242) belong to the group of resorcinol/acetone reaction products even though they belong to different structural classes. This rationalization can be extended to the remaining unknown components by using the numerical relationships apparent in the list of molecular weights. The molecular weights of the resorcinol oligomers with dimethyl xanthene end groups should form the series llO + 92n + 40. Just such a series is present; and the first member of the series, , M.W. 242, was completely identified. Therefore, it seems reasonable to assign unknowns 334 and 426 to this structural class. Thus, structure below is proposed for unknown 334, though the exact mass is the only bit of hard data available for this component. Other molecular weights also appear to be acetone/resorcinol products, but not members of the 110 + 92n + 40 series. Besides isopropyl resorcinol (M.W. 152), unknown 374 is an example; it is assigned to structure 1, based on the likelyhood that can further condense with acetone and resorcinol to form 11. BFG06377 LIST OF TABLES Table I - Predicted Versus Found 13C NMR Chemical Shifts for Aromatics in Resorcinol Oligomers...................... Table II - Relative Amounts of TFAA Derivatives of Code IOC Components Detected by GC/MS (represents the most volatile 30%)............................................................ Table III - Elemental Compositions of Code 10C TFAA Deriv atives Determined from Medium Resolution GC/MS Table IV - Molecular Weights of Code 10C Components Iden tified Through GC/MS and/or FD/MS Data .... Table V - Classification of Components Based on Reac tants.............................................................................................................. PAGE 8 17 20 30 30 20768006 BFG06347 31 Discussion - (continued) The same argument was used to assign members and structures in the acetaldehyde group. A supporting set of evidence for this method was obtained from mass analysis of the crude products formed in attempts to synthesize compounds 4 and 9. From acetone and resorcinol, products of molecular weight lS2, 242, and 374 were obtained. From acetaldehyde and resorcinol, products of mass 138, 228, 256, 346, 360, and many others were obtained. The assignment of molecular weights to structures is further demonstrated in Figure XI. The structures of the intended resorcinol oligomers and the unintended by-products produced tell much about the chemistry and conditions inside the Code 10C reactor. The structure of the resorcinol dimer, 5^, identifies the resorcinol "self-condensation" as first an aldol condensation between two moles of resorcinol in their keto-form, then a dehydration, and finally a proton rearrange ment to form the final stable tautomer, 5. This process is drawn in Figure XII. The presence of the by-products in 10C indicates a basecatalyzed decomposition reaction of resorcinol which is competitive with the condensation reactions at 280C. In Figure XIII, this reaction is shown as a double rearrangement reaction of resorcinol's diketo-form to produce acetone and 1-oxopropanoic acid. The acetone and resorcinol then produce isopropenyl resorcinol following another aldol condensation and dehydration. This intermediate then either reacts with resorcinol to form the 9,9-dimethylxanthene-3,6-diol (M.W. 242) or is reduced in an unknown process to form isopropylresorcinol . Isopropenyl resorcinol can react with any member of the resorcinol oligomer series (110 + 92n) to form any particular member of the "acetone series" 110 + 92n + 40. The isopropyl resorcinol is probably much less reactive and may be thought of as a dead end in this reaction scheme. In Figure XIV, the rest of the possibilities are indicated through reactions of the other by-product, 1-oxopropanoic acid. Condensation with resorcinol leads to 7-hydroxycoumarin, 6, while the apparently favored further decomposition to carbon dioxide and acetaldehyde leads to a myriad of acetaldehyde/resorcinol products. Acetaldehyde appears to be the most reactive component in the system. 20768033 TABLE OF CONTENTS Introduction............................................................................................................................. Experimental Results....................................................................................................... A.) 33C and 3H NMRAnalysis-- J. L. Dorsch......................... B.) Gas Chromatographic/MassSpectrometricResults. . C.) Code 10F Results -- J. A. Nikora............................................ D.) Liquid Chromatographic Results -- D. A. Ernes . . Discussion................................................................................................................................... References................................................................................................................................... Appendix 1................................................................................................................................... PAGE 1 3 3 13 22 25 29 38 39 20768005 BFG06346 33 Figure xii: PROPOSED REACTION PATHWAY "RESORCINOL SERIES" 202 20768040 - iii - Future Work 1. ) A detailed report on the syntheses of Code IOC-like oligomers from resorcinol and acetone, acetaldehyde, benzaldehyde, 4methoxybenzaldehyde, and 4-hydroxybenzaldehyde is in prepar ation by J. L. Dorsch and D. J. Smith. 2. ) Analytical characterization of Code IOC is proceeding with the help of R. P. Lattimer of BRDC. Acknowledgements A large amount of mass spectrometric data required for this analysis was contributed by R. P. Lattimer of BRDC. We appreciate his continuing support. D. J. Smith and D. Hanshumaker did the laboratory separations and syntheses required in this work. All of the mechanistic descriptions of the IOC reactions were developed in conversations with R. F. Koebel. We thank Louis Cohen for his many helpful comments on this manuscript. 20768004 BFG06345 35 Figure xiv: PROPOSED REACTION PATHWAY "ACETALDEHYDE SERIES" 228 138 9 20768042 - ii - Significant Results - (continued) 5. ) A large fraction by weight of the Code IOC mixture is com posed of components whose structures are not disclosed in the Code IOC patent. We have found an alternate route to a surface-active IOC-like mixture of oligomers starting from resorcinol and acetaldehyde using very mild conditions An invention record has been written for this preparation. 6. ) Many of the alkylated compounds in Code IOC contain the xanthene structure and so are highly fluorescent. From ultraviolet excitation, they fluoresce an intense green or blue-green in dilute aqueous or methanol solutions. This phenomenon may be of some use in analytical determinations. 7. ) The Code 10F mixture was found to be even less volatile and higher in molecular weight than IOC. Those components observed by mass spectrometry were found to be chlorinated versions of IOC components. Conclusions 1.) The formation of oligomers in Code IOC production proceeds to a large extent because of the thermal decomposition (at 280C) of resorcinol to the more reactive carbonyl contain ing compounds which condense rapidly with resorcinol and other poly(hydroxyphenyl) compounds. -2.) An alternate route to IOC-like oligomers exists through the condensations of resorcinol with aldehydes. These conden sations require relatively mild conditions ( 'v 120C, atmo spheric pressure) and so have an advantage over current IOC technology. However, these products are highly colored, just like IOC; the condensation of resorcinol with an alde hyde is not a route to a "white Code 10". Recommendations 1. ) We should be prepared to answer detailed questions concern ing Code 10C composition from our licensees who may be doing their own analytical work on IOC and who may be surprised at the results. 2. ) We should pursue a patent on the use of condensation products of resorcinol with aldehydes as reactor coating materials. 20768003 BFG06344 Discussion - (continued) Louis Cohen does not feel that oxidation of the initial com ponents, by itself, is an important factor in increasing the surface acitivity of the mixtures based on numerous experiments with Code 10 materials and antioxidants1. He feels that high molecular weight is the most desirable characteristic in any Code 10 component and that oxidative coupling, which leads to an increase in molecular weight, is useful. Oxidative coupling is his rationale for the improved activity of Code 10F over Code 10C and Code 10G over 4,4'thiobis(phenol). Oxidative coupling has not yet been confirmed analytically in Code 10F, but has been detected in Code 10G in R. P. Lattimer's recent FD/MS spectra. Whatever the relative effects of structure, oxidation, and molecular weight on the activity of resorcinol condensation products, their vulnerability to oxidation guarantees that they will all be highly colored. JLD-DAE-JAN/dmw John L. Dorsch David A. Ernes John A. Nikora 207C 8044 i- EXECUTIVE SUMMARY Objectives 1. ) To identify the structures of the components in Code IOC and Code 10F in support of licensing activity and continued development of reactor coating materials. 2. ) 3. ) To study the relationship of structure to the required surface activity in PVC reactors. To study the chemical reactions occurring in the production of Code IOC and 10F. Significant Results 1. ) Code IOC is a dark brown-colored solid, soluble in aqueous base and dimethyl sulfoxide. It is a mixture of compounds including residual starting material, resorcinol; alkylated resorcinols; alkyl hydroxy chromanes and chromenes; alkylated hydroxy xanthenes; poly(hydroxyphenyl) oligomers; and other compounds of increasing molecular weight and complexity. 2. ) A typical sample of Brecksville pilot plant production Code IOC contains 12.4 weight percent resorcinol and 4.9 percent 9,9-dimethyl xanthene-3,6-diol as determined by liquid chromatography. The third largest component is the resor cinol dimer, 2,3',4-trihydroxybiphenyl, which is present at 5-10%. The amounts of this and remaining components can only be estimated because authentic, pure samples are not available for calibration. Approximately 30% of Code 10C is volatile enough for mass spectrometric analysis. The molecular weights and structures for most of the components in this group have been identified. The structures in the volatile and non-volatile fractions have been compared by analysis of the NMR data and mechanistic descriptions of the reactions. 3. ) The Code 10C mixture is produced by a set of competing reac tions which include the intended self-condensation of resor cinol and the unintended decomposition of resorcinol into carbonyl compounds which further react to form the alkylated phenols, xanthenes, and higher molecular weight products observed. 4. ) It has not been determined which particular components in Code IOC are the desired surface-active species. Several low molecular weight components have been shown to be not surface-active. It is not known whether the alkylated com pounds are more or less surface-active than the poly(hydroxy phenyl) compounds. BFG06343 20768002 39 APPENDIX I Procedure for Preparation of Code IOC--TFAA Derivatives The two derivatizing reagents used were: 1. ) Hexamethyldilazane (HMDS), PCR Research Chemicals, Inc., Gainesville, Florida (Cat. 34020-8) 2. ) Trifluoroacetic Anhydride (TFAA); Aldrich Chemical Co., Milwaukee, Wis. and Supelco, Beliefonte, Pa. The derivatization reactions were typically carried out irw Supelco Micro Reaction Vessels (1,2, or 5 mL) capped with Teflon coated septa. Generally, a small quantity of material was dissolved in acetonitrile in the micro vessel and an excess of the derivatizing agent was added. The HMDS reactions were carried out at room temper ature and the ammonia gas hy-products were either blown off or removed by vacuum. The TFAA derivatization reactions were carried out at 60-80C and the excess TFAA was removed by vacuum distilla tion through a syringe needle. Caution must be exercised when using the TFAA derivatization technique! TFAA is very volatile and combines with any water available to form the very strong trifluoro acetic acid. Pressure build-up in the heated micro-reaction vessels can (and did) result in rupture of the septa. 20768046 BFG TECHNICAL DOCUMENT DIFGoodrich Chemical Division RESEARCH AND DEVELOPMENT REPORT djjjrOj THE COMPOSITION OF CODE IOC AND CODE 10F by J. L. Dorsch D. A. Ernes J. A. Nikora Project No. 3508/7721ReportType CLOSURE - STS #458 PLASTIC MATERIALS, Profit CenterlicensingDate ..................................... ..10 > 1980_____________ Copy Approval Authority R- F' K0EBEL------ ------------------------------------------- DISTRIBUTION Cleveland \ Akron *R. A. Krueger A. J. Hughes Powell, Jr. R. Wymb s \F. E./iesause X. C .Hblbrfcok - R.IC.mjntJTxej ALTC "TTF - (6) R. R. Bloor B. K. Mikofalvy L. B. Crider *C. E. Fleming L. Cohen D. E. Witenhafer R. L. Bowles C. A. Daniels GNU/ Schaaf M.W.Roha - G.A.Lindsay *A. W. Clements *H. 0. Jacobsen *B. A. DiLiddo *J.C.Healy - E.J.Sehm F. J. Donat *J. F. Malone M. M. O'Mara *R. M. Kreager R. F. Koebel D. J. Smith D. Hanshumaker R. J. Meyer/Int'l. - (2) J.A.TePas - R.D.Hardesty J.W.Ryan - R.C.Williams - Brecksville C . H . Lut'ter - J. B. Pausch R.Komoroski - J.C.Westfahl *R. J. Fawcett E.D.Truscott R.K.Schlatzer R.P.Lattimer E.R.Hooser * Summary Copy only. RD File - (2) BFG*t0805-A fl/78 PRINT ED IN U S A. BFG06342 20768001